Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Cloud mapping from the earth’s surface using an infrared radiance contrast technique Hertzman, Owen 1979

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


831-UBC_1979_A6_7 H37.pdf [ 7.25MB ]
JSON: 831-1.0094533.json
JSON-LD: 831-1.0094533-ld.json
RDF/XML (Pretty): 831-1.0094533-rdf.xml
RDF/JSON: 831-1.0094533-rdf.json
Turtle: 831-1.0094533-turtle.txt
N-Triples: 831-1.0094533-rdf-ntriples.txt
Original Record: 831-1.0094533-source.json
Full Text

Full Text

CLOUD MAPPING FROM THE EARTH'S SURFACE USING AN INFRARED RADIANCE CONTRAST TECHNIQUE by OWEN HERTZMAN B.A.Sc, University of British Columbia, 1973 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Dept. of Geography) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA July, 1979 0 Owen Hertzman, 1979 In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f t h e r e q u i r e m e n t s f o r a n a d v a n c e d d e g r e e a t t h e U n i v e r s i t y o f B r i t i s h C o l u m b i a , I a g r e e t h a t t h e L i b r a r y s h a l l m a k e i t f r e e l y a v a i l a b l e f o r r e f e r e n c e a n d s t u d y . I f u r t h e r a g r e e t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y p u r p o s e s may be g r a n t e d by t h e H e a d o f my D e p a r t m e n t o r by h i s r e p r e s e n t a t i v e s . I t i s u n d e r s t o o d t h a t c o p y i n g o r p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l n o t be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n . D e p a r t m e n t o f The U n i v e r s i t y o f B r i t i s h C o l u m b i a 2075 Wesbrook Place Vancouver, Canada V6T 1W5 i i ABSTRACT The assumptions and performance of an infrared radiance contrast method for mapping clouds from the earth's surface are evaluated, with a view to using the technique for objective cloud observation. The sky i s observed using a narrow view infrared window radiometer and a scanning mirror system. The radiometer signal is discriminated into 3 cloud categories (low, middle and high) and clear sky, using the assumption that cloudbase radiance decreases unambiguously with height. The discrimination radiances are obtained by modelling clouds represen-tative of marginal conditions, in a range of atmospheres. A multi-layer wavenumber specific infrared radiative transfer model i s used. Once discriminated, the radiance data are mapped in polar coor-dinates and compared with simultaneous all-sky photographs both quali-tatively and quantitatively. The latter involves gridding both maps and photos and comparing them using contingency matrices. Cloud amount, integrated by height class over the sky hemisphere, is also used for map/photo comparison. Cloud maps for 5 days in July 1978 indicate the mapping technique shows promise under a wide range of sky conditions (including cirrus, cumulus and layered cloud situations) but that certain problems remain to be solved. On some days large map errors may occur because of rapid changes in the clear sky radiance (which affect the values of the discrimi-nation radiances). The quality of the maps degrades rapidly with time after their creation because of the effects of cloud motions during the scanning of the sky. This loss of quality does not in general make the maps useless for synoptic analysis purposes, but the time interval at which they must be made for such use is about 20 minutes, 1/3 the current i i i standard hourly interval for network cloud observations. The only rejected assumption of the technique (as stated in Werner (1972, 1973a,b)) i s that clear sky radiances need only be computed twice a year for a given location. Changes in water vapour mixing ratio over time scales from hours to weeks cause changes in the clear sky radiance over these same scales. To adequately map high and middle clouds, and improve the mapping of low clouds, these changes must be included in the mapping methodology. Some options to