UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Carbon dioxide flux within and above a boreal aspen forest Yang, Paul Chenggang 1998

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1998-346471.pdf [ 11.12MB ]
Metadata
JSON: 831-1.0099326.json
JSON-LD: 831-1.0099326-ld.json
RDF/XML (Pretty): 831-1.0099326-rdf.xml
RDF/JSON: 831-1.0099326-rdf.json
Turtle: 831-1.0099326-turtle.txt
N-Triples: 831-1.0099326-rdf-ntriples.txt
Original Record: 831-1.0099326-source.json
Full Text
831-1.0099326-fulltext.txt
Citation
831-1.0099326.ris

Full Text

CARBON DIOXIDE FLUX WITHIN AND ABOVE A BOREAL ASPEN FOREST by PAUL CHENGGANG YANG B.Sc. (Agrometeorology), Nanjing Institute of Meteorology, China, 1983 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES Department of Soil Science We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA August 1998 © Paul Chenggang Yang, 1998 ln presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia Vancouver, Canada DE-6 (2/88) \ ABSTRACT Carbon dioxide, water vapour, sensible heat and momentum fluxes were continuously measured using the eddy covariance technique above and below the overstory in a 70-year old aspen (OA) stand in northern Saskatchewan from October to November 1993 and from February to September 1994, and above the overstory from April to December 1996 as a part of the Boreal Ecosystem-Atmosphere Study (BOREAS). Due to the relative openness of the aspen canopy, the air within the forest was usually stably stratified at night and unstable during the daytime. The relationships of the variances of the vertical velocity and scalars (air temperature, CO2 concentration and specific humidity) to the stability parameter above the forest followed the Monin-Obukhov similarity (MOS) relationships, while the applicability of MOS theory in the trunk space was poor, especially for CO2 concentration. On average there was no significant enhanced CO2 transport above that estimated using MOS theory both above the forest and in the trunk space. The rate of change in CO2 storage in the air column (ASa/At) beneath the above-canopy eddy covariance system could be well estimated with concentrations measured at one height above the forest and at one height (2.3 m) in the trunk space. ASa/At was significantly large near sunrise (6-9 CST) and sunset (18-22 CST). Within the trunk space, eddy covariance sensible and latent heat flux measurements at one position were representative of an area extending for at least two tree heights. The same was the case for C 0 2 flux and concentration during the daytime. At night, however, they exhibited significant horizontal variability but were representative of the above area when averaged over several days. ii Evidence supporting the hypothesis that the low nighttime CO2 fluxes resulted from the short-term changes in CO2 storage in the air-filled pores of soil/snow was presented. The rate of change of this storage (ASs/At) was estimated as ASJkt = (l-M)Rslta where Rsha (the forest respiration) is a function of the soil temperature and M is a function of the friction velocity. Long-term carbon sequestration was estimated by summing the eddy covariance CO2 fluxes (Fc) because changes in storage average to zero over periods of a week or more. Photosynthetic rates (P) were modelled as a product of P i , Pi and F 3 . P\ is a rectangular hyperbolic function of the absorbed photosynthetic photon flux density (PPFD), and Pz and Pi are second order polynomial functions of saturation deficit and air temperature, respectively. This empirical model explained about 80%, 76% and 26% of the variance in the measured half-hourly photosynthesis of the forest (Pe), aspen overstory and hazelnut understory, respectively, in 1994. The corresponding percentage of the variances explained by absorbed PPFD were 74%, 68% and 25%, respectively. The model explained 73% of the variance in half-hourly Pe obtained at the OA site during the 1996 growing season. In 1994, the OA forest photosynthesized about 1140 g C m"2, of which 83% was accounted for by the aspen overstory. Total forest respiration was about 920 g C m~2, of which 53% was estimated to be soil respiration. Thus, carbon sequestration by the forest was about 220 g C m"2, which is slightly higher than the value (200 g C m"2) obtained by directly summing Fc. Assuming that half of the soil respiration was heterotrophic, net primary productivity in 1994 was estimated to be 450 g C m" . iii TABLE OF CONTENTS A B S T R A C T i i T A B L E O F C O N T E N T S iv LIST O F S Y M B O L S A N D A C R O N Y M S v i i LIST O F T A B L E S xv i i LIST O F F I G U R E S xix A C K N O W L E D G M E N T S xxv i i 1. INTRODUCTION 1 1.1 R E F E R E N C E S 5 2. SIMILARITY RELATIONSHIPS ABOVE AND IN THE TRUNK SPACE OF A BOREAL ASPEN FOREST 7 2.1 I N T R O D U C T I O N 7 2.2 M E T H O D O L O G Y 10 2.2.1 Site, instrumentation and measurements 10 2.2.1.1 Site description 10 2.2.1.2 Flux measurements 11 2.2.1.3 Supporting measurements 14 2.2.2 Local similarity theory 15 2.2.2.1 Variance similarity relationships 15 2.2.2.2 Flux-gradient similarity relationships 17 2.3 R E S U L T S A N D D I S C U S S I O N 18 2.3.1 Stability within and above the forest. 18 2.3.2 Turbulence within and above the forest. 22 2.3.3 Variance similarity relationships above the forest 26 2.3.3.1 Variance of vertical velocity 27 2.3.3.2 Variance of temperature 30 2.3.3.3 Variances of C 0 2 concentration and specific humidity 33 2.3.4 Variance similarity relationships in the trunk space 39 2.3.5 Flux-gradient similarity relationships above the forest and in the trunk space 46 2.3.5.1 Diabatic influence function ( § F ) and the stability parameter ( r) 46 2.3.5.2 Comparison of calculated and measured above-forest C 0 2 fluxes during the full-leaf period 49 2.3.5.3 Comparison of calculated and measured trunk-space C 0 2 fluxes during the full-leaf period 52 2.3.5.4 Comparison of calculated and measured C 0 2 fluxes during leaf-free and full-leaf periods55 2.4 C O N C L U S I O N S 57 2.5 R E F E R E N C E S 60 3. ESTIMATING NET C0 2 EXCHANGE BETWEEN THE ATMOSPHERE AND A BOREAL ASPEN FOREST 65 3.1 I N T R O D U C T I O N 65 iv 3.2 E X P E R I M E N T A L M E T H O D S 67 3.2.1 Flux measurements 67 3.2.2 Trunk-space eddy flux comparison experiment 68 3.2.3 Comparison of eddy covariance systems 68 3.3 R E S U L T S A N D D I S C U S S I O N 73 3.3.1 CO2 concentration profiles and the change in the storage of CO2 in the air column beneath the eddy flux sensors (ASa/At) 73 3.3.1.1 CO2 concentration profiles and the estimation of ASJAt 73 3.3.1.2 Diurnal changes in ASJAt and effects of turbulent mixing 79 3.3.2 Estimating net CO2 exchange between the atmosphere and the forest. 83 3.3.2.1 Diurnal patterns of Fc, ASJAt and NEE for the forest 83 3.3.2.2 Seasonal changes in C 0 2 exchanges above and below the aspen overstory 88 3.3.3 Horizontal variability of CO2 concentration and fluxes within the trunk space 93 3.3.3.1 Horizontal variability of the C 0 2 concentration within the trunk space 93 3.3.3.2 Horizontal variability of the fluxes within the trunk space 96 3.4 C O N C L U S I O N S 101 3.5 R E F E R E N C E S 103 4. ESTIMATING N I G H T T I M E RESPIRATION USING E D D Y C O V A R I A N C E M E A S U R E M E N T S O F C 0 2 F L U X A B O V E A B O R E A L ASPEN FOREST 107 4.1 I N T R O D U C T I O N 107 4.2 T H E O R E T I C A L C O N S I D E R A T I O N S 113 4.3 M E T H O D O L O G Y 116 4.3.1 Instrumentation and measurements 116 4.3.1.1 Flux measurements 116 4.3.1.2 Supporting measurements 120 4.3.2 Determining Rsna andM. 121 4.4 R E S U L T S A N D D I S C U S S I O N 127 4.4.1 Evidence for and against the mass flow hypothesis 127 4.4.2 Evidence for and against the soil CO2 storage hypothesis 132 4.4.2A C 0 2 accumulation at ground level on calm nights 133 4.4.2.2 Enhanced C 0 2 efflux in windy conditions during winter 138 4.4.2.3 Enhanced C 0 2 efflux by rainfall 141 4.4.2.4 Evaluation of the soil C 0 2 storage model 143 4.4.3 Comparison of carbon sequestration estimates using the three approaches 146 4.4.3.1 Diurnal carbon sequestration 146 4.4.3.2 Annual carbon sequestration 151 4.5 C O N C L U S I O N S 153 4.6 R E F E R E N C E S 156 5. F A C T O R S A F F E C T I N G C A N O P Y PHOTOSYNTHESIS O F T H E O V E R S T O R Y AND U N D E R S T O R Y OF A B O R E A L ASPEN F O R E S T 161 5.1 I N T R O D U C T I O N 161 5.2 M A T E R I A L S A N D M E T H O D S 162 5.2.1 Eddy Flux Measurements '. 162 5.2.2 Supporting Measurements 163 5.2.2.1 Leaf area index measurements 163 5.2.2.2 Photosynthetic photon flux density measurements 165 5.2.2.3 Other measurements 168 5.2.3 Analytical procedure 169 5.2.3.1 Estimation of photosynthesis (P) 169 5.2.3.2 Estimation of respiration (R) 170 5.2.3.3 Relationship of photosynthesis to environmental variables 172 5.3 R E S U L T S A N D D I S C U S S I O N 173 5.3.1 Typical full-leaf diurnal patterns of photosynthesis of the forest, aspen overstory and hazelnut understory 173 5.3.2 Effects of environmental factors on the rate of photosynthesis of the forest 177 5.3.2.1 Photosynthetic photon flux density (PPFD) 177 5.3.2.2 Saturation deficit (£>) 181 5.3.2.3 Air temperature (T) 184 5.3.2.4 Cloudiness and other factors 186 5.3.3 Water use efficiency of the forest, aspen overstory and hazelnut understory 189 5.3.4 Evaluation of a simple empirical model of forest photosynthesis 192 5.3.5 Contributions of aspen overstory and hazelnut understory to forest photosynthesis 199 5.3.6 Carbon balance at the OA site in 1994. 203 5.4 S U M M A R Y A N D C O N C L U S I O N S 206 5.5 R E F E R E N C E S 208 6. S U M M A R Y A N D C O N C L U S I O N S 215 6.1 R E F E R E N C E S 222 A P P E N D I X A P R O F I L E S O F B U L K D E N S I T Y O F T H E SOIL A T T H E O A SITE 225 A P P E N D I X B N I G H T T I M E C O S P E C T R A O F W-T AND W-Xc 226 A P P E N D I X C D E R I V A T I O N S O F S O M E U S E F U L R A T I O S O F R E S P I R A T I O N F O R T H E O A SITE 230 A P P E N D I X D T I M E SERIES O F R„, HAND AE D U R I N G T H E 6 - D A Y P E R I O D S S H O W N I N FIG. 4.14 A N D FIG. 4.15 233 vi LIST OF SYMBOLS AND ACRONYMS C Ca Cr Cs (c) Cwx(f) cL CO2 concentration CO2 concentration in the air near ground CO2 concentration at reference height CO2 concentration in bulk soil mean CO2 concentration within the air column beneath the reference height cospectrum of variable x and w as a function of frequency (/) CO2 similarity function pmol mol Limol mol Limol mol Limol mol Limol mol dimensionless [imol mol"1 D Do De E Ea <cal saturation deficit CO2 diffusivity in the air effective CO2 diffusivity in the soil depth I evaporation flux density transpiration flux density of the aspen overstory evaporation flux density of the forest ecosystem evaporation flux density of the hazelnut-soil system C 0 2 eddy flux density at the 39.5-m height C 0 2 eddy flux density at the 4-m height eddy flux of CO2 calculated from concentration gradient using MOS theory , eddy flux of C 0 2 measured using the eddy covariance system kPa m V m V mmol m"2 s"1 or mg mmol m"2 s"1 or mg m"2 s"1 mmol m"2 s"1 or mg m"2 s"1 mmol m"2 s"1 or mg m"2 s"1 ixmol m"2 s"1 Lxmol m"2 s"1 Limol m"2 s"1 iimol m"2 s"1 VII Fs H K L L La Lh M Ma P Pa Pal Pal Pe Pel Pel Pel CO2 efflux from soil surface sensible heat flux density eddy diffusivity scaling (Monin-Obukhov) length leaf area index leaf area index of the aspen overstory leaf area index of the hazelnut understory wind function described the effect of turbulence on CO2 transfer molecular weight of moist air (~ 29 g mol"1) wind function described the effect of turbulence on CO2 transfer at the 4-m height photosynthetic rate photosynthetic rate of the aspen overstory calculated photosynthetic rate of the aspen overstory from the absorbed PPFD calculated ratio of photosynthetic rate of the aspen overstory to Pai (Pa2 = Pa /Pal) from saturation deficit calculated ratio of photosynthetic rate of the aspen overstory t o P a 1 P f l 2 (Pa3= Pa/{PalPa2)) from air temperature photosynthetic rate of the forest calculated photosynthetic rate of the forest from the absorbed PPFD calculated ratio of photosynthetic rate of the forest to Pei (PeZ = Pe/Pel) from saturation deficit calculated ratio of photosynthetic rate of the forest to PeXPei (Pe3 = Pe/{PelPe2)) from air u.mol m"2 s"1 W m " m V m m leaf m" ground 2 2 m leaf m" ground 2 2 m leaf m" ground dimensionless g mol"1 dimensionless u.mol m"2 s"1 uxnol m"2 s"1 Ltmol m"2 s"1 dimensionless dimensionless Ltmol m"2 s"1 fxmol m"2 s"1 dimensionless dimensionless viii Ph Phi temperature photosynthetic rate of the hazelnut understory calculated photosynthetic rate of the hazelnut understory from the absorbed PPFD urnol m~2 s"1 u.mol m"2 s"1 hi calculated ratio of photosynthetic rate of the hazelnut understory to Phi (Ph2 - P h / P h i ) f r o m saturation deficit dimensionless A3 Q Qo Qo calculated ratio of photosynthetic rate of the hazelnut understory to P h \ P h i (Ph3 = Phl{PhXPh2)) from air temperature canopy photosynthetic capacity (i.e., the photosynthetic rate at saturating PPFD) on ground basis photosynthetic photon flux density incident photosynthetic photon flux density above the forest incident photosynthetic photon flux density to the aspen overstory (Q^ = Q fQ - ) dimensionless u.mol m"2 s"1 u.mol m"2 s"1 |j,mol m"2 s"1 u.mol m"2 s"1 Ql incident photosynthetic photon flux density above the forest u,mol m"2 s"1 Ql Qio Qa Q: incident photosynthetic photon flux density above the hazelnut understory the ratio of the rate of a process at one temperature to that at a temperature 10°C lower absorbed photosynthetic photon flux density absorbed photosynthetic photon flux density by the aspen overstory {Q aa = Q fa -Qha) umol m~2 s"1 dimensionless u.mol m~2 s"1 LAmol m"2 s"1 Ql Ql absorbed photosynthetic photon flux density by the forest absorbed photosynthetic photon flux density above the hazelnut understory umol m"2 s"1 \imo\ m"2 s"1 ix Raa respiration rate of the aspen foliage, branches Limol m"2 s"1 and trunks Ra respiration rate of the aspen foliage, branches ixmol m"2 s"1 and trunks from the 4- to 21-m height Rsh respiration rate of the soil, hazelnut understory \x.mo\ m~2 s"1 and the aspen trunk from the ground to the 4-m height Rsha respiration rate of the soil, hazelnut understory pimol m"2 s"1 and the aspen overstory (Rsha = Rsh + Ra) Rs soil respiration rate (including microorganisms (jmol m"2 s"1 and roots) 2 S solar radiation W m" T air temperature °C Ts temperature of the soil at the 2-cm depth °C U Streamline wind speed m s"1 Uhc Streamline wind speed at the canopy height m s"1 Wa woody area index of the aspen overstory (0.620) m 2 woody part m" ground a one of the coefficients of the respiration \imo\ m"2 s"1 relationship to the soil temperature (Eq. (4.9)), it is the maximum respiration rate a one of the coefficients of the photosynthesis kPa"2 relationship to saturation deficit (Eq. (5.11)) aw one of the coefficients of (Eq. (2.9)) dimensionless ax one of the coefficients of (Eq. (2.10)) for scalar x dimensionless (T, c, q) b one of the coefficients of the respiration 0 C _ 1 relationship to the soil temperature (Eq. (4.9)) b one of the coefficients of the photosynthesis kPa"1 X cw d d dx f f f relationship to saturation deficit (Eq. (5.11)) one of the coefficients of (Eq. (2.9)) one of the coefficients of (Eq. (2.10)) for scalar* (Z c, q) one of the coefficients of the respiration relationship to the soil temperature (Eq. (4.9)), it is the temperature when the respiration rate is half of the maximum respiration rate. one of the coefficients of the photosynthesis relationship to saturation deficit (Eq. (5.11)) specific heat at constant pressure for moist air one of the coefficients of (Eq. (2.9)) one of the coefficients of (Eq. (2.10)) for scalar* (T, c, q) displacement height one of the coefficients of the wind function (Eq. (4.10)) one of the coefficients of the photosynthesis relationship to air temperature (Eq. (5.12)) one of the coefficients of (Eq. (2.9)) one of the coefficients of (Eq. (2.10)) for scalar* one of the coefficients of the wind function (Eq. (4.10)) one of the coefficients of the photosynthesis relationship to air temperature (Eq. (5.12)) one of the coefficients of the wind function (Eq. (4.10)) one of the coefficients of the photosynthesis relationship to air temperature (Eq. (5.12)) frequency dimensionless dimensionless °C dimensionless m 2 s-2 K"1 dimensionless dimensionless m dimensionless oc-l dimensionless dimensionless dimensionless dimensionless dimensionless dimensionless Hz xi fz vertical mass flow of CO2 at the instrument Lxmol m2 s"1 height (not measurable by the eddy covariance system) fzh the vertical mass flow of CO2 at the 4-m height lAmol m"2 s"1 (not measurable by the eddy covariance system) g gravitational constant m s" hc height of canopy m hc transfer coefficient of CO2 mol m"2 s"1 k von Karman's constant (0.4) dimensionless ka effective extinction coefficient of the aspen dimensionless overstory (0.540) kh effective extinction coefficient of the hazelnut dimensionless understory (0.756) I soil depth m n number of observations dimensionless q specific humidity g H2O g"1 moist air q'^ humidity similarity function g m~3 r mixing ratio g H2O g"1 dry air r 2 coefficient of determination dimensionless syx standard error of estimate dimensionless u longitudinal (streamline) velocity component m s" u measured longitudinal velocity m s' " * friction velocity (-J-u'w')  m s u' local velocity similarity function m s" Uy. 8-day running mean of m s" v lateral (crosswind) velocity component m s" xii vv vv vv z AC 4 _ 5 . 9 m 4k ASahlAt ASJAt A ay-vertical velocity component m s"1 measured vertical velocity m s"1 calculated 'true' vertical velocity m s"1 height m difference in CO2 concentrations measured in the umol mol"1 trunk space at two locations (40-m apart) net horizontal CO2 advection flux density umol m"2 s"1 net horizontal C 0 2 advection in the 0-4 m umol m"2 s"1 column, andfZh is the vertical mass flow of CO2 at the 4-m height half-hourly change rate in CO2 storage in the air umol m"2 s"1 column beneath eddy covariance system half-hourly change rate in CO2 storage in the air umol m"2 s"1 column from the 4-m height to the 39.5-m height half-hourly change rate in CO2 storage in the air umol m"2 s"1 column from the ground to the 4-m height half-hourly change rate in CO2 storage in the air- uinol m"2 s"1 filled pores in the soil column to a depth where the CO2 flux is zero carbon sequestration (photosynthesis - umol m"2 s"1 respiration) carbon sequestration (photosynthesis - umol m"2 s"1 respiration) of the hazelnut understory local Monin-Obukhov length. m quantum yield mol C 0 2 mol"1 photons reflectivity of the forest to the incident PPFD dimensionless xiii ah reflectivity of the hazelnut understory to the dimensionless incident PPFD 0 potential temperature °C 0^  temperature similarity function °C B0 soil volume fraction of organic m 3 organic per m 3 soil 0 v virtual potential temperature °C Or standard deviation of temperature °C ow standard deviation of vertical velocity m s"1 ow the 5-day running mean of the standard deviation m s"1 of vertical velocity at the 4-m height AE latent heat flux density W m" % tortuosity factor dimensionless p air density kg m"3 pa dry air density kg m"3 pc CO2 concentration \imo\ mol"1 x momentum flux density (shear stress) N m"2 a) wind direction 0 <|)c normalized standard deviation of the CO2 dimensionless concentration (j>? normalized standard deviation of the water dimensionless vapour density (J)r normalized standard deviation of temperature dimensionless tyw normalized standard deviation of vertical dimensionless velocity <]>£ diabatic influence function for latent heat dimensionless fl)/r diabatic influence function for CO2 dimensionless x i v § H diabatic influence function for sensible heat dimensionless WFc integration of diabatic influence function for dimensionless C 0 2 integration of diabatic influence function for dimensionless latent heat yH integration of diabatic influence function for dimensionless sensible heat £ stability parameter dimensionless Xc CO2 concentration ^imol mol"1 time average N/A deviation from time average N/A AES Atmospheric Environment Service N/A A N F All-Nighttime-Flux approach where all the N/A nighttime CO2 flux measurements are accepted and averaged over a long-term in order to relate them to soil temperature (Chapter 4) BOREAS Boreal Ecosystem-Atmosphere Study N/A C3 species fixing CO2 largely into 3-carbon N/A phosphoglyeric acid C4 species fixing CO2 largely into 4-carbon malic N/A and aspartic acid CSI Campbell Scientific Inc CST Central Standard Time D O Y Day of the year GMT Greenwich Mean Time HWS N/A s, min, h, day N/A s, min, h, day High-Wind-Speed approach where the low CO2 N/A flux measurements made on calm night are replaced by estimates from high wind speed half-XV hours (Chapter 4) IRGA Infrared Gas Analyzer N/A LAI projected (one side) leaf area index m 2 leaf m"2 ground MESONET BOREAS mesoscale meteorological network N/A MOS Monin-Obukhov Similarity N/A NEE Net ecosystem exchange of CO2 |lmol m"2 s"1 NHE Net hazelnut exchange of CO2 (imol m"2 s"1 OA Old Aspen N/A PPFD photosynthetic photon flux density umol m"2 s"1 SW Southwest N/A TDR Time Domain Reflectrometry N/A WPL Webb, Pearman and Leuning N/A WSW West Southwest N/A WUE Water Use Efficiency umol C 0 2 mol"1 H 2 0 , mg C 0 2 g"1 H 2 0 xvi LIST OF TABLES Table 2.1 The percentage of the nighttime half hours when the local stability parameter C, was less than 0.02, that is the air was statically unstable or neutral at the OA site in 1994 21 Table 2.2 The coefficients of the relationship between § w and C, above the forest (39.5-m height) for different months at the OA site in 1994 29 Table 2.3 The coefficients of the relationship between § T and C, above the forest (39.5-m height) for different months at the OA site in 1994 32 Table 2.4 The coefficients of the relationship between § c and £ above the forest (39.5-m height) for different months at the OA site in 1994 36 Table 2.5 The coefficients of the relationship between (j)9 and C, above the forest (39.5-m height) for different months at the OA site in 1994 37 Table 2.6 The coefficients of the relationship between a)w and £ within the trunk space (4-m height) for different months at the OA site in 1994 40 Table 2.7 The coefficients of the relationship between fa ar>d C, within the trunk space (4-m height) for different months at the OA site in 1994 42 Table 2.8 The coefficients of the relationship between c)c and C, within the trunk space (4-m height) for different months at the OA site in 1994 45 Table 2.9 The coefficients of the relationship between <j)? and £ within the trunk space (4-m height) for different months at the OA site in 1994 45 Table 2.10 Coefficients of the relationship between the C O 2 flux calculated using the MOS relationship from C O 2 gradient and the flux measured using the eddy covariance system above the forest and within the trunk space in the summer 1994 at the OA site.54 Table 3.1 Coefficients of the linear best fits (ASa/At\g_ht = a(ASa/At)\nj,t + b ) shown in Fig. 3.7 77 Table 3.2 Comparison of the means of scalar fluxes obtained on the 6-m scaffold tower (A) and the main tower (B) in the trunk-space eddy flux experiment, August 12-22, 1994.100 Table 4.1 The comparison of cumulative carbon sequestration (g C m"2) obtained using different approaches for the OA forest, Prince Albert National Park, Saskatchewan in 1994. The growing season is from May 20 - September 10. The values of the uncertainties are standard errors of the estimates. In the HWS approach, the nighttime flux measurements when < 0.45 m s"1 were replaced with calculated respiration rates estimated using an empirical equation relating nighttime C O 2 flux at w+> 0.45 m s"1 to soil temperature at the 2 cm depth. In the A N F approach, all the nighttime C O 2 flux measurements were used without correction using the wind function M. 146 Table 5.1 Parameters in Eq. (5.10) for the forest, aspen overstory and hazelnut understory for the period April 22-September 19 (DOY 112-262), 1994 using bin-averaged PPFD values. The values in brackets are for leaf-area based photosynthetic rates (P/L) and absorbed PPFD (QJL) for the period May 28-September 19 (DOY 148-262), 1994. The second line in the r2 and n columns are the values when the half-hourly data were used 179 xvii Table 5.2 Parameters in Eq. (5.11) for the forest, aspen overstory and hazelnut understory for the period of April 22-September 19 (DOY 112-262), 1994 using bin-averaged D values 183 Table 5.3 Parameters in Eq. (5.12) for the forest, aspen overstory and hazelnut understory for the period of April 22-September 19 (DOY 112-262), 1994 using bin-averaged T values 185 Table 5.4 Comparison of the light response of photosynthesis of the forest, hazelnut understory and aspen overstory under different sky conditions at the OA site, 1994188 Table 5.5 Performance of the simple photosynthesis model (P = PjP2P3) for the forest, aspen overstory and hazelnut understory using 1994 data (April 22-September 19). a and b are the intercept and the slope, respectively, of the equation: Pmeas = a + bPm0Cjei. Also shown are the slope, intercept and r using 1996 forest photosynthesis data 192 Table 5.6 Annual carbon fluxes (g C m"2) in the OA forest in 1994, where negative value means the biomass (or soil) losing carbon. The values in bracket are the total uptake (-NEE and -NHE) calculated as the minus summation of Fc and Fch (Chapter 4). Shaded rows indicate the eddy covariance measurement levels 205 Table C . l Some useful ratios of the respiration at the OA site in 1994 232 xviii LIST OF FIGURES Fig. 2.1 Profiles of the local stability parameter t,=z/L within and above the forest at the OA site on September 10 (DOY 253), 1994. Panel (a) is the daytime profiles with time intervals of 13:00-13:30 (circle), 13:30-14:00 (square), 14:00-14:30 (triangle) and 14:30-15:00 CST (downward triangle). Panel (b) is the nighttime profiles with time intervals of 1:00-1:30 (circle), 1:30-2:00 (square), 2:00-2:30 (triangle) and 2:30-3:00 CST (downward triangle) 20 Fig. 2.2 Relationship between the friction velocities and the standard deviations of the vertical velocities at the 4-m height to the friction velocity at the 39.5-m height at the OA site, 1994 21 Fig. 2.3 Vertical profiles of some turbulent statistics and the standard deviation of the temperature measured at the OA site, 1994 for neutral conditions which is defined here as |£| < 0.005. The symbols are the ensemble averages of the statistics during the period of August 10-September 19. The error bars were ± 1 standard deviation. The friction velocity (u ) was measured at the 39.5-m height. UhC is the interpolated wind speed at the top of the aspen canopy (21 m), T is the momentum flux (shear stress) and p is the air density 23 Fig. 2.4 Same as Fig. 2.3 except for unstable conditions which is defined as t, < -0.005 24 Fig. 2.5 Same as Fig. 2.3 except for stable conditions which is defined as £ > 0.005 25 Fig. 2.6 Normalized standard deviation of vertical velocity (§w) as a function of the stability parameter C, above the forest (39.5-m height) at the OA site in August. The dots are half-hourly values of § w with sensible heat flux H great than 10 W m"2. The solid line is the best 1/3 power fit as described in Eq. (2.9) and Table 2.2. The dash-dotted line is (j), = (1 + 2£) V 3 for <; > 0 and <j>w = (l - oXfR for c; < 0 following Shao and Hacker (1990) 28 Fig. 2.7 The calculated values of (j)w from the coefficients in Table 2.2 for different months in 1994 30 Fig. 2.8 Normalized standard deviation of temperature (§T) as a function of the stability parameter t, = (z-d)/L above the forest (39.5-m height) at the OA site in August. The dots are half-hourly values of (j)r with sensible heat flux H greater than 10 W m"2. The solid line is the best 1/3 power fit as described in Eq. (2.10) and Table 2.3. The dash-dotted line is <j)r = 2.45(1 - 20i;)"1/3 for t, < 0 and <j>r = 3.46 for t, > 0 following Shao and Hacker (1990) 31 Fig. 2.9 The calculated values of <\>T from the coefficients in Table 2.3 for different months in 1994 33 Fig. 2.10 Normalized standard deviation of CO2 concentration (Q)C) as a function of the stability parameter t, - (z-d)/L above the forest (39.5-m height) at the OA site in August. The dots are half-hourly values of a)c with sensible heat flux H great than 10 W m"2. The solid 1 ine is the best 1/3 power fit as described in Eq. (2.10) and Table 2.3. The dash-dotted line is § c = 2.45(1 - 20c;)_1/3 for £ < 0 and (}>c = 3.46 for t, > 0 following Shao and Hacker (1990) 34 XIX Fig. 2.11 Normalized standard deviation of specific humidity (<\>q) as a function of the stability parameter t, - (z-d)/L above the forest (39.5-m height) at the OA site in August. The dots are half-hourly values of (j)c with sensible heat flux H greater than l O W m " 2 . The solid line is the best 1/3 power fit as described in Eq. (2.10) and Table 2.3. The dash-dotted line is <j)? = 2.45(1 - 20Q~ 1 / 3 for £ < 0 and <J>? = 3.46 for £ > 0 following Shao and Hacker (1990) 35 Fig. 2.12 The calculated values of (f>c from the coefficients in Table 2.4 for different months in 1994 38 Fig. 2.13 Normalized standard deviation of vertical velocity (§w) as a function of the stability parameter t, = z/L within the trunk space (4-m height) at the OA site in August. The dots are half-hourly values of fJ)w. The solid line is the best 1/3 power fit as described in Eq. (2.9) with the coefficient given in the text. The dash-dotted line is <L = (1 + 2 £ ) V 3 for l, > 0 and a>w = (l - 6c)yi for <; < 0 following Shao and Hacker (1990) 39 Fig. 2.14 Normalized standard deviation of temperature (<J)j) as a function of the stability parameter t, = z/L within the trunk space (4-m height) at the OA site in August. The dots are half-hourly values of §T- The solid line is the best 1/3 power fit as described in Eq. (2.11). The dash-dotted line is <j>r = 2.45(1 - 20<;)"1/3 forc;<0 and <j>r =3.46 for £ > 0 following Shao and Hacker (1990) 41 Fig. 2.15 Normalized standard deviation of C 0 2 concentration (§c) as a function of the stability parameter t, = z/L within the trunk space (4-m height) at the OA site in August. The dots are half-hourly values of tyc- The solid line is the best -1/3 power fit as described in Eq. (2.10). The dash-dotted line is <J)C = 2.45(1 - 20Q~ 1 / 3 for £ < 0 and o)c = 3.46 for £ > 0 following Shao and Hacker (1990) 43 Fig. 2.16 Normalized standard deviation of specific humidity ((j)9) as a function of the stability parameter t, = z/L within the trunk space (4-m height) at the OA site in August. The dots are half-hourly values of cj)?. The solid line is the best -1/3 power fit as described in Eq. (2.10) 44 Fig. 2.17 Relationship between the reciprocal of the diabatic influence function for CO2 and the stability parameter in the full-leaf period (June-August) of 1994 at the OA site, (a) is above the forest and (b) is within the trunk space. The lines are obtained using Eq. (2.4) 47 Fig. 2.18 Comparison of CO2 flux calculated using Eq. (2.5) and that measured using the eddy covariance system at the 39.5-m height in June, July and August, 1994 at OA. The data were 2-hour averages. The circles are data under unstable conditions (t, < -0.02). The triangles are the data under neutral conditions (|£,| < 0.02). The crosses are the data under stable conditions (t, > 0.02). The values of flux calculated from the gradient which were higher that 10 umol m"2 s"1 were eliminated. The dot-dashed line is the regression, and the solid line is the one-to-one line. The coefficients of the regression equation are shown in Table 2.10 50 Fig. 2.19 Comparison of C 0 2 flux calculated using Eq. (2.3) and that measured using the eddy covariance system at the 4-m height in June, July and August, 1994 at OA. The XX data were 2-hour averages. The circles are data under unstable condition which is £ > -0.02. The triangles are the data under neutral condition which is \t\ < 0.02. The crosses are the data under stable condition which is t, > 0.02. The values of flux calculated from the gradient which were higher that 10 umol m"2 s"1 were eliminated. The heavy line is the regression, and the thin line is the one-to-one line. The coefficients of the regression equation are shown in Table 2.10 53 Fig. 2.20 CO2 fluxes calculated from the CO2 concentration gradient and measured using the eddy covariance systems during the period of April 16-20, 1994 at OA. Panel (a) shows the PAR (Q0, solid line) and the friction velocity (u^ dashed line). Panel (b) shows the calculated (solid line) and measured (circles) CO2 fluxes at the 39.5-m height. Panel (c ) is the same as (b) except at the 4-m height on the canopy tower. The data were 2-hour averages 55 Fig. 2.21 Same as Fig. 2.20 in except July 1-5. Panel c) is the fluxes at the 5.9-m height on the main tower 56 Fig. 3.1 Comparison of the sensible heat fluxes measured at the 39.5- and 28.6-m heights using eddy covariance systems at the OA site from June 12-17, 1994. The dotted line is the one-to-one line. The solid line is the regression through the origin 69 Fig. 3.2 Same as in Fig. 3.1 except for latent heat flux 71 Fig. 3.3 Same as in Fig. 3.1 except for eddy CO2 flux 72 Fig. 3.4 Ensemble-averaged CO2 profiles for the leafless period (February 4-April 10, 1994) at the OA site 73 Fig. 3.5 Ensemble-averaged CO2 profiles for the summer (June 1-August 31, 1994) at the OA site 74 Fig. 3.6 Diurnal course of CO2 concentration measurements above and below the aspen overstory during spring and summer, 1994. The thin line is the concentration above the aspen overstory at the 39.5-m height and the heavy line is that within the canopy at the 4-m height, a) measurements during the spring (DOY 105-109, April 15-19), b) measurements during the summer (DOY 161-165, June 10-14), and c) measurements late in the growing season (DOY 228-232, August 16-20) 75 Fig. 3.7 Comparison of the values of ASJAt calculated using CO2 concentrations at all eight heights with those calculated from concentrations measured at selected heights at the OA site in the summer 1994. The selected heights were the 34.2 m (a), 2.3 and 34.2 m (b), 2.3, 9.5 and 34.2 m (c) and 2.3, 9.5, 21.9 and 34.2 m (d) heights. The dash-dotted lines are the linear best fits. The solid lines are one-to-one lines 76 Fig. 3.8 The CO2 concentration change from the previous half-hour (calculated as ACJAt = (C(.+1 - C M ) / 2 (umol mol"1 half-hour"1), where C is the concentration, / is the half-hour concerned) at different heights calculated from the data in Fig. 3.5 78 Fig. 3.9 The rates of change of C 0 2 storage in the air column beneath the eddy covariance systems a) at the 39.5-m height and b) at the 4-m height during the leafless period (DOY 35 -110). The dots are the half-hourly measurements and the line is the ensemble average 79 Fig. 3.10 The rates of change of CO2 storage in the air column beneath the eddy covariance systems a) at the 39.5-m height, and b) at the 4-m height in the summer months (DOY xxi 152-243). The dots are the half-hourly measurements and the line is the ensemble average 81 Fig. 3.11 Ensemble-averaged standard deviation of the vertical velocity at the 18.6-m height during the full-leaf period at the OA site, in 1994 82 Fig. 3.12 Diurnal courses of a) net ecosystem exchange (NEE) which is the sum of measured flux (Fc) at 39-m height and storage change, b) storage change and c) Fc during D O Y 224-233. The storage change was estimated with profile data 84 Fig. 3.13 Ensemble-averaged CO2 flux measured using the eddy covariance system at the 39.5-m height Fc, and the change rate in the CO2 storage (ASa/At) in the air column beneath the eddy covariance system and NEE during the summer months (June-August) at the OA site in 1994. The thin line is ASa/At, the thick line is Fc, and the dash-dotted line is the sum of the two 85 Fig. 3.14 CO2 concentration profiles on July 16-17, 1996 at the OA site 86 Fig. 3.15 The eddy CO2 fluxes (solid line) measured above the forest and ASJAt in the 0-39.5-m air column (dash-dotted line) at the OA site on July 16-17, 1996 88 Fig. 3.16 A l l half-hourly eddy CO2 fluxes measured at the 39.5-m and 4-m heights at the OA site in 1993 and 1994. During the growing season, most of the positive values shown are C 0 2 fluxes at night as a result of forest respiration, while the negative values which mainly occurred during the daytime are the result of photosynthesis exceeding daytime respiration 90 Fig. 3.17 Diurnal courses of the eddy CO2 fluxes corrected with ASa/At at the 39.5-m height (the thick line) and at the 4-m height (the thin line) (a) in the spring (DOY 105 -109), (b) in the summer (DOY 161 -165) and (c) late of the growing season (DOY 228 -232). The missing data were replaced by using linear interpolation 92 Fig. 3.18 Diurnal changes of C 0 2 concentration at the 4-m (on the 6-m scaffold tower) and 5.9-m heights (on the main tower) in the trunk space on two towers 40-m apart during August 12-22,1994 (DOY 224-234). The dotted line was the measurements made on the main tower and the solid line was the measurements made on the 6-m scaffold tower 93 Fig. 3.19 Comparison of the half-hourly CO2 concentrations measured on two towers using the same data as in Fig. 3.18. The triangles are nighttime data and the circles are daytime data. The dash-dotted line is the daytime regression line (C 4 m = 0.878Cs.9m + 43 p.mol mol"1, r 2 = 0.87, syx = 9.4 umol mol"1, C 4 m - C59m - 0.9 n_mol mol"1), and the dotted line is the nighttime regression line (C4m.= 0.670Cs.9m + 120 (irnol mol"1, r2 = 0.48, syx = 26.2 umol mol"1, C4m - C59m = 10.9 umol mol"1) 95 Fig. 3.20 Comparison of half-hourly values of the air-storage corrected CO2 flux measured with the eddy covariance systems at the 4- (thin line) and 5.9-m (thick line) heights on the scaffold and main towers (40-m apart) during August 12-22, 1994 (DOY 224-234). The rates of change of CO2 storage were estimated using profile data 96 Fig. 3.21 Comparison of half-hourly sensible heat fluxes measured with eddy covariance system at 4 ~ 6 m height in the canopy on two towers 40-m apart during August 12-22,1994 (DOY 224-234). The thick line is the flux measured at the 6-m height on the main tower and the thin line is that at the 4-m height on the 6-m scaffold tower. 98 Fig. 3.22 Same as Fig. 3.21 except latent heat fluxes 99 xxii Fig. 4.1 Typical diurnal course of air-storage corrected CO2 eddy flux measured at the 39.5-m height using the eddy covariance method (solid line) and the air CO2 storage beneath the instrument height (dotted line) at the OA site on August 13 - 14, 1994. The nighttime data points (18:00 - 6:00 CST) are two-hour running means 108 Fig. 4.2 Diagram illustrates the mass flow hypothesis at a sloping site during the nighttime, ui and U2 are the horizontal velocities, w is the subsidence due to cold air drainage. Mass flow as a result of cold air drainage transports CO2 downwind I l l Fig. 4.3 Schematic description of the conservation equation for CO2 in a forest ecosystem. The control volume is from the height of the eddy covariance sensors down to a depth where CO2 flux becomes negligible. For the definition of symbols, see the text.... 114 Fig. 4.4 Comparison of ensemble averaged normalized (by the area under the curve) cospectra of w and T (the solid line) and w and %c (the dot-dashed line) measured at the 39.5-m height at the OA site during (a) the daytime and (b) nighttime of seven days in 1996 118 Fig. 4.5 Relationship of the daily averaged (from February 4 to May 10) and the bin averaged (from May 11 to September 20), air-storage corrected nighttime eddy CO2 flux, representing the soil, hazelnut and aspen respiration (RSha), to the soil temperature measured at the 2-cm depth at the OA site in 1994. Vertical bars are ± 1 standard deviation. This relationship is independent of turbulence levels 125 Fig. 4.6 Relationship between the bin averaged values of the nighttime wind function (M) and the ratio of current half-hour um to the mean over the previous eight days from May 10 (DOY 130) to September 20 (DOY 263) in 1994. The vertical bars are ± 1 standard deviation. This function is independent of soil temperature 126 Fig. 4.7 Histogram of the calculated w using Lee's procedure at the 39.5-m height at the OA site over the period of June-August 1996 128 Fig. 4.8 Relationship between the energy imbalance (J = Rn + H + AE + Go + Jt) and vv at the 39.5-m height at the OA site in summer 1996 129 Fig. 4.9 Comparison of the nighttime energy imbalance and the mass flow term for sensible heat at the OA site in summer 1996 129 Fig. 4.10 Relationship between Fc (solid line) measured above the forest and calculated w at the 39.5-m height (dashed line) at the O A site on the 89 nights of summer 1996. The percentage is the occurrence of the case 131 Fig. 4.11 Comparison of the air CO2 storage corrected nighttime Fc and the calculated mass flow term at the OA site in summer 1996. The solid line is the expected relationship between these two fluxes 132 Fig. 4.12 Comparisons of air temperature profiles and CO2 concentration {\m\o\ mol"1 of moist air) profiles at the OA site on a windy night (September 4; DOY 247) and calm night (September 8; D O Y 251), 1994. The narrow panels are for DOY 247 and the wider panels are for DOY 251. The time intervals are 1:00 - 1:30 (•), 1:30 - 2:00 (•), 2:00 - 2:30 (A), and 2:30 - 3:00 CST (•) 133 Fig. 4.13 The nighttime rate of change of air CO2 storage (A) and Fc (•) in relation to w+ at the 39.5-m height at the OA site from June to August, 1994. A l l the data are bin averaged with an bin-width of 0.05 m s"1. The lines are the corresponding 3-order best fits. The top heavy line is the best fit of the sum of the half-hourly values of Fc xxiii and air CO2 storage (not the sum of the two other l ines). The vert ical bars are ± 1 standard deviat ion 136 F i g . 4.14 Measurements made at the O A site w i th snow cover on the ground f r om December 11 ( D O Y 346) to December 17 ( D O Y 352), 1996. The eddy C 0 2 f luxes and « + were measured at the 39.5-m height and the so i l temperatures were measured at the 2-cm depth 138 F i g . 4.15 Measurements made at the O A site without snow cover on the ground f r om October 27 ( D O Y 301) to November 1 ( D O Y 306), 1996. The eddy C 0 2 f luxes and w+ were measured at the 39.5-m height and the so i l temperatures were measured at the 2-cm depth : 141 F i g . 