implement this suggestion are examined here, others are suggested for future work. iv TABLE OF CONTENTS Page ABSTRACT i i TABLE OF CONTENTS iv LIST OF FIGURES ix LIST OF TABLES x i LIST OF SLIDES x i i LIST OF SYMBOLS x i i i ACKNOWLEDGEMENTS xv CHAPTER ONE - PROBLEM AND OBJECTIVES 1 1.1 Purpose 1 1.2 Rationale 2 1.3 The Problem 3 CHAPTER TWO - THEORY AND METHODOLOGY 4 2.1 Theory 4 2.1.1 Overview 4 2.1.2 Radiance Equations 5 2.1.3 Continuum Absorption Parameters 8 2.1.4 Cloud and Aerosol Formulation 8 2.1.5 Geometry 10 2.2 Methodology 10 2.2.1 Constraints 10 2.2.2 Map Verification 12 2.2.3 Mapping Techniques 16 CHAPTER THREE - SITE, INSTRUMENTATION AND DATA ACQUISITION 18 3.1 Site 18 V 3.1.1 C r i t e r i a 3.1.2 Buchanan Tower L o c a t i o n 3.2 Instrumentation and I t s Deployment 3.2.1 Sky Photography 3.2.2 Barnes PRT-5 Radiometer System The Radiometer The R e f l e c t i n g System The Recording System Experimental Set Up 3.3 Data A c q u i s i t i o n and Proces s i n g 3.3.1 Temporal Considerations 3.3.2 S p a t i a l R e s o l u t i o n 3.3.3 Reduction of Radiometer Data 3.3.4 Other Data Reduction 3.3.5 E r r o r A n a l y s i s M i r r o r Radiometer and Recorder Radiance Traces D i s c r i m i n a t i o n Radiances Other 3.4 The Werner Experiment CHAPTER FOUR - RESULTS—CLEAR SKY (BACKGROUND) 4.1 Ratio n a l e 4.2 Time V a r i a t i o n s of N o 4.2.1 Within a Day 4.2.2 D i u r n a l and Day to Day 4.2.3 Seasonal Page 18 20 22 22 25 25 26 29 29 30 30 33 34 35 36 36 37 37 37 40 40 RADIANCE—N 42 o 42 42 42 43 46 v i Page 4.2.4 I m p l i c a t i o n s 49 4.3 A n g u l a r Dependence of N Q 49 4.3.1 M o d e l l e d N q R e s u l t s — J u l y 1978 50 4.3.2 Measured N R e s u l t s — J u l y 1978 56 o J 4.4 Comparison o f M o d e l l e d and Measured N q V a l u e s 56 4.5 S e n s i t i v i t y A n a l y s i s o f N q t o A t m o s p h e r i c F a c t o r s 58 4.5.1 Water Vapour M i x i n g R a t i o (q) 58 4.5.2 Temperature (T) 60 4.5.3 Ozone 60 4.5.4 Na s c e n t C l o u d 63 4.5.5 Urban E f f e c t s 64 4.5.6 E l e v a t i o n , Near t h e S u r f a c e 68 4.6 Summary 68 CHAPTER FIVE - RESULTS—CLOUDY SKIES 70 5.1 C o n t e x t f o r C l o u d M o d e l l i n g 70 5.1.1 M e t h o d o l o g y 70 5.1.2 M o d e l l e d C l o u d s 75 5.2 M o d e l l e d C l o u d b a s e R a d i a n c e s (N , ) 75 zb 5.2.1 Low and M a r g i n a l Low/Middle C l o u d s 75 5.2.2 H i g h and M a r g i n a l H i g h / M i d d l e C l o u d s . 79 5.3 M o d e l l e d S u r f a c e R a d i a n c e s (N ) 79 s 5.3.1 Low and M a r g i n a l Low/Middle C l o u d s 79 5.3.2 H i g h and M a r g i n a l H i g h / M i d d l e C l o u d s 81 5.4 D e t e r m i n a t i o n o f D i s c r i m i n a t i o n R a d i a n c e s 81 5.4.1 N Q 1 — C l e a r S k y / H i g h C l o u d Boundary R a d i a n c e 81 5.4.2 N 1 2 — H i g h / M i d d l e C l o u d Boundary R a d i a n c e 82 5.4.3 N„_—Middle/Low C l o u d Boundary R a d i a n c e 84 v i i Page 5.5 Maps for Skies Dominated by High (Cirrus) Clouds 86 5.5.1 Effect of N Variations 86 o 5.5.2 Case Studies 92 108 B 92 108 F 96 107 I 98 5.5.3 Comparisons of Cloud Integrated by Class (High, Middle, Low) 99 5.6 Maps for Skies Dominated by Low and Middle Clouds 102 5.6.1 Map Variation With Changes in Clear Sky Radiance (N ) 103 o 5.6.2 Map Variation With Changes in Method 107 5.6.3 Case Studies 112 106 F 112 111 G 112 111 D 114 112 D 114 112 E 115 5.6.4 Comparisons of Cloud Integrated by Class (High, Middle, Low) 115 5.6.5 'Maximum' Low Cloud Radiance 120 5.7 A Time Series of Maps 121 5.8 Overall Performance of the Maps 125 CHAPTER SIX - CONCLUSIONS 127 6.1 Map Quality 127 6.2 Validity of Technique Assumptions 129 6.3 Suggestions for Future Work 130 BIBLIOGRAPHY 133 v i i i Page APPENDIX A - ADDITIONAL TABULATED INFORMATION FOR INTERPRETATION OF MODELLED RESULTS 135 APPENDIX B - SYNOPTIC CONDITIONS 138 B.l August 8-15, 1977 138 B.2 July 11-14 and 24-27, 1978 138 i x LIST OF FIGURES Page 3.1 Experimental Location (Mesoscale) 19 3.2 Experimental Location (View of Site) 21 3.3 View of Radiometer System 23 3.4 View of Photographic System 23 3.5 Experimental Location (Synoptic Scale) 24 3.6 Spectral Response of Barnes PRT-5 27 3.7 Radiometer System—Elevation View 31 4.1 Clear Sky Radiance Time Series at Port Hardy (YZT)—August 1977 44 4.2 Clear Sky Radiance P r o f i l e Time Series 47 4.3 Seasonal Changes i n Modelled and Measured Clear Sky Radiance Versus Zenith Angle 48 4.4 Clear Sky Radiance Versus Zenith Angle—Day 106 51 4.5 " —Day 107 52 4.6 " —Day 108 53 4.7 " —Day 111 54 4.8 " —Day 112 55 4.9 Measured Limb Brightening Parameter (6) Versus Measured Near Zenith Clear Sky Radiance (N q) 57 4.10 Dependence of Modelled N q on 59 4.11 Dependence of Modelled N q on T^ 61 4.12 Ozone Mixing Ratio Versus Atmospheric Pressure 62 4.13 Azimuth Change i n N q Versus Zenith Angle 64 4.14 Synthetic St. Louis Area Clear Sky Radiances 66 5.1 Discrimination Radiances Versus Zenith Angle—Day 107 71 5.2 Discrimination Radiances Versus Zenith Angle—Day 112 72 X 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13 5.14 5. 15 N^^ Versus T (Modelled Atmospheres) Modelled Cloud Radiances (N) Versus Clear Sky Radiances (N ) o Werner D i s c r i m i n a t i o n Radiances Versus Z e n i t h Angle Map 107 IB Map 107 I J Map 108 BA Map 108 FA b) Map 107 IBS b) Map 108 BAS b) Map 108 FAS Comparisons of T o t a l Cloud Integrated by Height C l a s s — D a y 107 Comparisons of T o t a l Cloud Integrated by Height C l a s s — D a y 108 Map 112 EB-A Map 112 DB-A Map 106 F-C Map 111 D-C b) Map 112 ED-A b) Map 112 DD-a b) Map 106 F-B b) Map 111 G-C Comparisons of T o t a l Cloud Integrated by Height C l a s s — D a y 106 Comparisons of T o t a l Cloud Integrated by Height C l a s s — D a y 111 Comparisons of T o t a l Cloud Integrated by Height C l a s s — D a y 112 Map 112 BB-C Map 112 HE-C b) Map 112 FE-C d) Map 112 JE-C Page 74 83 86 90 91 93 94 100 101 105 106 111 113 116 117 118 122 123 x i LIST OF TABLES Page 2.1 Water Vapour Continuum Parameters for Infrared Radiance Calculation 9 3.1 Probable Absolute and Relat ive Error in the Radiance Traces at Different Radiance Values 38 3.2 Errors in Discrimination Radiances 38 4.1 Modelled Radiance Changes Over 12 Hour Intervals (Ordered) .45 4.2 Measured Clear Sky Radiance N on Days When Cloud Maps Drawn 45 5.1 Modelled Clouds 76 5.2 Surface and Cloudbase Radiances for Marginal Low/Middle Cloud Cases 77 5.3 Cloudbase and Surface Downwelling Radiances as Functions of Cloudbase Height and Zenith Angle 78 5.4 Surface and Cloudbase Radiances for High and Marginal High/Middle Cloud Cases 80 5.5 Cloud Map Statistics for Objective and Subjective Analyses for Predominantly High Cloud Scans 88 5.6 Contingency Matrices for Selected Maps of Skies Dominated by High Clouds 89 2 5.7 X Statistics for Selected Objectively Analyzed Maps 95 5.8 Cloud Map Statistics for Predominantly Low and Middle Cloud Scans, Objective Analyses Only 104 5.9 Contingency Matrices for Selected Maps of Skies Dominated by Low and Middle Clouds 108 A. 1 Pressure/Height Conversion 135 A.2 Pressure/Temperature/Humidity Data for Sounding SD1 135 A.3 Pressure/Temperature/Humidity Data for Sounding SN2 136 A.4 Pressure/Temperature/Humidity Data for Sounding SD7 136 A.5 Ozone Mixing Ratio Profiles 137 X l l LIST OF SLIDES Slide Photograph 1 108 Bl 2 108 Fl 3 108 F2 4 107 I 5 112 DI 6 112 D2 7 112 El 8 112 E2 9 106 F 10 111 Gl 11 111 DI 12 112 Bl 13 112 Fl 14 112 HI 15 112 J The slides follow page 141 and are numbered. They should e viewed facing the number, with the buildings to the l e f t and the tand (thin black object) to the right on the screen or slide viewer. x i i i LIST OF SYMBOLS Units Planck Function (w -2 m sr 1 Error Parameters i n Map/Photo Analysis Cloud Thickness m or k: T o t a l Water Path i n a Cloud -2 g m Recorded Radiance at the Surface w -2 m sr l Clear Sky Radiance (Measured and