4.16 Ef fect of ra infal l on the difference between the measured CO2 e f f lux and the estimated ef f lux (RshaM) dur ing the nightt ime f rom the spring to fa l l of 1994. The data are the b in averages using a rainfal l b in w idth of 0.6 m m half-hour" 1 . The vert ical bars are ± 1 standard deviat ion 143 F i g . 4.17 Compar i son of the relationships between Fc + ASa/At and RSha (a) and between Fc + ASaIAt andMRsha (b) at the O A site in 1994. Va lues are nightt ime means. The dash-dotted lines are the regressions: Fc + ASa/At = 0.85Rsha + 0.31 (a) and Fc + ASa/At = 0.97MRsha + 0.27 (b) and the so l id l ines are the 1:1 lines 144 F i g . 4.18 The relat ionship between C 0 2 eff luxes measured ( F c + ASa/At) and calculated (MRsha) us ing the 1994 mode l at the O A site in 1996. Va lues are nightt ime means. The dash-dotted l ine is the regression: Fc + ASa/At = \.Q\MRsha + 0.03 and the so l id l ine is the 1:1 l ine 145 F i g . 4.19 Effects of the three approaches on the diurnal carbon sequestration at the O A site on the same day as in F i g . 4.1. The so l id l ine is based on the new approach, the dotted line is based on the H W S approach, and the dashed-dotted l ine is based on the A N F approach. The dotted l ine (HWS ) and the dashed-dotted l ine ( A N F ) are the same dur ing the dayt ime. (See dai ly totals of carbon sequestration in Table 1) 148 F i g . 4.20 Compar i son of relationships between nightt ime respiration rate and so i l temperature at the 2-cm depth, developed in the three approaches 149 F i g . 4.21 Compar i son of the cumulat ive carbon sequestration at the O A site for 1994, calculated us ing the three approaches. The so l id l ine is a result of the new approach, the dotted l ine is a result of the H W S approach and the dashed-dotted l ine is a result of the A N F approach 151 F i g . 5.1 Lea f area indices (LAIs ) of the aspen overstory and the hazelnut understory at the O A site in 1994. These L A I s were measured us ing an LA I -2000 Plant Canopy Ana l yze r . 164 F i g . 5.2 The reflectivit ies of P P F D of the forest (&y) and the hazelnut understory (CXA) at the O A site in 1994. M i n i m u m values usual ly occurred at noon 167 F i g . 5.3 D iurna l courses of Fc at the 39.5-m height and Fch at the 4-m height measured at the O A site on Ju ly 27 ( D O Y 208), 1994 174 F i g . 5.4 Typ i c a l dayt ime photosynthetic rates of the forest, aspen overstory and the hazelnut understory on the same clear day as in F i g . 5.5 (July 27, 1994, D O Y 208) at the O A site 175 F i g . 5.5 Typ i c a l diurnal patterns of incident P P F D (Qo) and air temperatures (T) at the 39.5-m and 4-m heights, and saturation def ic i t (D) on a clear day in the summer 1994 (July xxiv 27, DOY 208) at the OA site. Sunrise was at about 5:00 CST and sunset was at about 21:00 CST on this day 176 Fig. 5.6 Relationship between the photosynthetic rates (Pe, Ph and Pa) and absorbed PPFD (Qfa, Qha and Qaa) at the OA site during period of April 22-September 19 (DOY 112-262), 1994 178 Fig. 5.7 Relationship between the bin-averaged ratio of the measured photosynthetic rates {Pe, Ph and Pa) to their estimates from absorbed PPFD {Qfa, Qha and Qaa) and saturation deficit/) at the OA site during the period of April 22-September 19 (DOY 112-262), 1994. The bin-width AD = 0.1 kPa. The vertical bars are ±1 standard deviation 181 Fig. 5.8 Relationship between the bin-averaged ratio of the measured photosynthetic rate of the forest to its estimate from Qfa 8c D and air temperature T at the OA site during April 22-September 19 (DOY 112-262), 1994. The bin-width AT = 2 °C. The vertical bars are ± 1 standard deviation 184 Fig. 5.9 The light response curves of the photosynthetic rate of the forest from April 22 to September 19 (DOY 112-262), 1994 at the OA site in different sky conditions 186 Fig. 5.10 Relationship between photosynthesis (Pe, Ph and Pa) and evaporation (Ee, Eh and Ea) during the summer in 1994 at the OA site (e = forest ecosystem, h = hazelnut understory and a = aspen overstory). The solid line is the linear best fit. The dot-dashed line is the relationship for a wheat crop (Baldocchi 1994) 189 Fig. 5.11 Relationship between the eddy CO2 flux (-Fc) and evaporation (Ee) above the forest during summer in 1994 at the OA site. The solid line is the linear best fit. The dot-dashed line is the relationship for a wheat crop (Baldocchi 1994) 191 Fig. 5.12 Comparison of the modelled and the measured half-hourly photosynthetic rates (Pe) from April 22 to September 19 (DOY 112-262), 1994 at the OA site. The dashed line is the regression line and the solid line is the one-to-one line 194 Fig. 5.13 Comparison of the modelled (solid line) and measured (open circles) photosynthetic rates of the forest, the aspen and the hazelnut on the dates of July 9-14, 1994 at the OA site 195 Fig. 5.14 Comparison of the predicted (solid lines) and measured (circles) daytime mean photosynthesis of the forest, aspen overstory and the hazelnut understory in the growing season 1994 at the OA site. The dotted line is the sum of the calculated aspen photosynthesis and hazelnut photosynthesis using the model (Pa = PaiPa2Pa3, and Ph = PhiPh2Ph3) which is almost indistinguishable from the solid line in panel (a). 196 Fig. 5.15 Comparison of the half-hourly photosynthetic rates modelled using the 1994 model and measured using the eddy covariance system in 1996 at the OA site. The dashed line is the regression and the solid line is the one-to-one line 197 Fig. 5.16 Comparison of the calculated photosynthetic rate estimated using the models and measured using the eddy covariance system in 1996 at the OA site. The line is the values calculated using the models discussed above. The circles are the measured photosynthetic rates 198 Fig. 5.17 The half-hourly photosynthetic rates of the forest, the aspen and the hazelnut in 1994 at the OA site 199 XXV Fig. 5.18 The cumulative photosynthesis of the forest, aspen and hazelnut in 1994 at the OA site 202 Fig. 5.19 Cumulative ecosystem respiration and its partitioning between the aspen overstory and the hazelnut understory at the OA site for 1994. Totals are given in Table 5.6.203 Fig. A . l Profile of the bulk density measured at the OA site in 1994 225 Fig. B . l comparison of normalized (by the area under the lines) cospectra of w and T (solid line) and w and c c (the dot-dashed line) measured at 39.5-m height at the OA site during period of 21:00-23:00 CST on June 10 (DOY 162), June 21 (DOY 163) and July 1 (DOY 183), 1996. The values of w^on those nights were around 0.1 m s"1.. 226 Fig. B.2 Same as in Fig. B . l except on July 19 (DOY 201), June 27 (DOY 207) and August 27 (DOY 240), 1996 227 Fig. B.3 Same as in Fig. B . l except on September 10 (DOY 254), October 8 (DOY 282) and October 16 (DOY 290), 1996 228 Fig. B.4 Same as in Fig. B . l except on October 20 (DOY 294), October 31 (DOY 305) and November 11 (DOY 316), 1996 229 Fig. D . l Net radiation, sensible and latent heat fluxes measured above the OA forest with snow cover on the ground from December 11 (DOY 346) to December 17 (DOY 352), 1996 233 Fig. D.2 Net radiation, sensible and latent heat fluxes measured above the OA forest without snow cover on the ground from October 27 (DOY 301) to November 1 (DOY 306), 1996 234 xxvi ACKNOWLEDGMENTS I would like to thank my supervisor, Dr. T. A . Black for his inspirational teaching, his constructive comments on the drafts of this dissertation, his patience and friendship, and to thank his family for their help. Thanks also to the members of my supervisory committee, Drs. M . D. Novak, T. M . Ballard and I. McKendry, and Dr. A . G. Barr for their guidance, encouragement and interest in my research. I wish to acknowledge the University Graduate Fellowship of the University of British Columbia and the teaching and research assistantships provided by the Department of Soil Science. The funding of this research was provided in forms of a four-year Natural Science and Engineering Research Council (NSERC) Collaborative Special Project Grant in support of BOREAS and grants from the AES/NSERC Joint Science Subvention Program and the Canadian Forest Service. This research could not have been possible in its form without the freely-provided above-canopy data from Drs. Gerry den Hartog and Harold N. Neumann and their group at the Atmospheric Environment Service (AES), Downs view, Ontario. Their advice and friendship are also highly appreciated. The friendship of Peter Blanken both here in Vancouver and in the field are very much treasured and appreciated. I greatly appreciate the support given by numerous other individuals. Zoran Nesic's expertise on engineering and computing issues, his friendship and good sense of humor definitely made my life much easier here at U B C . On-site assistance was provided by John Deary, Tom Hertzog and Monica Eberle. Field operations were made possible by the support of the Prince Albert National Park personnel, especially Mary Dahlman, Paula Pacholek and Murray Heap. Janusz Olejnik and Marian Breazu assisted with instrument construction and electronics. Xuhui Lee, Ralf Staebler, Joe xxvii Eley, Rick Ketler, Siguo Chen, Uwe Gramann, Ralph Adams, Craig Russell, Aisheng Wu, Wenjun Chen, Isobel Simpson, Grant Edwards and Jose Fuentes provided various forms of field assistance. The long-time friendship, encouragement and advice regarding both academic and personal matters of Wenjun Chen, Xuhui Lee, Jing Chen, and Jane Liu are highly valued and treasured. Thanks also due to Zhong Chen, Gordon Drewitt, Bandi Hermawan, Ellyn Humphreys, Eva-Maria Jork, Kayuum Mansor, Alberto Orchansky, Bob Sager, Rob Swanson, Jon Warland for providing me friendship and a comfortable work environment here at the Soil Annex 3. I would also like to express my sincere thanks to Rev. David Morrison and the people at West Point Grey Baptist Church, and to everyone at the UBC Chinese Bible Study Group for providing me a spiritual home and for their prayer and fellowship. Thanks also to Mr. Paul Birch, Mr. and Mrs. Frank and Vera Burnham, Mr. and Mrs. Neil and Margaret MacPhee for proofreading this thesis. Last but not the least, thanks and love to my parents and the rest of the family and my in-laws in China for their love, caring and support all these years of my studies in school. Without them, I would not be able to be here to write these words. xxviii To my wife, Mary Duanni Qian and our son, Andrew Zijiang Yang xxix 1. INTRODUCTION Climate is changing. The Intergovernmental Panel on Climate Change has reported that global surface temperatures have increased by 0.3-0.6 °C since the late 19th century, with the greatest warming over the continents between 40 °N and 70 °N (Nicholls et al. 1996). It is very probable that this increase has been caused by an increase in the heat-trapping ability of the atmosphere due to the increased concentrations of CO2 and other greenhouse gases (Schimel et al. 1996). During the 1980s, emission of CO2 to the atmosphere was 7.1 ±1.1 Pg C yr"1, of which 5.5 ±0.5 Pg C yr"1 was from fossil fuel combustion and cement production, and the other 1.6 ±1.0 Pg C yr"1 was from changes in tropical land-use. During the same period, CO2 storage in the atmosphere increased by 3.3 ±0.2 Pg C yr"1 and the oceans absorbed 2.0 ± 0.8 Pg C yr"1. What is happening to the remaining 1.8 ± 2.1 Pg C yr"1 has yet to be determined (Schimel et al. 1996). Many scientists have suggested that the temperate and boreal forests are likely a sink for carbon (e.g., Tans et al. 1990; Enting and Mansbridge 1991; Ciais et al. 1995; Denning et al. 1995). The processes involved and the spatial distribution of carbon sequestration by the boreal forest, which covers about 8-10% of the Earth's land surface (i.e., 12-15xl0 6 km2) (Lieth 1975; Bonan and Shugart 1989), was the focus of the Boreal Ecosystem-Atmosphere Study (BOREAS) (BOREAS Experimental Plan 1994). BOREAS was a large scale study of the exchanges of radiation, heat, water, CO2 and trace gases between the boreal forest and the atmosphere over a large region of 1000x1000 km. A primary goal of BOREAS was to integrate data obtained from leaf and soil chambers, 1 Chapter 1. Introduction 2 micrometeorological towers, aircraft and satellites that are needed to improve models of these exchange processes. The 1994 field experiment was conducted with coordinated 3-4 week long Intensive Field Campaigns (IFCs) that took place mainly during the winter, thaw and growing season. There were ten tower sites, four and six, respectively, in northern (near Thompson, Manitoba) and southern (near Prince Albert, Saskatchewan) study areas both of which were about 100x100 km. These sites were located in old and young aspen, old and young jack pine, old black spruce and fen ecosystems (BOREAS Experimental Plan 1994, Sellers etal. 1995). As part of one of the tower flux groups (TF 1) of the BOREAS science team, the U B C Biometeorology/Soil Physics Group conducted research at the Old Aspen (OA) site (53.63 °N, 106.20 °W) in Prince Albert National Park, Saskatchewan, cooperatively with scientists from the University of Guelph, Guelph, ON and another group (TF 2) from the Atmospheric Environment Service, Downsview, ON. Using data from this field experiment, this dissertation focuses on carbon dioxide exchange measured using the eddy covariance technique, between this aspen forest and the atmosphere. The energy and water vapour exchange at the OA site is discussed elsewhere (e.g., Black et al. 1996; Blanken et al. 1997; Blanken era/. 1998; Blanken 1997). The Monin-Obukhov empirical functions (MO similarity theory, MOS) are now used routinely in many practical applications involving turbulent exchange above land surfaces (Hill 1989). There are, however, still questions regarding the applicability of MOS functions above very rough vegetation. In particular an important issue is the apparent enhancement of the eddy diffusivity of scalars (e.g., air temperature, CO2 concentration and water vapour density) within the roughness sublayer above forests. In Chapter 2 the distribution of the Chapter 1. Introduction 3 turbulent statistics within and above the forest at the O A site is briefly discussed. M O S theory is used to describe the variances of the wind speed and scalars. The application of the M O S flux-gradient relationship to estimate CO2 fluxes above and within the canopy is also examined in this chapter. The rate of change in the CO2 storage of the air column beneath the eddy covariance system is essential in the calculation of half-hourly carbon sequestration by the forest. How many air sampling heights are required to reasonably measure these rates of changes in storage when eddy covariance sensors are many metres above the ground? H o w important is the correction using these rates of change to eddy covariance measurements of C 0 2 flux above and below the aspen overstory? What is the spatial variability of concentration and flux measurements in the canopy? The latter concerns the size of the flux footprint above the forest floor (Baldocchi 1997). These are questions that are addressed in Chapter 3. Some workers have found that the CO2 fluxes measured using the eddy covariance technique on calm nights are often 'unreasonably' low (e.g., Black et al. 1996; Goulden et al. 1996). Do these low fluxes mean that CO2 is lost from the forest via some route (e.g. mass flow) other than vertical diffusion? This is often assumed to be the case and a correction is made to account for this loss. In Chapter 4, an alternative hypothesis is presented which suggests that these low fluxes are a result of CO2 accumulation in the soil and/or snow. Evidence is presented to support this hypothesis. This is an important issue in terrestrial ecosystem carbon balance research because depending on how these low nighttime fluxes are interpreted there can be considerable different in the annual and seasonal estimates of carbon sequestration. These differences can be large enough to result in uncertainty as to whether a site is a carbon sink or source. A n approach for estimating short-term eddy fluxes at night is Chapter 1. Introduction 4 also presented and used to estimate long-term carbon sequestration at the OA site in this chapter. There is a very dense hazelnut understory at the OA site. In order to develop and test stand level photosynthesis models, it is necessary to make field measurements of the CO2 being taken up by the overstory and understory strata (Jarvis and Sandford 1986). Therefore, the photosynthesis of the ecosystem is estimated and partitioned between the aspen overstory and the hazelnut understory in Chapter 5. The rates of photosynthesis thus obtained are then related to environmental factors including photosynthetic photon flux density, water vapour pressure deficit and air temperature. Chapter 5 is completed with an analysis of the 1994 carbon balance of the OA site. This includes estimates of soil respiration and net primary productivity of the site in 1994. Chapter 6 gives the summary and conclusions of this dissertation. Chapter 1. Introduction 5 1.1 REFERENCES Baldocchi DD (1997) Flux footprints within and over forest canopy. Boundary-Layer Meteorology, 85, 273-292. Bonan GB, Shugart H H (1989) Environmental factors and ecological processes in boreal forest. Annual Review of Ecology and Systematics, 20, 1-28. Black TA, den Hartog G, Neumann H H , Blanken PD, Yang PC, Russell C, Nesic Z, Lee X , Chen SG, Staebler R, Novak M D (1996) Annual cycle of water vapour and carbon dioxide above a boreal aspen forest. Global Change Biology, 2, 219-229. Blanken PD, Black TA, Neumann, HH, den Hartog G, Yang PC, Nesic Z, Staebler R, Chen W, Novak M D (1998) Turbulent flux measurements above and below the overstory of a boreal aspen forest. Boundary-Layer Meteorology, (in press). Blanken PD, Black TA, Yang PC, den Hartog G, Neumann, HH, Nesic Z, Staebler R, Novak M D , Lee X (1997) Energy balance and canopy conductance of a boreal aspen forest: partitioning overstory and understory components. Journal of Geophysical Research, 24, 28915-28927. Blanken PD (1997) Evaporation within and above a boreal aspen forest. Ph.D. Thesis, University of British Columbia, British Columbia, Canada, 179 pp. BOREAS Experimental Plan, Chapters 1-3, Version 3.0 (1994) (eds: Sellers PJ, Hall FG, Baldocchi DD, Cihlar J, Crill P, den Hartog G, Goodison B, Kelly RD, Lettenmaier D, Margolis H, Ranson J, Ryan M). NASA, Greenbelt, M D . Ciais P, Tans PP, Trolier M , White JWC, Francey RJ (1995) A large northern hemisphere terrestrial CO2 sink indicated by the 1 3 C / 1 2 C ratio of atmospheric CO2. Science, 269, 1098-1102. Denning AS, Fung IY, Randall D (1995) Latitudinal gradient of atmospheric CO2 due to seasonal exchange with land biota. Nature, 376, 240-243. Chapter 1. Introduction 6 Enting IG, Mansbridge JV (1991) Latitudinal distribution of sources and sinks of CO2: results of an inversion study. Tellus, 43B, 156-170. Goulden M L , Munger JW, Fan S-M, Daube BC, Wofsy SC (1996) Measurements of carbon sequestration by long-term eddy covariance: methods and a critical evaluation of accuracy. Global Change Biology, 2, 169-182. Hil l RJ (1989) Implications of Monin-Obukhov Similarity theory for scalar quantities. Journal of the Atmospheric Sciences, 46, 2236-2245. Jarvis PG, Sandford AP (1986) Temperate forests. In: Photosynthesis in Contrasting Environments (eds: Baker NR, Long SP). Elsevier Publishers B.V., Amsterdam, pp. 199-236 Lieth H (1975) Primary production of the major vegetation units of the world. In: Primary Productivity of the Biosphere (eds Lieth H , Whittaker, R.H.), pp. 203-216. Springer-Verlag, New York. Nicholls N , Gruza GV, Jouzel J, Karl TR, Ogallo L A , Parker DE (1996) Observed climate variability and change. In: Climate Change 1995: The Science of Climate Change, (eds. Houghton JH, Meira Filho LG, Callander BA, Harries N, Kattenberg A, Maskell K), Cambridge University Press, Cambridge, pp. 132-192 Schimel D, Alves D, Enting I, Heimann M , Joos F, Raynaud D, Wigley T, Prather M , Derwent R, Ehhalt D, Fraser P, Sanhueza E, Zhou X , Jonas P, Charlson R, Rodhe H , Sadasivan S, Shine KP, Fouquart Y , Ramaswamy V, Solomn S, Srinivasan J, Albritton D, Isaksen I, Lai M , Wuebbles D (1996) Radiative forcing of climate change. In: Climate Change 1995: The Science of Climate Change, (eds. Houghton JH, Meira Filho L G , Callander BA, Harries N, Kattenberg A , Maskell K), Cambridge University Press, Cambridge, pp 65-131 Sellers P, Hall F, Margolis H, Kelly B, Baldocchi D, den Hartog G, Cihlar J, Ryan M G , Goodison B, Crill P, Ranson KJ, Lettenmaier D, Wickland DE (1995) The Boreal Ecosystem-Atmosphere Study (BOREAS): an overview and early results from the 1994 field year. Bulletin of the American Meteorological Society, 76, 1549-1577. Tans, PP, Fung, IY, Takahashi, T (1990) Observational constraints on the global atmospheric C 0 2 budget. Science, 247, 1431-1438. 2. SIMILARITY RELATIONSHIPS ABOVE AND IN THE TRUNK SPACE OF A BOREAL ASPEN FOREST 2.1 INTRODUCTION Turbulent motions within and above the canopy greatly affect the exchange of scalars (e.g., air temperature, water vapour density and CO2 concentration) between the forest and the atmosphere (Raupach 1989; Wilson 1989). Monin-Obukhov similarity (MOS) theory is a powerful tool used to describe the turbulent properties and the transfer of scalars in the lower atmosphere. Relationships between various similarity functions and dimensionless length scales have been successfully developed and verified for the convective boundary layer (see Stull 1988). These relationships have been applied not only to the mean profiles of meteorological variables, such as wind speed and temperature, but also to the statistical quantities of turbulence (Panofsky and Dutton 1984; Kaimal and Finnigan 1994). These MOS empirical functions are now used routinely in many practical applications involving turbulent exchange above land surfaces (Hill 1989). Until the mid-1980s, however, the original similarity theory was not as successfully applied to the stable nocturnal boundary layers as it was to the daytime convective boundary layer (Nieuwstadt 1984). Based on a large number of turbulence observations made on a tall mast in the Netherlands and the idea of 'local scaling' in stable conditions (e.g., Brost and Wyngaard 1978), Nieuwstadt (1984) developed a complete formulation of local similarity theory. He 7 Chapter 2. Local similarity in an aspen forest 8 found that local fluxes and shear depended only on z/A where z is the height and A is the local Monin-Obukhov length. Since then, scientists have adopted this idea of local similarity and successfully conducted experiments over different surfaces in various parts of the world. Shao and Hacker (1990) successfully applied this theory to airborne observations in a heterogeneous coastal area of South Australia. They found that the variances of velocity components and potential temperature were similar in a local sense both in stable and locally convective conditions although the result for the variance of specific humidity was less satisfactory. Wang and Mitsuta (1991; 1992) concluded that the variances of vertical velocity, temperature and specific humidity were both well described by the surface layer similarity theory over an oasis in the Gobi area in China. Sivaramakrishnan et al. (1993) also found that the standard deviations of vertical velocity and temperature scaled with z/L in accordance with the MOS theory in a gently rolling agricultural field in India under both unstable and stable conditions. Xu et al. (1997) found that the similarity relationships of variances of velocity components and temperature, measured on two towers over urban and suburban areas in central China, to z/A agreed with the local similarity law well. Liu et al. (1998) found that the normalized standard deviations of vv, T and specific humidity q changed with z/L according to MOS theory over bare soil in Japan. These results strongly suggest that local similarity theory can be applied over various surfaces. Although Ohtaki (1984; 1985) has discussed the normalized standard deviation of the C 0 2 concentration and concluded that it followed the MOS relationship, none of these recent papers studied this important scalar. Chapter 2. Local similarity in an aspen forest 9 An important issue is the applicability of MOS theory to estimate fluxes above vegetation. Thorn et al. (1975) were the first to report an apparent enhancement in the eddy diffusivity of heat and water vapour (computed using MOS theory). This has been referred to as the Thetford anomaly (Raupach 1979). Since then others have observed similar effects (e.g., Viswanadham et al. 1987). Since air temperature and water vapour gradients above rough vegetation are difficult to measure, the measurement of C 0 2 gradients provides a better experimental approach to detecting an enhanced eddy diffusivity. Within the canopy, the applicability of MOS theory has been questioned since counter-gradient fluxes often occur (Denmead and Bradley 1985). In the Old Aspen (OA) forest, where the data used in this thesis were obtained, Blanken (1997) found that counter-gradient flux of sensible heat frequently occurred within the trunk space. However, Denmead and Bradley (1985) argued that the predicted fluxes using MOS theory should be similar to the measured ones when very close to the ground. Is this true just above the hazelnut understory in the OA forest with its clearly defined tall trunk space? Therefore, the objectives of this chapter are (1) to describe the profiles (vertical distributions) of the cup wind speed, the variances of the vertical velocity, temperature, and the air stability from the ground to approximately twice the tree height, (2) to apply M O S theory to the variances of the vertical velocity and scalars (air temperature, C 0 2 concentration and specific humidity) at twice the tree height and in the trunk space just above the understory, and (3) to determine the degree of enhancement of the eddy diffusivity of C 0 2 above that calculated using MOS theory at the above two heights. Chapter 2. Local similarity in an aspen forest 10 2.2 METHODOLOGY 2.2.1 Site, instrumentation and measurements 2.2.1.1 Site description As part of BOREAS, the OA site (53.63 °N, 106.20 °W) is located in the southern part of Prince Albert National Park, Saskatchewan. A natural fire occurred in 1919 (Weir 1996), resulting in an even-aged aspen overstory {Populus tremuloides Michx.) with a mean canopy height of 21.5 m, a diameter at the 1.3-m height of 20 cm and a stem density of 830 stems ha"1. The trunk space with almost no branches extends up to about 15 m. The understory is mainly hazelnut {Corylus cornuta Marsh.) about 2 m high with occasional clumps of alder {Alnus crispa (Ait.) Pursch). The ground cover is 50% leaf litter with a variety of shrubs (e.g., prickly rose, Rosa acicularis Lindl.) and grasses (BOREAS Experimental Plan 1994). The soil at the site is an Orthic Gray Luvisol which commonly occurs in the central to northern Interior Plains Region in Canada extending into the Boreal Forest Region (Expert Committee on Soil Survey (Canada), 1987). The surface organic layer is about 8-10 cm deep in which litter (mainly undecomposed leaves) is about 2 cm deep. The average bulk density of this layer is 160 kg m"3 (see Appendix A) indicating a porosity of about 88% assuming the soil organic particle density is 1300 kg m"3 (Brady and Weil, 1996). The corresponding values for the mineral soil to a depth of 40 cm are, respectively, 1300 kg m"3 and 50% Chapter 2. Local similarity in an aspen forest 11 assuming the mineral particle density is 2650 kg m"3 (Brady and Weil, 1996). The topography is relatively level and the fetch is at least 3 km in all direction. 2.2.1.2 Flux measurements As reported by Chen et al. (1998), above-canopy (39.5-m height) fluxes were measured using the eddy covariance technique on a 37-m walk-up scaffold tower (main tower) from October to November 1993, from April to September 1994. This system consisted of an omni-directional three-dimensional sonic anemometer/thermometer (model DAT-310, probe model TR-61B, Kaijo-Denki Co., Tokyo, Japan) and a closed-path infrared gas analyzer (IRGA, model 6262, LI-COR Inc., Lincoln, NE) in a temperature-controlled box. The path length of this sonic anemometer is 20 cm. Air was drawn at 6.5 L min"1 down 6 m of heated 3.2-mm inner diameter tubing (model Bev-a-line, Thermoplastic Processes Inc., Sterling, NJ), then pumped through the sample cell using two diaphragm pumps (model TD-4X2N, Brailsford Co. Rye, NY) connected in parallel. A l l the signals, at each input, passed a passive low-pass RC filter with a cutoff frequency of 50 Hz. They were sampled at a frequency of 100 Hz by an A/D board (National Instruments AT-MIO-16X), and then reduced to 20 Hz using a 5-point block average. This procedure and the input RC filter prevent signal aliasing and very effectively rejects 60 Hz A C power line noise. Half-hourly eddy fluxes of C 0 2 dumol m"2 s"1) were obtained on-line by calculating the covariance of the vertical velocity (w) and mole fraction of CO2 in moist air {y^c), and making the Webb, Pearman and Leuning (WPL) correction half-hourly for the effect of air density fluctuations (Webb et al. 1980). This procedure is equivalent to obtaining the CO2 flux by calculating the covariance of w and the CO2 mixing ratio (i.e., moles of CO2 per mole of dry air) as shown Chapter 2. Local similarity in an aspen forest 12 by Webb etal. (1980). During the summer of 1994, the WPL correction decreases downward daytime CO2 flux by 1-5% and decreases nighttime upward CO2 flux by 5-15%. The three-dimensional coordinate system of the anemometer was mathematically rotated to make the half-hour average values of the vertical and lateral velocity components equal to zero, thus giving fluxes perpendicular to the stream lines of the half-hourly mean air flow above the forest (Tanner and Thurtell 1969). Within-canopy (4-m height) fluxes were measured using an eddy covariance system on a 6-m scaffold tower (canopy tower) which was 40 m away from the main tower. Like the system on the main tower, the eddy covariance system on the canopy tower consisted of a three-dimensional sonic anemometer/thermometer (model 1012R2A (Solent), Gi l l Instruments, Lymington, England) with a path length of 15 cm and a temperature-controlled IRGA (model 6262, LI-COR Inc., Lincoln, NE). Air was drawn at 8.0 L min"1 down 3 m of heated 3.2-mm i.d. Bev-a-line tubing, then down 1.7 m of copper tubing (3 mm i.d.) coiled and sandwiched between two aluminum plates within the same housing as the analyzer and then through the analyzer's sample cell. The pump (model DOA-V191-AA diaphragm pump, Gast Inc., Dayton, OH) was located downstream of the sample cell resulting in the sample cell pressure being about 22 kPa less than atmospheric pressure. The delay time was about 0.8 s. This IRGA was operated in absolute mode with ultra-pure nitrogen at zero CO2 3 1 concentration flowing through the reference cell at 25 cm min" . An open path H2O analyzer (model KH20 hygrometer, Campbell Scientific Inc, Logan, UT) was operated continuously with this unit to evaluate signal delay time and any attenuation resulting from the sample tubing (Leuning and King 1992; Lee etal. 1994). A 0.0025 cm thermocouple was also operated with this unit in order to compare with the measurements from the Solent Chapter 2. Local similarity in an aspen forest 13 thermometer. The wind and temperature data were measured at 20.83 Hz. The signals of other scalars (CO2 and water vapour) had been passed through a passive filter with a cut-off frequency of 10 Hz, and then the Solent A/D converter sampled them at 20.83 Hz. The raw data were recorded using two 486 PC computer systems with back-up tape drives. Half-hour fluxes were calculated on-line with W P L correction but with a one-dimensional rotation to bring the lateral velocity component to zero because non-zero mean vertical velocities are possible within the trunk space (Baldocchi and Hutchison 1987). Since April 1996, this system has been operated at the 39.5-m height on the main tower. For CO2 concentration and flux measurements above the forest and within the trunk space, the water vapour pressure broadening correction (McDermitt et al. 1994; LI-COR, Inc. 2 1 1996) was also applied. The change in CO2 flux in umol m" s" caused by the broadening effect was approximately equal to the latent heat flux in W m"2 divided by -740. Since the half-hourly latent heat fluxes were usually positive at the OA site, this correction was mostly negative, which means that the annual carbon sequestration would have been underestimated by about 5-8% of the amount of carbon sequestered had the broadening correction not been applied (Chen etal. 1998). According to Blanken (1997), the distance at which the peak flux footprint occurs was 129 m, 99 m and 296 m under neutral conditions, typical daytime and nighttime, respectively above the aspen canopy at the 39.5-m height. It was 15 m, 10 m and 37 m under neutral conditions, typical daytime and nighttime, respectively, at the 4-m height. He further pointed out that the cumulative flux reached 80% of the total flux at an upwind distance of 900 to 2660 m at the 39.5-m height, and 90 to 333 m at the 4-m height. Thus, the micrometeorological rule-of-thumb of a minimum fetch-to-height ratio of 100:1 is almost Chapter 2. Local similarity in an aspen forest 14 sufficient at both heights to capture 80% of the upwind source area regardless of atmospheric stability. 2.2.1.3 Supporting measurements A CO2 concentration profile system, consisting of 8 levels: 0.8, 2.3, 9.5, 15.7, 18.8, 21.9, 25 and 34.2 m, was operated for the whole field experiment period in 1994. Air was drawn through heated Dekoron tubing (9.3-mm inner diameter) by a rotary pump and pushed through a LI-COR 6262 IRGA by a small diaphragm pump. Therefore, sample cell air was approximately at atmospheric pressure. A similar system (with a flow rate of about 24 L min"1) but with unheated tubing was used in July, August and part of September 1996 during the BOREAS-1996 Summer Intensive Field Campaign. The air CO2 storage was calculated from these C 0 2 concentration profile data. The rate of change for a given half-hour was estimated by calculating the difference between the mean air CO2 storage in the previous and following half-hours. In addition to the sonic anemometer/thermometer at the 39.5-m height, there are three other sonic anemometers at the 28.6-, 18.6- and the 5.9-m heights on the main tower. The models are omni-directional Kaijo-Denki DAT310 (probe model TR61B) and directional Kaijo-Denki DAT310 (probe model TR61A) at the 5.9-m height. A l l these anemometers have a 20-cm path length. The upper sonic (39.5 m) was mounted on top of a 4-m tall mast attached to the top deck of the main tower. The one at the 28.6-m height was mounted at the end of a folding boom about 4 m off the WSW face of the tower. The ones at the 18.6-m and 5.9-m heights were on shorter booms about 1.5 m off the WSW face of the tower. Another Kaijo-Denki miniature sonic anemometer, DA600 (probe model TR90AH) with a 5-cm path Chapter 2. Local similarity in an aspen forest 15 length, was mounted at the 0.5-m height on a separate support 8 m away SW from the SW corner of the main tower. Wind speeds and temperatures were sampled at 100 Hz and then down sampled to 20 Hz as mentioned above. The wind speed at the canopy height UhC was calculated using a polynomial regression obtained from the wind speed profile using sonic anemometers at the above mentioned five heights. This wind speed was then used to get the dimensionless wind speed U/Uf,c. 2.2.2 Local similarity theory 2.2.2.1 Variance similarity relationships Monin and Obukhov (1954) developed a similarity theory to describe the turbulence for the surface layer. They introduced two scaling parameters, namely, the friction velocity and the length L, which is commonly referred to as Monin-Obukhov length. The length L is defined as -Tul with (2.1) w^Ql = tff' + 0 . 6 1 7 W + 0.6 l r w T + OMw'TV + 0.61— zw7? where k is von Karman's constant (0.4), g is the gravitational constant (9.8 m s"2), T is the mean temperature, 9V is the virtual potential temperature, cp is the specific heat of moist air, z is the height, r is the mixing ratio of H 2 0 (r = (0.018e/P)/[28.97(l-0.001e/P) + 0.018e/P], Chapter 2. Local similarity in an aspen forest 16 in which e/P is the mole fraction measured using an IRGA), and (w'Q'v)s is the kinematic sensible heat flux at the surface. Local similarity is a direct analogy to Monin-Obukhov Similarity (MOS) theory except that the similarity functions are local ones, i.e., at the specific height or a horizontal location within a boundary layer. These functions are given in various places (e.g., Nieuwstadt 1984; Stull 1988; Shao and Hacker 1990): K ^ = « V + v'w C^~Wc/4 (2.2) A = -T(u'J/(kgwWv) where u'^ is the local velocity similarity function, 0^ is the local temperature similarity function, C£ is the local CO2 function, q'^ is the local humidity function and A is the local length function. In this set of scaling parameters, u, v and w are the longitudinal, lateral and vertical components of the velocity vector, respectively, p c is the CO2 concentration, and q is the specific humidity. Since the fluxes in the surface layer change very little with height, A can be replaced by L in this layer. Therefore, the scaling factors given above are identical to M - 0 scaling factors in the surface layer. As Shao and Hacker (1990) pointed out, the local similarity relationships are comparable with existing ones for the surface layer. Therefore, L is used to denote local length scaling parameter in the later section of this chapter. Chapter 2. Local similarity in an aspen forest 17 2.2.2.2 Flux-gradient similarity relationships Analogous to molecular diffusion, the flux within a turbulent boundary layer can be considered proportional to the gradient (K-theory). Using MOS theory, CO2 flux can be calculated as (Kaimal and Finnigan 1994): 3p c _ ku^z~d>>dpc F=-K-dz 4>, dz (2.3) where K is the eddy diffusivity, z is the height, d is the displacement height (13.6 m at the OA site, Simpson 1996), and a)f is the diabatic influence function for CO2. The function § F is assumed to be 'universal', and is often assumed to be the same for all scalars (Panofsky and Dutton 1984). It has been estimated empirically to be (Dyer and Hicks 1970; Hiigstrflm 1988 for unstable conditions and Webb 1970; Hicks 1976 for stable conditions): _ui-i6cyy2,c<o 0E. =<t>E= (l )H = l + 5 £ C>0 (2.4) where L, -z/L, tyE and §H are the diabatic influence functions for latent and sensible heat, respectively. Fc can also be calculated as -U^{Qc2-Qcl) In fz2-d> yZx-dj (2.5) -^F(z2) + ^Fc(zx) where t j^ is the integrated form of Q)^  (Paulson 1970) given by 1J> „ = $ E = ^ H = \ 2 In I(l + VI^l6X) .2 -st:, t,>o , C<o (2.6) Chapter 2. Local similarity in an aspen forest 18 where \\fE and \\fH are the integrated forms of the diabatic influence functions for latent and sensible heat, respectively. These two ways of calculating Fc (Eqs (2.3) and (2.5)) are used for the within- and above-canopy cases, respectively, later in the chapter. 2.3 RESULTS AND DISCUSSION 2.3.1 Stability within and above the forest Fig. 2.1 shows typical daytime and nighttime profiles of the local stability parameter £ at the OA site (September 10, 1994). During the daytime, the air above the forest and within the trunk space was usually unstable. During the nighttime, it was usually stable all the way down to the top of the hazelnut canopy. Sometimes it was stable even down to the ground (1:00-1:30 CST). During the period of August 10-September 20, 1994, the air in the lower canopy of the OA site was generally unstable during the daytime when the air above canopy was unstable. At night, the air above the forest was unstable or neutral (c^  < 0.02) for only about 7-9% of the time; this increased to about 30-35% within the trunk and the crown space. At the 0.5-m height, which is under the dense hazelnut understory, the air was unstable for about 85% of the half-hours (Fig. 2.1 and Table 2.1). In other words, the air within the canopy was generally stable at night with strong stratification in the lower trunk space above the hazelnut understory (see also Fig. 4.12, Chapter 4). This is very different from previous findings in tall tropical forest canopies that air conditions during the daytime are often strongly stable in the lower canopy while during the nighttime the lower canopy is typically unstable and the upper canopy stable (Fitzjarrald et al. 1990; Fitzjarrald and Moore 1990; Chapter 2. Local similarity in an aspen forest 19 Kaimal and Finnigan 1994; Raupach et al. 1996). The reason for this difference was that the structure of the Amazon forest is such that only 1-3% of incident solar radiation reaches the forest floor (Shuttleworth et al. 1984) while the aspen forest is so open that about 25-30% of the incident solar radiation reaches the top of the hazelnut understory (Blanken 1997). At night, the active radiative cooling surface is at the top of the canopy in the tropical forest while it is at the top of the hazelnut understory in the aspen forest. Therefore, the nighttime cold air at the crown-layer of the tropical forest sinks and generates turbulence within the canopy while the turbulence at the OA forest is strongly suppressed by the stable stratification due to the openness of the canopy. Chapter 2. Local similarity in an aspen forest 20 Fig. 2.1 Profiles of the local stability parameter t, = z/L within and above the forest at the OA site on September 10 (DOY 253), 1994. Panel (a) is the daytime profiles with time intervals of 13:00-13:30 (circle), 13:30-14:00 (square), 14:00-14:30 (triangle) and 14:30-15:00 CST (downward triangle). Panel (b) is the nighttime profiles with time intervals of 1:00-1:30 (circle), 1:30-2:00 (square), 2:00-2:30 (triangle) and 2:30-3:00 CST (downward triangle). Table 2.1 also shows that the air around the 18.6-m height was unstable or neutral for more half-hours. The data obtained at the 4-m height were not included due to its separation of 40 m from the main tower. This suggests that the air in the trunk space has more instability due to the cooling of the aspen crown layer. This is, however, much smaller in scale (time and length) than the mixing in the trunk space of a tropical forest, or in a closed tall canopy. During the winter, the air was more stably stratified within than above the forest. Chapter 2. Local similarity in an aspen forest 21 Table 2.1 The percentage of the nighttime half hours when the local stability parameter L, was less than 0.02, that is the air was statically unstable or neutral at the OA site in 1994. Height (m) August 10 - September 19 February 2 - April 10 0.5 84.9% no data 5.85 34.6% 24.4% 18.6 28.4% no data 28.6 7.8% 43.3% 39.5 6.9% 35.2% 0 0.4 E i § 0 . 