Modelled) w -2 m sr l Cloudbase Radiance (Modelled) w -2 m sr l Surface Radiance (Modelled) w -2 m sr l Cloud Discrimination Radiances for Respectively: Clear Sky/High Cloud w -2 m sr l High/Middle Cloud w -2 m sr l Middle/Low Cloud w -2 m sr l 'Maximum' Low Cloud Radiance w -2 m sr l ( = N - N (z, )) Downwelling Cloud zb o b Radiance Minus Clear Sky Radiance at Cloudbase w -2 m sr l (= N - N ) Downwelling Cloud Radiance s o Minus Clear Sky Radiance at the Surface w -2 m sr l Temperature K or °C • Mean Temperature 70 to 90 kPa: °C. (5 Value Average) Mean Temperature 50 to 100 kPa: (11 Value Average) °C. Equivalent Blackbody Temperature i n Barnes PRT-5 9.5-11.5 um Operating Band °C. Mirror Surface Temperature °c. Transmissivity w -2 m sr l Planck Constant (W -2 m sr l Planck Constant (cm)K Vapour Pressure kPa Layer and Matrix Indices Mass Absorption C o e f f i c i e n t f o r Ice and Water Clouds 2 -1 m g XIV Foreign Broadening Absorption C o e f f i c i e n t S e l f Broadening Absorption C o e f f i c i e n t T o t a l Water Vapour Absorption C o e f f i c i e n t M a t r i x Element i n the i t h Column, j t h row Symmetry P a i r (Derived From m^  ) Atmospheric Pressure Humidity Mixing R a t i o Mean q 80 to 100 kPa: 5 Value Average Mean q 50 to 100 kPa: 11 Value Average Time Int r a s c a n Time InterScan Time Cloud T r a n s i t Time Wind V e l o c i t y at 50 kPa (Mid-Troposphere) Cloudbase V e l o c i t y Wind V e l o c i t y at the Surface (Nominal 10 m) Cloud L i q u i d Water Content Cloudbase Height Height Detector and F i l t e r Response Function of the Radiometer Volume Absorption C o e f f i c i e n t f o r Water and Ice Clouds (= N(.6=62°)- N(6=8°)) Limb B r i g h t e n i n g Parameter: Limb B r i g h t e n i n g From 6=8 to 6=62 Degrees (Use w i t h N ,N ,N , ,N) o s zb E m i s s i v i t y (of a Cloud or Atmospheric Layer) M i r r o r E m i s s i v i t y Constant i n N ^ Equation (5.2) M i r r o r R e f l e c t i v i t y Wavenumber R e f l e c t i v i t y (of a Cloud or Atmospheric Layer) O p t i c a l Depth Zeni t h Angle Azimuth Angle U n i t s 2 -1 1 m g atm 2 -1 — 1 m g atm 2 -1 _ 1 m g atm kPa -3 -1 x 10 kg kg -3 -1 x 10 kg kg -3 -1 x 10 kg kg standard u n i t s min min min  sm s ,-1 3- l 3 - l m s -3 g m m or kPa m or kPa (km) -1 W m s r W m s r (cm) km -1 X V ACKNOWLEDGEMENTS I acknowledge the support of the University of British Columbia in providing both access to the experimental location and the financial support necessary to complete the work. I would like to thank my supervisor, Dr. J.E. Hay, for providing me c r i t i c a l support and scientific impetus during this study. I also thank the other members of my committee, Dr. T.R. Oke and Dr. M.A. Church, for their advice on particular problems during the project. I acknow-ledge the excellent work of Mr. Richard Leslie who designed and built the scanning table used in this work. Special thanks go to Ludwig Braun who translated two key scien-t i f i c reports from German to English, to the UBC Physics Department for the laser used in testing the mirror, to Dr. P.M. Kuhn and Mrs. L. Stearns who provided me a copy of their excellent model RADIANCE, and to Mr. Peter Haering and his staff at the Pacific Weather Centre who gave me guidance and feedback during the project and access to their archived meteorological data. Throughout the study, the other climatology graduate students in the UBC Geography Department provided me with c r i t i c a l comments and suggestions. I thank each of them: Bruce McArthur, Brian Kalanda, Douw Steyn and John Knox. I also thank Bruce McArthur and Steve Lamble for designing and constructing some of the support equipment used, and the UBC History Department for providing work space for data processing. I thank a l l my various office and collective house partners who endured, cajoled and inspired me through to the completion of this thesis. Most of a l l , I thank my companera, Margarita, without whom I doubt I would be writing any of this. CHAPTER ONE PROBLEM AND OBJECTIVES 1.1 Purpose The purpose of t h i s thesis i s to examine the assumptions, and e v a l -uate the performance of the i n f r a r e d radiance contrast technique f o r map-ping clouds. The main assumptions to be examined are those which deter-mine the d i s c r i m i n a t i o n radiances used i n converting an inf r a r e d s i g n a l from the sky into a map of the sky showing the d i s t r i b u t i o n of clouds of d i f f e r e n t height cl a s s e s . The influence of the behaviour of atmospheric water vapour, ozone, and aerosols on the technique i s discussed. The v a l i d i t y of the mapping technique i s judged by determining how we l l the maps approximate simultaneous a l l - s k y photographs under various sky con-d i t i o n s . Integrated cloud t o t a l s by height c l a s s from the maps and photos are compared with nearby cloud observer r e s u l t s i n an examination of mesoscale cloud f i e l d v a r i a t i o n s . An a d d i t i o n a l purpose i s to evaluate the radiance contrast tech-nique as a remote cloud detection system. Remote here implies that the radiometer and mirror systems are operating unattended some tens or hundreds of kilometres from the l o c a t i o n where a trained operator would interrogate the s i g n a l . This involves asking the questions: 1) how much information i s contained i n a se r i e s of maps made at i n t e r v a l s and, 2) how e f f i c i e n t i s the technique i n providing a continuous p i c t u r e of the cloud f i e l d ? Questions of the engineering of such a system are beyond the scope of th i s study. 1 ( 2 1.2 Rationale Routine operational cloud information i s now gathered v i a the sub-jective observations of trained, ground-based observers for a l l hours. In addition, satellites operating in both the visible and infrared spectral bands provide, at various spatial scales, multiple grey-scale photographs of clouds, which may be objectively analyzed from an above-cloud perspec-tive. It would be desirable to have an objective, ground-based technique which could supplement and/or verify s a t e l l i t e information at a local (site) scale. If such a technique could be developed, the spatial uni-formity of the information in the present cloud observation network could be checked. Once developed, the technique could be used for 24 hour operation at sites where this i s impossible manually. Additionally, by learning about the angular distribution of clouds in the sky as viewed from a point on the earth's surface, important information about the sky distribution of short- and long-wave radiation at that point may become available. Also, by use of this technique, subjective observers would be freed for other tasks, or the present cloud observation network could be extended. Werner (1973b) describes and tests a methodology for what is called an 'Automatic Cloud Cover Indicator System'. To the knowledge of this author, Werner's work (1972,1973a,b) represents the only other past or present work on infrared cloud mapping over the sky hemisphere, from the ground. Many workers in the USA and Europe are doing multichannel cloud discrimination experiments from s a t e l l i t e data (mainly infrared). This present work i s much simpler in concept since only a single infrared window signal i s employed. 