2 o 0 Daytime Nighttime » • X 1 : . - , .• v.* ••• «• • ' -s?v •• 0 0.5 1 1.5 0 0.5 1 1.5 u, (m s"\ 39.5 m) Fig. 2.2 Relationship between the friction velocities and the standard deviations of the vertical velocities at the 4-m height to the friction velocity at the 39.5-m height at the OA site, 1994 Chapter 2. Local similarity in an aspen forest 22 2.3.2 Turbulence within and above the forest Fig. 2.2 shows the relationships between friction velocity and standard deviation of the vertical velocity (ow) at the 4-m height and the friction velocity measured at the 39.5-m height. It is seen that the within-canopy aw has a better correlation with the above-canopy friction velocity both day and night than the within-canopy friction velocity. Fig. 2.3 shows profiles of some turbulence statistics within the canopy measured for near-neutral conditions on the main tower during the period of August 10 - September 19, 1994 at the OA site. Mean wind speed (U) varies significantly with depth into the canopy (Fig. 2.3a). This change of normalized wind speed with height was within the range of scatter of the 'family portrait' of the canopy wind speed in different canopies (Figure 1 of Raupach et al. 1996). Mean wind (Fig. 2.3a) also showed a strong inflection near the top of the OA canopy although it is smaller than that reported by Baldocchi and Meyers (1988) for a deciduous forest in Tennessee. The latter is probably due to the limited measurement heights at the O A site or the exaggerated concentration of foliage in the crown space of the Tennessee forest. Another difference between the Tennessee and the OA forests was that the mean wind at the OA site did not show a second maximum within the trunk space as in the Tennessee forest. This again may due to the inadequate measurement heights (five in total) at the OA site. Chapter 2. Local similarity in an aspen forest 23 Fig. 2.3b shows that almost all the momentum was absorbed by the top layer of the canopy. The transmitted stress to the ground is virtually zero. This is similar to the U/U, he 4 5 T/(Qllf) 0.2 0.4 oTCC) (d) 0.6 Fig. 2.3 Vertical profiles of some turbulent statistics and the standard deviation of the temperature measured at the OA site, 1994 for neutral conditions which is defined here as |^| < 0.005. The symbols are the ensemble averages of the statistics during the period of August 10-September 19. The error bars were + 1 standard deviation. The friction velocity (u^ was measured at the 39.5-m height. Uhc is the interpolated wind speed at the top of the aspen canopy (21 m), T is the momentum flux (shear stress) and p is the air density. observation in other forests (Kaimal and Finnigan 1994; Raupach etal. 1996). The normalized standard deviation of vertical velocity shows a similar decreasing trend with the depth into the canopy (Fig. 2.3c). Its rate of decrease within the canopy, however, was smaller than that of the momentum. The normalized standard deviation of the vertical Chapter 2. Local similarity in an aspen forest 24 velocity did approach 1.25 above the forest and about 1.1 at the canopy height as reported by Kaimal and Finnigan (1994). Fig. 2.3d is the profile of the standard deviation of the virtual temperature. It is seen that the temperature fluctuated the most near the ground level. It decreased with increasing height. The amplitude of the change under neutral conditions was about only 0.2 °C. Fig. 2.4 shows profiles of the same turbulence statistics under unstable conditions. It indicates that the wind speed profile under these conditions changed little from that for neutral conditions (Fig. 2.4a). The momentum profile was also very similar (Fig. 2.4b) Chapter 2. Local similarity in an aspen forest 2 5 except the transmitted momentum to the ground was larger than that in the neutral case. This was probably because the thermal uplifting near the ground increased the vertical velocity. This assumption can also be supported by the observations in Fig. 2.4c. The standard deviation of the vertical velocity followed the same pattern as in Fig. 2.3c but is about 15-30% higher than that in neutral conditions. The standard deviation of the temperature was also larger than that in neutral conditions (Fig. 2.4d). Under stable conditions, i.e., during most of the nighttime hours, the statistics except for Fig. 2.5 Same as Fig. 2.3 except for stable conditions which is defined as t, > 0.005. the standard deviation of temperature also followed the patterns mentioned above. The wind speed above the forest was much higher than that at the canopy height, compared to that in Chapter 2. Local similarity in an aspen forest 26 the cases of neutral and unstable conditions (Fig. 2.5a). The value of the normalized wind speed was, however, the same within the canopy as that under neutral and unstable conditions. This indicates that the air above the forest and within the trunk space was more decoupled than under the previous two conditions. At night, there was a more constant wind speed layer within the trunk space than in the daytime. Occasionally, the wind speeds at the 5.85-m height were higher than elsewhere within the canopy. The standard deviations of the vertical velocity at all the heights were the highest in these three cases (Fig. 2.5c). The standard deviation o~w at the 28.6-m height was much higher than at any other heights. This is one of the reasons why the momentum flux was highest at this level (Fig. 2.5b). Mahrt et al. (1998) suggested that this large stress may be due to (1) a very thin boundary layer, (2) stress divergence associated with gravity waves and/or (3) change of footprint with height. Different to the neutral and unstable cases, the value of Or slightly increased with increasing height (Fig. 2.5d). Its magnitude was similar to the daytime value of about 0.25-0.3 °C. 2.3.3 Variance similarity relationships above the forest Using Eq. (2.2), the normalized standard deviations (square root of the variance) of vertical velocity w and the scalars (T, C, and q) are defined as follows: tyT = oTITit, (2.8) <t>c = °c/  C* The stability parameter is defined as t, = (z-d)/L at the 39.5-m height, and t, = z/L elsewhere. According to MOS theory, these standard deviations are functions of the stability Chapter 2. Local similarity in an aspen forest 27 parameter only. They will be examined briefly in this section. In this analysis, the half-hourly (j>w, cj>r, <j)c, and <j)9 whose difference from the mean of the month were larger than three times its standard deviation were disregarded (1-3%). 2.3.3.1 Variance of vertical velocity At the OA site, the standard deviation of the vertical velocity above the forest (39.5-m height) could be well described by the MOS relationships. Fig. 2.6 shows the half-hourly d)w versus the stability parameter at the OA site in August, 1994. It is seen that d)w can be described as k ( l + ^ ) ' / 3 f o r 0 < c ; < 1 0 cp = s (^ .9) [ c w ( l - ^ ) V 3 f o r - 1 0 < c ; < 0 where aw = 1.32, bw = 0.54, cw = 1.24 and dw = 1.37. This equation explained about 32-44% of the variance of the half-hourly § w data set for this month. Equation (2.9) suggests that tj>w approached 1.28 when the air was neutral at the OA site in August. This is very similar to the values of 1.23 over an urban area and 1.35 over a rural area in central China reported by Xu et al. (1997). The corresponding values for other surfaces were about 1.4 at the Cabauw tower site in the Netherlands (Nieuwstadt 1984), 1 for the coastal area in Australia (Shao and Hacker 1990), 1.14 for the Gobi dessert in China (Wang and Mituta 1991), 1.1 for an agricultural field in India (Sivaramakrishnan et al. 1992), and 1.3 over a bare soil in Japan (Liu etal. 1998). Under unstable conditions (t, < 0, mostly during the daytime at the OA site), d)w agreed fairly well with that reported by Shao and Hacker (1990). This suggests that the o)w at the OA site behaved following the 'universal' similarity theory. In his pioneer work, Nieuwstadt Chapter 2. Local similarity in an aspen forest 28 (1984) suggested that <|)w was independent of £ under stable conditions. Most recent research, however, suggests that tyw in stable conditions increases with £ following the '1/3' law (e.g., Shao and Hacker 1990; Wang and Mitsuta 1991, 1992; Xu etal. 1997). Our results are very similar to these findings. Close examination of Liu et al.'s (1998) results also revealed an increase in rjv with increasing t, in stable conditions. These findings suggest that the turbulence at our site was not different from that over other surfaces. Comparing the patterns of the tyw at night and during the daytime, we can see that they were also very similar, or symmetric to each other. This suggests that the sonic anemometer 10[ • • • ][ • • « ] '• i c i • • i I i— — i -10 -1 -0.1 -0.01 0.01 0.1 1 10 (z-d)/L (z-d)/L Fig. 2.6 Normalized standard deviation of vertical velocity (0J as a function of the stability parameter C, above the forest (39.5-m height) at the OA site in August. The dots are half-hourly values of <|)w with sensible heat flux H great than 10 W m"2. The solid line is the best 1/3 power fit as described in Eq. (2.9) and Table 2.2. The dash-dotted line is <j)w = (l + 2C,)Vi for £ > 0 and (pw = (1 - 6£,)1/3 for t; < 0 following Shao and Hacker (1990). Chapter 2. Local similarity in an aspen forest 29 used at the OA site could well measure the variation of the vertical velocity above the forest even when the sensible heat flux was small. The nighttime data, however, were more scattered than those of the daytime. A l l these results give us confidence in the vertical velocity measurements made during the nighttime. The results for other months are shown in Table 2.2 and Fig. 2.7. The value of tyw approached 1.15-1.33 when the air was neutral. The coefficients of determination were about 30-40% under stable conditions and about 30-60% under unstable conditions. These relationships were fairly similar between these months. This was more so for the nighttime than for the daytime. Table 2.2 The coefficients of the relationship between % and C, above the forest (39.5-m height) for different months at the OA site in 1994. C>o ;<o Month aw bw 2 r cw dw rz April 1.3062 0.5668 0.34 1.1814 2.4083 0.56 May 1.3028 0.6025 0.28 1.1546 2.9771 0.61 June 1.2562 0.8392 0.43 1.2253 1.2985 0.42 July 1.2732 0.6583 0.30 1.2635 0.7478 0.25 August 1.3179 0.5369 0.32 1.2397 1.3741 0.43 September 1.3340 0.4015 0.28 1.2592 1.4149 0.43 Chapter 2. Local similarity in an aspen forest 30 ~i 1 1 r " i 1 r April May June July August September _ l I I I I L . _ l L . -10 -8 -6 -4 -2 0 2 4 (z-d)/L 6 8 10 Fig. 2.7 The calculated values of § w from the coefficients in Table 2.2 for different months in 1994. 2.3.3.2 Variance of temperature Fig. 2.8 shows the change of the half-hourly normalized standard deviations of the air temperature with the stability parameter t, at the 39.5-m height at the OA site in August 1994. This relationship, as well as the relationship of other scalars, was described in the form of Eq. (2.10): | a t ( l + ^ ) _ 1 / 3 f o r 0 < c ; < 1 0 cj) = " ' (2.10) C ; c ( l - ^ V 3 f o r - 1 0 < c ; < 0 Chapter 2. Local similarity in an aspen forest 31 where x is temperature T, CO2 concentration C, or specific humidity q. In the case of the temperature, the coefficient aj was about 2.59, br was 0, c^was 2.63 and dr was 9.50. The relationships between cpr and the stability parameter both during the day and at night were similar to those reported by Shao and Hacker (1990). Fig. 2.8 Normalized standard deviation of temperature (cb^ ) as a function of the stability parameter t, - (z—d)/L above the forest (39.5-m height) at the OA site in August. The dots are half-hourly values of cbr with sensible heat flux H greater than 10 W m"2. The solid line is the best 1/3 power fit as described in Eq. (2.10) and Table 2.3. The dash-dotted line is <pr = 2.45(l - 20t,)_I/3 for t, < 0 and <pr = 3.46 for £ > 0 following Shao and Hacker (1990). Fig. 2.8 suggests that (pr approached 2.62 when the air was neutral. The corresponding values were about 3 at the Cabauw tower site in the Netherlands (Nieuwstadt 1984), about Chapter 2. Local similarity in an aspen forest 32 2.5 above a wheat field in Japan (Ohtaki 1985), about 2.96 for the coastal area in Australia (Shao and Hacker 1990), about 3.0 for an oasis in Gobi desert (Wang and Mitsuta 1991; 1992), about 6.56 for a agricultural field in India (Sivaramakrishnan et al. 1992), about 2.17 at an urban and a suburb tower site in central China (Xu et al. 1997), and about 2.0 over bare soil in Japan (Liu et al. 1998). This suggests the at our site behaved similarly to that over various other surfaces. The coefficients in Eq. (2.10) for temperatures in other months are listed in Table 2.3. The comparison of these relationships is shown in Fig. 2.9. It suggests that <$T approached 2.67 on average when the air was neutral. This is within the range of the values reported previously. The value of br varied from 0 to almost 3. Zero values in August and September were also observed by Shao and Hacker (1990), Wang and Mitsuta (1991; 1992), and Liu et Table 2.3 The coefficients of the relationship between § T and C, above the forest (39.5-m height) for different months at the OA site in 1994 £ > 0 t < 0 Month ar br 2 r cT dr 2 r April 3.4827 2.9814 0.19 2.1182 7.6403 0.64 May 2.6278 0.8874 0.09 2.5343 13.7894 0.72 June 2.7480 0.4382 0.03 2.5737 9.0734 0.74 July 2.8198 0.4165 0.05 2.4990 4.6069 0.60 August 2.5924 0 0.00 2.6292 9.5036 0.80 September 2.5527 0 0.02 2.8273 9.4214 0.66 Chapter 2. Local similarity in an aspen forest 33 al. (1998) in stable conditions. On the other hand, Xu et al. (1997) observed decreasing § T with increasing C. for stable conditions. The value of a)r approached about 1 which is similar to what we found in Fig. 2.9. -10 -8 -2 0 2 (z-d)/L 10 Fig. 2.9 The calculated values of <J>rfrom the coefficients in Table 2.3 for different months in 1994. 2.3.3.3 Variances of C 0 2 concentration and specific humidity Fig. 2.10 and Fig. 2.11 show the normalized standard deviations of C 0 2 concentration and specific humidity, respectively, under different stability conditions at the OA site in August, 1994. Similar to temperature, these relationships can be described by Eq. (2.10) with an ac of 2.90, a be of 0 , a cc of 2.84 , and a dc of 10.10 (Table 2.4) for C 0 2 concentration Chapter 2. Local similarity in an aspen forest 34 and an aq of 2.91, a bq of 0 , a cq of 2.66 , and a dq of 10.43 (Table 2.5). The relationships between (|)c and t^ , and between G)? and were very similar, however, there was a little more scatter for CO2 data. Comparing Table 2.4 and Table 2.5, we can also see that these two relationships were very similar to each other for July to September. Fig. 2.10 Normalized standard deviation of C0 2 concentration (<t>c) as a function of the stability parameter t, = (z-d)/L above the forest (39.5-m height) at the OA site in August. The dots are half-hourly values of <$>c with sensible heat flux H great than 10 W m"2. The solid line is the best 1/3 power fit as described in Eq. (2.10) and Table 2.3. The dash-dotted line is rj)c = 2.45(1 - 20Q"1 / 3 for £ < 0 and cj>c = 3.46 for t; > 0 following Shao and Hacker (1990). Chapter 2. Local similarity in an aspen forest Fig. 2.11 Normalized standard deviation of specific humidity (c^ ) as a function of the stability parameter t, = (z-d)/L above the forest (39.5-m height) at the OA site in August. The dots are half-hourly values of <j)c with sensible heat flux H greater than 10 W m"2. The solid line is the best 1/3 power fit as described in Eq. (2.10) and Table 2.3. The dash-dotted line is G) = 2.45(1 - 20Q"V3 for C, < 0 and cj>9 = 3.46 for t, > 0 following Shao and Hacker (1990). Chapter 2. Local similarity in an aspen forest 36 These relationships compare well with those reported by other researchers. Ohtaki (1985) found that <J)C approached 2.5 when the air was neutral over wheat fields in Japan. This agreed with our value of 2.7 fairly well. Wang and Mitsuta (1991) found that <j)? was very scattered and no definite functional relation was obtained over the Gobi desert because of the low specific humidity. They found, however, that the specific humidity variance similarity function was easily defined and was the same as that for air temperature over an oasis in the same region (Wang and Mitsuta 1992). Their value of o)g approached 3.0 under neutral conditions. Shao and Hacker (1990) also found that the application of the local similarity theory to humidity variance was less successful than to temperature over the Australian Table 2.4 The coefficients of the relationship between <bc and t, above the forest (39.5-m height) for different months at the OA site in 1994 i>0 C<0 Month ac bc 2 r cc dc r2 April 2.7363 1.4587 0.04 3.7156 4.7659 0.31 May 2.4627 0 0.00 2.7200 1.8275 0.17 June 2.7123 0 0.03 2.4522 4.2475 0.50 July 3.0100 0.2161 0.01 2.5119 4.9399 0.44 August 2.9036 0 0.01 2.8400 10.0969 0.67 September 2.7103 0 0.07 2.5201 2.8347 0.30 Chapter 2. Local similarity in an aspen forest 37 coastal area although their data suggested that the value of <J>g approached 5.5 when the air was neutral. However, Ohtaki (1985) found that the o)9 over a wheat field in Japan was well defined under unstable conditions. It approached 2.5 when the air was neutral. Liu et al. (1998) have also found a very plausible relationship between.d)9 and C, over bare soil in the same experimental farm as used by Ohtaki (1985) in Japan. Their value of <j)9 approached 2.4. Thus, our values of § q at neutrality were very similar to these findings Table 2.5 The coefficients of the relationship between <b? and C, above the forest (39.5-m height) for different months at the OA site in 1994 C<o Month aq r l cq r2 April 4.3990 1.8850 0.06 3.3258 6.0464 0.30 May 3.5535 0.5045 0.02 3.1363 6.4127 0.40 June 3.2479 0 0.00 2.4663 5.6366 0.61 July 3.0085 0 0.00 2.4414 7.0279 0.68 August 2.9140 0 0.00 2.6589 10.4289 0.76 September 2.8479 0 0.00 2.4637 7.4534 0.72 Chapter 2. Local similarity in an aspen forest 38 Fig. 2.12 compares the relationships between tyc and C, calculated using the coefficients listed in Table 2.4 in different months at the OA site. Except for May and August, these lines are fairly similar. Relationships for cp? versus C, are very similar to those in Fig. 2.9 for temperature. Furthermore, there is similarity between cpc, cp? and cpr. In summary, the MOS theory describes the relationships between the variances of the vertical velocity, scalars and the stability parameter in the surface layer at the OA site very. well. -10 -8 -2 0 2 (z-d)/L 10 Fig. 2.12 The calculated values of (bc from the coefficients in Table 2.4 for different months in 1994. Chapter 2. Local similarity in an aspen forest 39 2.3.4 Variance similarity relationships in the trunk space Fig. 2.13 shows the change of the half-hourly normalized standard deviation of the vertical velocity with the local stability parameter Z,=z/L calculated from the sensible heat 10 Q.1I- n , , • i -10 -1 -0.1 -0.01 0.01 0.1 1 10 z/L z/L Fig. 2.13 Normalized standard deviation of vertical velocity (cbj as a function of the stability parameter t, - z/L within the trunk space (4-m height) at the OA site in August. The dots are half-hourly values of The solid line is the best 1/3 power fit as described in Eq. (2.9) with the coefficient given in the text. The dash-dotted line is <J)W = (l + 2^) for t, > 0 and <j>w = (l - 6t,)V3 for £ < 0 following Shao and Hacker (1990). flux and at the 4-m height in August, 1994 at the OA site. Surprisingly, cj^  within the trunk space followed the 1/3 power law described in Eq. (2.9) with an aw = 1.49, bw = 0.40, Chapter 2. Local similarity in an aspen forest 40 cw = 1.48 and 6^ = 0.55 although the scatter of the data points was about one order of magnitude greater than that of above canopy tyw (Fig. 2.6). When the air stability was neutral, cpH, approached 1.49 which is a little bit higher than the 1.28 above the forest. Results in other months are given in Table 2.6. The values of aw are 1.5-1.85 compared Table 2.6 The coefficients of the relationship between § w and C, within the trunk space (4-m height) for different months at the OA site in 1994 C<0 Month aw bw 2 r cw dw 2 r April 1.7348 0.3196 0.21 1.5142 0.9228 0.41 May 1.8517 0.2214 0.08 1.6507 0.6441 0.30 June 1.5105 0.3081 0.15 1.4579 0.6735 0.30 July 1.5532 0.2594 0.17 1.4939 0.5665 0.28 August 1.4900 0.4041 0.27 1.4822 0.5475 0.25 September 1.5906 0.2709 0.20 1.5562 0.4752 0.31 to 1.26-1.33 above the forest. The values of cw are 1.46-1.65, compared to 1.15-1.26 above the forest. In the summer of 1994, mean a> approached 1.57 at the 4-m height when the air was neutral. This higher value at the 4-m height indicates that the wind within the trunk space fluctuated more than that above canopy due to the trees. Like the one above the forest, this value was in the range reported by other researchers over a variety of surfaces (e.g., Ohtaki 1984, 1985; Shao and Hacker 1990; Wang and Mitsuta 1991, 1992; Sivaramakrishnan etal. 1993; Xuetal. 1997; Liu et al. 1998). Chapter 2. Local similarity in an aspen forest 41 Fig. 2.14 shows the relationship between § T and t, at the 4-m height in August, 1994 at the OA site. Again it changed following the -1/3 power law. The relationship was f3.9917(1 + 0.8745<;)"1/3, £ > 0 cbj. = (2.11) 3.7802(1 -0.8745Lj V 3 , £ < 0 This equation explained about 40-50% of the variance in the half-hourly data set of this month. It shows that § T approached 3.89 when the air was neutral, compared to 2.61 for the above canopy §T- The results in other months are given in Table 2.7. On average, § T Fig. 2.14 Normalized standard deviation of temperature (tyT) as a function of the stability parameter t,=z/L within the trunk space (4-m height) at the OA site in August. The dots are half-hourly values of <\>T. The solid line is the best 1/3 power fit as described in Eq. (2.11). The dash-dotted line is Q)r =2.45(1-20<;)"I/3 fort;<0 and <py =3.46 for t, > 0 following Shao and Hacker (1990). Chapter 2. Local similarity in an aspen forest 4 2 approached 3.25 when it was neutral in the summer of 1994. The data were, however, very scattered. Table 2.7 The coefficients of the relationship between (j)?- and C, within the trunk space (4-m height) for different months at the OA site in 1994 C<o Month ar br r l d.T r2 April 3.1505 0.6648 0.25 3.4391 15.6795 0.70 May 2.6999 0.4030 0.13 2.1253 1.9262 0.43 June 3.6671 0.3148 0.08 2.6670 0.9101 0.19 July 3.4355 0.6621 0.24 2.5515 0.6076 0.28 August 3.9917 0.8745 0.47 3.7802 3.2951 0.40 September 3.1296 0.2181 0.13 2.7714 1.2473 0.28 Chapter 2. Local similarity in an aspen forest 43 Half-hourly standard deviations of CO2 concentration were very scattered in August (Fig. 2.15). They were slightly less scattered in other months, but the coefficient of determination was never higher than 50% (Table 2.8) under unstable conditions. Specific humidity showed a better picture (Fig. 2.16 and Table 2.9) with an r 2 of 31-60% under unstable conditions. Values of both <j)C and fya under neutral conditions were about 2.0-4.0 except in September for specific humidity. These values were not very different from 2.1-4.0 for the standard deviation of the temperature within the trunk space. This similarity suggests that the scalars Fig. 2.15 Normalized standard deviation of C0 2 concentration (<\>c) as a function of the stability parameter t>=z/L within the trunk space (4-m height) at the OA site in August. The dots are half-hourly values of tyc- The solid line is the best -1/3 power fit as described in Eq. (2.10). The dash-dotted line is fj>c = 2.45(1 - 20£)~V 3 for £ < 0 and cj>c = 3.46 for £ > 0 following Shao and Hacker (1990). Chapter 2. Local similarity in an aspen forest 44 at the 4-m height behaved similarly in 1994. Their standard deviations changed with the 10 -e-0.1 • • : •:. ' • ' t • , U> —i 1 — ' • « 1 . Ml , , 1 C i t — -10 -1 -0.1 -0.01 0.01 0.1 1 10 z/L z/L Fig. 2.16 Normalized standard deviation of specific humidity (cb?) as a function of the stability parameter t,=z/L within the trunk space (4-m height) at the OA site in August. The dots are half-hourly values of (p,. The solid line is the best -1/3 power fit as described in Eq. (2.10). stability parameter according to the -1/3 power law with considerable scatter. These results clearly show that whereas there is marked similarity between all scalars above the forest, this is not the case within the trunk space. In this case, w is similar to that above the forest. Temperature shows similar results. However, CO2 and H2O (particularly C0 2 ) have poor similarity relationships. This suggests that the behaviours of these two are very different to those above the forest and different to that of the temperature within the trunk space. Chapter 2. Local similarity in an aspen forest Table 2.8 The coefficients of the relationship between (bc and ^  within the trunk space (4-m height) for different months at the OA site in 1994 C>0 C<0 Month ac be r 2 cc dc r 2 April 2.4957 — 0.07 2.4237 2.4259 0.39 May 2.3330 1.3315 0.13 2.4070 2.6761 0.48 June 2.8131 — 0.02 2.9114 0.9891 0.14 July 3.4532 0.0131 0.00 4.0047 2.5047 0.31 August 3.7738 0.0923 0.01 3.0589 0.4615 0.10 September 3.3856 0.0484 0.00 2.2913 0.6252 0.19 Table 2.9 The coefficients of the relationship between § q and C, within the trunk space (4-m height) for different months at the OA site in 1994 £ > 0 C<0 Month aq \ r 2 cq dq r 2 April 3.1237 0.3200 0.08 4.3165 27.4396 0.59 May 3.2517 0.7800 0.11 2.4266 2.9331 0.51 June 3.6369 0.1208 0.01 2.1401 0.9346 0.31 July 3.7229 0.3102 0.06 2.1852 1.0264 0.47 August 2.7229 0.1854 0.04 2.0617 1.0499 0.34 September 7.3393 5.4515 0.20 4.2457 13.9492 0.60 Chapter 2. Local similarity in an aspen forest 4 6 2.3.5 Flux-gradient similarity relationships above the forest and in the trunk space 2.3.5.1 Diabatic influence function (cpF) and the stability parameter (C) According to MOS theory, the diabatic influence function, or the dimensionless concentration gradient in the inertial sublayer should change with the stability parameter following Eq. (2.4). This was tested by calculating cpFc using Eq. (2.3). The gradient above the forest was calculated using the CO2 concentrations measured at the 21.9- and 34.2-m heights using the CO2 profile system, while that within the trunk space was calculated using the CO2 concentrations measured at the 2.3- and 9.5-m heights using the same profile system. Fc and were measured at the 39.5-m and the 4-m heights, respectively. This assumes that the fluxes were approximately constant with height in the air layer above the forest and within the trunk space. In applying Eq. (2.3) above the forest, the height z was assumed to be the logarithmic average of the two CO2 concentration measurement heights (26.1 m), while for the trunk space, z was assumed to be 4 m. The values of d were 13.3 m for the forest (Simpson 1996) and 2 m (the understory height) for the trunk space. It is clear at the outset that conducting this test would be very difficult because the vertical gradients of CO2 concentration were small. For example, typical daytime gradients above the forest and in the trunk space were about 0.1 and 0.2 p:mol mol"1 m"1, respectively. Data points with Fc > 10 p.mol m"2 s"1 (unreasonably high) or IFCI < 0.5 itmol m"2 s"1 (CO2 flux resolution) were eliminated from the calculation. If the CO2 concentration difference (Apc) between the two pairs of heights was less than 0.2 u.mol mol"1 (CO2 concentration resolution/accuracy), the Chapter 2. Local similarity in an aspen forest 47 data point was also eliminated. Using these criteria about 31% of the data points in the full-leaf period were eliminated. Fig. 2.17 shows the relationship of the reciprocal of the diabatic influence function, or inverse stability function for CO2 both above the forest and within the trunk space for the full-leaf period at the OA site. The above-forest value of cj)^ 1 was about 0.93 at neutrality and increased to about 1.8 when £ dropped to -0.2, compared to the values of 1.0 and 2.0, respectively, from Eq. (2.4). In other words, the values of the measured enhancement factor, 5 4 3 2 1 0 -1 -2 ^ 4 3 2 1 0 -1 -2 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 (z-d)/L Fig. 2.17 Relationship between the reciprocal of the diabatic influence function for C0 2 and the stability parameter in the full-leaf period (June-August) of 1994 at the OA site, (a) is above the forest and (b) is within the trunk space. The lines are obtained using Eq. (2.4). Chapter 2. Local similarity in an aspen forest 48 i.e., the ratio of the measured to calculated cp~' at the 26.1-m height (about 1.2 hc, where hc is canopy height) were about 0.90 and 0.90. The mean values of the enhancement factor under unstable (-0.5 < t, < -0.05), near neutral (-0.05 < £ < 0.05) and stable (0.05 < £ < 0.4) conditions were about 0.72, 0.88 and 2.3, respectively. This indicates no enhanced turbulence transport above that based on the MOS theory at the OA site except in stable conditions. This is surprisingly different to that reported by other researchers (e.g., Garratt (1978) over a savannah, Raupach (1979) over a pine forest, Denmead and Bradley (1985) over a pine forest, Cellier and Brunet (1992) over a tall maize canopy). They found that the enhancement factor for heat transfer was 1.7-2.5 above the forest (see Cellier and Brunet (1992) for a thorough review). Cellier and Brunet (1992) also found that the enhancement factor decreased with height and reached about unity when z/hc ~ 1.5. Furthermore, they found a striking similarity between the enhancement factors for heat and water vapour. Using a tunable diode laser trace gas analysis system, Simpson (1996) found the enhancement factors for CO2 were about 1.60 (-2 < £, < -0.05), 1.57 and 1.66 respectively, under the unstable, near neutral and stable conditions for z/hc = 1.2-1.4 in a little denser deciduous forest at Camp Borden, Ontario, Canada during full-leaf period in 1995. The enhancement factors decreased to 1.10 when z/hc reached to 1.9-2.2. Furthermore, she observed that the enhancement factor at different heights (1.4-2.2 hc), especially during post-senescence period, was sometimes less than unity. In the trunk space (Fig. 2.17b), cp"1 shows a remarkably similar dependence on l\ as that above the forest. The enhancement was more evident here than above the forest. The corresponding values of the enhancement factor within the trunk space were 0.82,1.58 and 2.05, respectively. Chapter 2. Local similarity in an aspen forest 4 9 To take a close look at the enhancement factor, a manual calculation was done. On a typical summer day (DOY 240), the average daytime difference in CO2 concentration between the 21.9- and 34.2-m heights was about 2 umol mol"1, mean Fc was about -30 umol m"2 s"1, was about 0.5 m s"1, and t, was about -0.2. This gives an enhancement factor of about 0.76 (z = 28.1 m, the arithmetic centre of the two heights) and 0.87 (z = 26.1 m, the logarithmic centre of the two heights). To get an enhancement factor of 1.7, a CO2 concentration difference of 1.14 umol mol"1 is needed (z = 26.1 m). To get an enhancement factor of 2.0, a CO2 concentration difference of 0.87 umol mol"1 is needed. If the low enhancement factor observed at the OA site was mainly due to errors in CO2 concentration gradient measurements, the errors would be likely due to error in the sensitivity of the IRGA since the offset has been cancelled in the calculation of the CO2 concentration difference because one IRGA sampled all the eight heights. The error in the sensitivity would have to be more than 70%. This is very unlikely because the CO2 concentration profile system used at the OA site was well maintained and calibrated every half-hour automatically. However, the scatter of the data were very likely because of the measurement error in CO2 concentration. 2.3.5.2 Comparison of calculated and measured above-forest C 0 2 fluxes during the full-leaf period Another way of presenting the flux-gradient data reported in the previous section is to make a direct comparison of the calculated and measures fluxes. From the concentration difference at the 21.9-m and the 34.2-m heights, the half-hourly CO2 flux above the forest was calculated using Eq. (2.5) rather than Eq. (2.3) because of the large separation of the two Chapter 2. Local similarity in an aspen forest 50 measurements heights. A l l of the calculated half-hourly fluxes were then averaged in 2-hour increments. This averaging was to eliminate spatial variability since the half-hourly CO2 concentration measurements at a certain height were made in a period of 2-3 minutes in a half-hour. Fig. 2.18 compares the 2-hour mean C 0 2 flux (Fc cal) calculated using Eq. (2.5) to the 2-hour mean flux (Fc meas) measured at the 39.5-m height under different stability conditions in -40 -30 -20 -10 0 10 20 F calculated (nmol r r f 2 s"1) Fig. 2.18 Comparison of C0 2 flux calculated using Eq. (2.5) and that measured using the eddy covariance system at the 39.5-m height in June, July and August, 1994 at OA. The data were 2-hour averages. The circles are data under unstable conditions (t, < -0.02). The triangles are the data under neutral conditions (|c;|<0.02). The crosses are the data under stable conditions (£, > 0.02). The values of flux calculated from the gradient which were higher that 10 umol m"2 s"' were eliminated. The dot-dashed line is the regression, and the solid line is the one-to-one line. The coefficients of the regression equation are shown in Table 2.10. Chapter 2. Local similarity in an aspen forest 51 the summer of 1994. Using all the data shown on the graph, the following relationship was obtained Fmeas = j Qg ^cal _ ^ ^ About 64% of the variance in this 2-hour mean data set could be explained by this equation. This agrees with the findings that the MOS relationship underestimates scalar fluxes in the roughness sublayer reported by other researchers (e.g., Thorn et al. 1975; Garratt 1978; Raupach 1979; Denmead and Bradley 1985; Shuttleworth 1989; Cellier and Brunet 1992; Simpson 1996). This slight enhancement is somewhat different from the little and no enhancement reported in the previous section. By categorizing the data according to stability, it was found that the relationship between the calculated and the measured fluxes was better under stable and neutral conditions than that under unstable conditions. The calculated flux explained 49-57% of the variance in the measured flux data set (Table 2.10). This agrees with that reported by Thorn etal. (1975) over a pine forest and by Shuttleworth (1989) over a tropical forest in Brazil. Under unstable conditions, the calculated fluxes were very scattered. The relationship between the calculated and the measured fluxes explained only about 28% of the variance (Table 2.10). This is likely because eddies were so large under unstable conditions that the fluxes of scalars were no longer so driven by the local gradients as they were under stable conditions. To confirm this hypothesis, the flux calculated from the CO2 gradient between the 15.6-m height and 34.2-m height was compared to the measured CO2 flux. The coefficient of determination r2 increased substantially to 41%. The new relationship, however, became Fcmeas = 1.2956FC"" -10.5654. During the summer in 1994 (Fig. 2.18), counter-gradient flow was observed above the forest at the OA site 9.5% of the time (77 two-hour periods) in which 18%, 13% and 69% Chapter 2. Local similarity in an aspen forest 52 occurred under unstable, neutral and stable (nighttime) conditions, respectively. Only 4.6% of the unstable periods had counter-gradient flow. It increased to about 12 and 13% under stable and neutral conditions, respectively. 2.3.5.3 Comparison of calculated and measured trunk-space CO2 fluxes during the full-leaf period For the transport of scalars within the canopy, Denmead and Bradley (1985) reported, in the Uriarra forest in Australia: It is evident that the profiles provide a very distorted picture of transport everywhere in the canopy. Their indications are worst near the bottom of the needle space, where they differ largely from reality, even to the extent of predicting fluxes up to twenty times the observed values, and for the heat flux of the wrong sign. Only very close to the ground do the predicted fluxes come close to matching observation. Blanken (1997) also reported that counter-gradient flux of sensible heat occurred within the trunk space at the OA site. What about the CO2 flux? How often did counter-gradient flux of CO2 occur within the trunk space? Can flux-gradient similarity relationship be used in the trunk space? To answer these questions, trunk-space CO2 fluxes calculated using similarity theory from the CO2 concentration gradient between the 2.3-m and the 9.5-m heights was compared to the measured fluxes in the trunk space. In this case, Eq. (2.3) was used rather than Eq. (2.5) because of lack of local X, data at these two heights. Using the flux data obtained at the 5.9-m height on the main tower during the period August 12-21, 1994 (Fig. 2.19), it is found that the counter-gradient flow was observed slightly more frequently in the trunk space than above the forest. Over this period, counter-gradient and zero-gradient flow Chapter 2. Local similarity in an aspen forest 53 was observed for about 11.7% of the time (nine two-hour periods) which all occurred under stable conditions. This is much less than the percentage of 70% reported by Denmead and Fig. 2.19 Comparison of C0 2 flux calculated using Eq. (2.3) and that measured using the eddy covariance system at the 4-m height in June, July and August, 1994 at OA. The data were 2-hour averages. The circles are data under unstable condition which is t, > -0.02. The triangles are the data under neutral condition which is |£|<0.02. The crosses are the data under stable condition which is i ; > 0.02. The values of flux calculated from the gradient which were higher that 10 umol m"2 s"1 were eliminated. The heavy line is the regression, and the thin line is the one-to-one line. The coefficients of the regression equation are shown in Table 2.10. Bradley (1985) probably because of the relative openness of the OA forest. Chapter 2. Local similarity in an aspen forest 54 Table 2.10 Coefficients of the relationship between the C0 2 flux calculated using the MOS relationship from C0 2 gradient and the flux measured using the eddy covariance system above the forest and within the trunk space in the summer 1994 at the OA site. Stability condition a b 2 r Above the forest A l l data 1.0812 -1.0750 0.6430 Unstable 0.4976 -9.4573 0.2830 Neutral 1.0881 -2.3924 0.4928 Stable 0.9636 1.2951 0.5684 Within the trunk space (at the 5.9-m height on the main tower) A l l data 0.2252 3.4781 0.1176 Unstable 0.2122 3.4576 0.1081 Neutral — — — Stable 0.2780 3.4029 0.1568 Within the trunk space (at the 4-m height on the canopy tower) A l l data 0.3328 0.8730 0.2475 Unstable 0.3977 0.4240 0.4452 Neutral 0.5321 1.0270 0.5670 Stable 0.1905 1.5699 0.0643 The relationship between the calculated and the measured C 0 2 fluxes was very poor (Fig. 2.19 and Table 2.10). The coefficient of the determination was only about 10-16%. There Chapter 2. Local similarity in an aspen forest 55 was no difference in rz values when separating the data on the basis of stability. These results indicate that MOS theory on flux-gradient relationship was not applicable within the trunk space. 