3 The present work i s not an explicit test of Werner's work, but rather an attempt to examine and use the underlying methodology under different circumstances. Werner, i n fact, does not produce maps, but integrates the cloud cover information over the sky hemisphere. Compar-isons are then made with the cloud cover observations of trained obser-vers. In the present study such integrated comparisons are made, but emphasis i s placed on the representation of the spatial distribution of clouds in the sky. This work i s necessarily restricted to a small number of cases observed on 5 days in July 1978. Therefore, conclusions about the long term performance of the system are not made. In short, this i s a study to determine i f the infrared cloud mapping technique can work and not a test of whether i t w i l l work. 1.3 The Problem The essential mapping objective i s to record, using a narrow view radiometer, enough information quickly enough from as much of the sky as possible—then to represent the present cloud situation as accurately as possible. Questions of stationarity and resolution must be treated. The problem is whether a single infrared signal contains sufficient unambigu-ous information so that the occurrence of cloud and even different height classes of clouds in the radiometer field-of-view can be inferred by objective discrimination of that signal alone. The cloud classes employed here are the standard height classes (AES, 1970, pp.157-163) (high, middle and low), with the latter including cumiliform clouds because of similar cloudbase heights. CHAPTER TWO THEORY AND METHODOLOGY 2.1 Theory 2.1.1 Overview The radiance contrast technique assumes that the downwelling radiances from clouds of different classes (determined by base height, z^) are significantly different from each other and from the radiance of a cloudless sky. That assumption has l i t t l e chance of being correct (at least for the nominal 870-1053 cm 1 band employed in this experiment). The sky actually exhibits a continuum of radiance sources. The lower limi t at any location i s a cold, dry, clear atmosphere, the upper lim i t is an 'ideal' thick cloud of emissivity 1.0 based at the elevation of maximum actual temperature, which i s presumably close to the ground. Between those limits l i e the perturbations caused by non-black clouds, situated in atmospheres of varying constituents, and based at a continuum of elevations and hence temperatures. The radiance difference between -2 -1 the limiting conditions at the experimental site i s about 15 W m sr (hereinafter abbreviated W) on an annual basis, and about 12 W on a summer day. The 12 W value represents measurements made on a number of summer days with both thick cumulus clouds and clear skies. The 15 W value represents the signal from a blackbody at maximum cloudbase temper-ature (^ 20 °C), minus the modelled results for the coldest, driest clear sky observed in November, 1977. 4 5 The radiance contrast technique assumes that the climatologically averaged decrease in temperature with height in the atmosphere causes the radiance from clouds to decrease with increasing cloudbase height (z^). It also assumes that changes of effective emissivity (which i s strongly zb+H dependent on liquid water path LWP = /w dz ) with height do not offset z b the radiance changes due to temperature. In the absence of deep inver-sions and increases of emissivity with height, the method would be . excellent. However, the latter condition does occur frequently enough to cause a considerable problem, especially when discriminating high from middle clouds (Figure 5.4). Historically, clouds have been assumed to be effective black bodies (e "v 1) in the broadband long-wave spectral region. This may be approximately correct in other infrared bands, but i n the experimental band used here, most clouds except cumulonimbus, nimbostra-tus and some cumulus are not black bodies (Piatt, 1976 and Piatt and Bartusek, 1974). Further, they are a l l somewhat spatially inhomogen-eous, in both the vertical and horizontal, and time variable. Thus, one can observe higher clouds emitting more radiance than lower ones (see 5.4.3). Cloud discrimination using a single signal, therefore, is often reduced to a process of minimizing errors inherent in the method. 