2.3.5.4 Comparison of calculated and measured C O 2 fluxes during leaf-free and full-leaf periods In this section, comparison of calculated and measured CO2 fluxes are shown for leaf-free and full-leaf periods. Rather that using 1:1 type plots as in the previous section, comparisons Fig. 2.20 C0 2 fluxes calculated from the C0 2 concentration gradient and measured using the eddy covariance systems during the period of April 16-20,1994 at OA. Panel (a) shows the PAR (Q0> solid line) and the friction velocity (u^ dashed line). Panel (b) shows the calculated (solid line) and measured (circles) C0 2 fluxes at the 39.5-m height. Panel (c ) is the same as (b) except at the 4-m height on the canopy tower. The data were 2-hour averages. Chapter 2. Local similarity in an aspen forest 5 6 are made using plots showing how fluxes change as a function of time, i.e., time series plot. This type of plot shows how well the calculated flux follows the hour-to-hour changes in the measured flux. Fig. 2.20 and Fig. 2.21 compares the two fluxes during a period of five clear days before leaf emergence and during the full-leaf period in 1994, respectively. Fig. 2.20 shows that the agreement between the calculated and measured fluxes within the trunk space was very good both in magnitude and in pattern. Surprisingly the agreement was not as good above the forest, particularly during the daytime. In the case of the full-leaf period, there was good agreement between calculated and measured fluxes above the forest, especially during the 226 227 228 229 230 DOY, 1994 Fig. 2.21 Same as Fig. 2.20 in except July 1-5. Panel c) is the fluxes at the 5.9-m height on the main tower Chapter 2. Local similarity in an aspen forest 57 nighttime (Fig. 2.21b). Within the trunk space, these two fluxes hardly agreed with each other during the period of August 14-18, 1994. However, there was good agreement at the 4-m height both day and night on some of the days such as D O Y 130, 154, 164-166, 183, 190, 205 (data not shown) when was high. 2.4 CONCLUSIONS In this chapter, the turbulence regime and the applicability of the MOS relationships at the OA forest have been examined. Unlike within a closed canopy, the air within the OA forest was usually stably stratified during the nighttime and unstable during the daytime. Although stable stratification was strongest in the trunk space, a larger proportion of nighttime half-hours with unstable conditions occurred there than at other heights. The air beneath the hazelnut understory was usually unstable during both daytime and nighttime. Wind speed showed a strong inflection at the top of the forest canopy and decreased gradually with the depth into the trunk space. A constant wind speed layer with occasional secondary wind speed maxima existed in the trunk space during the nighttime. Both the momentum and standard deviation of the vertical velocity decreased with depth into the forest. Air temperature showed more variation at the ground level than above the forest under neutral and unstable conditions while the opposite occurred under stable conditions. The relationships of the normalized standard deviation of the vertical velocity {§w) and scalars (air temperature, CO2 concentration and specific humidity) to the local stability parameter (C) above the forest followed MOS theory. tyw changed with C, according to the 1/3 Chapter 2. Local similarity in an aspen forest 58 power law both during the daytime and at night. The relationship between the normalized standard deviations of the scalars (air temperature, CO2 concentration and specific humidity) and C, followed the -1/3 power law under unstable conditions. They were often invariant with t, during stable conditions. The parameters in the power-law relationships for the CO2 and specific humidity were very similar. The values of the normalized standard deviations for all three scalars in neutral conditions were similar to those reported by other researchers. The applicability of the local similarity theory within the trunk space was surprisingly good. tyw changed with following the 1/3 power law although the data points were fairly scattered and the value in neutral conditions was higher than that above the forest. The normalized standard deviations of air temperature and specific humidity were more scattered but had a reasonably well-defined relationship in unstable conditions. The normalized standard deviations of CO2 concentration were, however, too scattered to give a reasonable relationship. At the OA site, no significant enhanced turbulent transport of CO2 above that based on MOS theory both above the forest and within the trunk space was observed. Within the trunk space, the inverse stability function for CO2 ((j)^ 1) showed a remarkably similar dependence on £ to that above the forest. Counter-gradient flux of C 0 2 was observed both above the forest and in the trunk space at the O A site. It occurred slightly more frequently within the trunk space than above the forest. However, the occurrence was not as frequent as reported by Denmead and Bradley (1985). The MOS flux-gradient relationship appeared to be applicable above the forest. On average, calculated fluxes using Eq. (2.5) were about 8% Chapter 2. Local similarity in an aspen forest 5 9 lower than that measured at the 39.5-m height during the summer in 1994; however, for individual half-hours there were considerable differences. These findings point to the importance of making accurate concentration gradient measurements to resolve this issue (Simpson 1996). The relationship was better under stable and neutral conditions than under unstable conditions. Within the trunk space, the flux-gradient relationship was not applicable. Chapter 2. Local similarity in an aspen forest 60 2.5 REFERENCES Baldocchi DD, Meyers TP (1991) Trace gas exchange above the forest floor of a deciduous forest. 1. Evaporation and CO2 flux. Journal of Geophysical Research, 96, 7271-7285. Baldocchi DD, Hutchison B A (1987) Turbulence in an almond orchard: Vertical variations in turbulent statistics. Boundary-Layer Meteorology, 40, 127-146. Blanken PD (1997) Evaporation within and above a boreal aspen forest. Ph.D. Thesis, University of British Columbia, British Columbia, Canada, 179 pp. BOREAS Experimental Plan, Chapters 1-3, Version 3.0 (1994) (eds: Sellers PJ, Hall FG, Baldocchi DD, Cihlar J, Crill P, den Hartog G, Goodison B, Kelly RD, Lettenmaier D, Margolis H, Ranson J, Ryan M). NASA, Greenbelt, M D . Brady NC, Weil RR (1996) The Nature and Properties of Soils. Prentice Hall, Upper Saddle River, N . J., USA. Brost RA, Wyngaard JC (1978) A model study of the stably stratified planetary boundary layer. Journal of Atmospheric Science, 35, 1427-1440. Cellier P, Brunet Y (1992) Flux-gradient relationships above tall plant canopies. Agricultural and Forest Meteorology, 58, 93-117. Chen WJ, Black TA, Yang PC, Barr A G , Neumann HH, Nesic Z, Novak M D , Eley J, Ketler RJ, Cuenca R (1998) Effects of climatic variability on the annual carbon sequestration by a boreal aspen forest. Global Change Biology (in press). Denmead OT, Bradley EF (1985) Flux-gradient relationships in a forest canopy. In: The Forest-Atmosphere Interaction (eds: Hutchison BA, Hicks BB). Reidel, Dordrecht, pp. 421-442. Dyer AJ , Hicks BB (1970) Flux-gradient relationships in the constant flux layer. Quarterly Journal of the Royal Meteorological Society, 96, 715-721. Chapter 2. Local similarity in an aspen forest 61 Expert Committee on Soil Survey (Canada) (1987) The Canadian System of Soil Classification (2nd ed.). Publication/Agriculture Canada 1646, 164 pp. Fitzjarrald DR, Moore K E (1990) Mechanisms of nocturnal exchange between the rain forest and the atmosphere. Journal of Geophysical Research, 95, 16839-16850. Fitzjarrald DR, Moore K E Cabral OMR, Scolar J, Manzi AO, De Abreu Sa L D (1990) Daytime turbulent exchange between the Amazon forest and the atmosphere. Journal of Geophysical Research, 95, 16825-16838. Garratt JR (1978) Flux-profile relationships above tall vegetation. Quarterly Journal of the Royal Meteorological Society, 104, 199-211. Hicks BB (1976) Wind profile relationships from the Wangara experiment. Quarterly Journal of the Royal Meteorological Society•, 102, 535-551. Hil l RJ (1989) Implications of Monin-Obukhov similarity theory for scalar quantities. Journal of the Atmospheric Sciences, 46, 2236-2245. Hogstrom U (1988) Nondimensional wind and temperature profiles. Boundary-Layer Meteorology, 42, 55-78. Kaimal JC, Finnigan JJ (1994) Atmospheric Boundary Layer Flows: Their Structure and Measurement. Oxford University Press, Inc., New York, 289 pp. Lee X , Black TA, Novak M D (1994). Comparison of flux measurements with open-and closed-path gas analyzers above an agricultural field and a forest floor. Boundary-Layer Meteorology, 67, 195-202. Leuning R, King K M (1992). Comparison of eddy covariance measurements of CO2 fluxes by open- and closed-path CO2 analyzers. Boundary-Layer Meteorology, 59, 297-311. LI-COR, Inc. (1996) Some recommendations for using LI-COR gas analyzer in eddy correlation measurements. LI-COR Application Note No. 118, 7 pp. Liu X , Tsukamoto O, Oikawa T, Ohtaki E (1998) A study of correlation of scalar quantities in the atmospheric surface layer. Boundary-Layer meteorology, 87, 499-508. Chapter 2. Local similarity in an aspen forest 6 2 Mahrt L, Lee X , Black TA, Staebler R, Neumann H H (1998) Vertical mixing in the subcanopy. Extended abstract will be presented at 23 r d Conference on Agricultural and Forest Meteorology, 2-6 November 1998, Albuquerque, New Mexico. McDermitt DK, Welles JM, Eckles, RD (1994) Effects of temperature, pressure and water vapor on gas phase infrared absorption by CO2. LI-COR Application Note No. 116, 5 pp. Monin AS, Obukhov A M (1954) Basic laws of turbulent mixing in the atmosphere near the ground. Tr. Akad. Nauk., SSSR Geophiz. Inst, 24(151), 1963-1987 Nieuwstadt F T M (1984) The turbulent structure of the stable, nocturnal boundary layer. Journal of the Atmospheric Sciences, 41, 2202-2216. Ohtaki E (1985) On the similarity in atmospheric fluctuations of carbon dioxide, water vapor and temperature over vegetated fields. Boundary-Layer Meteorology, 32, 25-37. Ohtaki E (1984) The budget of carbon dioxide variance in the surface layer over vegetated fields. Boundary-Layer Meteorology, 29, 251-261. Paulson C A (1970) The mathematical representation of wind speed and temperature profiles in the unstable atmospheric layer. Journal of Applied Meteorology, 9, 857-861. Panofsky HP, Dutton JA (1984) Atmospheric Turbulence: Models and Methods for Engineering Applications. John Wiley & Sons, Inc. New York, 397 pp. Raupach MR, Finnigan JJ, Brunet Y (1996) Coherent eddies and turbulence in vegetation canopies: the mixing-layer analogy. Boundary-Layer Meteorology, 78, 351-382. Raupach M R (1989) Stand overstorey processes. Philosophical Transactions of the Royal Society, London, B324,175-190. Raupach M R (1979) Anomalies in flux-gradient relationships over forest. Boundary-Layer Meteorology, 16, 467-486. Shao Y, Hacker J M (1990) Local similarity relationships in a horizontally inhomogeneous boundary layer. Boundary-Layer Meteorology, 52, 17-40. Chapter 2. Local similarity in an aspen forest 6 3 Shuttleworth WJ (1989) Micrometeorology of temperate and tropical forest. Philosophic Transactions of the Royal Society of London, B324, 299-334. Shuttleworth WJ, Gash JHC, Lloyd CR, Moore CJ, Roberts J, Marques AdeO, Fisch G, Silva Filho VdeP, Ribeiro M N G , Molion LCB, de Saskatchewan L D A , Nobre JC, Caral OMR, Patel SR, de Moraes JC (1984) Observations of radiation exchange above and below Amazonian forest. Quarterly Journal of the Royal Meteorological Society, 110, 1163-1169. Simpson IJ (1996) Trace gas exchange and the validity of similarity theory in the roughness sublayer above forests. Ph.D. thesis, University of Guelph, Ontario, Canada, 205 pp. Sivaramkrishnan S, Saxena S, Vernekar K G (1992) Characteristics of turbulent fluxes of sensible heat and momentum in the surface boundary layer during the Indian summer monsoon. Boundary-Layer Meteorology, 60, 95-108. Stull RB (1988) An Introduction to Boundary Layer Meteorology. Kluwer Academic Publishers, Dordrecht, The Netherlands, 666 pp. Tanner CB, Thurtell GW (1969) Anemoclinometer measurements of Reynolds stress and heat transport in the atmospheric boundary layer. Research and Development Technical Report ECOM-66-G22F, University of Wisconsin, Madison, Wisconsin. Thorn AS, Stewart JB, Oliver HR, Gash JHC (1975) Comparison of aerodynamic and energy budget estimates of fluxes over a pine forest. Quarterly Journal of the Royal Meteorological Society, 101, 93-105. Viswanadham Y, LDdeA Sa, VdeP Silva Filho, A O Manzi (1987) Ratios of eddy transfer coefficients over Amazon forest. In: Forest hydrology and watershed management, IAHS Press, Institute of Hydrology. Wallingford, U K . Webb EK (1970) Profile relationships: The log-linear range and extension to strong stability. Quarterly Journal of the Royal Meteorological Society, 96, 67-90. Wang J, Mitsuta Y (1991) Turbulence structure and transfer characteristics in the surface layer of the HEIFE Gobi area. Journal of the Meteorological Society of Japan, 69, 587-593. Chapter 2. Local similarity in an aspen forest 6 4 Wang J, Mitsuta Y (1992) An observational study of turbulent structure and transfer characteristics in Heihe Oasis. Journal of the Meteorological Society of Japan, 70, 1147-1154. Webb EK, Pearman GI, Leuning R (1980) Correction of flux measurements for density effects due to heat and water vapor transfer. Quarterly Journal of the Royal Meteorological Society, 106, 85-100. Weir J (1996) The fire frequency and age mosaic of a mixedwood boreal forest. MSc Thesis, University of Calgary, Calgary, Canada. Wilson JD (1989) Turbulent transport within the plant canopy. In: Estimation of Areal Evapotranspiration. (eds. Black TA, Splittlehouse DL, Novak M D , Price DT), IAHS Publication No. 177, pp. 43-80. Xu Y , Zhuo C, L i Z, Zhang W (1997) Turbulent structure and local similarity in the tower layer over the Nanjing area. Boundary-Layer Meteorology, 82, 1-21. 3. ESTIMATING NET C 0 2 EXCHANGE BETWEEN THE ATMOSPHERE AND A BOREAL ASPEN FOREST 3.1 INTRODUCTION The eddy covariance technique has been widely and successfully used in studies of C O 2 exchange between various forests and the atmosphere (e.g. Verma et al. 1986; Wofsy et al. 1993; Hollinger et al. 1994, 1998; Grace etal. 1995; Fan etal. 1995; Goulden etal. 1996b, 1998; Baldocchi and Vogel 1996; Black et al. 1996; Baldocchi et al. 1997; Jarvis et al. 1997). Some researchers have also successfully used this technique to study C O 2 exchange above the forest floor (e.g., Baldocchi and Meyers (1991) in an oak-hickory forest in Tennessee, Lee et al. (1994) in a hemlock-Douglas-fir forest in coastal B.C.; Baldocchi and Vogel (1996) in a boreal jack pine forest; and Black et al. (1996) in a boreal aspen forest). The relatively small contributions of small eddies to turbulent transport above forests make it feasible to use the eddy covariance technique to measure C O 2 fluxes above the forest, even using a closed-path IRGA system with a relatively long sampling tube (Black et al. 1996). Large, intermittent eddy structures that penetrate through the overstory enable eddy covariance measurements to be made near the forest floor (Baldocchi and Meyers 1991; Blanken et al. 1998). Furthermore, cospectral analysis has shown that turbulent transfer mechanisms in the trunk space are similar to above the forest at the OA site in this study (Blanken etal. 1998). There is, however, concern that eddy covariance measurements both above and below the overstory on calm nights may underestimate C O 2 fluxes (see Chapter 4). 65 Chapter 3. Net C02 exchange of an aspen forest 6 6 One of the important uses of CO2 fluxes measured using the eddy covariance technique (eddy CO2 flux, Fc) is to estimate the net ecosystem exchange (NEE) between the atmosphere and the soil or vegetation (e.g., Wofsy et al. 1993; Hollinger et al. 1994, 1998; Fan et al. 1995; Grace etal. 1995; Black etal. 1996; Greco and Baldocchi 1996; Baldocchi and Vogel 1996; Jarvis et al. 1997). The latter, which is often referred to as the biotic flux, can be expressed as NEE = Fc +ASa/At where ASJAt is the change of the CO2 storage in the air column beneath the eddy flux sensors. The estimation of ASJAt requires measurements of C 0 2 concentration change at various heights between the eddy covariance instruments and the ground (referred to as the CO2 concentration profile). Several questions need to be answered in carrying out this procedure. First, how many sampling levels in the profile system are needed to properly estimate ASJAtl Second, how important are the half-hourly changes of CO2 storage in the air column in determining the pattern of NEE during the day? Third, how well does one CO2 concentration profile represent a large forest area, or what are the implications of the eddy flux and concentration profile footprints being different? The answers to these questions are important in obtaining reliable diurnal NEE patterns at forest CO2 exchange sites like those in the AmeriFlux network. Subtraction of the CO2 fluxes measured in the trunk space from those above the forest provides an estimate of the C 0 2 source or sink strength of the overstory. However, footprints in the trunk space are much smaller than above the forest. For example, Blanken (1997), using the footprint model developed by Schuepp et al. (1990), found that at the Old Aspen site the upwind distance where the vegetation makes the maximum contribution to the measured flux at the 39.5-m height was 100 m and 300 m for typical daytime and nighttime conditions, respectively. In contrast, the corresponding values for the 4-m height (in the Chapter 3. Net C0 2 exchange of an aspen forest 6 7 trunk space) were about 10 m and 40 m. Using a Lagrangian random walk model, Baldocchi (1997) found that the flux footprints detected beneath the overstory are even more contracted when horizontal wind velocity fluctuations were considered. This means that the validity of subtracting the two fluxes is questionable when there is marked horizontal variability of the vertical fluxes in the trunk space. One way of assessing this is to determine the horizontal variability of eddy fluxes in the trunk space. Therefore, the specific objectives of this chapter are (1) to describe the CO2 concentration as a function of height diurnally and seasonally, (2) to show the dependence of AS/At on the number of sampling heights in the concentration profile system, (3) to report the relative importance of AoVAr in the process of CO2 exchange between the atmosphere and the forest, and (4) to report the results of an experiment designed to assess the horizontal variability of sensible heat, latent heat and CO2 fluxes in the trunk space. 3.2 EXPERIMENTAL METHODS 3.2.1 Flux measurements In 1994, the sensible heat, latent heat and CO2 fluxes were measured with two eddy covariance systems at the 39.5-m and 4-m heights continuously from February to September and from April to September, respectively at the BOREAS OA site (see the previous chapter for details). The latter system has been operating continuously at the 39.5 m height from April 1996 to the present time. Chapter 3. Net C02 exchange of an aspen forest 6 8 3.2.2 Trunk-space eddy flux comparison experiment The experiment to determine the horizontal variability of eddy fluxes in the trunk space was conducted from August 12-22, 1994 (DOY 224-234). In addition to the eddy covariance system which operated continuously at the 4-m height on the 6-m scaffold tower (see Chapter 2), a system was installed at the 5.9-m height on the main tower. The instruments used at the 5.9-m height are a Kaijo-Denki 3-D sonic anemometer/thermometer and a LI-COR 6262 analyzer system identical to the one on the 6-m scaffold tower. 3.2.3 Comparison of eddy covariance systems Before the 1994 BOREAS field experiment, the eddy covariance systems used by the Atmospheric Environment Service (AES) and the UBC teams were compared above a mixed deciduous forest at Camp Borden, Canada (Lee et al. 1996). Both systems were mounted at the 34.5-m height. The two systems agreed with each other very well. The velocity statistics agreed within 8%, while sensible heat and CO2 fluxes agreed within 5%. Chapter 3. Net C0 2 exchange of an aspen forest 6 9 During the period June 12-17, 1994, a LI-COR 6262 IRGA system similar to that used at the 4-m height (and at Camp Borden) was compared with the one at the 39.5-m height. This IRGA was mounted at the 28.6-m height, adjacent to and operated with a Kaijo-Denki DAT 310 sonic anemometer/thermometer. The purpose of the comparison was to ensure accurate Fig. 3.1 Comparison of the sensible heat fluxes measured at the 39.5- and 28.6-m heights using eddy covariance systems at the OA site from June 12-17, 1994. The dotted line is the one-to-one line. The solid line is the regression through the origin. Chapter 3. Net C0 2 exchange of an aspen forest 70 partitioning of fluxes above and below the overstory. The sensible heat, latent heat and CO2 fluxes at the 28.6-m height were about 7%, 14% and 6% higher than that at the 39.5-m height (Fig. 3.1, Fig. 3.2 and Fig. 3.3). Since the values of ow were virtually identical (see Chapter 2), these discrepancies in the fluxes are probably the result of differences in temperature (obtained from the speed of sound measured by the sonic anemometer), humidity and CO2 concentration measurements. In subsequent analyses in this thesis, latent heat and CO2 flux at the 39.5-m height were increased by the above percentages. This was done because the calibration of the IRGA at the 39.5-m height was carried out assuming a linear relationship between mole fraction and output voltage over 100 p:mol mol"1 intervals rather than using the recommended polynomial relationship (LI-COR 1996). Chapter 3. Net C0 2 exchange of an aspen forest Fig. 3.2 Same as in Fig. 3.1 except for latent heat flux. Chapter 3. Net C0 2 exchange of an aspen forest 72 Fig. 3.3 Same as in Fig. 3.1 except for eddy C0 2 flux. Chapter 3. Net CO2 exchange of an aspen forest 73 3.3 RESULTS AND DISCUSSION 3.3.1 C 0 2 concentration profiles and the change in the storage of C 0 2 in the air column beneath the eddy flux sensors (ASa/Af) 3.3.1.1 C 0 2 concentration profiles and the estimation of ASJAt Fig. 3.4 and Fig. 3.5 show ensemble-averaged diurnal C 0 2 profiles during the leafless and Fig. 3.4 Ensemble-averaged C0 2 profiles for the leafless period (February 4-April 10, 1994) at the OA site. Chapter 3. Net C0 2 exchange of an aspen forest 74 full-leaf periods at the OA site in 1994. Typically nighttime C 0 2 concentrations increased under stable conditions as a result of soil, understory and tree respiration. Minimum concentrations occurred during the daytime and were nearly constant with height as a result of Fig. 3.5 Ensemble-averaged C0 2 profiles for the summer (June 1-August 31, 1994) at the OA site. turbulent mixing. During the leafless period, the accumulation at night mainly occurred below the 5-m level, and concentrations near the ground reached about 380 |imol mol"1. In contrast, during the full-leaf period C O 2 accumulations at night extended to the 10-m level and concentrations near the ground usually exceeded 500 (imol mol"1. Chapter 3. Net C0 2 exchange of an aspen forest 75 228 229 230 231 232 DOY 1994 Fig. 3.6 Diurnal course of C0 2 concentration measurements above and below the aspen overstory during spring and summer, 1994. The thin line is the concentration above the aspen overstory at the 39.5-m height and the heavy line is that within the canopy at the 4-m height, a) measurements during the spring (DOY 105-109, April 15-19) , b) measurements during the summer (DOY 161-165, June 10-14), and c) measurements late in the growing season (DOY 228-232, August 16-20). Chapter 3. Net C0 2 exchange of an aspen forest 76 During the leafless period, differences of CO2 concentration between the 39.5- and 4-m heights were usually less than 5 umol mol"1, even during the nighttime (Fig. 3.6a). In the mid-growing season, concentration differences between the two levels could be as much as 100 umol mol"1 (Fig. 3.6c). Fig. 3.7 shows ASJAt in the 0-39.5 m column approximated using CO2 concentration measurements made at (1) the 34.5-m height, (2) the 34.5- and 2.3-m heights, (3) the 34.5-, 3r -20 -10 0 10 20 -20 -10 0 10 20 Is A S /At (1 ht) (pmol m"2 s"1) A S /Af (2 hts) (umol m"2 s"1) 0 0 -20 -10 0 10 20 -20 -10 0 10 20 A S /At (3 hts) (^ imol rrf2 s"1) A S /Af (4 hts) (pmol rrf2 s"1) Fig. 3.7 Comparison of the values of ASJAt calculated using C0 2 concentrations at all eight heights with those calculated from concentrations measured at selected heights at the OA site in the summer 1994. The selected heights were the 34.2 m (a), 2.3 and 34.2 m (b), 2.3, 9.5 and 34.2 m (c) and 2.3, 9.5, 21.9 and 34.2 m (d) heights. The dash-dotted lines are the linear best fits. The solid lines are one-to-one lines. Chapter 3. Net C 0 2 exchange of an aspen forest 77 9.5- and 2.3-m heights, and (4) the 34.5-, 21.9-, 9.5- and 2.3-m heights. ASa/At was calculated using ASjAt = ( l / M o ) X Q a A / ACjAt (3.1) 1=1 where Ma is the molecular weight of wet air (=29 g mol"1), pai is the density of the air in layer /, AZJ is the depth of layer i, and C, is the CO2 concentration (ptmol mol"1 wet air) in layer /. The weighting factors (the depth of the layer assumed to be represented by the sampling heights) are given in Table 3.1. The results of the regression analysis are also given in Table 3.1. If only the sampling height corresponding to where the eddy covariance system located (above the canopy) was used to calculate ASaIAt, then only about 50% of the variance of the storage was explained. In contrast, the use of the four heights selected explained about 97% of the variance with the slope of the regression (a) being 1.04. Surprisingly, use of only two heights as indicated explained about 86% with a being 1.11. Table 3.1 Coefficients of the linear best fits (ASJM&_ht = a(ASJAt)\n_ht + b ) shown in Fig. 3.7. Number of Heights (n) a b r1 Heights (m) of Measurement Weighting Factors Az (m) 1 1.0075 IO"4 0.504 34.2 39.5 2 1.1087 0.0029 0.863 2.3, 34.2 5, 34.5 3 1.0504 0.0011 0.954 2.3, 9.5, 34.2 5, 10, 24.5 4 1.0377 10"5 0.968 2.3, 9.5, 21.9, 34.2 5, 10, 10, 14.5 Chapter 3. Net C 0 2 exchange of an aspen forest 7 8 Fig. 3.8 shows the ensemble-averaged profiles of the change in C 0 2 concentration per half-hour during the full-leaf period. It shows that the C 0 2 concentration near the ground generally increased steadily during the early evening at rates of 10-20 pimol mol"1 half-hour"1. Between midnight and 5:00 CST, concentrations tended to remain nearly constant. Following the break-up of the stable boundary layer (SBL), C 0 2 concentration usually decreased sharply at rates of about 10-25 p.mol mol"1 half-hour"1 near the ground. This occurred as the convective boundary layer (CBL) quickly developed in the morning. Fig. 3.8 The C 0 2 concentration change from the previous half-hour (calculated as ACjAt = (Cj+] - C M ) / 2 (M-mol mol"1 half-hour"1), where C is the concentration, i is the half-hour concerned) at different heights calculated from the data in Fig. 3.5. Chapter 3. Net C 0 2 exchange of an aspen forest 7 9 3.3.1.2 Diurnal changes in ASJAt and effects of turbulent mixing Fig. 3.9 and Fig . 3.10 show the diurnal patterns of ASa/At beneath the 39.5-m level and the 4-m level during the leafless and full-leaf periods at the O A site, 1994. Fig . 3.10 presents the same results as in Fig . 3.8 except integrated with respect to height. During the leafless period, AS a /Ar for the 0-39.5 m air column generally reached a maximum of about 0.8 umol m" 2 s"1 at about 20:00 C S T and reached a minimum of about -1.1 j imol m" 2 s"1 at 11:00 CST. 7 2 05 •* o o E a. §-2 CO < -4 1 7 05 <V 0.5 o E co-0.5 i j l i j i i j i i j i i j i i i i a) above canopy a • • * • . I . nwnr iTTjr f • i i Mi • i i i i i 1 j 1 j 1 i : j * * * i i i b) within canopy •. j I i : ; j ; : . 4||f+jH-r+^ ! 1 • M !TTr ; ' • • i i i 12 CST (hours) 16 20 24 Fig. 3.9 The rates of change of C 0 2 storage in the air column beneath the eddy covariance systems a) at the 39.5-m height and b) at the 4-m height during the leafless period (DOY 35 - 110). The dots are the half-hourly measurements and the line is the ensemble average. 2 1 For the 0-4 m air column, the corresponding values were 0.1 and -0.1 umol rrf s" with the Chapter 3. Net C02 exchange of an aspen forest 80 maximum also occurring at 20:00 CST and the minimum occurring between 9:00 and 10:00 CST. The corresponding maximum and minimum values for the full-leaf period for the 0-39.5 m air column were about 4 and -5 p:mol m"2 s"1, while for the 0-4 m air column they were 1 and -1 pimol m"2 s"1 (Fig. 3.10). In this case, the occurrence of the minimum values of ASJAt (between 8:00 and 9:00 CST) tended to be earlier than the leafless period, corresponding to the earlier sunrise. As indicated in the previous section, CO2 accumulation reached a maximum between 19:00 and 22:00 CST (Fig. 3.4, Fig. 3.5 and Fig. 3.8). During the rest of the night, ASJAt was generally lower. The reason for this was that before midnight following the sunset, conditions in the canopy were relatively stable, i.e., little mixing, whereas after midnight, probably as a result of the cooling of the leaves and branches of the trees, there was mixing as a result of cold air sinking through the canopy. This effect was confirmed by examining the pattern of half-hourly values of the standard deviation of the vertical velocity (ow). Fig. 3.11 shows the ensemble-averaged diurnal pattern of aw at the 18.6-m height (in the aspen canopy layer) during the full-leaf period. It shows that ow decreased to around 0.20 m s"1 shortly after the break-down of the C B L (near sunset), and slowly increased to about 0.25 m s"1 at about midnight. It remained relatively constant or increased slightly during the rest of the night. This increasing mixing was generally sufficient to prevent the continued accumulation of CO2 in the understory and the lower trunk space. The effect of the mixing during the latter part of the night can also be seen in the CO2 concentration and temperature profiles shown in Fig. 4.12 (Chapter 4). Chapter 3. Net CO2 exchange of an aspen forest 81 Fig. 3.10 The rates of change of C0 2 storage in the air column beneath the eddy covariance systems a) at the 39.5-m height, and b) at the 4-m height in the summer months (DOY 152-243). The dots are the half-hourly measurements and the line is the ensemble average. Chapter 3. Net C0 2 exchange of an aspen forest 82 CST (Hours) Fig. 3.11 Ensemble-averaged standard deviation of the vertical velocity at the 18.6-m height during the full-leaf period at the OA site, in 1994. Chapter 3. Net C02 exchange of an aspen forest 8 3 3.3.2 Estimating net C 0 2 exchange between the atmosphere and the forest 3.3.2.1 Diurnal patterns of Fc, ASJAt and NEE for the forest Fig. 3.12 shows Fc, ASJAt and NEE (= Fc + ASJAt ) for an eight-day period in the middle of the 1994 growing season. In spite of considerable scatter, ASa/At was usually negative in the early morning and near zero during the rest of the daytime. On some nights, such as DOY 228 (August 16) when mixing was strong (u^ = 0.41 m s"1), the eddy fluxes measured above the forest were high, and ASJAt accounted for little of the net exchange of C 0 2 with the atmosphere while on other nights it made a moderate contribution. With storage (ASa/At) taken into account (Fig. 3.12c), about half of the nighttime values of z 1 Fc +ASa/At were in the range of 4.5-6 pimol m" s" . On some very calm nights, however, both ASJAt and Fc were small which will be discussed in the next chapter. Chapter 3. Net C0 2 exchange of an aspen forest 84 Fig. 3.12 Diurnal courses of a) net ecosystem exchange (NEE) which is the sum of measured flux (Fc) at 39-m height and storage change, b) storage change and c) F c during DOY 224-233. The storage change was estimated with profile data. Chapter 3. Net C0 2 exchange of an aspen forest 85 Fig. 3.13 shows the ensemble-averaged diurnal pattern of ASJAt for the 0-39.5 m air column, the eddy CO2 flux (Fc) at the 39.5-m height and NEE for the summer months in 1994 at the OA site. During this period, sunrise occurred between 5:00 and 6:00 CST and sunset occurred between 20:30 and 22:00 CST. During the early morning hours (08:00-09:00 CST), ASJAt accounted for about 50% of the NEE. Photosynthesis resulted in the reduction of CO2 concentration in the 0-39.5 m air column as well as a strong downward C 0 2 flux at T 1 1 1 r Hours (CST) Fig. 3.13 Ensemble-averaged C0 2 flux measured using the eddy covariance system at the 39.5-m height Fc, and the change rate in the C0 2 storage (ASa/At) in the air column beneath the eddy covariance system and NEE during the summer months (June-August) at the OA site in 1994. The thin line is ASa/At, the thick line is Fc, and the dash-dotted line is the sum of the two. Chapter 3. Net C0 2 exchange of an aspen forest 86 the 39.5-m height. As indicated above, between mid-morning and mid-afternoon, ASJAt made negligible contributions to NEE. During the early evening with a sharp drop of photosynthesis, there was a marked accumulation of C O 2 as respiration continued while temperature was high. After midnight, the upward eddy fluxes of C 0 2 increased slightly while the accumulation of C 0 2 tended to be small as the mixing generally increased. As a result, N E E tended to be (on average) fairly constant during the night. On some calm nights, C 0 2 accumulation continued to occur after midnight while eddy fluxes above the forest were relatively small. An example of this (July 16-17, 1996) is shown Fig. 3.14 C0 2 concentration profiles on July 16-17, 1996 at the OA site. Chapter 3. Net C02 exchange of an aspen forest 8 7 in Fig. 3.14 and Fig. 3.15. In this case, CO2 concentration continued to increase somewhat sporadically until sunrise (6:30 CST). Around 9:00 CST, the SBL broke up resulting in a strong mixing of C 0 2 throughout the 39.5-m air column (clearly seen in Fig. 3.14). At this time, there was a strong upward flux (approximately 26 umol m~2 s"1) at the 39.5-m height (Fig. 3.15). About a dozen such events were observed during the growing season in both 1994 and 1996. This phenomenon of loss of C 0 2 to the atmosphere in the early morning was commonly observed in a tropical rain forest by Grace et al. (1995). In their case, it was almost the only time during the day that CO2 was lost to the atmosphere because during the nighttime stability above the dense canopy appeared to trap respired CO2. Hollinger et al. (1998) also reported that the CO2 storage in the air column accounted for the low nighttime eddy fluxes over a Siberian Larch forest. In the case of the old aspen, this phenomenon generally occurred following very calm nights. However, even on such nights, there was generally some upward eddy flux measured at the 39.5-m height as shown in Fig. 3.15. In contrast to Grace et al. (1995) and Hollinger et al. (1998), on these calm nights Fc+AS</At (or just AS a /Ar in their case) did not appear to entirely account for the likely rate of respired CO2. Other possible explanations for this 'missing' CO2 will be discussed in the next chapter. Chapter 3. Net C0 2 exchange of an aspen forest 88 30 _301 i i 1 1 1 1 18 22 2 6 10 14 18 CST (Hours) Fig. 3.15 The eddy C0 2 fluxes (solid line) measured above the forest and ASJAt in the 0-39.5-m air column (dash-dotted line) at the OA site on July 16-17,1996. 3.3.2.2 Seasonal changes in C O 2 exchanges above and below the aspen overstory Fig. 3.16 shows the seasonal changes in half-hourly eddy C O 2 fluxes measured above and below the aspen overstory at the OA site in 1993 and 1994. The positive values are mainly half-hour fluxes at night, while negative values are mainly half-hour fluxes during the daytime. C O 2 flux above the forest showed a very clear seasonal change. The forest was a large source of C 0 2 until mid-May. C O 2 fluxes in winter were as high as 0.5 |imol m"2 s"1, 2 1 and then increased to about 2 umol m" s" before leaf emergence as a result of increasing soil temperature. The upward fluxes of C 0 2 measured at the 4-m height were generally very Chapter 3. Net C0 2 exchange of an aspen forest 8 9 similar to the above-forest values (except for some outliers). This indicates that a large proportion of the respired CO2 at this site is from the soil. The forest was a strong sink for CO2 during the growing season. The magnitude of downward CO2 fluxes was as large as 35 Umol m"2 s"1 occurring mainly from late June to mid July. The forest became a CO2 source again in the fall. The source strength in the fall was greater (24-h mean flux was higher) than in the spring and winter due to higher soil temperature in the fall (Chen et al. 1998). The CO2 flux in the trunk space did not show a very strong seasonal trend. It was usually positive except during some of the daytime half-hours in the summer months when hazelnut photosynthesis exceeded soil respiration. Chapter 3. Net C0 2 exchange of an aspen forest 90 Fig. 3.16 All half-hourly eddy C0 2 fluxes measured at the 39.5-m and 4-m heights at the OA site in 1993 and 1994. During the growing season, most of the positive values shown are C0 2 fluxes at night as a result of forest respiration, while the negative values which mainly occurred during the daytime are the result of photosynthesis exceeding daytime respiration. Fig. 3.17 shows the diurnal patterns of Fc + ASaIAt, i.e. NEE above and below the aspen overstory during different seasons. The CO2 fluxes both above and below the overstory were small in early spring. The small negative CO2 fluxes that occurred during the daytime were likely the result of photosynthesis of shrubs and grasses which leafed-out shortly after snow-melt, which occurred around April 10 (DOY 100). The five-day average of CO2 fluxes in Fig. 3.17a were about 1.1 and 0.8 pimol m*2 s"1 above and below the aspen overstory, respectively. This shows (as suggested earlier) that the major part (about 76%) of the upward Chapter 3. Net C0 2 exchange of an aspen forest 91 CO2 flux at this time of year was the result of soil respiration. During the growing season, the diurnal variations in the CO2 flux above the forest were much more pronounced (Fig. 3.17b & c) than in early spring. Even the CO2 fluxes in the trunk space were negative during some daytime hours (Fig. 3.17b). In mid-June, daytime mean CO2 fluxes at the 39.5-m and 2 1 4-m heights were -7.4 and 0.7 p.mol m" s" , respectively, while the corresponding nighttime values were 4.5 and 1.8 jxmol m"2 s"1 (Fig. 3.17b). Daytime fluxes at the 39.5-m height remained much the same through August. Nighttime flux, however, increased to 5.3 umol m"2 s"1 due to increasing soil temperature (Fig. 3.17c). At the 4-m height, daytime and nighttime fluxes also increased (to 2.7 and 3.0 umol m"2 s"1). This increasing soil respiration over this period resulted in the occurrence of almost no negative daytime CO2 fluxes at the 4-m height in August although there was still hazelnut photosynthesis. The daytime values above the forest agree with those for temperate forests (Verma et al. 1986; Baldocchi et al. 1987; Hollinger et al. 1994; Goulden et al. 1996b) but were almost twice as high (more negative) as those for other boreal forests (Baldocchi and Vogel 1996; Baldocchi etal. 1997; Goulden et al. 1997; Jarvis etal. 1997). They were also much higher than those measured in tropical forests (Fan etal. 1990; Wofsy etal. 1993; Grace et al. 1995) likely because tropical forests are in a near-equilibrium condition, i.e., ecosystem respiration approximately balances photosynthesis. Nighttime values were similar to those for the above mentioned forests possibly because of the similarity of growing season soil temperatures. Fig. 3.17 reinforces the point made earlier that a large proportion of the CO2 lost from the forest at night is a result of the soil respiration. This is particularly evident on the windy nights of D O Y 164-165 and 229-230. Chapter 3. Net C0 2 exchange of an aspen forest 9 2 10 5h DOY 1994 Fig. 3.17 Diurnal courses of the eddy C0 2 fluxes corrected with ASa/At at the 39.5-m height (the thick line) and at the 4-m height (the thin line) (a) in the spring (DOY 105 -109) , (b) in the summer (DOY 161 -165) and (c) late of the growing season (DOY 228 - 232). The missing data were replaced by using linear interpolation. Chapter 3. Net C0 2 exchange of an aspen forest 9 3 3.3.3 Horizontal variability of C 0 2 concentration and fluxes within the trunk space 3.3.3.1 Horizontal variability of the C 0 2 concentration within the trunk space In making the comparison of C 0 2 concentrations between the two towers, it was recognized that the heights of the sample-tube inlets were 5.9 and 4 m at the main and 6-m Fig. 3.18 Diurnal changes of C0 2 concentration at the 4-m (on the 6-m scaffold tower) and 5.9-m heights (on the main tower) in the trunk space on two towers 40-m apart during August 12-22, 1994 (DOY 224-234). The dotted line was the measurements made on the main tower and the solid line was the measurements made on the 6-m scaffold tower. Chapter 3. Net C02 exchange of an aspen forest 9 4 scaffold tower, respectively. This height difference was because the two sonic anemometers used in comparing the fluxes at the two locations had been mounted at these levels for some time, and it was undesirable to change their heights. The ground level at the main tower was 0.3-0.5 m lower than at the 6-m scaffold tower, so that the elevation difference between the two inlets was 1.4-1.6 m. This should probably result in the concentration at the main tower being systematically lower than at the scaffold tower. In spite of this, the variation in the differences in concentration between the two towers should provide an indication of horizontal variability of CO2 concentration in the trunk space. Fig. 3.18 and Fig. 3.19, which compare the CO2 concentration in umol mol"1 moist air at the 4-m height on the 6-m scaffold tower and at the 5.9-m height on the main tower over a ten-day period (August 12-22, 1994), confirm that concentrations at 5.9-m height at the main tower were slightly lower than at the 4-m height on the scaffold tower. The mean difference between the concentrations at the two towers (AC4.5.9m = C 4 m - Cs.9m) during the daytime (solar radiation S > 50 W m"2 above the forest) was 0.9 umol mol"1, while at night it was 10.9 umol mol"'. There was, however, considerable half-hour to half-hour variability in the concentration difference (the standard error of estimate syx = 9.4 umol mol"1 during the daytime and syx = 26.2 umol mol"1 at night). These differences suggest that a single profile can provide only an approximate estimate of ASa/At on calm nights. For aw > 0.07 m s"1 at the 4-m height (w^ ~ 0.15 m s"1 at the 39.5-m height), the mean difference in the concentration between the two towers at night decreased to 6.0 u.mol mol"1. However, only 17% of the nighttime data met this criterion. These results suggest that ASa/At can be reasonably well estimated with a single profile system when atmospheric mixing is adequate. Of course under such conditions, ASa/At is relatively small. Chapter 3. Net C0 2 exchange of an aspen forest 9 5 Fig. 3.19 Comparison of the half-hourly C0 2 concentrations measured on two towers using the same data as in Fig. 3.18. The triangles are nighttime data and the circles are daytime data. The dash-dotted line is the daytime regression line (C4m = 0.878C5.9m + 43 umol mol"1, r2 = 0.87, syx = 9.4 umol mol'1, C 4 m - C5 9 m = 0.9 umol mol"1), and the dotted line is the nighttime regression line (C4 m = 0.670C5.9m +120 tlmol mol"1, r2 = 0.48, syx = 26.2 |imol mol"1, C 4 m - C 5 9 m = 10.9 umol mol"1). Chapter 3. Net C0 2 exchange of an aspen forest 9 6 3.3.3.2 Horizontal variability of the fluxes within the trunk space Fig. 3.20 shows the air-storage corrected C O 2 fluxes (= Fc + AS a /Ar, Black et al. 1996 referred to this flux as net hazelnut exchange (NHE) analogous to NEE for the above-forest 229 230 231 DOY 1994 232 233 Fig. 3.20 Comparison of half-hourly values of the air-storage corrected C0 2 flux measured with the eddy covariance systems at the 4- (thin line) and 5.9-m (thick line) heights on the scaffold and main towers (40-m apart) during August 12-22, 1994 (DOY 224-234). The rates of change of C0 2 storage were estimated using profile data. Chapter 3. Net C0 2 exchange of an aspen forest 9 7 flux) on these two towers. They agreed fairly well during the daytime just as in the case of CO2 concentration. These fluxes, however, often differed greatly during the nighttime, to the degree that at times they were of opposite sign. This suggests that the CO2 flux measured at one point can not represent a large area of forest floor or understory during a calm night because the canopy or the soil is horizontally patchy, as concluded by Fan et al. (1995), or because the turbulence in the canopy is patchy during the night. An important observation is that negative CO2 fluxes were observed at both locations simultaneously on calm nights (e.g., D O Y 232-233). This indicates that negative eddy CO2 flux during the nighttime in the trunk space was not an artifact. This observation and the fact that significant difference in the C 0 2 concentrations at the two towers could persist for six or more hours suggest that advective transfer in the canopy is very likely. As discussed in the next chapter, this advective transfer term tends to average to zero over periods of 1-2 days. Thus, long-term average values of the CO2 fluxes measured at different locations within the canopy should be similar. In this case, the averaged values of the CO2 fluxes, over a period of 10 days, were almost identical: 2.3 and 2.2 p:mol m"2 s"1 (Table 3.2) on the 6-m scaffold and main towers, respectively. The daytime and nighttime means were also very similar. Fig. 3.21 compares the sensible heat flux (H) in the trunk space at the two towers. The results differ from the CO2 case. The diurnal patterns are more clear. The sensible heat fluxes measured at these two positions agreed during most daytime hours, especially on D O Y 225, 229 and 230 (August 13, 17-18) when they were almost identical. At night, H showed horizontal variability similar to that of the CO2 flux. In contrast to the CO2 flux, the magnitude of H during the nighttime was small so that these differences did not affect Chapter 3. Net C0 2 exchange of an aspen forest 9 8 60 229 230 231 232 233 D O Y 1994 Fig. 3.21 Comparison of half-hourly sensible heat fluxes measured with eddy covariance system at 4 ~ 6 m height in the canopy on two towers 40-m apart during August 12-22, 1994 (DOY 224-234). The thick line is the flux measured at the 6-m height on the main tower and the thin line is that at the 4-m height on the 6-m scaffold tower. average fluxes very much. Over the 10-day period, the 24-h averages were 1.6 and 0.9 W m (Table 3.2) on the 6-m scaffold and main towers, respectively. Chapter 3. Net C0 2 exchange of an aspen forest 9 9 Fig. 3.22 compares the latent heat flux (XE) measurements at the two towers. Both eddy covariance systems observed the same pattern day and night although half-hourly values did not always agree. On some days, such as DOY 228-232 (August 16-20), latent heat fluxes were almost identical. This suggests that half-hourly XE measured at one place in the trunk T r 229 230 231 232 233 DOY 1994 Fig. 3.22 Same as Fig. 3.21 except latent heat fluxes. Chapter 3. Net C0 2 exchange of an aspen forest 100 space can represent a relatively large area. Averaging over the entire period, XE was 19 and 21 W m"2 (Table 3.2) on the 6-m scaffold and main towers, respectively. This comparison suggests that sensible and latent heat fluxes measured at one point in the trunk space can represent the short-term (half-hour to one-hour) value over a relatively large area during the daytime and nighttime. CO2 flux measurement at night has a considerable short-term horizontal variability. It has, however, a good representativeness for a large area during the daytime (short-term) and for the long-term average. To our knowledge, this type of comparison has not been done previously. In a similar type of experiment above a uniform pine stand (Duke forest, USA), Katul et al. (1997) found similarly that Fc had the worst horizontal homogeneity among the scalar fluxes. They found that the H was most homogeneous of the three (Fc, H and XE) particularly during the daytime, whereas we found that XE is the most homogeneous in the trunk space. Table 3.2 Comparison of the means of scalar fluxes obtained on the 6-m scaffold tower (A) and the main tower (B) in the trunk-space eddy flux experiment, August 12-22,1994. Daytime Nighttime 24-hour Tower A Tower B Tower A Tower B Tower A Tower B Fc+ASa/At (umol m"2 s"1) 2.0 1.9 2.8 2.7 2.3 2.2 H(W mz) 3.1 2.6 -0.9 -1.9 1.6 0.9 XE (W m"2) 30.0 34.0 0.5 0.5 18.8 21.4 Chapter 3. Net C0 2 exchange of an aspen forest 101 3.4 CONCLUSIONS This study shows that changes in the CO2 storage in the air column beneath the above-forest eddy covariance system (ASa/At) occurred mainly in the 0-10 m layer during the nighttime. Therefore it could not be estimated very well using only the CO2 concentration measured at one level above the forest. It could, however, be reasonably well estimated with the CO2 concentration measured at a level above the forest and at one additional height within the canopy (2.3-m height in this case). Values of ASa/At were about 1-2 times the measured eddy flux Fc during the early evening while they were at the same order of magnitude as Fc during the rest of the nighttime because of increased mixing in the canopy. It was negative (1-2 times Fc) during the early morning and was negligible late in the morning and in the afternoon. The half-hourly air-storage corrected C 0 2 fluxes at the 39.5-m and the 4-m heights (Fc + ASa/At) were small during the spring. The nighttime value of upward CO2 fluxes above the forest, of which major part was from the soil just as in the case of other seasons, were in the range of 4.5-6 jimol m"2 s"1 during the growing season. The daytime values of the CO2 fluxes 2 1 above and below the aspen overstory reached -35.0 and -4.6 fjmol m" s" at the 39.5- and 4-m heights, respectively. Negative Fc +ASa/At at the 4-m height was only observed in June, July and early August when the hazelnut was growing and CO2 flux from soil had not peaked. Within the trunk space, eddy covariance sensible and latent heat flux measurements at one position were representative of an area extending for at least two tree heights. The same was the case for CO2 flux and concentration during the daytime. At night, however, they Chapter 3. Net CO2 exchange of an aspen forest 102 exhibited significant horizontal variability but they were representative of an area extending for at least two tree heights when averaged over several days. This suggests either a patchiness of CO2 source strength or complex large scale horizontal motion resulting in short-term CO2 advection. Chapter 3. Net C0 2 exchange of an aspen forest 103 3.5 R E F E R E N C E S Baldocchi DD (1997) Flux footprints within and over forest canopy. Boundary-Layer Meteorology, 85, 273-292. Baldocchi DD, Vogel C, Hall, B (1997) Seasonal variation of carbon dioxide exchange rates above and below a boreal jack pine forest. Agricultural and Forest Meteorology, 83, 147-170. Baldocchi DD, Vogel C (1996) A comparative study of water vapor, energy and C 0 2 flux densities above and below a temperate broadleaf and a boreal pine forest. Tree Physiology, 16, 5-16. Baldocchi DD, Meyers TP (1991) Trace gas exchange above the forest floor of a deciduous forest. 1. Evaporation and C 0 2 flux. Journal of Geophysical Research, 96, 7271-7285. Black TA, den Hartog G, Neumann HH, Blanken PD, Yang PC, Russell C, Nesic Z, Lee X, Chen SG, Staebler R, Novak MD (1996) Annual cycle of water vapour and carbon dioxide above a boreal aspen forest. Global Change Biology, 2, 219-229. Blanken PD, Black TA, Neumann, HH, den Hartog G, Yang PC, Nesic Z, Staebler R, Chen W, Novak MD (1998). Turbulent flux measurements above and below the overstory of a boreal aspen forest. Boundary-Layer Meteorology, (in press). Blanken PD (1997) Evaporation within and above a boreal aspen forest. Ph.D. Thesis, University of British Columbia, British Columbia, Canada, 179 pp. Blanken PD, Black TA, Yang PC, den Hartog G, Neumann, HH, Nesic Z, Staebler R, Novak MD, Lee X (1997). Energy balance and canopy conductance of a boreal aspen forest: partitioning overstory and understory components. Journal of Geophysical Research, 24, 28915-28927. Chen WJ, Black TA, Yang PC, Barr AG, Neumann HH, Nesic Z, Novak MD, Eley J, Ketler RJ, Cuenca R (1998) Effects of climatic variability on the annual carbon sequestration by a boreal aspen forest. Global Change Biology (in press). Chapter 3. Net C0 2 exchange of an aspen forest 104 Fan S-M, Goulden ML, Munger JW, Daube BC, Bakwin PS, Wofsy ST, Amthor JS, Fitzjarrald DR, Moore KE, Moore TR (1995) Environmental controls on the photosynthesis and respiration of a boreal lichen woodland: a growing season of whole-ecosystem exchange measurements by eddy correlation. Oecologia, 102, 443-452. Fan S-M, Wofsy ST, Bakwin PS, Jacob DJ (1990) Atmosphere-biosphere exchange of C 0 2 and O 3 in the Central Amazon forest. Journal of Geophysical Research, 95, 16851-16864. Goulden ML, Wofsy SC, Harden JW, Trumbore SE, Crill PM, Gower ST, Fries T, Daube BC, Fan S-M, Sutton DJ, Bazzaz A, Munger JW (1998) Sensitivity of boreal forest carbon balance to soil thaw. Science, 279, 214-217. Goulden ML, Daube BC, Fan S-M, Sutton DJ, Bazzaz A, Munger JW, Wofsy SC (1997) Physiological responses of a black spruce forest to weather. Journal of Geophysical Research, 102, 28987-28996. Goulden ML, Munger JW, Fan S-M, Daube BC, Wofsy SC (1996a) Measurements of carbon sequestration by long-term eddy covariance: methods and a critical evaluation of accuracy. Global Change Biology, 2, 169-182. Goulden ML, Munger JW, Fan S-M, Daube BC, Wofsy SC (1996b). Carbon dioxide exchange by a temperate deciduous forest: response to interannual changes in climate. Science, 271, 1576-1578. Grace J, Lloyd J, Mclntyre J, Miranda AC, Meir P, Miranda HS, Nobre C, Moncrieff J, Massheder J, Malhi Y, Wright I, Gash J (1995) Carbon dioxide uptake by an undisturbed tropical rain forest in Southwest Amazonia, 1992 to 1993. Science, 270, 778-780. Greco S, Baldocchi DD (1996) Seasonal variations of CO2 and water vapor exchange rates over a temperate deciduous forest. Global Change Biology, 2, 183-197. Hollinger DY, Kelliher FN, Byers JN, Hunt JE, McSeveny TM, Weir PL (1994). Carbon dioxide exchange between an undisturbed old-growth temperate forest and the atmosphere. Ecology, 75, 134-150. Hollinger DY, Kelliher FN, Schulze E-D, Bauer G, Arneth A, Byers JN, Hunt JE, McSeveny TM, Kobak KI, Milukova I, Sogatchev A, Tatarinov F, Varlarigin A, Ziegler W, Chapter 3. Net C 0 2 exchange of an aspen forest 105 Vygodskaya NN (1998) Forest-atmosphere carbon dioxide exchange in eastern Siberia. Agricultural and Forest Meteorology, 90 , 291-306. Jarvis PG, Massheder JM, Hale SE, Moncrieff JB, Rayment M, Scott SL (1997) Seasonal variation of carbon dioxide, water vapour and energy exchanges of a boreal black spruce forest. Journal of Geophysical Research, 102, 28953-28966. Katul G, Albertson J, Anderson D, Bowling D, Clark K, Evans B., Hollinger D, Hsieh C-I, Lee J, Oren R, Orff B, Shurpali N, Turnipseed A, Tu K (1997) Spatial variability of turbulent fluxes in the canopy sublayer of a uniform pine stand. Presented at the 2nd Annual AmeriFlux Meeting, St. Louis, MO. October 28-29, 1997. Lee X, Black TA, den Hartog G, Neumann HH, Nesic Z, Olejnik J (1996). Carbon dioxide exchange and nocturnal processes over a mixed deciduous forest Agricultural and Forest Meteorology, 81 , 13-29. Lee X, Black TA, Novak MD (1994). Comparison of flux measurements with open-and closed-path gas analyzers above an agricultural field and a forest floor. Boundary-Layer Meteorology, 67 , 195-202. LI-COR, Inc. (1996) Some recommendations for using LI-COR gas analyzer in eddy correlation measurements. LI-COR Application Note No. 118, 7 pp. Schuepp PH, Leclerc MY, MacPerson JI, Desjardins RL (1990) Footprint prediction of scalar fluxes from analytical solutions of the diffusion equation. Boundary-Layer Meteorology, 50, 355-373. Verma SB, Baldocchi DD, Anderson DE, Matt DR, Clement RE (1986) Eddy fluxes of C0 2 , water vapor and sensible heat over a deciduous forest. Boundary-Layer Meteorology, 36, 71-97. Wofsy SC, Goulden ML, Munger JW, Fan S-M, Bakwin PS, Daube BC, Bassow SL, Bazzaz FA (1993) Net exchange of C 0 2 in a mid-latitude forest. Science, 260, 1314-1317. Chapter 3. Net C0 2 exchange of an aspen forest 106 4. ESTIMATING NIGHTTIME RESPIRATION USING EDDY COVARIANCE MEASUREMENTS OF C0 2 FLUX ABOVE A BOREAL ASPEN FOREST 4.1 I N T R O D U C T I O N Approximately 12-14.7xl06 km2 (8-11% ) of the Earth's land surface is covered by boreal forest (Whittaker and Likens 1975; Bonan and Shugart 1989). Its plant mass is about 13% of the total carbon of terrestrial ecosystems (Whittaker and Likens 1975; Schlesinger 1997), and its soil stores about 43% of the world soil organic carbon (Schlesinger 1997). The boreal forest is likely a major sink for fossil fuel carbon (e.g., Tans et al. 1990; Enting and Mansbridge 1991; Ciais etal. 1995; Denning etal. 1995); therefore long-term monitoring of CO2 exchange between this forest and the atmosphere is essential to help resolve global change issues. Recently, various papers have been published on carbon exchange between the boreal forest and the atmosphere using the eddy covariance technique (e.g., Baldocchi and Vogel 1996; Blacked/. 1996; Baldocchi et al. 1997; Goulden et al. 1997; Jarvis etal. 1997). All of these authors, along with others such as Valentini et al. (1996), working in a beech forest in Italy, have raised concerns about the variable and often low CO2 fluxes that occur on calm nights. 107 Chapter 4. Estimating nighttime respiration in an aspen forest 108 Fig. 4.1 shows a typical diurnal course of the CO2 eddy covariance flux measured at a height of 39.5 m (about twice tree height) at the BOREAS (Boreal Ecosystem-Atmosphere Study) Old Aspen forest site (OA) on August 13 - 14, 1994. The flux has been corrected by the rate of change of CO2 storage in the air column beneath the eddy covariance system. During the night, the mean friction velocity was about 0.2 m s"1 (about 17% of the wind speed), mean soil temperature was about 12 °C and the mean C 0 2 flux corrected for the rate Fig. 4.1 Typical diurnal course of air-storage corrected C0 2 eddy flux measured at the 39.5-m height using the eddy covariance method (solid line) and the air C0 2 storage beneath the instrument height (dotted line) at the OA site on August 13 - 14, 1994. The nighttime data points (18:00 - 6:00 CST) are two-hour running means. Chapter 4. Estimating nighttime respiration in an aspen forest 109 of change in air CO2 storage was about 1.7 umol m"2 s"1. It shows that the CO2 fluxes were zero or even negative around midnight when dropped to 0.03 - 0.1 m s"1. These flux values were much lower than expected respiration rates. The rates of change in air CO2 storage were also near zero (Fig. 4.1). What causes such 'unreasonably low' measurements? Are they really unreasonable and do they need to be corrected? Two main approaches dealing with this nighttime CO2 flux measurement concern have been reported. The first approach, which is commonly used, assumes that all the respired C 0 2 in the ecosystem, regardless of the wind regime, is released to the air because soil storage of CO2 is assumed to be zero; therefore all C 0 2 flux measurements on calm nights are underestimates and must be corrected. It also assumes that on windy nights the eddy covariance CO2 flux corrected for the rate of change in air C 0 2 storage adequately estimates the respiration rate of the ecosystem; thus nighttime CO2 fluxes measured in high wind speed conditions are correlated to soil or air temperature, and an empirical equation developed. Low fluxes measured at low wind speeds are replaced with values estimated using this equation (hereafter referred to as the high-wind-speed (HWS) approach) (e.g., Black et al. 1996; Goulden et al. 1996; Jarvis et al. 1997). The usual criterion for determining which fluxes should be replaced is when the friction velocity (u^ falls below a critical value. Critical values of reported in the literature range from 0.17 to 0.45 m s"1 (Blanken 1997). One difficulty with this approach is that the critical chosen affects the relationship obtained between the high wind speed CO2 flux and temperature. The second approach accepts as true all nighttime CO2 flux measurements with the exception of those associated y Chapter 4. Estimating nighttime respiration in an aspen forest 110 with instrument problems. Missing nighttime CO2 flux data are replaced with values calculated using an equation which relates either long-term averaged nighttime respiratory fluxes (Valentini et al. 1996; hereafter referred to as the all-nighttime-flux (ANF) approach) or CO2 fluxes measured on windy nights (Greco and Baldocchi 1996; Baldocchi et al. 1997) to temperature. These approaches provide alternative ways of obtaining long-term estimates of CO2 exchange, but they will give significantly different results for the same forest because of the different ways of estimating respiration rate and dealing with nighttime data. As Greco and Baldocchi (1996) point out, a 1 pmol rrf2 s"1 difference in the nighttime respiration rate 0 1 estimation for a year with 365 12-hr nights results in a difference of about 200 g C m" y" in carbon sequestration. Therefore, accurately assessing the nighttime respiration rate is essential for annual carbon sequestration estimation. Recently, Lee (1998) suggested a mechanism for the unknown escape of CO2 from the control volume (the 'mysterious route' referred to by Goulden et al. 1996). He argued that the low nighttime fluxes of scalars possibly resulted from mass flow fz (see Eq. (4.3)) which is not measurable by the eddy covariance technique. Using Mahrt's (1982) model, Lee assumed that the horizontal velocity was accelerated by the cold air drainage at a sloping site which resulted in subsidence at night. This subsidence brought relatively CCVfree air into the control volume and took C02-rich air downwind which results in the low C 0 2 fluxes measured above the forest (Fig. 4.2). The opposite occurred to the sensible heat. He proposed that this mass flow term of CO2 could be estimated using (4.1) Chapter 4. Estimating nighttime respiration in an aspen forest 111 where Cr is the C O 2 concentration at the reference height (39.5 m in our case), {C^j is the mean C O 2 concentration within the air column beneath the reference height, and w is the true vertical velocity at the reference height. He further proposed a procedure to calculate this vv as: w=w-(a(<b)+H<k)u) (4.2) where vv and u are the vertical velocity and streamline wind measured using the sonic anemometer, § is the wind direction, and a and b are constants dependent on the wind-Fig. 4.2 Diagram illustrates the mass flow hypothesis at a sloping site during the nighttime, uj and u2 are the horizontal velocities, vv" is the subsidence due to cold air drainage. Mass flow as a result of cold air drainage transports C0 2 downwind. direction. Can chamber measurements of nighttime respiration rates help resolve this issue of low nighttime C O 2 flux? Measurement of leaf and bole respiration by chambers is relatively Chapter 4. Estimating nighttime respiration in an aspen forest 112 straight forward (Ryan etal. 1997), but the measurement of soil CO2 efflux using chambers is difficult (Denmead and Raupach 1993). It is easy to damp out the nighttime pumping effects of atmospheric turbulence and fluctuations in atmospheric pressure on CO2 diffusion from the soil (Norman etal. 1992; Rayment and Jarvis 1997). Norman etal. (1997) recently found that systematic differences existed between the six chamber methods used by participants in BOREAS. In another recent study, Lavigne etal. (1997) found too much variability in scaled up chamber (soil, bole and leaf) measurements to draw any conclusion as to whether or not eddy covariance measurements underestimate nighttime CO2 fluxes. Maintaining an automatic chamber system in a long-term flux monitoring program is very difficult. This chapter is restricted to an analysis of eddy covariance data because of their long-term continuity and applicability at the stand scale. In this chapter, an alternative hypothesis is presented that there is no loss of CO2 other than what was measured by eddy covariance. This hypothesis states that CO2 accumulates in the air-filled pores of the forest floor and soil/snow under calm conditions and releases to the atmosphere when the wind speed is high. Consequently the objectives of this chapter are (i) to present an approach for estimating short-term net ecosystem exchange at night based on this hypothesis, (ii) to show evidence for and against both the mass flow and soil storage hypotheses, and (iii) to compare the estimates of long-term carbon sequestration at the O A site using the soil storage, HWS and A N F approaches. Chapter 4. Estimating nighttime respiration in an aspen forest 113 4.2 THEORETICAL CONSIDERATIONS Carbon sequestration (<J>, photosynthesis minus respiration) is the gain of carbon by the vegetation in a control volume of forest. It equals minus the sum of the net flux out of, and the rate of CO2 storage increase within the forest volume (Fig. 4.3), i.e., * = -{Fc + ASa I At + ASSI At + Afx + fz) (4.3) where F c is the vertical eddy diffusive CO2 flux (transport as a result of the vertical motion of atmospheric eddies, a positive value corresponding to an upward flux), ASa/At and ASs/At are, respectively, the rates of change in air CO2 storage and soil CO2 storage to a depth beneath the root zone where the CO2 flux equals zero, Afx is the net horizontal CO2 advection and fz is the vertical mass flow of CO2 at the instrument height (referred to earlier). Molecular diffusive transport is negligible compared with the eddy flux. At night, in the absence of photosynthesis, O is negative and equals minus the respiration rate. The horizontal advection term in Eq. (4.3) is difficult to measure or parameterize due to the limitation of instrumentation and the complexity of the process. Since the OA site is relatively homogeneous and within 1-2 kilometres of the tower, there is no significant change in elevation, advective fluxes likely vary randomly and are assumed to average to zero over the long-term. In order to test the soil CO2 storage hypothesis,^ is assumed to be zero, i.e., the mass flow term averages to zero over the long-term as suggested by Mahrt (1997). (Use of Lee's theoretical estimate of fz would invalidate the test of the soil C 0 2 storage hypothesis.) The air and soil C 0 2 storage terms are assumed to average to zero over the long-Chapter 4. Estimating nighttime respiration in an aspen forest 114 term. Hence, over the long term, e.g., growing season or year, it is reasonable to assume that total carbon sequestration is well approximated by minus the sum of the eddy covariance C O 2 flux measurements, i.e., 2> = -5>c (4.4) 4 ASJAt Fig. 4.3 Schematic description of the conservation equation for C0 2 in a forest ecosystem. The control volume is from the height of the eddy covariance sensors down to a depth where C0 2 flux becomes negligible. For the definition of symbols, see the text. Chapter 4. Estimating nighttime respiration in an aspen forest 115 Over the short-term, such as a few hours, the change in soil CO2 storage has to be taken into account to estimate 0. ASs/At can be expressed as ASjAt = Rs-Fs (4.5) where Rs is soil respiration (CO2 produced by soil microorganisms and plant roots) and Fs is the efflux of CO2 from the soil surface. This equation assumes no horizontal flux divergence and zero vertical CO2 flux at the base of the control volume (Fig. 4.3). The efflux can be expressed as the product of a CO2 concentration difference (between the bulk soil (Cs) and the atmosphere (CJ) and a transfer coefficient (hc) for the surface soil-atmosphere interface, i.e., Fs = hc (Cs-Ca). Since Cs was not available, Fs was parameterized by assuming it was proportional to the product of Rs and a transfer coefficient M which is a function of u^, i.e., Fs = MRS. The justification for using Rs is that CO2 concentration will tend to be driven by Rs although it would probably not be directly proportional to Rs (Suarez and Simunek 1993). Since the soil dominates ecosystem respiration (RSha) (Black et al. 1996), Rs was replaced with Rsha resulting in Fs = MRsha (4.6) where M is a function of at the 39.5-m height. Substituting Eq. (4.6) into Eq. (4.5) gives ASJAt = {\-M)Rsha (4.7) When wind conditions are steady, M = 1 and soil CO2 storage remains unchanged (ASJAt = 0), i.e., CO2 is released from the soil at the same rate as it is generated by respiration. At night when the wind speed drops, CO2 accumulates in the soil (ASJAt>0) Chapter 4. Estimating nighttime respiration in an aspen forest 116 and M < 1. During the daytime when wind speed increases, soil CO2 storage likely decreases (ASJAt < 0) and M > 1. It is very unlikely that ASJAt > Rsha, i.e., M < 0. Combining Eqs (4.3) and (4.7) for nighttime conditions (® = -Rsha) and assuming that Afx = fz = 0, we have: F +AS I At Rh = ^— (4.8 M Equation (4.8) indicates that the respiration rate of the ecosystem (soil and vegetation) can be estimated by correcting the nighttime CO2 efflux measurements with the wind function (M). In other words, the product RshaM estimates the efflux of CO2 from the soil and vegetation. By obtaining the relationships between Rsha and temperature, and between M and (discussed later), we can estimate the change in soil CO2 storage and the carbon sequestration over the short-term. 4.3 METHODOLOGY 4.3.1 Instrumentation and measurements 4.3.1.1 Flux measurements Above-canopy (39.5-m height) CO2 fluxes were measured using the eddy covariance technique on a 37-m high walk-up tower. This system consisted of a three-dimensional sonic anemometer/thermometer and a closed-path infrared gas analyzer in a temperature-controlled box. This eddy covariance system was operated continuously from mid-October to mid-Chapter 4. Estimating nighttime respiration in an aspen forest 117 November in 1993, and from early February to late September in 1994. No eddy covariance measurements were made from late September 1994 until mid-April 1996, when a similar system was installed at the same height (39.5-m) and continues to operate at the site to the present time. This system had operated at the site in 1993 and 1994 at the 6-m and 4-m heights, respectively, to determine the contribution of soil and the understory to the forest CO2 fluxes. The quality of the CO2 flux data was assessed by examining the w and CO2 concentration (Xc) power spectra and cospectra. We selected 12 nights when the CO2 fluxes were very low. On each night, two 2-h sets of high frequency (20 Hz) data were selected. One set was early in the night (21:00 - 23:00 CST), and the other was late at night (2:00 - 4:00 CST). In order to make comparisons with daytime cospectra, a two-hour set of high frequency data from 12:00 - 14:00 CST on the following day was also chosen. Using the normalized cospectrum of w and air temperature (T, calculated from the speed of sound measured by the sonic anemometer) as the input signal (i.e., assuming no frequency loss in the w and T signals) and that of w and x c as the output signal, the transfer function of the eddy covariance system was also calculated as the ratio of these two cospectra. Chapter 4. Estimating nighttime respiration in an aspen forest 118 Fig. 4.4 shows the seven day ensemble average of the cospectra of vv and T, and vv and %c measured at the 39.5-m height at the OA site from 21:00 - 23:00 and 12:00 - 14:00 CST (see Appendix B for the cospectra on each of the 12 nights). The 2-h mean values of wind speed, u^. and CO2 flux for nighttime and daytime were 2.1 and 3.1 m s"1, 0.08 and 0.67 m s"\ and 0.2 and -8.5 umol m"2 s"\ respectively. Comparison of the 21:00-23:00 CST w-T and vv-^c cospectra with those at 2:00 - 4:00 CST (not shown) indicated little change during the course .0 0.5 0.4 0.3 0.2 0.1 0 -0.1 S 0.5 0.4 0.3 0.2 0.1 0 -0.1 -1 J M v \ •/ V 1 (a) 12:00-1 14:00 CST — A \~~ -1 1 1 —1 1— 1 1 ' ' ' ' ' ' (b) 21:00- 23:00 CST - / A / ^—'V/X / \ -1 I x " — ^ ' \ , _ J 1 1 . . 1 1 10" 10 -2 10 f(Hz) 10" 10' Fig. 4.4 Comparison of ensemble averaged normalized (by the area under the curve) cospectra of w and T (the solid line) and w and %c (the dot-dashed line) measured at the 39.5-m height at the OA site during (a) the daytime and (b) nighttime of seven days in 1996. Chapter 4. Estimating nighttime respiration in an aspen forest 119 of the night. Fig. 4.4a shows that the w-T and w-Xc cospectra during the daytime at the OA site were very similar in shape and in the frequency of the peak spectral fluxes. This indicates a similarity in the transfer mechanisms of sensible heat and CO2 in unstable conditions, and also suggests that the eddy covariance system recorded nearly all of the fluctuations in CO2 associated with turbulent transport during the daytime. At night, the shapes of the two cospectra (Fig. 4.4b) were also very similar to one another which strongly suggests that the nighttime CO2 flux measurements were as reliable as the nighttime sensible heat flux measurements. Fig. 4.4 further shows that the roll-off in w-xc cospectra at high frequency was more noticeable at night than during the daytime. The analysis of the transfer function (the ratio of w-%c to w-T) of the eddy covariance system showed that the cut-off frequency of the eddy covariance system was about 0.4 Hz, i.e., w-fc is about half of w-T at this frequency. From Fig. 4.4, the mean contribution when / > 0.1 Hz to the total sensible heat flux was about 11% during the daytime and 34% during the nighttime for the seven days. The corresponding figures for the CO2 flux were 9% and 24%. As these values were calculated on low CO2 flux nights, the maximum high frequency loss for CO2 is very unlikely to exceed 2% during the daytime and 10% at night. This high frequency loss could have been corrected following Moore (1986); however, as the effect on the seasonal NEE was estimated to be 1-2% of Rsha, the correction was not applied. Chapter 4. Estimating nighttime respiration in an aspen forest 120 4.3.1.2 Supporting measurements Supporting measurements made during the BOREAS experiment included a soil temperature profile from soil surface down to a depth of 1 m, an air temperature profile from the 0.1-m to the 39.5-m height, incident solar radiation and photosynthetic photon flux density (Qo) above the forest, air humidity above and below the overstory, wind speed and direction (vaned propeller anemometer) above and below the overstory, wind speed (sonic anemometer) at 0.5, 5.9, 18.6 and 28.6 m above the ground, soil water content (time-domain reflectometry) to a depth of 1.2 m, precipitation (tipping bucket and weighing rain gauge), and leaf area index (LAI) of the overstory and understory (Black et al. 1996). To get a complete climate data set, missing soil temperature measurements atthe 2-cm depth were estimated using a relationship developed between these measurements and soil temperatures measured at the 10-cm depth about 100 m away at a BOREAS mesoscale meteorological network (MESONET) tower site (Shewchuk 1997) and/or linear interpolation. Missing QQ measurements were estimated using a relationship between the QQ measurements on the main tower and those on the MESONET tower, and/or linear interpolation. About 10% of air-storage corrected CO2 flux data at the 39.5-m height were missing in 1994. Missing data during the leafless periods and at night were replaced by the soil efflux values calculated using the product of Eqs (4.9) and (4.10) below. Those during the daytime in the growing season were replaced by values calculated using a rectangular hyperbolic relationship between air-storage corrected C 0 2 flux and QQ (e.g., Ruimy et al. 1995). The Chapter 4. Estimating nighttime respiration in an aspen forest 121 complete u^. data set at the 39.5-m height was obtained by filling the data gaps using a linear relationship between the measured using sonic anemometer and the standard deviation of wind speed (ou) measured using the sonic or vaned propeller anemometers, was well correlated to ou measured using the propeller-vane anemometer with an r of around 0.9. 4.3.2 Determining Rsha and M Following Bunnell etal. (1977) and Schlentner and Van Cleve (1985), the respiration rate was described using a logistic function of soil temperature (Ts) at the 2-cm depth, ^ a = l + exp[Mc-7;)] ( 4 * 9 ) where the values of Ts were bin averages, and the minimum respiration rate was assumed to be zero. To obtain the wind function M, the variable chosen to describe the degree of change in the turbulence regime was the ratio of the current friction velocity (u^ to the mean w^over the previous several days (u^). The wind function M was described by a rectangular hyperbolic function of this ratio, i.e., M = =f==—+ / (4.10) where the values of j were also bin averages. By bin averaging, we assumed that (1) the average value of Rsha (or M) within each bin represented the average effect of soil temperature (or change in the turbulence regime) by eliminating the short-term effects of advection and other factors, and (2) enough data points were collected in each bin to make Chapter 4. Estimating nighttime respiration in an aspen forest 122 the first point valid. Applying Eqs (4.9) and (4.10) to each half-hour implies that it provides a spatially averaged effect of soil temperature and turbulence on CO2 production and transport. When there are no leaves on the trees, i.e., no photosynthesis, Eq. (4.4) indicates that mean respiration rates can be obtained by averaging the eddy CO2 fluxes of enough sequential half-hours (in practice, at least 24 h). These mean respiration rates can then be related to the corresponding mean soil temperatures. In this case, correction using the wind function is not necessary, i.e., Eq. (4.8) is not applied. This procedure, however, can not be used during the growing season because of photosynthesis, and the correction of nighttime CO2 efflux data using the M function is essential to obtaining the true relationship between respiration rate and soil temperature. Consequently, the data used to obtain the coefficients in Eqs (4.9) and (4.10) included 24-hour averaged Fc +ASa/&t and Ts during the leafless period (fall in 1993, winter and spring in 1994, 128 days in total) as well as the nighttime Fc+ASa/lU, Ts and u^ju^ during the 1994 growing season (92% of the nighttime half-hours, 2005 half-hours in total). An iterative procedure, written using Matlab® (MathWorks, Inc., Natick, M A , USA), was used to calculate the coefficients a, b, c, d, e, and/ in Eqs (4.9) and (4.10). The first step in the iteration was to assume M = 1, and then to calculate half-hourly Rsha values during the growing season using Eq. (4.8). These calculated Rsha values were bin averaged with a Ts bin-width of 0.5 °C and combined with the 24-h averages of Fc +ASa/At during the leafless periods. These data were then used to obtain coefficients a, b and c in Eq (4.9) by doing a Chapter 4. Estimating nighttime respiration in an aspen forest 123 non-linear least squares fit. Next we calculated all M values during the growing season also using Eq. (4.8) but with the computed Rsf,a values from Eq (4.9) with the coefficients obtained in the previous step. These calculated M values were then bin averaged with a " * / M * bin-width of 0.1 and used to obtain the coefficients d, e and / i n Eq. (4.10) by doing a non-linear least squares fit. At the end of the cycle, we checked to see if Suw">=5X (4.ii) where summations extended over all half-hour data making up the growing season. (From Eq. (4.7), we see that Eq. (4.11) is equivalent to ^ (ASJAt) = 0 over the long term.) If Eq. (4.11) was not met, we then recalculated the growing season Rs/,a values using Eq. (4.8) with the M values calculated from Eq. (4.10) with the coefficients d, e and / obtained in the previous step, and again combined them with the 24-h averages of Fc + ASa /At during the leafless periods. Using these data, we repeated the non-linear least squares fit to obtain better estimates of a, b and c. I then recalculated all M values during the growing season and obtained improved estimates of d, e and/. This procedure was continued until Eq. (4.11) was met within 0.0005 g rrf2 for the growing season. Normally, stable values of the coefficients were obtained within 30 cycles. One- to ten-day running means of were tested to obtain in this iterative procedure. The coefficient of determination (r 2) of M reached its maximum value with 8-day running means of u^, however, r 2 did not change significantly for 5-day to 8-day running means. The assumption underlying Eq. (4.11) is supported by the calculations of Suarez and Simunek (1993). They found that soil respiration in an irrigated fine sandy loam was well Chapter 4. Estimating nighttime respiration in an aspen forest 124 approximated by the measured efflux of CO2 when integrated over weekly or longer intervals. This can be confirmed by estimating the order of magnitude of the time constant ( T ~ t2/De) for a depth I in a soil with an effective CO2 diffusivity De to equilibrate to a step change in CO2 concentration of the air at the soil surface. We estimated De using D0^g, where D0 is the diffusivity of C 0 2 in air (= 1.5xl0~5m"2 s"1) and £,g is the tortuosity factor approximated as 0.6 times the volumetric air content (= 0.2 m 3 air m"3 soil) (Jury etal. 1991). Assuming I = 0.5 m results in x ~ 1.6 days, i.e., about 5 days (~ 3t) for 95% equilibration. The resulting relationship between the respiration rate (RSha) and Ts is shown in Fig. 4.5 where a = 6.9594, b = 0.2462 and c = 8.5147. About 92% of the variance of the data was explained by the solid line in Fig. 4.5. Maximum respiration rate was about 7.0 p:mol m" s" (coefficient a), and half the maximum respiration rate occurred at 8.5 °C (coefficient c). Although respiration is also a function of soil moisture (Bunnell et al. 1977; Chen et al. 1998), it has not been included here since the soil remained moist through out the growing season. Chapter 4. Estimating nighttime respiration in an aspen forest 125 Fig. 4.5 Relationship of the daily averaged (from February 4 to May 10) and the bin averaged (from May 11 to September 20), air-storage corrected nighttime eddy CO2 flux, representing the soil, hazelnut and aspen respiration (RShJ, to the soil temperature measured at the 2-cm depth at the OA site in 1994. Vertical bars are ± 1 standard deviation. This relationship is independent of turbulence levels Raich and Schlesinger (1992) reported that the median value of the Q10 coefficient (the ratio of the rate of a process at one temperature to that at a temperature 10 °C lower) for soil respiration rates is about 2.4 though it varies from 1.3 - 3.3 as reported by numerous authors. The value of the Q10 coefficient for the regression in Fig. 4.5 is 2.8 (when Ts changes from 5 °C to 15 °C) and 1.6 (when 7; changes from 10 °C to 20 °C) which agrees well with above values. The corresponding value for the RSha(Ts) equation reported by Black et al. (1996) was 5.4 which is much higher than expected probably because they used only high wind speed Chapter 4. Estimating nighttime respiration in an aspen forest 126 nighttime eddy covariance data, which probably overestimated respiration rates. Russell and Voroney (1998), who manually measured soil CO2 efflux using a dynamic closed chamber (LI-COR model 6000-09) at the OA site, reported a Q10 value of 3.9. The maximum values of 6 umol m"2 s"1 shown in Fig. 4.5 are about 25% less than the maximum values (8 umol m"2 s"1) obtained using the dynamic closed chamber (Russell etal. 1998). Fig. 4.6 shows the resulting wind function M (the solid line), where d = 0.7888, e = 0.4076 and/ = 0.2385 when an 8-day running mean was used. Approximately 85% (r2) of <1 CO < + Fig. 4.6 Relationship between the bin averaged values of the nighttime wind function (M) and the ratio of current half-hour to the mean over the previous eight days from May 10 (DOY 130) to September 20 (DOY 263) in 1994. The vertical bars are ± 1 standard deviation. This function is independent of soil temperature. Chapter 4. Estimating nighttime respiration in an aspen forest 127 the variance in the data was explained by this relationship. Since the C 0 2 efflux equals RshaM, Fig. 4.6 indicates that when the air was completely still (u^ = 0), about 24% (coefficient f) of the respired C 0 2 was either released to the atmosphere from the soil or respired by vegetation, while 76% was stored in the soil. When > (generally during the daytime), some of the C 0 2 stored in the soil at night is released to the atmosphere, and the air-storage corrected eddy covariance C 0 2 flux overestimates the respiration rate. The value of M was about 1.4 for u^j = 2.3. This indicates that the eddy C 0 2 flux overestimated the respiration rate by about 50% when the wind speed was extremely high. 4.4 R E S U L T S A N D D I S C U S S I O N 4.4.1 E v i d e n c e for and against the m a s s f low h y p o t h e s i s The most compelling evidence supporting the mass flow hypothesis is the occurrence of slightly negative values of vv at night and slightly positive values during the daytime reported by Lee (1998) in his analysis of 1994 OA eddy covariance data. He found that the median nighttime value of w from August to mid September 1994 was about -1.0 cm s"1. Fig. 4.7 shows the nighttime and daytime frequency distributions of vv calculated using Lee's procedure for June-August 1996 for which the average values of vv were -1.1 and +0.7 cm s~\ respectively. This shows that Lee's results were not instrument specific as different sonic anemometers were used in 1994 and 1996. Assuming these negative values of vv are realistic, mass flow at night might account for the loss of C 0 2 from the control volume. Chapter 4. Estimating nighttime respiration in an aspen forest 128 Fig. 4.7 Histogram of the calculated vv using Lee's procedure at the 39.5-m height at the OA site over the period of June-August 1996. Another test of this hypothesis would be the degree of correlation between vv and the residual in the energy balance. This residual or imbalance is given by / = Rn - H - AE - Go - Jt where Rn is the net radiation, H is the sensible heat flux, AE is the latent heat flux, Go is the soil heat flux and Jt is the rate of change in the total heat storage in the control volume. The imbalance was commonly observed to be negative on calm nights (Blanken et al. 1998). Fig. 4.8 shows the relationship between / and vv for June-August 1996. The positive correlation in Fig. 4.8 (r2 = 0.005) which is slightly worse than that found by Lee (1998) (r2 = 0.024) provides some but weak support for the mass flow hypothesis. Perhaps somewhat more importantly, vv was negative in only 57% of the half-hours when / < 0. Fig. 4.9 compares the energy imbalance with that estimated from mass flow {wocp(Tr - (T)). This figure clearly shows that mass flow provides a relatively poor estimate of the energy imbalance. Chapter 4. Estimating nighttime respiration in an aspen forest w(m s_1) Fig. 4.8 Relationship between the energy imbalance (J = R„ + H + AE + G0 + J,) and w at the 39.5-m height at the OA site in summer 1996. Sensible heat transfer by mass flow (W nrr2) Fig. 4.9 Comparison of the nighttime energy imbalance and the mass flow term for sensible heat at the OA site in summer 1996. Chapter 4. Estimating nighttime respiration in an aspen forest 130 In order to determine whether mass flow could account for the small eddy flux of CO2, we compared the patterns of vv and Fc on 86 nights between June 1-August 31 1996. The objective was to see if the direction and magnitude of w were consistent with the measured eddy fluxes. For example, when Fc was zero, we expect to see vv< 0, and when Fc was positive and of reasonable magnitude, vv is expected to be close to zero. Fig. 4.10 summarizes the results of this analysis by showing patterns for selected nights that were representative of a significant proportion of the 86 nights. In 31% of the cases (represented by D O Y 207) Fc was close to zero {u^ < 0.15 m s"1) and vv was negative as expected. In the other 13% of the cases (represented by DOY 229), Fc was reasonably high (u^> 0.3m s"1) and vv oscillated around zero. These two categories, accounting for 44% of the nights, can be considered evidence for the occurrence of mass flow. However, in the other 56% of the nights there was no consistent relationship between vv and Fc. For example, in 11% of the cases (represented by DOY 211) Fc was zero but vv was zero or even positive. This clearly shows that vv as calculated using Lee's procedure did not account for the zero flux. In the other 24% of the cases (represented by DOY 171) the magnitude of Fc was reasonable but vv was strongly negative. In this case, a mass flow correction will lead to an unreasonably high respiratory flux by the ecosystem. In summary, this analysis suggests that either mass flow is not significant or vv should be calculated using a different procedure. Chapter 4. Estimating nighttime respiration in an aspen forest 131 - / ^ 3- DOY 229 u_u ' 1 ' 1 1 ' J J I 1-i r -11% DOY 211 -i i i i 1 — i 1 1 1 i 6% ^ DOY 223 i i 1-31% \ 5 0 DOY 207 18 20 22 0 2 4 6 18 20 22 0 2 4 6 Hour (CST) CO E Fig. 4.10 Relationship between Fc (solid line) measured above the forest and calculated vv at the 39.5-m height (dashed line) at the OA site on the 89 nights of summer 1996. The percentage is the occurrence of the case. Fig. 4.11 compares the air storage corrected Fc with CO2 mass flow term fz. The solid line represents the expected approximate relationship between Fc + ASa/At and fz. For example, when Fc + ASalAt is about 7 umol m"2 s"1 (expected summer value of the respiration of the ecosystem), the expected mass flow term would be approximately zero. On the other hand, when Fc + ASa/At is near zero (i.e., extremely calm nights), the mass flow term would be expected to be 7 umol m"2 s"1. The figure shows very large scatter around this line with virtually no relationship for the data. This shows even more clearly that either the mass flow hypothesis is not applicable or its formulation is not yet proper. Perhaps the assumption of Lee (1998) that w decreases linearly to zero from the reference height to the Chapter 4. Estimating nighttime respiration in an aspen forest 132 ground is too simplified. There may be a difference in flow divergence above and below the overstory in the forest. 10 w CNJ £ o E co + -2 -60 . 1 I- • ' . « . 1 • >• • • • • • • • • • • • • V a** • ?*•% • • • • • • • * • • •V . ^ % • * • • • • • « • * J * • * • • * • • • * • -40 -20 0 20 40 60 C 0 2 transfer by mass flow (umol nr 2 s1) Fig. 4.11 Comparison of the air C0 2 storage corrected nighttime Fc and the calculated mass flow term at the OA site in summer 1996. The solid line is the expected relationship between these two fluxes. 4.4.2 Evidence for and against the soil C0 2 storage hypothesis In this section, three pieces of evidence are presented that support the view that changes in soil CO2 storage can account for a significant proportion of respired CO2 on calm nights. This evidence involves the occurrence of strong nighttime atmospheric stability within the stand, the effect of wind speed on CO2 efflux during the winter, and the effect of rainfall on C 0 2 efflux. Chapter 4. Estimating nighttime respiration in an aspen forest 133 4.4.2.1 C O 2 accumulation at ground level on calm nights Fig. 4.12 compares the air temperature and CO2 concentration profile measurements during 2 hours (1:00 - 3:00 CST) on two typical nights at the OA site in 1994. The first night (September 4; DOY 247) was windy with = 0.34 m s"1 and a stability index (z-d)/L = 0.018 (z is the height, d is the displacement height (13.2 m at the OA site, 40 30 —• 20 O ) ' © X 10 0 Air temperature ( C) — I • I 1—1 1 r - , 1 1 (0 • (d) • I • r 1 1 ; < w, , 1 1 1 — 40 30 'E 20 ••-» _ c D ) '<D 10 I 0 400 500 600 400 500 600 C 0 2 concentration (umol mol ) F i g . 4.12 Comparisons of air temperature profiles and C 0 2 concentration (umol mol" 1 of moist air) profiles at the O A site on a windy night (September 4; D O Y 247) and calm night (September 8; D O Y 251), 1994. The narrow panels are for D O Y 247 and the wider panels are for D O Y 251. The time intervals are 1:00 - 1:30 ( • ) , 1:30 - 2:00 ( • ) , 2:00 - 2:30 ( • ) , and 2:30 - 3:00 C S T ( • ) . Chapter 4. Estimating nighttime respiration in an aspen forest 134 Simpson 1996) and L is the Monin-Obukhov length) at the 39.5-m height, and the second night (September 8; D O Y 251) was calm with = 0.13 m s_1and (z-d)/L = 3.2 at the 39.5-m height. Only 9% of the nighttime half-hours were windy ((z-d)/L < 0.02 ) in the summer (June - August) of 1994. On the windy night, mixing was strong (Fig. 4.12a and Fig . 4.12c). The air temperature gradient was very small except near the ground where it was about -0.2 °C m" 1. The CO2 gradient was also small from the forest floor to the 39.5-m height. On the calm night, however, the temperature gradient was large (Fig. 4.12b). There was a strong inversion (about 0.1 °C m"1) within the stand almost down to the ground as a result of the radiative cooling of the hazelnut understory beneath the relatively open aspen canopy. Therefore, turbulence within the canopy was strongly suppressed. The effects of this strong inversion can be seen in the CO2 concentration profile (Fig. 4.12d). The CO2 concentration profile shows a strong lapse, especially in the trunk space. Thus, a significant amount of CO2 accumulated near the ground and then slowly diffused up into and above the canopy. However, this amount was not large enough to give an efflux value (i.e., Fc + ASa/At) which was comparable with the expected respiration rate (only about 60% when = 0.1 m s"1) (Fig. 4.13). From June to August 1994, when was low (u^. < 0.17 m s - 1), the values of Fc +ASa/At were much lower than the expected respiration rate (4 - 5 p:mol m" 2 s"1), although A 5 a / A / tended to be higher than when was lower. Similar patterns have been observed in a temperate deciduous forest (Harvard forest, Goulden et al. 1996) and in boreal coniferous forests (Jarvis et al. 1997). Cold air drainage, or an undetected escape route of CO2 has been suggested as possible explanations for this Chapter 4. Estimating nighttime respiration in an aspen forest 135 phenomenon. Another equally possible explanation, however, is that much of the respired CO2 was stored in the soil rather than in the air layer beneath the eddy covariance sensors. The fact that considerable CO2 accumulation occurred in the air layer near the ground in Fig. 4.12d strongly suggests the latter explanation is very possible. Our calculations indicate that the mean half-hourly values of the changes in soil and air CO2 storage at night during the summer of 1994 were 0.5 umol m"2 s"1 (using Eqs (4.7) and (4.9)) and 1.3 umol m"2 s"\ respectively. This rate of change in soil CO2 storage is equivalent to a soil CO2 concentration change of 300 mg m"3 (about 180 umol mol"1 of moist air) over a half hour in a soil column a half metre deep with an air-filled porosity of 25% (see Appendix A). This rate of change would be very likely considering that the daytime average of the measured soil air C 0 2 concentration reached 1600 umol mol"1 in the surface organic layer and 3600 umol mol"1 at the 15-cm depth in the underlying mineral layer at the OA site over the summer of 1994 (Russell and Voroney 1998; Russell etal. 1998). Chapter 4. Estimating nighttime respiration in an aspen forest 136 Fig. 4.13 The nighttime rate of change of air C0 2 storage (A) and Fc (•) in relation to at the 39.5-m height at the OA site from June to August, 1994. All the data are bin averaged with an bin-width of 0.05 m s"1. The lines are the corresponding 3-order best fits. The top heavy line is the best fit of the sum of the half-hourly values of Fc and air C0 2 storage (not the sum of the two other lines). The vertical bars are +1 standard deviation. Chapter 4. Estimating nighttime respiration in an aspen forest 1 3 7 The situation in tropical rain forests seems to be very different. Grace et al. (1995) reported that at an undisturbed site in northern Brazil the CO2 stored in the air accounted for most of the respiration at night. This indicates that there was no significant CO2 storage in the soil. There are two likely reasons for this difference. The first is the difference in nighttime turbulence between the two types of forest. In tropical rain forests, the atmosphere is usually unstable within the canopy mainly because of the cooling of the upper part of the dense canopy, and stably stratified above the canopy which acts as a lid preventing turbulent transport of CO2 from the canopy. Thus, nighttime vertical mixing frequently occurs within the rain forest canopy (Wofsy etal. 1988; Fitzjarrald and Moore 1990; Grace etal. 1995). In the OA forest with its relatively open canopy, however, the stable stratification tends to suppress mixing within the canopy. Mixing was usually confined to the 2-m deep layer beneath the top of the understory, due to the cooling of the upper part of the hazelnut canopy, although as shown in Fig. 4.12 it occasionally extends into the trunk space and aspen canopy later in the night. As a result, some of the CO2 respired by roots and soil microbes tends to be stored in the soil. The second reason is the significant difference in the soils. The mean soil organic matter in the tropical forest is about 10.4 kg C m" compared to about 14.9 kg C m"2 in the boreal forest (Schlesinger 1997). Examining the specific soils, the oxisolic soils in the tropical rain forests have little organic matter in the mineral layers and virtually no surface organic layer whereas the luvisolic soil at the OA site has a relatively thick surface organic layer (see Appendix A). Consequently, the much reduced porosity of the tropical soil results in lower CO2 storage in the soil, and higher storage in the air. Direct measurements of Chapter 4. Estimating nighttime respiration in an aspen forest 138 short-term changes in soil CO2 concentration profiles are necessary to confirm this hypothesis. 4.4.2.2 Enhanced CO2 efflux in windy conditions during winter Baldocchi and Meyers (1991) showed that the CO2 efflux rates from a temperate deciduous forest floor were 'modulated' by fluctuations in the static pressure, which were well correlated with the standard deviation of the vertical velocity w (ow). The 1996/97 winter measurements at the OA site and winter observations in the Harvard forest (Goulden et al. 1996) are consistent with these earlier findings. Fig. 4.14 shows the relationship 10 o E O ^--0.5 -1 soil & snow 346 347 348 349 350 351 352 DOY 1996 1.5 1 E 0.5 ^ * 0 Fig. 4.14 Measurements made at the OA site with snow cover on the ground from December 11 (DOY 346) to December 17 (DOY 352), 1996. The eddy C 0 2 fluxes and were measured at the 39.5-m height and the soil temperatures were measured at the 2-cm depth. Chapter 4. Estimating nighttime respiration in an aspen forest 139 between and soil CO2 efflux measured above the OA canopy when the ground was covered with snow. The efflux was low at low u^, and increased significantly with increasing Wjj, on December 16 (DOY 351), 1996, even though the soil temperature dropped slightly. Similar phenomena were also observed when the soil was not covered with snow (e.g., Fig. 4.15). The quality of these eddy covariance measurements was checked carefully. The w-%c cospectra were similar to the daytime cospectra shown in Fig. 4.4a. Independent measurements of wind speed using an R M Young propeller anemometer also showed sudden increase at the time of the increase in u^. During both episodes of high u^, sensible heat also increased to about 300 W m"2 on DOY 303 and 50 W m"2 on D O Y 351 (see Appendix D). Latent heat increased to 100-220 W m"2 and then slowly decreased as the surface soil dried out (DOY 303), while it increased to about 50 W m"2 and remained unchanged until the event ended (DOY 351). During the daytime portion of these periods, these heat fluxes were generally consistent with net radiation. The disagreement at night indicates the importance of the heat storage and soil heat flux terms in the energy balance. CO2 concentration before the events tended to have a diurnal pattern whereas it was steady and lower (about 360 umol mol"1) when was high. These observations of high Fc strongly support the hypothesis that significant CO2 is stored in the soil and snow, and a high turbulence level enhances the efflux while a low level suppresses the efflux. Goulden etal. (1996) also reported: "The increase in CO2 efflux during windy periods was especially pronounced in winter 1993. Rates of efflux exceeding 10 umol m"2 s"1 were repeatedly observed when the friction velocity exceeded 0.8 m s"1." They also attributed this to the aspiration of C02-rich air from soil and snow pore Chapter 4. Estimating nighttime respiration in an aspen forest 140 space rather than a short-term increase in C 0 2 production. These bursts of high C 0 2 flux could last for 2/3 of the day which gave us a sense of the capacity of the soil for storing C0 2 . The calculated decrease in the C 0 2 concentration in the air-filled pores of the soil during those periods was in the order of 10000 p.mol mol"1. This is higher than expected (see C A . Russell's measurements in the previous section). There are two possible explanations. Firstly, a significant amount of C 0 2 is dissolved in water (Suarez and Simunek 1993). This can vaporize and be released to the atmosphere so that the actual decrease of C 0 2 in the air-filled pores is smaller. Secondly, C 0 2 concentrations deeper in the soil are much higher than near the surface. Trumbore et al. (1995) observed in a tropical forest soil that C 0 2 concentrations were 20000-40000 itmol mol"1 for the 1-2 m depth while it was about 11000 p.mol mol"1 in the first 50-cm layer. The C 0 2 deeper in the soil diffuses upward faster due to the C 0 2 lost in the upper layers of soil, which results a smaller decrease in the CO2 concentration in this upper layer. Therefore, accurately measured C 0 2 profiles in the soil on an hourly or a two-hourly basis (e.g. Fang and Moncrieff 1998) would help to fully understand this issue. Chapter 4. Estimating nighttime respiration in an aspen forest 141 10 o E ' »0 6 4 O ^ 2 0 soil 1.5 1 ^ E 0.5 ^* 0 301 302 3 0 3 DOY199§° 4 3 0 5 3 0 6 Fig. 4.15 Measurements made at the OA site without snow cover on the ground from October 27 (DOY 301) to November 1 (DOY 306), 1996. The eddy C 0 2 fluxes and were measured at the 39.5-m height and the soil temperatures were measured at the 2-cm depth. 4.4.2.3 Enhanced C 0 2 efflux by rainfall Although there is debate about the reliability of the eddy sensible heat flux measurements during rain, Mizutani et al. (1997) found that the standard deviation and power spectra of u, v, and w did not change for rainfall rates up to 30 mm h"1. On the basis of this and our own Chapter 4. Estimating nighttime respiration in an aspen forest 142 direct observation of vv signals dur ing rainfal l events, we have conf idence in our CO2 f l ux measurements dur ing rainfal ls up to 10 m m per half-hour. F i g . 4.16 shows the effect of rainfal l on the CO2 e f f lux dur ing the same half-hour at the O A site on nights between 16 M a y and 20 September 1994. The measured C 0 2 e f f lux (Fc + ASJAt) was 0.8 - 1.5 [imo\ m" 2 s"1 higher than the estimated value (RShaM) when it was ra in ing. Typ i ca l values of the ratio of these two f luxes were 1.1 to 1.4. The ratio also shows a tendency to increase w i th increasing rainfal l intensity w i th a large amount of scatter. Th is is ma in ly because C02-enriched soi l air is forced out by inf i l t rat ing water although there might be an enhancement of so i l respiration by the O2 d isso lved in the water (Jury et al. 1991). W e also observed a s ignif icant increase in the eddy CO2 f lux and concentration at the 4-m level dur ing periods of rain in 1994. These observations strongly suggest that a considerable amount of respired CO2 was stored in the air-fi l led pores of the so i l . These results are pre l iminary and need further conf i rmat ion. Chapter 4. Estimating nighttime respiration in an aspen forest 143 o E DC I < CO < Rainfall (mm half-hour ) Fig. 4.16 Effect of rainfall on the difference between the measured C 0 2 efflux and the estimated efflux {RShcM) during the nighttime from the spring to fall of 1994. The data are the bin averages using a rainfall bin width of 0.6 mm half-hour"1. The vertical bars are ± 1 standard deviation. 4.4.2.4 Evaluat ion of the soil C 0 2 storage model Because of the difficulty of obtaining the advective term (A/x) in Eq. (4.3) and the horizontal variability of C 0 2 concentration within the forest which resulted in variability of ASaIAt, it was almost impossible to compare the calculated C 0 2 efflux iMRsha) with the measured values (Fc + ASa/At) on a half-hourly basis. However, the nighttime means of the two decreased the effects of Afx and the horizontal variability of ASa/At on C 0 2 efflux estimation. Thus, the comparison of the nighttime means should be a good indicator of the Chapter 4. Estimating nighttime respiration in an aspen forest 144 performance of the soil CO2 storage model. Fig. 4.17 shows that the modeled efflux explained 78% of the variance of the measured efflux, whereas the respiration function alone (Eq. (4.9) and Fig. 4.5) was able to explain only 67%. (The similarity of the 1:1 line and the regression is, of course, due to the assumption that ^(Fc + ASa/At) = ^MRsha). Fig. 4.18 shows that the same model (i.e., using the 1994 coefficients) explained about 66% of the variance in 1996. Fig. 4.17 Comparison of the relationships between Fc + ASa/At and Rsha (a) and between Fc + ASa/At and MRsha (b) at the OA site in 1994. Values are nighttime means. The dash-dotted lines are the regressions: Fc + ASJAt = 0.85Rsha + 0.31 (a) and Fc + ASJAt = 0.97MRsha + 0.27 (b) and the solid lines are the 1:1 lines. Chapter 4. Estimating nighttime respiration in an aspen forest 145 Fig. 4.18 The relationship between C0 2 effluxes measured (Fc + ASa/Ar) and calculated (MRsha) using the 1994 model at the OA site in 1996. Values are nighttime means. The dash-dotted line is the regression: Fc + ASa/At = \.0\MRsha + 0.03 and the solid line is the 1:1 line. Chapter 4. Estimating nighttime respiration in an aspen forest 146 4.4.3 Comparison of carbon sequestration estimates using the three approaches 4.4.3.1 Diurnal carbon sequestration The carbon sequestration obtained using the new approach was much higher than the HWS approach. For example, on August 13 - 14 (24-h), the total sequestration was 6.4 g C m"2 using the new approach, compared to 5.5 g C m"2 using the HWS approach (Table 4.1 and Fig. 4.19). During the nighttime, the respiration (-O) estimated using the new approach was 0.4 g C m"2 lower (22% of the respiration obtained using the new approach) than the HWS approach. The other portion of the difference in the 24-h total sequestration was from Table 4.1 The comparison of cumulative carbon sequestration (g C m"2) obtained using different approaches for the OA forest, Prince Albert National Park, Saskatchewan in 1994. The growing season is from May 20 - September 10. The values of the uncertainties are standard errors of the estimates. In the HWS approach, the nighttime flux measurements when < 0.45 m s"1 were replaced with calculated respiration rates estimated using an empirical equation relating nighttime C0 2 flux at > 0.45 m s"1 to soil temperature at the 2 cm depth. In the ANF approach, all the nighttime C0 2 flux measurements were used without correction using the wind function M. Approach Aug. 13-14 (24-h) Growing Season Annual New 6.4 ±0.2 460 ±15 197 ±30 HWS 5.5 ±1.0 400 ±30 103 ± 4 0 A N F 6.2 ±0.2 460 ±15 188 ±30 Chapter 4. Estimating nighttime respiration in an aspen forest 147 the estimation of the daytime carbon sequestration. The daytime value obtained using the new approach was 0.5 g C m"2 higher (6% of the fixation obtained using the new approach) than the HWS approach due to the daytime correction of the soil CO2 storage. The difference between the 24-h total carbon sequestration obtained using the new approach and the A N F approach was not very significant, again due to the correction of the soil CO2 storage during the daytime, although the estimated respiration was higher using the new approach. Thus, the HWS approach overestimates nighttime respiration and underestimates daytime CO2 sequestration, while the A N F approach underestimates both values. Chapter 4. Estimating nighttime respiration in an aspen forest 148 Hour (CST) Fig. 4.19 Effects of the three approaches on the diurnal carbon sequestration at the OA site on the same day as in Fig. 4.1. The solid line is based on the new approach, the dotted line is based on the HWS approach, and the dashed-dotted line is based on the A N F approach. The dotted line (HWS) and the dashed-dotted line (ANF) are the same during the daytime. (See daily totals of carbon sequestration in Table 1). To further examine the reasons for these differences, the relationships of respiration rate to soil temperature obtained using the three approaches are shown in Fig. 4.20. The critical Uy, for the HWS approach here is 0.45 m s"1 which is approximately equivalent to the critical u of 3.5 m s"1 reported by Black etal. (1996). Clearly, the respiration rates obtained using the HWS approach are much higher than those using the new and the A N F approaches, especially at high temperatures. The difference (HWS - new) varies from 0.24 umol m"2 s"1 (86% of Rsha using the new approach) at 7^  = -4 °C to 1.4 umol m"2 s"1 (24% of R^ using the Chapter 4. Estimating nighttime respiration in an aspen forest 149 new approach) at 7^  = 16 °C. The fact that the HWS approach uses only those Fc + ASa/At values on windy nights but neglects nighttime CO2 release from the soil (ASJAt < 0) results in an overestimate of respiration (-<&) on those nights. On calm nights, which tend to predominate, use of a nighttime high wind speed CO2 flux vs. temperature relationship will further overestimate the respiration rate, and might be more than on windy nights if increasing soil CO2 suppresses soil respiration (Amthor 1991; Sierra and Renault 1995; Burton etal. 1997). Similarly, the A N F approach is also problematic. On calm nights, respired CO2 tends to T at the 2-cm depth ( C) Fig. 4.20 Comparison of relationships between nighttime respiration rate and soil temperature at the 2-cm depth, developed in the three approaches. Chapter 4. Estimating nighttime respiration in an aspen forest 150 be stored in the soil (AS^/Af > 0). Thus, the respiration relationship obtained using this approach underestimates the true respiration (-<£) because it ignores the soil CO2 storage. On the other hand, if only the effluxes on windy nights are used as Greco and Baldocchi (1996) suggested, the resulting respiration rate is an overestimate of respiration for the same reason as in the HWS approach. Therefore, the cumulative respiration will be either a significant underestimate (the A N F approach) or a significant overestimate (Greco and Baldocchi 1996). Although much less than in the HWS approach, the significance of the error in carbon sequestration estimation, associated with filling data gaps using these respiration relationships, depends on the length of missing data gaps. In the case of frequent and lengthy gaps, the error can be significant. At the OA site, the annual carbon sequestration estimated using the A N F approach did not show a significant difference (<10 g C m"2, Table 4.1) from the new approach because of the relative completeness of the data set at high temperature and the similarity of the respiration rates at low temperature obtained using the two approaches. The cumulative respiration, however, was 100 g C m ' lower than that using the new approach. Chapter 4. Estimating nighttime respiration in an aspen forest 151 4.4.3.2 Annual carbon sequestration Fig. 4.21 and Table 4.1 present the values of the annual carbon sequestration using the three approaches. The uncertainties in the new approach are the results of the replacement of missing data with calculated C O 2 effluxes at night for long-term estimates, and the soil C O 2 storage estimation on August 13 - 14. The uncertainties in the A N F approach are from filling missing data with calculated Rsha. The uncertainties in the HWS approach are the results of the substitutions of low wind speed C O 2 flux measurements which made the estimates more 400 I • 1 1 r -200 1 ' 1 ' 1 1 100 200 300 DOY 1994 Fig. 4.21 Comparison of the cumulative carbon sequestration at the OA site for 1994, calculated using the three approaches. The solid line is a result of the new approach, the dotted line is a result of the HWS approach and the dashed-dotted line is a result of the ANF approach. Chapter 4. Estimating nighttime respiration in an aspen forest 152 uncertain. Since the estimates of the carbon sequestration using both the new and the A N F approaches were very similar, the following discussion will focus only on the comparison of the new and the HWS approaches. Fig. 4.21 shows that the forest lost about 118 and 133 g C m"2 to the atmosphere in the first 148 days in 1994 using the new and the HWS approaches, respectively. Using the new approach, the cumulative sequestration of carbon by the forest reached its maximum of 360 g C m"2 on September 16 (DOY 259), whereas it reached its maximum of 288 g C m"2 on September 13 (DOY 256) using the HWS approach. The corresponding figures for the annual carbon sequestration were 197±30 and 103±40 g C m" , respectively. During the growing season (May 20 - September 10), as defined by Black et al. (1996), the values of the total carbon fixation obtained using the new and the HWS approaches were 460±15 and 400±30 g C m"2, respectively (Table 4.1). Compared to the carbon sequestration of 350 g C m"2 for the growing season and 130 g C m"2 for the whole year reported by Black et al. (1996), the estimates using the HWS approach here are significantly different. The reason was that they did not make a water vapour pressure broadening correction, and also underestimated the ecosystem respiration rate in the fall of 1994 when they filled in missing data. This indicates the importance of almost uninterrupted continuous long-term flux monitoring for obtaining the annual carbon sequestration of a forest. As shown above, the estimation of annual carbon sequestration at the OA site using the new approach was about 90 g C m" more than the HWS approach. The annual carbon sequestration at the Harvard forest estimated using the A N F approach (Wofsy et al. 1993) Chapter 4. Estimating nighttime respiration in an aspen forest 153 was similarly higher, being about 80 g C m"2 more than the HWS approach (Goulden et al. 1996) . The primary results of BOREAS have shown that the average carbon sequestration of the boreal forest is lower than expected (Sellers et al. 1995) probably because of the prevalence of the use of the HWS approach in analyzing nighttime tower CO2 flux data in BOREAS (e.g., Black etal. 1996; Goulden etal. 1997; Jarvis etal. 1997; McCaughey etal. 1997) . If the underestimation of 90 g C m"2 y"1 were applicable to the entire boreal forest, then its sink strength would be greater by roughly 1 Pg C y"1, assuming the area of the boreal forest is about 12 x 106 km"2 (Whittaker and Lekins 1975). 4.5 C O N C L U S I O N S In this chapter, a hypothesis that the low nighttime CO2 effluxes in boreal aspen forest result from the short-term changes in CO2 storage in the air-filled pores of soil has been presented. The fact that CO2 accumulation at the ground level with CO2 concentrations at the 0.5-m height reached 700-900 umol mol"1 on calm nights in the summer indicates that a considerable amount of C 0 2 probably accumulated in the air layer below the lowest sampling tube of the concentration profile and within the air-filled pores in the soil. Perhaps the best data showing the capacity of the soil to store CO2 is that obtained before and after high during the leafless period. It shows that CO2 flux above the background respiration flux of 3-4 umol m"2 s"1 can persist for 10-15 hours. Finally, enhanced C 0 2 efflux after and during rainfall, although debatable, is consistent with a significant amount of C 0 2 being stored in the soil. Direct measurements of short-term changes in soil CO2 concentration profiles, however, Chapter 4. Estimating nighttime respiration in an aspen forest 154 are necessary to confirm this hypothesis. The rate of change of soil CO2 storage is estimated by assuming it is proportional to soil, hazelnut and aspen respiration rate (Rs/,a) with a proportionality factor (1 - M), i.e., ASJAt = (l-M)Rsha. Rsna is expressed as a logistic function of the soil temperature at the 2-cm depth and M a rectangular hyperbolic function of the ratio of the current to the running mean of (u^) over the previous several days. Then the short-term carbon sequestration (O) is obtained by adding the rate of change in soil CO2 storage to the air-storage corrected CO2 eddy flux. Long-term carbon sequestration, however, can be estimated simply by summing the CO2 eddy fluxes because CO2 storage changes in both soil and canopy air, as well as advective effects, average to zero over periods of a week or more. Evaluation of the mass flow hypothesis using the 1996 data, independent to that used by Lee (1998) showed the mass flow hypothesis performed poorly in estimating the half-hourly energy imbalance values. For CO2, in only 44% of the cases did vv behave as predicted. In examining the relationship between air CO2 storage corrected Fc and the calculated mass flow term, the scatter of the points strongly suggests that mass flow does not account for the low C 0 2 fluxes on calm nights. Using this approach based on the soil CO2 storage, the annual total carbon sequestration at the OA site was estimated to be 197±30 g C m" , compared to 103±40 g C m" using the high wind speed approach. The corresponding values for the growing season were 460±15 and 400±30 g C m" , respectively, which suggests that this boreal deciduous forest is probably sequestering significantly more carbon than was indicated by our preliminary Chapter 4. Estimating nighttime respiration in an aspen forest 155 analysis (Black et al. 1996). If this underestimation of 9 0 g C m" 2 y"1 were applicable to the entire boreal forest, then its annual carbon sequestration would be greater by roughly 1 Pg C y 1 . Chapter 4. Estimating nighttime respiration in an aspen forest 156 4.6 R E F E R E N C E S Amthor JS (1991) Respiration in a future high CO2 world. Plant, Cell and Environment, 14, 13 - 20. Baldocchi DD, Vogel C, Hall, B (1997) Seasonal variation of carbon dioxide exchange rates above and below a boreal jack pine forest. Agricultural and Forest Meteorology, 83, 147-170. Baldocchi DD, Vogel C (1996) A comparative study of water vapor, energy and CO2 flux densities above and below a temperate broadleaf and a boreal pine forest. Tree Physiology, 16, 5-16. Baldocchi DD, Meyers TP (1991) Trace gas exchange above the forest floor of a deciduous forest. 1. Evaporation and CO2 flux. Journal of Geophysical Research, 96, 7271-7285. Black TA, den Hartog G, Neumann H H , Blanken PD, Yang PC, Russell C, Nesic Z, Lee X , Chen SG, Staebler R, Novak M D (1996) Annual cycle of water vapour and carbon dioxide above a boreal aspen forest. Global Change Biology, 2, 219-229. Blanken PD (1997) Evaporation within and above a boreal aspen forest. Ph.D. Thesis, University of British Columbia, British Columbia, Canada, 179 pp. Bonan GB, Shugart H H (1989) Environmental factors and ecological processes in boreal forest. Annual Review of Ecology and Systematics, 20, 1-28. Bunnell FL, Tait DEN, Flanagan PW, Van Cleve K (1977) Microbial respiration and substrate weight loss - I: a general model of the influences of abiotic variables. Soil Biology and Biochemistry, 9, 33-40. Burton AJ , Zogg GP, Pregitzer KS, Zak DR (1997) Effect of measurement CO2 concentration on sugar maple root respiration. Tree Physiology, 17, 421 - 427. Chen WJ, Black TA, Yang PC, Barr A G , Neumann H H , Nesic Z, Novak M D , Eley J, Ketler RJ, Cuenca R (1998) Effects of climatic variability on the annual carbon sequestration by a boreal aspen forest. Global Change Biology (in press). Chapter 4. Estimating nighttime respiration in an aspen forest 157 Ciais P, Tans PP, Trolier M , White JWC, Francey RJ (1995) A large northern hemisphere terrestrial CO2 sink indicated by the 1 3 C / 1 2 C ratio of atmospheric CO2. Science, 269, 1098-1102. Denmead OT, Raupach M R (1993) Methods for measuring gas transport in agricultural and forest systems. In: Agricultural ecosystem effects on trace gases and global climate change. ASA special publication No. 55. pp. 19-43. Denning AS, Fung IY, Randall D (1995) Latitudinal gradient of atmospheric CO2 due to • seasonal exchange with land biota. Nature, 376, 240-243. Enting IG, Mansbridge JV (1991) Latitudinal distribution of sources and sinks of CO2: results of an inversion study. Tellus, 4 3 B , 156-170. Fang C, Moncrieff JB (1998) A simple and fast technique to measure CO2 profile in soil. Soil Biology and Biochemistry, (in press). Fitzjarrald DR, Moore K E (1990) Mechanisms of nocturnal exchange between the rain forest and the atmosphere. Journal of Geophysical Research, 95, 16839-16850. Goulden M L , Daube BC, Fan S-M, Sutton DJ, Bazzaz A , Munger JW, Wofsy SC (1997) Physiological responses of a black spruce forest to weather. Journal of Geophysical Research, 102, 28987-28996. Goulden M L , Munger JW, Fan S-M, Daube BC, Wofsy SC (1996) Measurements of carbon sequestration by long-term eddy covariance: methods and a critical evaluation of accuracy. Global Change Biology, 2, 169-182. Grace J, Lloyd J, Mclntyre J, Miranda A C , Meir P, Miranda HS, Nobre C, Moncrieff J, Massheder J, Malhi Y , Wright I, Gash J (1995) Carbon dioxide uptake by an undisturbed tropical rain forest in Southwest Amazonia, 1992 to 1993. Science, 270, 778-780. Greco S, Baldocchi DD (1996) Seasonal variations of CO2 and water vapor exchange rates over a temperate deciduous forest. Global Change Biology, 2, 183-197. Jarvis PG, Massheder JM, Hale SE, Moncrieff JB, Rayment M , Scott SL (1997) Seasonal variation of carbon dioxide, water vapour and energy exchanges of a boreal black spruce forest. Journal of Geophysical Research, 102, 28953-28966. Chapter 4. Estimating nighttime respiration in an aspen forest 158 Jury W A , Gardner WR, Gardner W H (1991) Soil Physics. John Wiley & Sons Inc., New York, 328 pp. Lavigne M B , Ryan M G , Anderson DE, Baldocchi DD, Crill P M , Fitzjarrald DR, Goulden M L , Gower ST, Massheder JM, McCaughey JH, Rayment M , Striegl R G (1997) Comparing nocturnal eddy covariance measurements to estimates of ecosystem respiration made by scaling chamber measurements at six coniferous boreal sites. Journal of Geophysical Research, 102, 28977-28985. Lee X (1998) On micrometeorological observations of surface-air exchange over tall vegetation. Agricultural and Forest Meteorology, 91 , 39-49. Mahrt L (1982) Momentum balance of gravity flows. Journal of Atmospheric Sciences, 39, 2701-2711. Mahrt L (1997) Flux sampling errors for aircraft and towers. Journal of Atmospheric and Oceanic Technology (submitted). McCaughey JH, Lafleur P M , Joiner DW, Bartlett PA, Costello A M , Jelinski DE, Ryan M G (1997) Magnitudes and seasonal patterns of energy, water, and carbon exchanges at a boreal young jack pine forest in the BOREAS northern study area. Journal of Geophysical Research , 102, 28997-29007. Mizutani K, Yamanoi K, Ikeda T, Watanabe T (1997) Applicability of the eddy correlation method to measure sensible heat transfer to forest under rainfall conditions. Agricultural and Forest Meteorology, 86, 191-200. Moore CJ (1986) Frequency response corrections for eddy correlation systems. Boundary-Layer Meteorology, 37, 17-35. Norman J M , Kucharik CJ, Gower ST, Baldocchi DD, Crill P M , Rayment M , Savage K, Striegl R G (1997) A comparison of six methods for measuring soil-surface carbon dioxide fluxes. Journal of Geophysical Research, 102, 28771-28777. Norman JM, Garcia R, Verma SB (1992) Soil surface C 0 2 fluxes and the carbon budget of grassland. Journal of Geophysical Research, 97, 18845-18853. Chapter 4. Estimating nighttime respiration in an aspen forest 159 Raich JW, Schlesinger W H (1992) The global carbon dioxide flux in soil respiration and its relationship to vegetation and climate. Tellus, 44B, 81-90. Rayment M B , Jarvis PG (1997) An improved open chamber system for measuring soil CO2 effluxes in the field. Journal of Geophysical Research , 102, 28779-28784. Ruimy A , Jarvis PG, Baldocchi DD, Saugier B (1995) CO2 fluxes over plant canopies and solar radiation: a review. Advances in Ecological Research, 26, 1-63. Russell CA, Voroney RP (1998) Carbon dioxide efflux from the floor of a boreal aspen forest. I. Relationship to environmental variables and estimates of C respired. Canadian Journal of Soil Science, 78, 301-310. Russell CA, Voroney RP, Black TA, Blanken PD, Yang PC (1998) Carbon dioxide efflux from the floor of a boreal aspen forest. II. Evaluation of methods-verification by infra-red analysis of a dynamic closed chamber. Canadian Journal of Soil Science, 78, 311-316. Ryan M G , Lavigne M B , Gower ST (1997) Annual carbon cost of autotrophic respiration in boreal forest ecosystems in relation to species and climate. Journal of Geophysical Research, 102, 28871-28883. Schlentner RE, Van Cleve K (1985) Relationships between CO2 evolution from soil, substrate temperature, and substrate moisture in four mature forest types in interior Alaska. Canadian Journal of Forest Research, 15, 97 -106. Schlesinger W H , (1997) Biogeochemistry: an Analysis of Global Change (2nd ed.). Academic Press, New York. Sellers P, Hall F, Margolis H, Kelly B, Baldocchi D, den Hartog G, Cihlar J, Ryan M G , Goodison B, Crill P, Ranson KJ, Lettenmaier D, Wickland DE (1995) The Boreal Ecosystem-Atmosphere Study (BOREAS): an overview and early results from the 1994 field year. Bulletin of the American Meteorological Society, 76, 1549-1577. Shewchuk SR (1997) Surface mesonet for BOREAS. Journal of Geophysical Research, 102, 29077-29082. Sierra J, Renault P (1995) Oxygen consumption by soil microorganisms as affected by oxygen and carbon dioxide levels. Applied Soil Ecology, 2, 175-184. Chapter 4. Estimating nighttime respiration in an aspen forest 160 Simpson IJ (1996) Trace gas exchange and the validity of similarity theory in the roughness sublayer above forests. Ph. D. thesis, University of Guelph, Ontario, Canada, 205 pp. Suarez DL, Simtinek J (1993) Modelling of carbon dioxide transport and production in soil, 2, Parameter selection, sensitivity analysis, and comparison of model prediction to field data. Water Resources Research, 29, 499-513. Tans PP, Fung IY, Takahashi T (1990) Observational constraints on the global atmospheric C 0 2 budget. Science, 247, 1431-1438. Trumbore SE, Davidson EA, de Camargo PB, Nepstad DC, Martinelli L A (1995) Below-ground cycling of carbon in forests and pastures of eastern Amazonia. Global Biogeochemical Cycles, 9, 515-528. Valentini R, De Angelis P, Matteucci G, Monaco R, Dore S, Scarascia Mugnozza GE (1996) Seasonal net carbon dioxide exchange of a beech forest with the atmosphere. Global Change Biology, 2, 199-208. Whittaker R H , Lekins GE (1975) Primary production: The biosphere and man. In: Primary Productivity of the Biosphere (eds Lieth H, Whittaker, R.H.), pp. 305-328. Springer-Verlag, New York. Wofsy SC, Goulden M L , Munger JW, Fan S-M, Bakwin PS, Daube BC, Bassow SL, Bazzaz FA (1993) Net exchange of C 0 2 in a mid-latitude forest. Science, 260, 1314-1317. Wofsy SC, Harriss RC, Kaplan WR (1988) Carbon dioxide in the atmosphere over the Amazon Basin. Journal of Geophysical Research, 93, 1377-1387. 5. FACTORS AFFECTING CANOPY PHOTOSYNTHESIS OF THE OVERSTORY AND UNDERSTORY OF A BOREAL ASPEN FOREST 5.1 INTRODUCTION Canopy photosynthesis has long been of interest to scientists of many disciplines, and studied using various methods (e.g., Thomas and Hil l 1949; Lemon 1960; Musgrave and Moss 1961; Monteith 1962). Because eddy covariance provides a non-intrusive, temporally-continuous and spatially-averaging flux measurement, it is a highly desirable technique for use in studying canopy photosynthesis and the related processes of respiration and transpiration. Thus, with the development of reliable fast response C O 2 sensors in the 1980's, a number of studies using this method on canopy photosynthesis both for crops and forests have been successfully conducted (e.g., Anderson et al. 1984; Verma et al. 1986; Baldocchi etal. 1987; Fan etal. 1990; Hollinger etal. 1994, 1998; Wofsy etal. 1993; Fan et al. 1995; Grace et al. 1995; Goulden etal. 1996; Valentini et al. 1996; Lindroth et al. 1998). Recently, with the completion of the Boreal Ecosystem-Atmosphere Study (BOREAS), a significant number of papers on canopy photosynthesis or carbon sequestration in the boreal forests have been published- (e.g., Black et al. 1996; Baldocchi et al. 1997; Goulden et al. 1997; Jarvis et al. 1997; McCaughey et al. 1997; and Pattey et al. 1997). These papers discussed measurements of carbon sequestration using the eddy covariance method and the relationships of the carbon sequestration or photosynthesis to environmental factors. 161 Chapter 5. Canopy photosynthesis of an aspen forest 162 Most of the studies mentioned above were conducted in forest canopies without significant understory; thus, no partitioning of photosynthesis was needed. However, in forests with more than a single overstory tree canopy, the leaves are likely to be structurally and physiologically different and acclimated to local environments differing particularly in the flux density and quality of radiation. Therefore, it is desirable to determine the separate response functions of the component species or strata in order to develop a realistic two-layer model of photosynthesis of the forest (Jarvis and Sandford 1986). Since the aspen forest at the BOREAS Old Aspen (OA) site is such a forest with a hazelnut understory, the objectives of this chapter are (1) to obtain the photosynthetic rates of the forest ecosystem, the aspen overstory and the hazelnut understory, (2) to obtain separate response functions for the overstory and understory and (3) to partition the annual carbon sequestration between the overstory and understory. 5.2 MATERIALS AND METHODS 5.2.1 Eddy Flux Measurements Carbon dioxide and energy fluxes were measured at the 39.5-m (above-canopy) and 4-m (within-canopy) heights using the eddy covariance technique on 37-m and 6-m scaffold towers about 40-m apart. Both of the eddy covariance systems consist of a three-dimensional sonic anemometer/thermometer and a closed-path infrared gas analyzer (IRGA) within a temperature-controlled box. Half-hourly fluxes were calculated on-line with coordinate rotation using the procedure of Tanner and Thurtell (1969) and the WPL correction (Webb et Chapter 5. Canopy photosynthesis of an aspen forest 163 al. 1980). Both eddy covariance systems were operated continuously from early October to late November in 1993, and the above-canopy system was operated continuously from mid-February to late September in 1994, while the within-canopy system was operated from April to late September, (see Chapter 2). 5.2.2 Supporting Measurements 5.2.2.1 Leaf area index measurements Leaf area indices (LAIs) of the aspen overstory and hazelnut understory throughout 1994 were measured using a LI-COR Plant Canopy Analyzer (model LAI-2000, LI-COR Inc., Lincoln, NB). The aspen LAI was measured at two locations on the main walk-up tower, and the hazelnut LAI was measured at six locations along a 100-m east-west transect near the tower. The same locations were used throughout the year (Blanken, 1997). LAIs were also corrected for clumping (Chen et al. 1997) and adjusted for the pre-leaf wood area index (Blanken 1997). The forest LAI was then calculated as the sum of the LAIs of the aspen overstory and hazelnut understory. As Chen et al. (1997) noted, leaf emergence began on about April 20 (DOY 110) and senescence on about September 17 (DOY 260). The maximum values of the forest, aspen and hazelnut LAIs were 5.6, 2.3 and 3.3 m leaf m" ground in late July and early August (Fig. 5.1). Since the field experiment terminated on September 20 (DOY 263), the LAI curves were extrapolated to zero for times beyond that, the zero values being September 30 (DOY 273) for the hazelnut and October 3 (DOY 276) for the aspen. Chapter 5. Canopy photosynthesis of an aspen forest 164 Destructive measurements of the hazelnut LAI were made on August 6, 1994. Leaves harvested in three 3 mx3 m plots were immediately weighed on site. The area of about 200 g (fresh weight) of randomly sampled leaves was determined using a plant area meter (LI-3100, LI-COR Inc., Lincoln, NB), and the weight of this sample was determined to an accuracy of 1 Fig. 5.1 Leaf area indices (LAIs) of the aspen overstory and the hazelnut understory at the OA site in 1994. These LAIs were measured using an LAI-2000 Plant Canopy Analyzer. Chapter 5. Canopy photosynthesis of an aspen forest 165 mg using a 500 g balance (Mettler, Princeton, NJ). The LAI of each plot was then calculated from the relationship between the weight and the L A I (Blanken 1997; Chen et al. 1997). Hazelnut LAI measured using the destructive and LI-2000 methods agreed well with each other (insert in Fig. 5.1). 5.2.2.2 Photosynthetic photon flux density measurements Incident photosynthetic photon flux density (PPFD in umol m"2 s~\ denoted as Ql where the subscript 0 indicates incident and the superscript / stands for forest) and reflected PPFD above the canopy were measured using up-facing and down-facing quantum sensors (model 190-SB, LI-COR Inc., Lincoln, NB) at the 33-m height from the main tower and recorded on a datalogger (model 21X, Campbell Scientific Inc., Logan, UT). The within-canopy Ql, where the superscript h stands for hazelnut understory, and reflected PPFD were measured using up-facing and down-facing LI-COR 190-SB quantum sensors on an automatic tram system. This tram system moved horizontally at a speed of 7 cm s"1 along a 65-m path about 4-m above the ground (Blanken 1997; Chen et al. 1997). Using these tram measurements and Ql, a relationship between Ql and the aspen LAI (Z,a) was developed as follows (Blanken 1997): QhQ = 0.5719Q'e -° 4 4 5 0 t " 0 0 (5.1) Ql=Ql-Ql where superscript a denotes the aspen overstory. Missing Ql measurements were replaced using a regression relationship between the Ql measurements on the main tower and those (at the 31-m height) on a BOREAS mesoscale meteorological network (MESONET) tower about Chapter 5. Canopy photosynthesis of an aspen forest 166 100 m away (Shewchuk 1997), and/or linear interpolation. The complete data set of QQ was then obtained using Eq. (5.1). During the growing season, about 20% of Ql reached to the hazelnut understory (Blanken 1997) and less than 1% of Ql reached the forest floor. The absorbed PPFD (Qa) data were calculated from incident PPFD as follows (Chen etal. 1998): Ql = (1 - a, )Ql [1 - e~k°L° + e~k° u°+w°1 (1 - e~k^)] Ql = (l-ah)[(l-af)Qle-k'a'+w')(l-e-k^)] (5.2) Q a =Q f-Qh where ka and kh are extinction coefficients of the aspen and the hazelnut (0.540 and 0.756, respectively, Blanken 1997), Wa is the woody area of the aspen (0.620, Blanken 1997), a/ and at, are the reflectivities of the forest and the hazelnut to the incident PPFD. Qha was about 18-25% of the total absorbed PPFD by the forest during the summer. The reflectivity to Qo was calculated as the ratio of the reflected and the incident PPFD. The reflectivity of the forest cc/ was calculated using the measurements made above the canopy, while the reflectivity of the hazelnut understory eg, was calculated using the tram measurements. Fig. 5.2a shows the diurnal and the seasonal changes of a/. This ratio was about 0.05-0.06 around sunrise and sunset and about 0.04 at noon during the full-leaf period. Seasonally, it was about 0.033 when leaves were young and increased to about 0.04 when the leaf was old. It started to further increase around August 28 (DOY 240, the beginning of the senescence) and reached to about 0.07 when the experiment was ended. It was lower than the summer value of a young jack pine forest (-0.05, McCaughey et al. 1997). The 1996 and 1997 data at the OA site showed that a/of the OA forest reached its maximum value of about Chapter 5. Canopy photosynthesis of an aspen forest 167 0.08 0.06 0.04 0.02 0.08 0.06 0.04 0.02 (a) forest 1 (b) hazelnut + + + + + n I.+ 44 .+ 200 220 240 DOY, 1994 260 Fig. 5.2 The reflectivities of PPFD of the forest (af) and the hazelnut understory (txA) at the OA site in 1994. Minimum values usually occurred at noon. 0.2-0.25 during the winter, a/, showed a similar seasonal trend (Fig. 5.2b) although it was slightly higher than a/. In order to replace missing reflectivity data, a relationship with leaf area index and the time of day was developed using least squares best fitting. For the reflectivity of the forest, leaf area index and the local time explained about 85% of the variance of the 1532 half-hour measurements of the reflectivity in 1994. For the hazelnut, about 79% of the variance of cc/, was explained by the leaf area index and the local time. Chapter 5. Canopy photosynthesis of an aspen forest 168 5.2.2.3 Other measurements Air temperature and saturation deficit above the forest were measured using aspirated platinum resistance bulb thermometers and a dewpoint hygrometer (model M l and D2, respectively, General Instruments, Waterdown, MA) at the 37-m height on the main tower. Within the canopy, air temperature and saturation deficit were measured with a shield thermistor and humidity (model HMP-35C, Vaisala Inc., Woburnm MA) sensor, respectively. Missing air temperature and saturation deficit data above the canopy were replaced with the estimates from the MESONET data (Shewchuk 1997), and the missing temperature at the 4-m height were then replaced using the values calculated from a relationship between air temperatures above and below the overstory. Other supporting measurements made during the BOREAS experiment included a soil temperature profile from the soil surface down to a depth of 1 m, an air temperature profile from the 0.1-m to the 39.5-m height, incident solar radiation above the forest, wind speed and direction (vaned propeller anemometer) above and below the overstory, soil water content (time-domain reflectometry) to a depth of 1.2 m, precipitation (tipping bucket and weighing rain gauge). Chapter 5. Canopy photosynthesis of an aspen forest 169 5.2.3 Analytical procedure 5.2.3 .1 Est imation of photosynthesis ( P ) The rate of photosynthesis in the forest ecosystem (Pe) can be obtained using P. (5-3) where O is the carbon sequestration of the forest defined in Eq. (4.3) and Rsha is the respiration of the soil, hazelnut understory and aspen overstory (ecosystem respiration, Eq. (5.7)). Many authors (e.g., Black et al. 1996; Goulden et al. 1996) refer to Pe as gross ecosystem photosynthesis or gross ecosystem production (GEP). Similarly, the photosynthetic rate of the hazelnut {Ph) is Ph=Rsh+®h (5-4) where is the carbon sequestration of the hazelnut and RSh is the respiration of the soil, hazelnut understory and the aspen trunks below the 4-m height (Eq. (5.9)). <&/, is given by O A =-{Fch + ASah/At + ASJAt + Afxh +fzh) (5.5) where FCh is the eddy flux measured at the 4-m height, ASa/,/At and ASs/At (see next section) are the rates of changes in the CO2 storage in the 0-4 m air column and in the soil column, AfXh is the net horizontal C 0 2 advection, and fzh is the vertical mass flow of CO2 at the 4-m height. Both Afxh and fzh were assumed to be zero. The photosynthetic rate of the aspen (Pa), is then calculated by subtracting Eq. (5.4) from Eq. (5.3) using Eqs. (4.3) and (5.5): Pa=Fch-Fc-ASaJAt + Ra (5.6) Chapter 5. Canopy photosynthesis of an aspen forest 170 where ASaa/At is the rate of change in the CO2 storage in the air column from the 4-m height to the 39.5-m height (ASa/At - ASan/At), and Ra is the respiration rate of the aspen leaves, branches and the trunks from the 4-m height above (Rsha - Rsh). The advective and mass flow terms were neglected because they are generally small during the daytime. Note that changes in soil CO2 storage have no effect on the calculation of Pa. These calculations of photosynthesis (Pe, Pa and Ph) assume that the respiration relationships (Eqs (5.7) and (5.9)) obtained using the nighttime CO2 fluxes give a good approximation of daytime respiration using daytime soil temperatures. This commonly-used procedure (e.g., Hall and Moll 1975) slightly underestimates daytime respiration because it does not include photorespiration occurring in the leaves (Sharkey 1988; Schlesinger 1997). Consequently, the calculated values of photosynthesis (Pe, Pa and Ph) are gross photosynthesis minus photorespiration. These values closely approximate scaled-up leaf-gas-exchange (chamber) measurements of net assimilation or net photosynthesis (Pn) that have been corrected for dark respiration (Rd) (i.e., Pn + Rd) (Whittaker and Marks 1975). (Note, P n - P g - R p - Rd, where Pg is leaf gross photosynthesis and Rp is leaf photorespiration.) Leaf-area based photosynthetic rates were calculated by dividing Pe, Pa and Ph by the forest, aspen overstory and hazelnut understory leaf area indices, respectively. They were related to Qa, air temperature, and saturation deficit as done by Baldocchi (1994). 5.2.3.2 Estimation of respiration (R) Rsha was approximated by (Chapter 4) Chapter 5. Canopy photosynthesis of an aspen forest 171 where Ts is the soil temperature at the 2-cm depth. Since soil moisture remained high during most of the 1994 growing season, it had little effect on Rsha (Chen et al. 1998). ASal At was calculated using (\-M)Rsna (Eq. (2.5) in Chapter 4) where M is the wind function (Chapter 4) M = ^=1 + 0.2385 (5.8) 0.7888^/^+0.4076 where u^. is the friction velocity at the 39.5-m height, is the 8-day running mean of u^. The respiration rate of the soil and hazelnut (Rsh) was approximated by 6.7328 s h l + exp[0.1906(12.8939-7;)] Values of RSh used to obtain this equation were calculated using nighttime Fch measurements with Eq. (5.5) assuming AfXh and fzn were equal to zero. For the bin-averaged data set with a bin width ATS = 0.5 °C, this relationship explained about 88% of the variance which is less than that for Rsha (92%, see Chapter 4). The reason is likely related to the intermittent character of the trunk-space turbulence coupled with short-term advective transport (Chapter 3). Equation (5.9) indicates that respiration rate of the soil and hazelnut understory 2 i asymptotically approaches 6.7 umol m" s" as soil temperature increases and that half of this rate occurs at 12.9 °C. This equation indicates that the Qio coefficient (the ratio of the rate of a process at one temperature to that at a temperature 10 °C lower) is about 3.3 for 5 < 7^  < 15 °C and 2.2 for 10 < Ts < 20 °C. These values are not too different from the expected Qio of 2 and all within the range reported by Raich and Schlesinger (1992) for many terrestrial ecosystems. Chapter 5. Canopy photosynthesis of an aspen forest 172 5.2.3.3 Relationship of photosynthesis to environmental variables The response of photosynthesis to Qa (Pi) for the forest (Pe\), aspen (P„i) and hazelnut (Phi) was expressed using a rectangular hyperbola (Ruimy etal. 1995): P= a Q a P " (5.10) where Qa is Qfa, Qaa or Qha, a is the quantum yield (i.e., the slope dPxjdQa at Qa = 0) and Px in pmol m"2 s"1 is the canopy photosynthetic capacity (i.e., the photosynthetic rate at saturating Qa). For the same reasons discussed in Chapter 4, the values of P (i.e., Pe, Pa and Ph) and Qa were bin averaged using a Qa bin-width of 50 pmol m"2 s"1 for the forest and aspen, and 10 pmol m"2 s"1 for the hazelnut. These bin-averaged values were then used to obtain the parameters a and P„ in Eq. (5.10) using a least squares best fit. To analyze the effect of saturation deficit D measured at the 39.5-m height on photosynthetic rate, the ratio of the values of P (i.e., P & Pa and Ph) to the calculated values using Eq. (5.10) (Pz =P/Pl) was bin-averaged, using a D bin-width of 0.1 kPa, and then related (using a least square best fit) to D by a 2-order polynomial as follows P2=aD2+bD + c (5.11) where Pz is dimensionless and a, b, c are the coefficients obtained in the best-fit. By using the ratio instead of the values of P, the effect of light on photosynthetic rate is assumed to be removed. This is similar to Hollinger et al. (1993) and Sullivan et al. (1996) although they used residues instead of ratios. Similarly, the effect of air temperature T at the 39.5- and 4-m heights on photosynthetic rate was analyzed by relating (using a least square best fit) the ratio of the values of P to the Chapter 5. Canopy photosynthesis of an aspen forest 173 product of the calculated P\ and Pi (P3 -P/(P\P2)) to T, which were bin-averaged using a T bin-width of 2 °C, by a 2-order polynomial P3=dT z+eT + f (5.12) where P 3 is dimensionless and d, e, f are the coefficients obtained in the best-fit. It is assumed that PT, has removed the effects of light and saturation deficit on photosynthetic rate. Because of the interaction amongst the light, air temperature and saturation deficit, Eqs (5.10)-(5.12) are not the actual dependence of photosynthetic rate on Qa, D and T. Nevertheless, their product (P = P1P2P2,) should give a good estimate of photosynthetic rate from Q, D and T (Chen et al. 1998). This simple empirical model was evaluated using the 1994 and 1996 data. 5.3 RESULTS AND DISCUSSION 5.3.1 Typical full-leaf diurnal patterns of photosynthesis of the forest, aspen overstory and hazelnut understory Fig. 5.3 shows the diurnal courses of Fc and FCh at the OA site on July 27, 1994 (DOY 208). It was an almost cloudless day with PPFD reaching a maximum of 1610 umol m"2 s"1 (Fig. 5.5). LAI was at its maximum for the year (Fig. 5.1). CO2 uptake (-Fc) by the forest 2 1 reached a maximum of about 25 umol m" ground s" shortly after noon which was typical for clear days during the summer. Although Fc was strongly negative, Fch remained positive (1-3 umol m"2 s"1), i.e., CO2 flux upward just above the hazelnut understory, during most of the day. This upward flux is largely a result of the CO2 flux from the soil being at its yearly Chapter 5. Canopy photosynthesis of an aspen forest 174 maximum and providing more CO2 than could be taken up photosynthetically by the hazelnut understory. 39.5 m 11 1 1 1 1 1 1 0 4 8 12 16 20 24 CST (hour) Fig. 5.3 Diurnal courses of Fc at the 39.5-m height and Fch at the 4-m height measured at the OA site on July 27 (DOY 208), 1994. Chapter 5. Canopy photosynthesis of an aspen forest 175 Fig. 5.4 shows courses of the photosynthesis by the forest (Pe), the hazelnut understory (Ph) and the aspen overstory (Pa) on July 27, 1994. The values of Pe, Ph and Pa were calculated using Eqs. (5.3)-(5.9) and the values of Fc and Fch shown in Fig. 5.3. ASa/At was mainly important during 6:00-9:00 CST (-3 to -5 umol m2 s"1, Fig. 3.10 and Fig. 3.13), while ASs/At (i.e., RshcM) was about -0.5 to -1.5 umol m"2 s"1 during the daytime hours. Just as in the case of Fc, Pe reached its maximum of about 35 Mmol m"2 s"1 shortly after noon when PPFD reached its maximum. Forest respiratory C 0 2 contributed only about 17% to the C 0 2 CST (hour) Fig. 5.4 Typical daytime photosynthetic rates of the forest, aspen overstory and the hazelnut understory on the same clear day as in Fig. 5.5 (July 27,1994, DOY 208) at the OA site. Chapter 5. Canopy photosynthesis of an aspen forest 176 photosynthesized by the forest. In the case of the hazelnut understory, respiratory CO2 (largely from the soil) far exceeded the photosynthetic CO2. The maximum of Ph was about 5.5 umol m"2 s"1 around noon. Note that photosynthesis by the understory started later in the morning probably because light to the hazelnut was relatively low for early morning hours (Fig. 5.5). o E o o CL 2000 1500 1000 500 ggcj) fl)(D(D(Da)(Da)(j)(D(D-25 Day 208 39.5 m <D 0 0 Q Q m 12 16 CST (hour) 24 Fig. 5.5 Typical diurnal patterns of incident PPFD (Q0) and air temperatures (T) at the 39.5-m and 4-m heights, and saturation deficit (£>) on a clear day in the summer 1994 (July 27, DOY 208) at the OA site. Sunrise was at about 5:00 CST and sunset was at about 21:00 CST on this day. Chapter 5. Canopy photosynthesis of an aspen forest 177 5.3.2 Effects of environmental factors on the rate of photosynthesis of the forest The common environmental factors affecting photosynthesis are PPFD, temperature, saturation deficit, and soil moisture (Larcher 1995). In this section, effects of these variables on the photosynthetic rates of the forest, hazelnut understory and aspen overstory at the O A site were determined using the 1994 data. The applicability of these relationships to the site in 1996 was also investigated. 5.3.2.1 Photosynthetic photon flux density (PPFD) Fig. 5.6 and Table 5.1 show the relationships between the photosynthetic rates (forest, hazelnut understory and aspen overstory) and absorbed PPFD (Q fa , Qha and Q aa) using all the half-hourly values obtained from April 22 (DOY 112) to September 19 (DOY 262) in 1994. Fig. 5.6a shows the half-hour values for the forest as a whole, while Fig. 5.6b shows the bin averages using a bin width AQl = 50 jimol m"2 s"1 in order to eliminate the effect of uneven distribution of data on the regression as discussed in Chapter 4. Forest photosynthesis clearly shows a Michaelis-Menten (rectangular hyperbolic) response to Q fa (Fig. 5.6b). The coefficient of determination (r2) was 0.73 using the half-hourly data and 0.99 using the bin-averaged data (Table 5.1). Forest photosynthetic rate did not reach light-saturation even at Ql = 1600 pmol m"2 s"1 which is the maximum on a clear summer day. This contrasts with the light response curve of other boreal forest ecosystems (e.g. Fan et al. 1995; Baldocchi et al. 1997; Goulden et al. 1997; Jarvis et al. 1997, Hollinger et al. 1998) where the photosynthetic rates reached their canopy photosynthetic capacities. It is, however, similar to Chapter 5. Canopy photosynthesis of an aspen forest 178 many temperate and tropical forests (e.g., Wofsy et al. 1993; Amthor et al. 1994; Hollinger et al. 1994; Baldocchi etal. 1996; Goulden etal. 1996; Valentini etal. 1996). This feature of not reaching its canopy photosynthetic capacity seems to be a characteristic of broadleaf forests. The mean value of the quantum yield a of this ecosystem was about 0.040 mol CO2 mol"1 0 100 200 300 400 0 500 1000 1500 O" (nmol rrf2 s_1) Cf (umol rrf2 s"1) Fig. 5.6 Relationship between the photosynthetic rates (Pe, Ph and Pa) and absorbed PPFD {QJa , Qha and Ql) at the OA site during period of April 22-September 19 (DOY 112-262), 1994. photons obtained using the bin-averaged half-hourly values which is slightly lower than that obtained using daily (24-h) average data (0.045 mol CO2 mol"1 photons, Chen etal. 1998). It is within the range of values for deciduous broadleaf forests reported by Ruimy et al. (1995). Chapter 5. Canopy photosynthesis of an aspen forest 179 It is, however, about three times the values (0.017 mol CO2 mol"1 photons) for a boreal lichen woodland ecosystem (Fan etal. 1995) and slightly lower than the value (0.05 mol CO2 mol"1 photons) for a boreal black spruce forest (Goulden et al. 1997). On a leaf-area basis, the value of a was 0.076 mol CO2 mol"1 photons (Table 5.1) which means that about 13 moles of absorbed quanta were needed to reduce one mole of CO2. This agrees with the values (0.0662-0.0772 mol CO2 mol"1 photons) for Cornus florida and Acer pensylvanicum in the southern Appalachian area measured using leaf chambers (Sullivan et al. 1996). Table 5.1 Parameters in Eq. (5.10) for the forest, aspen overstory and hazelnut understory for the period April 22-September 19 (DOY 112-262), 1994 using bin-averaged PPFD values. The values in brackets are for leaf-area based photosynthetic rates (P/L) and absorbed PPFD (QJL) for the period May 28-September 19 (DOY 148-262), 1994. The second line in the r1 and n columns are the values when the half-hourly data were used. PPFD (jimol m"2 s"1) a (mol CO2 mol"1 photons) (umol m"2 s"1) 2 r n Pel Ql 0.040 (0.076) 52.88 (5.46) 0.99 (0.83) 0.73 33 (30) 4164 Pal Q: 0.040 (0.053) 46.86 (13.07) 0.98 (0.98) 0.67 27 (25) 3893 Phi Q: 0.027 (0.039) 13.37 (4.95) 0.98 (0.32) 0.42 36 (25) 3747 The photosynthetic capacity (Pj of the forest was 52.9 umol m"2 s"1 using bin-averaged half-hourly values (Table 5.1) which is within the range of the values for broadleaf forests Chapter 5. Canopy photosynthesis of an aspen forest 180 (Ruimy et al. 1995). The corresponding values were about 10-15 pmol m"2 s"1 for boreal coniferous forests (e.g. Baldocchi etal. 1997; Fan et al. 1995; Goulden et al. 1997; Jarvis et al. 1997) and about 30 pmol m"2 s"1 for temperate broadleaf forests (Baldocchi et al. 1997; Goulden et al. 1996; Valentini etal. 1996). On a leaf area basis, the photosynthetic capacity was about 5.5 pmol m"2 leaf s"1 (Table 5.1), which is similar to the value of 5.6 pmol m"2 s"1 for Nyssa sylvatica (Sullivan et al. 1996) and in the range of 3-15 pmol m"2 s"1 for deciduous trees (Table 2.4, Larcher 1995). There was almost a linear relationship between hazelnut photosynthesis (Ph) and Qha (Fig. 5.6c). The quantum yield was 0.027 mol C 0 2 mol"1 photons, or 0.039 mol CO2 mol"1 on a leaf-area basis (Table 5.1). This value on a ground area basis is about 68% of the forest value, and also much lower than the leaf-level values of 0.05-0.06 for other understory species in a Scots pine (Pinus sylvestris) plantation (Wedler et al. 1996) and the value of 0.069 for C 3 plants (Farquhar and von Caemmerer 1982). The photosynthetic capacity of the hazelnut understory on a ground-area basis was 13.4 pmol m"2 s"1 which is much lower than that for the forest. The photosynthetic characteristics of the aspen overstory are likely to be similar to those of the forest since it accounted for a large proportion of the photosynthesis of the forest (see Section 5.3.5). Not surprisingly, the aspen overstory exhibited a Michaelis-Menten response. (Fig. 5.6d). The quantum yield was 0.040 mol CO2 mol"1 photons which was similar to the forest value (Table 5.1). On a leaf-area basis, however, it was 0.053 which means that about 19 mol photons are needed to convert one mol of CO2. This is more than twice the theoretical value of 8-10 quanta per molecule of CO2 fixed for C3 plants, but similar to the commonly accepted value of 15-22 for C3 plants (p. 204 and Table 7.1, Jones 1992). The Chapter 5. Canopy photosynthesis of an aspen forest 181 photosynthetic capacity of the aspen overstory was 46.9 pmol m"2 s"1 which is similar to the forest value. On a leaf-area basis, it was 13.07 pmol m"2 s"1 which is within the range (3-15 pmol m"2 s"1) for deciduous trees (Table 2.4, Larcher 1995). 5.3.2.2 Saturation deficit (D) Fig. 5.7 and Table 5.2 show the effect of saturation deficit on the photosynthetic rates of the forest, hazelnut understory and aspen overstory. For the forest and aspen overstory, 1 2 D (39.5 m) (kPa) 0 1 2 3 D (39.5 m) (kPa) 1 2 3 D (4 m) (kPa) Fig. 5.7 Relationship between the bin-averaged ratio of the measured photosynthetic rates (Pe, Ph and Pa) to their estimates from absorbed PPFD (Q^, Qha and Q aa) and saturation deficit D at the OA site during the period of April 22-September 19 (DOY 112-262), 1994. The bin-width AD = 0.1 kPa. The vertical bars are +1 standard deviation. Chapter 5. Canopy photosynthesis of an aspen forest 182 photosynthesis decreased with increasing D (Fig. 5.7a and Fig. 5.7c). Blanken's (1997) finding that canopy conductance at the OA site also decreased significantly with increasing D suggests that the decrease in P was likely the result of stomatal closure (Hollinger et al. 1994; Fan et al. 1995). The best fit 2 n d order polynomial explained about 91% and 89%, respectively, of the variance of the bin-averaged data (Table 5.2), which is significantly lower than that for PPFD. These results resemble those of Fan et al. (1995) for a boreal lichen woodland in Canada and of Hollinger etal. (1994) for a temperate red beech in New Zealand. Some authors have found no response of forest photosynthesis to increasing D (deciduous forest, Verma et al. 1986) or a slight increase with increasing D (boreal old black spruce forest, Jarvis et al. 1997). The 20-30% reduction in photosynthetic rates of the forest and aspen overstory at high saturation deficit (1.5-2.5 kPa, Fig. 5.7a) observed in this study was also observed by Fan et al. (1995). Suyker et al. (1997) found that there was a 32% reduction in the photosynthesis of a boreal fen site in northern Saskatchewan caused by a combination of high temperature (25<r<28 °C) and high saturation deficit (1.7<D<2.2 kPa). In the case of the hazelnut understory the photosynthetic rate did not show a significant dependence on saturation deficit [r2 = 0.1) (Fig. 5.7b and Table 5.2). Chapter 5. Canopy photosynthesis of an aspen forest 183 Table 5.2 Parameters in Eq. (5.11) for the forest, aspen overstory and hazelnut understory for the period of April 22-September 19 (DOY 112-262), 1994 using bin-averaged D values. Saturation deficit (kPa) a (kPa2) b (kPa1) c (dimensionless) 2 r n Pel D (39.5 m) 0.125 -0.509 1.211 0.91 27 Pal D (39.5 m) 0.234 -0.862 1.446 0.89 27 Phi D (4 m) 0.099 -0.250 1.135 0.12 28 Chapter 5. Canopy photosynthesis of an aspen forest 184 5.3.2.3 A i r temperature (T) Effects of air temperature on the photosynthetic rates of the forest, hazelnut understory and aspen overstory are shown in Fig. 5.8 and Table 5.3. In the case of the forest, the best-fit explained about 94% of the variance of the bin-averaged data (Fig. 5.8a and Table 5.3). Very similar results were obtained for the aspen overstory (Fig. 5.8b and Table 5.3). The Fig. 5.8 Relationship between the bin-averaged ratio of the measured photosynthetic rate of the forest to its estimate from Q fa &.D and air temperature T at the OA site during April 22-September 19 (DOY 112-262), 1994. The bin-width AT = 2 °C. The vertical bars are ± 1 standard deviation. photosynthetic rate increased with air temperature, reached its maximum in the range of 14-24 °C, and then decreased very slightly with increasing temperature. This optimal air Chapter 5. Canopy photosynthesis of an aspen forest 185 Table 5.3 Parameters in Eq. (5.12) for the forest, aspen overstory and hazelnut understory for the period of April 22-September 19 (DOY 112-262), 1994 using bin-averaged rvalues. Air temperature (°c) d ( °C 2 ) e CC"1) / (dimensionless) 2 r n Pel T (39.5 m) -0.002 0.078 0.295 0.94 17 Pal T (39.5 m) -0.001 0.064 0.350 0.90 17 Ph3 7/(4 m) -0.003 0.114 0.079 0.85 16 temperature range is similar to the value of 20-25 °C for temperate deciduous trees (Table 2.11, Larcher 1995) and for other boreal forests (e.g., Fan et al. 1995 for the boreal lichen woodland; Baldocchi et al. 1997 for the boreal jack pine forest; Jarvis et al. 1997 for the boreal black spruce forest). When temperature dropped from the optimal range to 0 °C, photosynthetic rate decreased by about 70%. It decreased to about zero when temperature dropped to about -3.5 °C which is similar to the value of -3 °C for deciduous trees (Table 2.11, Larcher 1995). For the hazelnut understory, the best-fit polynomial explained about 85% of the variance of the bin-averaged data (Table 5.3). In contrast to the forest, the optimal temperature was 9-18 °C, (Fig. 5.8b). When temperature was higher, photosynthesis decreased significantly. This is similar to the temperature response of Brachypodium pinnatum, Cartex alba and Carex flacca, the understory species in a Scots pine {Pinus sylvestris) forest (Wedler et al. 1996). The low optimal temperature range is likely a result of the low light intensity (p. 108, Larcher 1995). The photosynthetic rate decreased by about 20% when temperature dropped to 0 °C. Chapter 5. Canopy photosynthesis of an aspen forest 186 5.3.2.4 Cloudiness and other factors Fig. 5.9 and Table 5.4 compare the photosynthesis-light response curves for the forest under different sky conditions. Table 5.4 also shows the results for the hazelnut understory and the aspen overstory. At a given PPFD, photosynthetic rates were much higher during overcast conditions than during clear sky conditions (e.g., about 40% higher at 40 35 30 clear partly cloudy overcast all data "0 200 400 600 800 1000 1200 1400 d (umol nrf2 s_1) 1600 Fig. 5.9 The light response curves of the photosynthetic rate of the forest from April 22 to September 19 (DOY 112-262), 1994 at the OA site in different sky conditions. Q FA ~ 600 pmol m"2 s"1). The photosynthetic rates during partly cloudy conditions were also higher than during clear sky conditions (e.g., about 10% higher at QFA ~ 600 pmol m"2 s"1). Other researchers (Hollinger et al. 1994; Fan et al. 1995; Baldocchi et al. 1997; Goulden et Chapter 5. Canopy photosynthesis of an aspen forest 187 al. 1997) reported similar findings. As shown in Table 5.4, the values of forest quantum yield under clear sky and partly cloudy conditions was very similar (about 0.03 mol C mol"1 absorbed photons). The value under overcast conditions was much higher (about 0.05 mol C mol"1 absorbed photons). The corresponding values for the hazelnut understory were 0.02, 0.02 and 0.03 mol CO2 mol"1 photons, and for the aspen overstory were 0.03, 0.03 and 0.05 mol CO2 mol"1 photons. The values of photosynthetic capacity under different cloudy conditions were also different. For the forest, they were about 58, 111 and 88 pmol m"2 s"1 for clear sky, partly cloudy and overcast, respectively. The corresponding values for the aspen overstory were 52, 89 and 87 pmol m"2 s"1, while they were 15, 31 and 75 pmol m"2 s"1 for the hazelnut understory The reason for the quantum yield and photosynthetic capacity being higher in partly cloudy and overcast conditions is probably that light is more evenly distributed through the overstory and understory canopies. (Under these conditions, there is a higher proportion of incident diffuse light and plant canopies have a lower extinction coefficient for diffuse light.) Another reason for the higher values in overcast conditions is that the saturation deficit is usually lower (Fan et al. 1995). The daytime mean values of D under clear sky (1033 half-hours), partly cloudy (635 half-hours) and overcast conditions (638 half-hours) were 0.93, 0.94 and 0.42 kPa during the summer of 1994 at the OA site. This indicates that there was a much lower chance of water restriction to photosynthesis when it was overcast. The higher r for the hyperbolic light response under overcast conditions (indicating reduced importance of D and T) also support this hypothesis. Chapter 5. Canopy photosynthesis of an aspen forest 188 Table 5.4 Comparison of the light response of photosynthesis of the forest, hazelnut understory and aspen overstory under different sky conditions at the OA site, 1994 Forest Hazelnut Aspen a P- a P o oc P o Clear 0.031 58.36 0.022 15.00 0.032 51.97 Partly cloudy 0.030 110.78 0.024 31.23 0.031 89.26 Overcast 0.050 87.88 0.030 74.86 0.052 86.95 A l l 0.040 52.88 0.027 13.37 0.040 46.86 No clear dependence of photosynthetic rate on ambient C 0 2 concentration was found at the OA site in 1994. This is likely because of the narrow range of concentration encountered. Similarly, little or no effect of soil water content on photosynthesis was found because of the short period of reduced rainfall occurred near the end of the growing season in 1994 (Chen et al. 1998). When the photosynthetic rates were separated into morning and afternoon values, no significant difference between their relationships to absorbed PPFD was found. Differences in photosynthetic rate at a given absorbed PPFD were less than 0.2 pmol m"2 s"1. Chapter 5. Canopy photosynthesis of an aspen forest 189 5.3.3 Water use efficiency of the forest, aspen overstory and hazelnut understory Fig. 5.10 shows the relationships between the photosynthetic rate and the water use for the forest, hazelnut understory and aspen overstory during the summer in 1994. The slope of the linear best fit, which is regarded as the water use efficiency (WUE), was 10.8 mg CO2 g"1 H2O for the forest (4.4 umol mmol"1, Fig. 5.10a). This is very similar to the value of 11.3 mg Fig. 5.10 Relationship between photosynthesis (Pe, Ph and Pa) and evaporation (Ee, Eh and Ea) during the summer in 1994 at the OA site (e = forest ecosystem, h = hazelnut understory and a = aspen overstory). The solid line is the linear best fit. The dot-dashed line is the relationship for a wheat crop (Baldocchi 1994). Chapter 5. Canopy photosynthesis of an aspen forest 190 CO2 g"1 H2O (4.6 pmol mmol'1) for a wheat crop (also a C3 plant) in Oregon (Baldocchi 1994). This value for the OA forest was twice that of the value for a juvenile Douglas-fir forest in coastal BC (2.0 pmol mmol"1, Price and Black 1991). Using the eddy C 0 2 flux (Fc) measured above the OA forest instead of Pe resulted in only a small change in W U E (9.6 mg CO2 g"1 H2O or 3.9 pmol mmol"1, Fig. 5.11). In the case of the aspen overstory, W U E was about 8.1 mg CO2 g"1 H-2O or 3.3 pmol mmol"1 (Fig. 5.10c) which is slightly less than that for the forest as a whole. W U E of the hazelnut understory was only 2.9 mg C 0 2 g H 2 0 or 1.2 pmol mmol"1 which is slightly lower than that (2-8 pmol mmol"1) of the understory in a Scots pine plantation (Wedler et al. 1996). This indicates that the hazelnut understory was significantly less water-use efficient than the aspen overstory at the OA site. Chapter 5. Canopy photosynthesis of an aspen forest 191 Fig. 5.11 Relationship between the eddy C0 2 flux (-Fc) and evaporation (Ee) above the forest during summer in 1994 at the OA site. The solid line is the linear best fit. The dot-dashed line is the relationship for a wheat crop (Baldocchi 1994). Chapter 5. Canopy photosynthesis of an aspen forest 192 5.3.4 Evaluation of a simple empirical model of forest photosynthesis The performance of the model (P =P1P2P3) is shown in Fig. 5.12 and Table 5.5. For the forest as a whole, using all the daytime half-hours over the period April 22-September 19 (DOY 112-263), 1994 at the OA site (n = 4180), modelled Pe explained about 80% of the variance in the measured Pe in which Pe\ (from Qfa) explained about 74%, Pe2 (from D) explained another 5% and Pei (from T) explained only 1%. The remaining 20% of the variance is accounted for by measurement error (in estimating the rates of change in air and soil CO2 storage, and omission of the advection and mass flow terms) and the inadequacy of the model. This coefficient of determination (r2) of 80% for this empirical model is comparable to the value of 72% for a similar empirical model used to estimate photosynthesis of a boreal old black spruce forest (Goulden et al. 1997). By comparison, Leuning et al. (1998) was able to explain 67-71% of the variance in photosynthesis of a winter wheat using Table 5.5 Performance of the simple photosynthesis model (P = for the forest, aspen overstory and hazelnut understory using 1994 data (April 22-September 19). a and b are the intercept and the slope, respectively, of the equation: Pmeas = a + bPmodel. Also shown are the slope, intercept and r2 using 1996 forest photosynthesis data. a (pmol m"2 s"1) b (dimensionless) 2 r Forest 1994 -0.96 1.13 0.80 1996 0.71 0.91 0.73 Aspen 1994 -0.78 1.18 0.76 Hazelnut 1994 0.31 0.83 0.26 Chapter 5. Canopy photosynthesis of an aspen forest 193 a physiological model which included canopy radiation, leaf conductance and photosynthesis sub-models. The small contributions of T and D to the explanation of the variance in photosynthesis agree with the findings of Sullivan et al. (1996), who used chambers to measure photosynthesis of twelve deciduous tree species in the southern Appalachian Mountains. In the case of the aspen overstory, the model (Pa = PalPa2Pa3) explained about 76% of the variance of which Pa\ explained about 69%, Pa2 explained another 5% and Pa3 explained only 1.6%. In the case of the hazelnut understory, the model (Ph = PhXPh2Ph3) explained only 37% of the variance where Ph\ explained about 26%, while Phi and Ph3 explained only 1%. This indicates that the rates of change in air and soil CO2 storage, the advective and mass flow terms played more important roles in accurately estimating the photosynthesis of the understory than in the case of the forest. Fig. 5.13 compares the diurnal patterns of modelled and measured photosynthesis (Pe, Pa and Ph) over a five-day period in 1994. In general, agreement was very good except during midday hours when modelled values for the forest and the aspen overstory were frequently 10-30% lower than measured values. However, the model was able to account for the hour-to-hour variations in photosynthesis (see D O Y 193 in Fig. 5.13). Chapter 5. Canopy photosynthesis of an aspen forest 194 Fig. 5.14 compares modelled and measured daytime mean values of photosynthesis (Pe, Pa and Ph) during 1994. In the case of the forest, values using the forest model (Pe = PeiPe2Pe3> were compared with values obtained by summing the modelled aspen-overstory values (Pa=PaiPa2Pa3) and modelled hazelnut understory values (Pa = PhiPh2Ph3) • Values using the latter procedure were on average less than 1% (0.07 g C m"2 d"1) higher than those obtained using the forest model. The modelled average photosynthesis for the forest and the aspen overstory over the growing season was 5% (0.44 g C m"2 d"1) and 7% (0.36 g C Modelled p (umol m " 2 s" 1) Fig. 5.12 Comparison of the modelled and the measured half-hourly photosynthetic rates (Pe) from April 22 to September 19 (DOY 112-262), 1994 at the OA site. The dashed line is the regression line and the solid line is the one-to-one line. Chapter 5. Canopy photosynthesis of an aspen forest 195 m"2 d"1) lower than the corresponding measured values. In the case of hazelnut understory, the modelled value was about 11% (0.17 g C m"2 d"1) higher. One of the main reasons for the differences between the measured and modelled values is probably the use of averaged absorbed PPFD rather than making the distinction between the sunlit and shaded leaves (Spitters 1986). DOY, 1994 Fig. 5.13 Comparison of the modelled (solid line) and measured (open circles) photosynthetic rates of the forest, the aspen and the hazelnut on the dates of July 9-14,1994 at the OA site. Chapter 5. Canopy photosynthesis of an aspen forest 196 Fig. 5.14 Comparison of the predicted (solid lines) and measured (circles) daytime mean photosynthesis of the forest, aspen overstory and the hazelnut understory in the growing season 1994 at the OA site. The dotted line is the sum of the calculated aspen photosynthesis and hazelnut photosynthesis using the model (Pa = PaiPa2Pa3, and Ph - PMPKIPKS) which is almost indistinguishable from the solid line in panel (a). Chapter 5. Canopy photosynthesis of an aspen forest 197 The model was further tested using an independent data set by comparing its estimates of forest photosynthesis with the values measured during the 1996 growing season at the OA site. Forest photosynthetic rates in 1996 were obtained similarly to those in 1994. The model was able to explain 73% of the variance in half-hourly photosynthetic rates (Fig. 5.15 and Table 5.5), slightly less than for the 1994 data. It slightly overestimated photosynthetic rates in contrast to the slight underestimate by the model for 1994. Precipitation during the 1996 growing season was slightly higher than in 1994 (about 280 and 250 mm, respectively). The Measured and modelled half-hourly P in 1996 at OA T 1 r— 1 1 1 o . <9 I _ Q Q__J Q , 1 1 1 1 0 5 10 15 20 25 30 35 Modelled P (umol rrf2 s"1) F i g . 5.15 Comparison of the half-hourly photosynthetic rates modelled using the 1994 model and measured using the eddy covariance system in 1996 at the O A site. The dashed line is the regression and the solid line is the one-to-one line. Chapter 5. Canopy photosynthesis of an aspen forest 198 main difference between the two years was the colder spring in 1996 resulting in a two-week delay in the leaf emergence (Chen et al. 1998). Chen et al. (1998) reported that the 190 192 DAY 1996 Fig. 5.16 Comparison of the calculated photosynthetic rate estimated using the models and measured using the eddy covariance system in 1996 at the OA site. The line is the values calculated using the models discussed above. The circles are the measured photosynthetic rates. respiration functions (Rsha versus Ts at the 2-cm depth) for both years were very similar. The model generally accounted for the major hour-to-hour variations (Fig. 5.16). However, there were some significant differences between the modelled and measured values during midday hours. Chapter 5. Canopy photosynthesis of an aspen forest 199 5.3.5 Contributions of aspen overstory and hazelnut understory to forest photosynthesis Fig. 5.17 shows the half-hourly values of forest, aspen-overstory and hazelnut-understory photosynthetic rates at the OA site in 1994, This figure shows the contributions of the aspen overstory and hazelnut understory to forest photosynthesis. Photosynthesis began on April 22 \- (c) hazelnut 20 120 150 180 210 DOY 1994 240 270 Fig. 5.17 The half-hourly photosynthetic rates of the forest, the aspen and the hazelnut in 1994 at the O A site (DOY 112) and ended on October 3 (DOY 276). Before May 25 (DOY 145), photosynthesis was very low because of the low leaf area (Fig. 5.1). Midday forest photosynthetic rate Chapter 5. Canopy photosynthesis of an aspen forest 200 increased significantly from about 6 pmol m"2 s"1 on May 26 (DOY 146) to about 30 pmol m" 2 s"1 in about two weeks. After that, it increased slowly and reached a maximum of about 40 2 1 pmol m" s" in late June-early July. This value was comparable to the maximum half-hourly photosynthetic rate of a temperate oak-hickory-pine forest near Oak Ridge, Tennessee, (Baldocchi and Vogel 1996), but it was about 10 pmol m' 2 s"1 higher than that of Harvard forest (Goulden et al. 1996). It was almost double the values for two old black spruce forests (Jarvis etal. 1997; Goulden etal. 1997), and it was about two and half times the values for an old jack pine forest (Baldocchi and Vogel 1996) and a boreal lichen woodland site (Fan et al. 1995). The pattern of half-hourly photosynthetic rates of the aspen overstory during the growing season was similar to that of the forest (Fig. 5.17b). The maximum value was about 35 pmol m"2 s"1. The maximum half-hourly photosynthetic rate of the hazelnut understory was also low (0.5-1 pmol m"2 s"1) during the first 30 days of the growing season and then on May 20 (DOY 141) it increased significantly to 7-10 pmol m"2 s"1 (Fig. 5.17c). It remained in this range until near the end of the growing season. During the first 30 days of the growing season (before May 20, D O Y 141), values of daily average forest photosynthesis were 0.5-2 g C m"2 d"1, of which 30-40% was accounted for by hazelnut photosynthesis (Fig. 5.14). During the remainder of the growing season (approximately the full-leaf period), values were 5-14 g C m"2 d"1, of which 15-20% was accounted for by hazelnut photosynthesis. The decrease in this proportion was not easily related to changes in the proportion of PPFD absorbed by the hazelnut understory (Q*/Qf) which was 18-25% during the growing season. Chapter 5. Canopy photosynthesis of an aspen forest 201 Note that there are some negative values in photosynthesis, especially in the case of the hazelnut understory. Those values mainly occurred around sunrise and sunset when the air shifted from stable to unstable or vice verse. During these periods, unusually high eddy CO2 fluxes (as that shown in Fig. 3.15) were frequently observed. At these times, changes in CO2 storage both in the air column beneath the eddy covariance instruments and in the soil often behaved strangely. In addition, the advective and mass flow terms, which were neglected in the calculation, could also be more important. All these effects resulted in the negative photosynthetic rates shown in Fig. 5.17, which accounted for about 4%, 6% and 15% of the growing-season half-hours in the case of the forest, aspen overstory and hazelnut understory, respectively. Chapter 5. Canopy photosynthesis of an aspen forest 202 Fig. 5.18 shows the course of cumulative photosynthesis of the forest, aspen overstory and hazelnut understory during 1994. Except for the first four weeks and last two weeks of the growing season, photosynthesis of the forest, aspen overstory and hazelnut understory 1200 1000 800 O 600 h 400 200 50 100 150 200 DOY, 1994 250 300 350 Fig. 5.18 The cumulative photosynthesis of the forest, aspen and hazelnut in 1994 at the OA site. was linear to a good approximation. During this stage of approximately constant growth, the rates of forest, aspen-overstory and hazelnut-understory photosynthesis were about 10.4, 8.8 and 1.6 g C m"2 d"1, respectively The annual totals of forest, aspen-overstory and hazelnut-understory photosynthesis were about 1140, 950 and 190 g C m"2, respectively (Fig. 5.18 and Table 5.6). Chapter 5. Canopy photosynthesis of an aspen forest 203 5.3.6 C a r b o n ba lance at the O A site in 1994 In order to calculate the annual carbon balance, daily values of respiration of the forest, aspen overstory and hazelnut understory were estimated. Fig. 5.19 shows the courses of the E o 1000 900 800 700 600 500 400 300 200 100 0 f o r e s t 50 100 150 200 DOY, 1994 250 300 350 Fig. 5.19 Cumulative ecosystem respiration and its partitioning between the aspen overstory and the hazelnut understory at the OA site for 1994. Totals are given in Table 5.6. corresponding cumulative values during 1994. During the leafless period, respiration of the soil, hazelnut understory and aspen trunks below the 4-m height (RSh) accounted for 65-80% of forest respiration with a maximum occurred in January. During the growing season, RSh accounted for 60-70% of forest respiration with a maximum occurring in mid-summer (DOY Chapter 5. Canopy photosynthesis of an aspen forest 204 190-220). As shown in Appendix C, soil respiration accounted for 80-90% of Rsh during the leafless period and 57-71% during the growing season. Using the latter value and the mid-summer value of Rsh (3.5-4.2 g C m"2 d"1) calculated using Eq. (5.9), soil respiration during 2 1 mid summer was estimated to be roughly 2.0-3.1 g C m " d" . This is slightly less than values of 2.7-3.5 g C m"2 d"1 obtained by Savage et al. (1997) using a static closed chamber in a similar aged aspen stand in northern Manitoba in 1994. During 1994, forest respiration was about 920 g C m"2 (Table 5.6), of which about 36% was accounted for by the aspen overstory (leaves, branches and trunks above the 4-m height) and 64% by the soil, hazelnut understory and aspen trunks (0-4 m). Using the RsIRSh ratios in Appendix C, soil respiration was estimated to be about 490 g C m"2 which was about 53% of the annual forest respiration. Chapter 5. Canopy photosynthesis of an aspen forest 205 Table 5.6 Annual carbon fluxes (g C m"2) in the OA forest in 1994, where negative value means the biomass (or soil) losing carbon. The values in bracket are the total uptake (-NEE and -NHE) calculated as the minus sum of Fc and Fch (Chapter 4). Shaded rows indicate the eddy covariance measurement levels. PHOTOSYNTHESIS RESPIRATION NET U P T A K E Forest 1140 ± 4 0 920130 220 ± 7 0 Aspen 950 ± 2 0 330 ± 1 5 * 620 ± 3 5 ll:i/i-liuil V K H - i i ) . inn-ion Soil 0 490 ±70** -490 ± 7 0 Including only aspen biomass from the 4- to 21-m heights. ""Estimated using the ratios in Appendix C (see text). Hazelnut 190 ± 5 0 100 ± 1 2 0 90 ± 1 7 0 Table 5.6 shows that the calculation of carbon sequestration by the forest (Pe - Rsha = 220 ± 70 g C m"2) was almost 10% greater than that measured value, i.e., = 197 ± 3 0 g C m"2 (see Chapter 4). This discrepancy, which is less than 2% of the annual forest photosynthesis, was a result of the overestimation of the daytime loss of CO2 from the storage in the soil (ASs/At) (Section 5.2.3). A similar effect was also observed in the case of CO2 exchange above the hazelnut understory. This illustrates the difficulty in calculating the difference between two large numbers (P and R) even if they have relatively small errors. The analysis of carbon uptake summarized in column 4 in Table 5.6 shows that in 1994 the forest sequestered about 220 g C m"2 (Pe - RSha) while the soil released about 490 g C m"2 Chapter 5. Canopy photosynthesis of an aspen forest 206 as a result of root and heterotrophic respiration. This means that the net uptake (roughly the net photosynthesis) by the vegetation was about 710 g C m"2 of which 87% was accounted for by the aspen overstory and 13% by the hazelnut understory. Making the same assumption as above that heterotrophic respiration (Rh) was equal to root respiration, the net primary productivity (NPP) of the forest was estimated to be roughly 450 g C m"2 (NPP = -NEE + Rh = 220 + 490/2). The estimate by Gower et al. (1997) that the 1994 above-ground NPP in the OA stand was about 350 g C m"2 suggests that below-ground NPP was roughly 100 g C m"2. 5.4 SUMMARY AND C O NCLUSIONS When the forest was in full-leaf, it absorbed 96-97% of the incident PPFD at noon, compared to 94-95% near sunrise and sunset. Corresponding values during the winter with snow cover were 75-80% and 70-75%. During the full-leaf period, the hazelnut understory absorbed 18-25% of the PPFD absorbed by the forest. The photosynthetic rates (P) of the forest (Pe), aspen overstory (Pa) and hazelnut understory (Ph) were modelled as a product of P\, Pi and Pz, where P i is a rectangular hyperbolic function of the absorbed PPFD, Pi is a second order polynomial function of saturation deficit D and P3 is a second order polynomial function of air temperature T. This empirical model explained about 80%, 76% and 26% of the variance in the measured half-hourly photosynthesis for the forest, aspen overstory and hazelnut understory, respectively, in 1994. Absorbed PPFD explained about 74%, 68% and 25% of the variance for the forest, aspen overstory and hazelnut understory, respectively while saturation explained about 5% of the variance in the cases of the forest and the aspen overstory. The model explained 73% of the variance of all growing season half-hourly Pe data obtained at the OA site in 1996. Chapter 5. Canopy photosynthesis of an aspen forest 207 Quantum yield for the forest and aspen overstory was about 0.04 mol CO2 mol"1 photons and 0.02 for the hazelnut understory. The photosynthetic capacity of the forest, aspen overstory and hazelnut understory was 53, 47 and 13 pmol m"2 s"1, respectively. For the same incident PPFD, Pe, Pa and Ph were higher in overcast and partly cloudy conditions than in clear sky conditions. This was probably due to combination of higher D in clear conditions and the higher proportion of diffuse light in overcast and partly cloudy conditions which penetrate the canopy more efficiently than direct light. The optimal temperature range for the photosynthesis of the forest and the aspen overstory was about 14-25 °C, while it was about 9-18 °C for the hazelnut understory. Photosynthesis decreased by 70-80% when temperature dropped to 0 °C. The photosynthetic rate of the forest and the aspen overstory decreased with increasing D , while it remained unchanged for the hazelnut understory. Net exchange of CO2 between the forest and the atmosphere (Fc) and Pe were moderately correlated with forest evaporation (V2 = 0.62 and 0.59, respectively). Water use efficiency of the forest was 4.4 pmol C mmol"1 H2O. The total photosynthesis of the OA forest in 1994 was about 1140 g C m"2, of which 83% was accounted for by the aspen overstory and 17% by the hazelnut understory. Total forest respiration was about 920 g C m"2 of which 490 g C m"2 (53%) was estimated to be soil respiration. Using these estimates, carbon sequestration by the forest in 1994 was about 220 g C m"2, which is slightly higher than the value (200 g C m"2) obtained by directly summing the eddy covariance CO2 flux measurements (Fc). Assuming that a half of the soil respiration was heterotrophic, net primary productivity in 1994 was estimated to be about 450 g C m" . Chapter 5. Canopy photosynthesis of an aspen forest 208 5.5 REFERENCES A m t h o r JS, Gou lden M L , M u n g e r JW , W o f s y S C (1994) Test ing mechanist ic mode l of forest-canopy mass and energy exchange us ing eddy correlat ion: carbon d iox ide and ozone uptake by a m i x e d oak-maple stand. Australian Journal of Plant Physiology, 21, 623-651. Anderson D E , V e r m a SB , Rosenberg N J (1984) Eddy correlat ion measurements of CO2, latent heat, and sensible heat f luxes over a crop surface. Boundary-Layer Meteorology, 29, 263-272. Ba ldocch i D D , V o g e l C , H a l l , B (1997) Seasonal variat ion of carbon d iox ide exchange rates above and be low a boreal jack pine forest. Agricultural and Forest Meteorology, 83, 147-170. Ba ldocch i D D , V o g e l C (1996) A comparat ive study of water vapor, energy and CO2 f l ux densities above and be low a temperate broadleaf and a boreal pine forest. Tree Physiology, 16, 5-16. Ba ldocch i D D (1994) A comparat ive study of mass and energy exchange rates over a c losed C 3 (wheat) and an open C 4 (corn) crop: II. CO2 exchange and water use ef f ic iency. Agricultural and Forest Meteorology, 67, 291-321. Ba ldocch i D D , V e r m a SB , Anderson D E (1987) Canopy photosynthesis and water-use ef f i c iency in a deciduous forest. Journal of Applied Ecology, 24, 251-260. B l a ck T A , den Hartog G , Neumann H H , B l anken P D , Y a n g P C , Russe l l C , Nes i c Z, Lee X , Chen S G , Staebler R, N o v a k M D (1996) A n n u a l cycle of water vapour and carbon d iox ide above a boreal aspen forest. Global Change Biology, 2, 219-229. B l anken P D (1997) Evaporation within and above a boreal aspen forest. Ph .D . Thesis , Un ivers i ty of B r i t i sh C o l u m b i a , B r i t i sh C o l u m b i a , Canada, 179 pp. Chen J M , B l anken P D , B l a ck T A , Gui lbeaul t M , Chen S (1997) Radia t ion regime and canopy architecture in a boreal aspen forest. Agricultural and Forest Meteorology, 86, 107-125. Chapter 5. Canopy photosynthesis of an aspen forest 209 Chen WJ, Black TA, Yang PC, Barr A G , Neumann H H , Nesic Z, Novak M D , Eley J, Ketler RJ, Cuenca R (1998) Effects of climatic variability on the annual carbon sequestration by a boreal aspen forest. Global Change Biology (in press). Fan S-M, Goulden M L , Munger JW, Daube BC, Bakwin PS, Wofsy SC, Amthor JS, Fitzjarrald DR, Moore K E , Moore TR (1995) Environmental control on the photosynthesis and respiration of a boreal lichen woodland: a growing season of whole-ecosystem exchange measurements by eddy correlation. Oecologia, 102, 443-452. Fan S, Wofsy SC, Bakwin PS, Jacob DJ (1990) Atmosphere-biosphere exchange of CO2 and O 3 in the central Amazon forest. Journal of Geophysical Research, 95, 16851-16864. Farquhar GD, von Caemmerer S (1982) Modeling of photosynthetic response to environmental conditions. In: Physiological plant ecology II. (Encyclopedia of plant physiology, NS, Vol. 12B) (eds, Lang OL, Nobel PS, Osmond CB, Zigeler H). Springer, Berlin Heidelberg New York, pp. 549-587. Goulden M L , Daube BC, Fan S-M, Sutton DJ, Bazzaz A, Munger JW, Wofsy SC (1997) Physiological responses of a black spruce forest to weather. Journal of Geophysical Research, 102, 28987-28996. Goulden M L , Munger JW, Fan S-M, Daube BC, Wofsy SC (1996) Measurements of carbon sequestration by long-term eddy covariance: methods and a critical evaluation of accuracy. Global Change Biology, 2, 169-182. Gower ST, Vogel JG, Norman J M , Kucharic CJ, Steele SJ, Stow TK (1997) Carbon distribution and aboveground net primary production in aspen, jack pine, and black spruce stands in Saskatchewan and Manitoba, Canada. Journal of Geophysical Research, 102, 29029-29041. Grace J, Lloyd J, Mclntyre J, Miranda A C , Meir P, Miranda HS, Nobre C, Moncrieff J, Massheder J, Malhi Y , Wright I, Gash J (1995) Carbon dioxide uptake by an undisturbed tropical rain forest in Southwest Amazonia, 1992 to 1993. Science, 270, 778-780. Hall CAS, Moll R (1975) Methods of assessing aquatic primary productivity. In: Primary Productivity of the Biosphere (eds Lieth H, Whittaker, R.H.), Springer-Verlag, New York, pp. 19-54. Chapter 5. Canopy photosynthesis of an aspen forest 210 Hollinger DY, Kelliher FN, Schulze E-D, Bauer G, Arneth A , Byers JN, Hunt JE, McSeveny T M , Kobak KI, Milukova I, Sogatchev A , Tatarinov F, Varlarigin A , Ziegler W, Vygodskaya N N (1998) Forest-atmosphere carbon dioxide exchange in eastern Siberia. Agricultural and Forest Meteorology, 90, 291-306. Hollinger DY, Kelliher FN, Byers JN, Hunt JE, McSeveny T M , Weir PL (1994) Carbon dioxide exchange between an undisturbed old-growth temperate forest and the atmosphere. Ecology, 75, 134-150. Jarvis PG, Sandford AP (1986) Temperate forests. In: Photosynthesis in Contrasting Environments (eds: Baker NR, Long SP). Elsevier Publishers B.V., Amsterdam, pp. 199-236. Jarvis PG, Massheder JM, Hale SE, Moncrieff JB, Rayment M , Scott SL (1997) Seasonal variation of carbon dioxide, water vapour and energy exchanges of a boreal black spruce forest. Journal of Geophysical Research, 102, 28953-28966. Jones H G (1992) Plants and Microclimate (2nd Edition). Cambridge University Press, New York, 428 pp. Larcher W (1995) Physiological Plant Ecology: Ecophysiology and Stress Physiology of Function Groups (3 rdEdition). Springer-Verlag New York Berlin Heidelberg, 506 pp. Lemon ER (1960) Photosynthesis under filed conditions. II. Agronomy Journal, 52, 697-703. Leuning R, Dunin FX, Wang Y-P (1998) A two-leaf model for canopy conductance, photosynthesis and partitioning of available energy. II. Comparison with measurements. Agricultural and Forest Meteorology, 91, 113-125. Lindroth A, Grelle A, Moren AS (1998) Long-term measurements of boreal forest carbon balance reveal large temperature sensitivity. Global Change Biology, 4, 443-450. McCaughey JH, Lafleur P M , Joiner DW, Bartlett PA, Costello A M , Jelinski DE, Ryan M G (1997) Magnitudes and seasonal patterns of energy, water, and carbon exchanges at a boreal young jack pine forest in the BOREAS northern study area. Journal of Geophysical Research, 102, 28997-29007. Chapter 5. Canopy photosynthesis of an aspen forest 211 Monteith, JL (1962) Measurements and interpretation of carbon dioxide fluxes in the field. Netherlands Journal of Agricultural Sciences, 10, 334-346. Moss DN, Musgrave RB, Lemon ER (1961) Photosynthesis under filed conditions. III. Some effects of light, carbon dioxide, temperature, and soil moisture on photosynthesis, respiration, and transpiration of corn. Crop Science, 1, 83-87. Musgrave RB, Moss D N (1961) Photosynthesis under field conditions. 1. A portable, closed system for determining net assimilation and respiration of corn. Crop Science, 1, 37-41. Pattey E, Desjardins RL, St-Amour G (1997) Mass and energy exchange over a black spruce forest during key periods of BOREAS 1994. Journal of Geophysical Research, 102, 28967-28975. Price DT, Black T A (1991) Effects of summertime changes in weather and root-zone soil water storage on canopy C O 2 flux and evapotranspiration of two juvenile Douglas-fir stands. Agricultural and Forest Meteorology; 53, 303-323. Raich JW, Schlesinger W H (1992) The global carbon dioxide flux in soil respiration and its relationship to vegetation and climate. Tellus, 44B, 81-90. Ruimy A , Jarvis PG, Baldocchi DD, Saugier B (1995) C O 2 fluxes over plant canopies and solar radiation: a review. Advances in Ecological Research, 26, 1-63. Savage K, Moore TR, Crill P M (1997) Methane and carbon dioxide exchange between the atmosphere and the northern boreal forest soils. Journal of Geophysical Research, 102, 29279-29288. Schlesinger W H , (1997) Biogeochemistry: An Analysis of Global Change (2nd Edition). Academic Press, New York. Sharkey TD (1988) Estimating the rate of photorespiration in leaves. Physiologia Plantarum, 73,147-152. Shewchuk SR (1997) Surface mesonet for BOREAS. Journal of Geophysical Research, 102, 29077-29082. Chapter 5. Canopy photosynthesis of an aspen forest 212 Spitters CJT (1986) Separating the diffuse and direct component of global radiation and its implications for modelling canopy photosynthesis. Part II. Calculation of canopy photosynthesis. Agricultural and Forest Meteorology, 38, 231-242. Sullivan N H , Bolstad PV, Vose J M (1996) Estimates of net photosynthetic parameters for twelve tree species in mature forests of the southern Appalachians. Tree Physiology, 16, 397-406. Suyker A E , Verma SB, Arkebauer TJ (1997) Season-long measurement of carbon dioxide exchange in a boreal fen. Journal of Geophysical Research, 102, 29021-29028. Tanner CB, Thurtell GW (1969) Anemoclinometer measurements of Reynolds stress and heat transport in the atmospheric boundary layer. Research and Development Technical Report ECOM-66-G22F, University of Wisconsin, Madison, Wisconsin. Thomas M D , Hil l GR (1949) 2. Photosynthesis under field conditions. In: Photosynthesis in Plants, (eds. Franck J and Loomis WE. A monograph of the American Society of Plant Physiologists), The Iowa State College Press, Ames, Iowa. Valentini R, De Angelis P, Matteucci G, Monaco R, Dore S, Scarascia Mugnozza GE (1996) Seasonal net carbon dioxide exchange of a beech forest with the atmosphere. Global Change Biology, 2, 199-208. Verma SB, Baldocchi DD, Anderson DE, Matt DR, Clement RJ (1986) Eddy fluxes of C 0 2 , water vapour and sensible heat over a deciduous forest. Boundary-Layer Meteorology, 36, 71-96. Webb EK, Pearman G l , Leuning R (1980) Correction of flux measurements for density effects due to heat and water vapor transfer. Quarterly Journal of the Royal Meteorological Society, 106, 85 -100. Wedler M , Geyer R, Heindl B, Hahn S, Tenhunen JD (1996) Leaf-level gas exchange and scalling-up of forest understory carbon fixation rates with a "patch-scale" canopy model. Theoretical and Applied Climatology, 53, 145-156. Whittaker R H , Marks PL (1975) Methods of assessing terrestrial productivity. In: Primary productivity of the biosphere (eds Lieth H, Whittaker, R.H.), pp. 55-118. Springer-Verlag, New York. Chapter 5. Canopy photosynthesis of an aspen forest 2 1 3 Wofsy SC, Goulden M L , Munger JW, Fan S-M, Bakwin PS, Daube BC, Bassow SL, Bazzaz FA (1993) Net exchange of C 0 2 in a mid-latitude forest. Science, 260, 1314-1317. Chapter 5. Canopy photosynthesis of an aspen forest 214 6 . S U M M A R Y A N D C O N C L U S I O N S As part of B O R E A S , carbon dioxide fluxes were continuously measured using the eddy covariance technique above (39.5-m height) and within (4-m height) the Old Aspen (OA) stand during October and November 1993 and from February to September 1994, and above the stand from Apr i l to December 1996. Using these measurements, carbon sequestration (net ecosystem productivity) of this forest in 1994 was estimated. Forest photosynthesis and respiration was also calculated, related to environmental variables, and partitioned between the aspen overstory, hazelnut understory and the soil. The major findings of this research are summarized in this chapter. Due to the relative openness of the aspen canopy, significant nighttime radiative cooling of the hazelnut-understory canopy occurred. As a result, the air within the aspen canopy and trunk space was usually stably stratified at night. Stratification was generally strongest in the lower part of the trunk space above the hazelnut understory. This is in contrast to forests with closed canopies (e.g., tropical forests) in which radiative cooling of the upper part of the forest canopy results in vertical drainage of cold air and mixing within the stand. The openness of the aspen-overstory canopy also resulted in unstable stratification (i.e., strong mixing) within the stand during the daytime. The relationships of the variance of the vertical velocity (vv), air temperature (7), CO2 concentration (C) and specific humidity (q) to the stability parameter (t) above the forest followed M O S theory. The relationship between the normalized standard deviation of w and ^ obeyed the 1/3 power law. The normalized standard deviations of T, C and q changed with 215 Chapter 6. Summary and Conclusions 216 ^ following the -1/3 power law under stable conditions and were often invariant with <^  under stable conditions as expected. Within the trunk space, the dependence of the normalized standard deviations of w, Tand q on C, followed MOS theory surprisingly well but with larger scatter than for above-forest standard deviations. The CO2 concentration, however, did not show a clear relationship with C. Evidence strongly suggests that at this site there was no significant enhanced CO2 transport above that based on MOS theory both above the forest and within the trunk space, compared to other sites in the literature. On average, CO2 fluxes above the forest calculated using the MOS flux-gradient relationship (Eq. (2.5)) was about 8% lower than the measured flux at the 39.5-m height during the 1994 growing season. Agreement was better under stable and neutral conditions than that under unstable conditions. However, it should be emphasized that for individual half-hours, MOS theory could over- or underestimate CO2 fluxes by as much as 150%. Although the flux footprint beneath the overstory was small (typically less than 20 m during the daytime and 60 m at night), eddy covariance sensible and latent heat flux measurements at one position were representative of an area extending for at least two tree heights. The same was the case for C 0 2 flux and concentration during the daytime. This suggests that the understory was relatively horizontally homogeneous. Furthermore, this suggests it was valid to estimate the carbon sequestration of the aspen overstory by subtracting the eddy covariance CO2 fluxes measured at the 4-m height from those at the 39.5-m height. At night, half-hourly CO2 concentration and flux measurements exhibited significant horizontal variability. This suggests either a patchiness of CO2 source strength or complex large scale horizontal motion resulting in short-term advection. When averaged over several days, however, they were representative of the above area. Chapter 6. Summary and Conclusions 217 At night, the half-hourly rate of change of the CO2 storage in the air column beneath the above-forest eddy covariance system (ASa/At) occurred mainly in the 0-10 m layer. Thus it could not be well estimated using only the CO2 concentration measured at one level above the forest. It could, however, be well approximated using the concentration at one additional height (2.3 m) within the trunk space. ASa/At was negative from sunrise to about 10:00 CST and negligible in the afternoon. It was one to two times the measured CO2 flux (Fc) early in the night, while it was less than Fc during the rest of the night. On calm nights, ASa/At, however, could not account for the low Fc. Evaluation of the mass flow theory using 1996 data, independent of that used by Lee (1998), showed that the theory performed poorly in explaining the half-hourly energy imbalance values during the nighttime. Similarly for CO2, only in 44% of the cases did calculated vertical velocity (w) behave as predicted, i.e., occurrence of large negative w when Fc was near zero. In examining the relationship between Fc corrected for air CO2 storage (Fc + ASa/At) and the calculated mass flow term, the scatter of the points strongly suggested that mass flow did not account for the low CO2 fluxes on calm nights. Therefore, low nighttime CO2 fluxes were hypothesized to result from the short-term changes in CO2 storage in the air-filled pores of the soil. Porosity in the surface 10-cm organic layer of 90-95%, and 50% in the underlying mineral soil down to the 40-cm depth in the generally well-drained soil at this site support this hypothesis. The fact that CO2 concentrations at the 0.5-m height reached 700-900 pmol mol"1 on calm nights in the summer indicates that considerable CO2 accumulated in the air layer below the 0.5-m height and within the air-filled pores of the soil. Perhaps the best data showing the capacity of the soil to store CO2 were the CO2 fluxes Chapter 6. Summary and Conclusions 218 measured before and after high friction velocity (u^ episodes during the leafless period. CO2 flux above the background respiration flux of 3-4 umol m"2 s"1 can persist for 10-15 hours. Finally, enhanced CO2 efflux during and after rainfall, although debatable, suggested a significant quantity of CO2 is stored in the soil. In order to confirm this hypothesis, it is recommended that measurements of the CO2 concentration profile in the soil are made on an hourly basis (e.g. Fang and Moncrieff 1998) at forest research sites at which eddy covariance measurements of CO2 flux and CO2 concentration profiles in the air column beneath the eddy covariance sensors are being made. The rate of change in soil CO2 storage was estimated by assuming that it was proportional to soil, hazelnut and aspen respiration rate (RSha) with a proportionality factor (1 - M), i.e., ASS jAt = (1 - M)Rsha. Rsha was expressed as a logistic function of the soil temperature at the 2-cm depth and M a rectangular hyperbolic function of the ratio of the current to the running mean of [u^) over the previous 8 days. Then the short-term carbon sequestration (<J>) was obtained by adding the rate of change in soil CO2 storage to the corrected C 0 2 eddy flux Fc+ASa/At. Long-term carbon sequestration, however, was estimated simply by summing the CO2 eddy fluxes because CO2 storage changes were assumed to average to zero over periods of a week or more. This empirical approach required no soil CO2 concentration measurements and physical properties (e.g., CO2 diffusivity and air-filled porosity). It would be desirable to combine a meteorological turbulent diffusion model with a physical model of soil CO2 diffusion and transport like that of Simunek and Suarez (1993). Chapter 6. Summary and Conclusions 219 Photosynthetic rates (P) of the forest ecosystem {Pe), aspen overstory (Pa) and hazelnut understory (Ph) were modelled as a product of P i , P 2 and P 3 . P i is a rectangular hyperbolic function of the absorbed photosynthetic photon flux density (Qa), P 2 is a second order polynomial function of saturation deficit D and P 3 is a second order polynomial function of air temperature T. This empirical model explained about 80%, 76% and 26% of the variance in the measured half-hourly photosynthesis for the forest, aspen overstory and hazelnut understory, respectively, in 1994. Qa explained about 74%, 68% and 25% of the variance for the forest, aspen overstory and hazelnut understory, respectively while saturation explained about 5% of the variance in the cases of the forest and the aspen overstory. The model explained 73% of the variance of all growing season half-hourly Pe data obtained at the OA site in 1996. Quantum yield was about 0.04 mol C 0 2 mol"1 photons for the forest and the aspen overstory, while it was about 0.02 mol C 0 2 mol"1 photons for the hazelnut. The photosynthetic capacities of the forest, aspen overstory and hazelnut understory were about 53, 47 and 13 pmol m"2 s"1, respectively. For the same incident PPFD, Pe, Pa and Ph were higher in overcast and partly cloudy conditions than in clear sky conditions. This was probably due to a combination of higher D in clear conditions and the higher proportion of diffuse light in overcast and partly cloudy conditions which penetrates the canopy more efficiently than direct light. The optimal temperature range for the photosynthesis of the forest and the aspen overstory was about 14-25 °C, while it was about 9-18 °C for the hazelnut understory. Photosynthesis decreased by 70-80% when temperature dropped to 0 °C. The photosynthetic rate of the forest and the aspen overstory decreased with increasing D, while it remained unchanged for the hazelnut understory. Blanken's (1997) finding that Chapter 6. Summary and Conclusions 220 canopy conductance of the OA forest also decreased significantly with increasing D suggests that the decrease in P was likely the result of stomatal closure. Several authors (Hollinger et al. 1994; Fan et al. 1995; Suyker et al. 1997) also found that photosynthesis decreased with increasing D, while some authors have found no response of forest photosynthesis to increasing/) (Verma etal. 1986) or a slight increase with increasing/) (Jarvis etal. 1997). Net exchange of CO2 between the forest and the atmosphere (Fc) and Pe were moderately correlated with forest evaporation (r2 = 0.62 and 0.59, respectively). Water use efficiency of the forest was 4.4 p.mol C mmol"1 H2O. This value was very similar to that of 5 pmol C mmol"1 H2O, which have been found for both a corn (C4 plant) and a wheat (C3 plant) crop in Oregon (Baldocchi 1994). The 1994 annual total photosynthesis of the OA forest was about 1140 g C m" , of which 83% was attributed to the aspen overstory and 17% to the hazelnut understory. Total forest respiration was about 920 g C m"2 of which 490 g C m"2 (53%) was estimated to be soil respiration. Using these estimates, carbon sequestration by the forest in 1994 was about 220 g C m"2, which is slightly higher than the value (200 g C m"2) obtained by directly summing the eddy covariance CO2 flux measurements. If this value is representative of all Canadian boreal deciduous forests, which occupy about 2 x l 0 5 km 2 (Porkland et al. 1991), then their total annual carbon sequestration would be roughly 40 Tg C (Chen et al. 1998). This accounts for about 2% of the current global missing carbon sink (about 1.9 Pg C y"1). Assuming that a half of soil respiration was heterotrophic, net primary productivity in 1994 was estimated to be 450 g C m" . The forest carbon sequestration estimate in this thesis was about 90-100 g C m"2 higher than using the approach in which calm-night fluxes were replaced by estimates based on nighttime fluxes measured during high wind speed conditions Chapter 6. Summary and Conclusions 221 (Black et al. 1996). This difference has a major implication for estimating the carbon balance of the boreal forest. A similar difference (50 g C m"2) changed a boreal old black spruce forest from a weak carbon sink to a weak carbon source (Goulden et al. 1998). If this underestimate of 90 g C m"2 were a common error in tower-based carbon-balance research in the boreal forest, which covers about 12-14.7xl06 km 2 (Whittaker and Likens 1975; Bonan and Shugart 1989), then annual carbon sequestration by this forest would be underestimated by roughly 1 Pg C. Chapter 6. Summary and Conclusions 222 6.1 REFERENCES Baldocchi DD (1994) A comparative study of mass and energy exchange rates over a closed C 3 (wheat) and an open C 4 (corn) crop: II. CO2 exchange and water use efficiency. Agricultural and Forest Meteorology, 67, 291-321. Black TA, den Hartog G, Neumann H H , Blanken PD, Yang PC, Russell C, Nesic Z, Lee X , Chen SG, Staebler R, Novak M D (1996) Annual cycle of water vapour and carbon dioxide above a boreal aspen forest. Global Change Biology, 2, 219-229. Blanken PD (1997) Evaporation within and above a boreal aspen forest. Ph. D. Thesis, University of British Columbia, British Columbia, Canada, 179 pp. Bonan GB, Shugart H H (1989) Environmental factors and ecological processes in boreal forest. Annual Review of Ecology and Systematics, 20, 1-28. Chen WJ, Black TA, Yang PC, Barr A G , Neumann H H , Nesic Z, Novak M D , Eley J, Ketler RJ, Cuenca R (1998) Effects of climatic variability on the annual carbon sequestration by a boreal aspen forest. Global Change Biology (in press). Fan S-M, Goulden M L , Munger JW, Daube BC, Bakwin PS, Wofsy SC, Amthor JS, Fitzjarrald DR, Moore K E , Moore TR (1995) Environmental control on the photosynthesis and respiration of a boreal lichen woodland: a growing season of whole-ecosystem exchange measurements by eddy correlation. Oecologia, 102, 443-452. Goulden M L , Wofsy SC, Harden JW, Trumbore SE, Crill P M , Gower ST, Fries T, Daube BC, Fan S-M, Sutton DJ, Bazzez A , Munger JW (1998) Sensitivity of boreal forest carbon balance to soil thaw. Science, 279, 214-217. Hollinger DY, Kelliher FN, Byers JN, Hunt JE, McSeveny T M , Weir PL (1994) Carbon dioxide exchange between an undisturbed old-growth temperate forest and the atmosphere. Ecology, 75,134-150. Jarvis PG, Massheder J M , Hale SE, Moncrieff JB, Rayment M , Scott SL (1997) Seasonal variation of carbon dioxide, water vapour and energy exchanges of a boreal black spruce forest. Journal of Geophysical Research, 102, 28953-28966. Chapter 6. Summary and Conclusions 223 Lee X (1998) On micrometeorological observations of surface-air exchange over tall vegetation. Agricultural and Forest Meteorology, 91 , 39-49. Porkland H, Palko S, Jowe J (1991) The use of remote sensing in producing the National Atlas of Canada. In: Geographic Information System Seminar, 25-26 November 1991. Canadian Institute of Surveying and Mapping, Ottawa. Simunek J, Suarez DL (1993) Modelling of carbon dioxide transport and production in soil, 1, model development. Water Resources Research, 29, 487-497. Suyker AE , Verma SB, Arkebauer TJ (1997) Season-long measurement of carbon dioxide exchange in a boreal fen. Journal of Geophysical Research, 102, 29021-29028. Verma SB, Baldocchi DD, Anderson DE, Matt DR, Clement RJ (1986) Eddy fluxes of C 0 2 , water vapour and sensible heat over a deciduous forest. Boundary-Layer Meteorology, 36, 71-96. Whittaker R H , Lekins GE (1975) Primary production: The biosphere and man. In: Primary Productivity of the Biosphere (eds Lieth H, Whittaker, R.H.), pp. 305-328. Springer-Verlag, New York. Chapter 6. Summary and Conclusions 224 Appendices 225 APPENDIX A PROFILES OF BULK DENSITY OF THE SOIL AT THE OA SITE Fig. A . l shows the bulk density profile of the Luvisolic soil measured at the OA site in 1994. It indicates that the porosity of the first 10-cm was about 90 - 95%, and dropped to about 50% at the 40-cm depth. 0.1 0.4 0.5 1 1 o 1 1 1 1 1 1 o - o -o _ o o ' 1 1 1 1 1 1 o 1 ;l I I 1 1 1 1 1 ' 1 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 Qb (Mg m"3) Fig. A . l Profile of the bulk density measured at the OA site in 1994. Appendices 226 APPENDIX B NIGHTTIME COSPECTRA OF W-T AND W-%C The following figures show the cospectra of w-T and w-%c on 12 calm nights at the OA site in 1996. The data used here were in the period of 21:00-23:00 CST. Subsets consisted of 214-point (about 13 minutes) were sampled using a symmetric Hanning window. There 0.6 i i i !\ . (a) DOY 162 0.4 0.2 -SV^ / V V v - - T ^ r 0 i i i 0.6 I i i (b) DOY 173 0.4 - / . A . 0.2 . / v < V ' / V — > ^ J~~=" . 0 i i i i 1 — ~ V* (c) DOY 183 JV/V\ W x ^ 0 -1 \ \ i ' M A J V I • , , , , , , , l . — I . — .1 . — . . . . . . . I 10"3 10~2 10"1 10° 101 r(Hz) Fig. B.l Comparison of normalized (by the area under the lines) cospectra of w and T (solid line) and w and cc (the dot-dashed line) measured at 39.5-m height at the OA site during period of 21:00-23:00 CST on June 10 (DOY 162), June 21 (DOY 163) and July 1 (DOY 183), 1996. The values of on those nights were around 0.1 m s"1. were no overlaps between these subsets. These subsets were linearly detrended and then were analyzed using discrete Fourier transform. Appendices 227 10~3 10"2 10"1 10° 101 f(Hz) Fig. B.2 Same as in Fig. B. except on July 19 (DOY 201), June 27 (DOY 207) and August 27 (DOY 240), 1996. Appendices 228 Fig. B.3 Same as in Fig. B ; except on September 10 (DOY 254), October 8 (DOY 282) and October 16 (DOY 290), 1996. Fig. B.4 Same as in Fig. B. except on October 20 (DOY 294), October 31 (DOY 305) and November 11 (DOY 316), 1996. Appendices 230 APPENDIX C DERIVATIONS OF SOME USEFUL RATIOS OF RESPIRATION FOR THE OA SITE Using Mike Ryan's OA chamber measurements (pers. comm. 1997), during the 1994 growing season, aspen-stem respiration accounted for about 26% of above-ground-aspen respiration (Raa), and hazelnut respiration (Rh) accounted for about 36% of the above-ground plant respiration (Rh + Raa)> i-e., Rh l.(Rh + Raa) ~ 0-36. Thus, the respiration of the aspen trunks below the 4-m height accounted for about 5% (0.26x4/21.5) of Raa, in turn, the respiration of the aspen overstory above the 4-m height (Ra), which is the difference of the respiration of the soil, hazelnut understory and aspen overstory (RSha) and the respiration of the soil, hazelnut understory and aspen trunks below the 4-m height (Rsh) (i.e., Ra = Rsha -RSh), accounted for about 95% of Raa- In other words, Raa is about 1.05 times Ra (Raa = \.05Ra). Furthermore, hazelnut-stem respiration was about 11% of aspen-stem respiration. Assuming the ratio of hazelnut-stem respiration to aspen-stem respiration remained the same during the leafless period, then R 2 1 - 5 ~ 4 P = 0.81*flfl or = 0.11xl.23£ a = 0.14Ra As discussed in Chapter 5, Rsh = (0.65 to 0.80)Rsha, thus, Ra = (0.20 to 035)Rsha, Rh = 0.14^, = (0-03 to 0.05)^ t e / . ^ (0.25 to 0.43)^. •'• Rs = Rsha ~ R h ~ Raa = Rsha - (0.03 to 0.05)Rsha - (0.25 to 0 . 4 3 ) ^ = (0.52 to 0.72)^ t e v Rsh = (0.65 to 0.80)/^ „ ,0.52 0.72. „ •'•*' = <a65 " W * * (0.80 to 0.90)i?, •sh During the growing season, - ^ — - 0 . 3 6 , Rh+Ra, ^ . = 1.05^,, and Ra = (0.30 to 0A0)Rsha :.Rh = 0.S6Raa = 0 .56xl .05£ a = 0.59 x (0.30 to 0.40)^ t o = (0.18 to 0.24)^, and ^.=1.054, = 1.05 x (0.30 to 0.40)^ t o = (0.32 to 0.42)^ t e. •'• Rs = Rsha ~ R h ~ Ka = Rsha - (0.18 to 024)Rsha - (0.32 to 0.42)*, t a = (0.34 to 0.50)^ t e v Rsh = (0.60 to 0.70)/?,ta „ ,0.34 0.50. .-. R = ( to )R,h s 0.60 0.70 s h = (0.57 to 0.71)3 'sh These ratios are summarized in the following table. Appendices 232 Table C.l Some useful ratios of the respiration at the OA site in 1994 R s / R s k a R h / R s h a R s h l R s h a R J R a a Leafless 0.80-0.90 0.52-0.72 0.03-0.05 0.65-0.80 0.11 Full-leaf 0.57-0.71 0.34-0.50 0.18-0.24 0.60-0.70 0.36 Appendices 233 Fig. D.l Net radiation, sensible and latent heat fluxes measured above the OA forest with snow cover on the ground from December 11 (DOY 346) to December 17 (DOY 352), 1996. Appendices 234 Fig. D.2 Net radiation, sensible and latent heat fluxes measured above the OA forest without snow cover on the ground from October 27 (DOY 301) to November 1 (DOY 306), 1996. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0099326/manifest

Comment

Related Items