2.1.2 Radiance Equations In simple terms, the infrared radiance received at the surface from a cloud consists of three streams: that from the cloud i t s e l f , that from behind (above) the cloud attenuated by the cloud's transmis-s i v i t y , and that from in front of (below) the cloud. That formulation is used in chapter 5 to describe in bulk terms the relative importance of the three terms mentioned. The downwelling infrared radiance observed or calculated at a reference level, r, i n the atmosphere i s more generally given by: v„ TR N - - / / a(v) B(v,T ) dTR dv (2.1) r ci V l TR= 1 where v is the wavenumber; TR i s the atmospheric transmissivity in specific spectral intervals, equal to the product of TR^ Q(vap)' TR , TR etc.; a(v) is the response function of the radiometer, U3 t U2 including the f i l t e r and detector transmissivity product; B(v,T ) is the Planck function of the form: B(v,T )= av 3 (exp((bv/T ) - l ) ) _ 1 (2.1a) 3- 3. with a and b the Planck constants; T i s the absolute temperature of an a air layer; dTR i s equivalent to TR +^^ -TR^, where i refers to successive atmospheric levels. This form of the standard equation i s adapted from Kuhn and Stearns (unpublished report) . If the radiance from the sky i s reflected in a mirror, as in this experiment, the integrand in (2.1) must be multiplied by u^(v), the mirror r e f l e c t i v i t y . A second (mirror emission) term i s added to the right side of (2.1): v„ • ' ^ < W £M ( V ) d V ( 2 ' l b )  V l Equation (2.1) can be used, by computing the transmission of each layer as a function of temperature, pressure and other variables, to determine the downwelling radiance at any pressure level in the atmosphere. Conceptually, the TR integration is replaced by a pressure (height) integration. Equation (2.1) is used by the radiance model of Kuhn and Stearns (RADIANCE), employed in this work, in the following form: i-1 N. .,.= ( EN. ... • TR..) + B.e, ( n 0. - v j = 1 j,j+l j i 1 i (2.2) i , or N. (...((N1TR0+N0)TR_+N.,)TR. ... N, ,)TR.+N. (2.2a) i,i+l 1 2 2 3 3 4 1-1 I i where is the downwelling radiance from layer i to layer i+1; B_^  i s the blackbody radiance for layer i ; T ^ j ^ ^ s t n e transmissivity from the bottom of layer j to the bottom of layer i ; and N^e^B^.. The N^ are also expressed as equivalent blackbody temperatures (T . ) in the model output, equiv The radiative transfer problem in this part of the infrared spectrum i s dominated by absorption. Reflectivity (integrated back-scatter) is usually small ( <gz .05) in the clear atmosphere. Therefore radiative extinction and absorption are essentially synonymous and absorptivity (equal to emissivity by Kirchoff's Law) approximates 1-TR. The infrared character of an atmospheric layer is reduced to the pro-blem of expressing the absorption by the various constituents important at that wavenumber. In the experimental band, which l i e s in the atmospheric window, emission by water vapour far exceeds that by CC^, ozone and other gases. (Liquid water is treated in the cloud section, 2.1.4.) The water vapour absorption in this 'continuum' spectral region is not well understood. The total mass absorption coefficient (k ) is expressed (after Bignell (1970)) in the form: k ^ ( T > p j e ) = k ^ ( T ) . p + k ^ T ) . e ( 2. 3) The f i r s t term on the right side of (2.3) represents absorption due to the combined effects of the wings of distant strong rotational lines, or so-called foreign-broadening. The second term (representing e-type or water vapour self-broadening) i s believed by many workers to be the 8 result of absorption by the water vapour dimer (R^O^ (Paltridge and Piatt, 1976, pp.161-2). 2.1.3 Continuum Absorption Parameters For this experiment, the choice of k^^ and values for modelling purposes is extremely important. The clear sky radiance (N q) depends strongly on them. Table 2.1 indicates three sets of k values derived from the given sources. The 'G' values are found to considerably under-estimate observed N values. The 'B' values considerably overestimate in o some cases (see chapter 4). The 'P' values, defined arbitrarily using the error limits quoted by Grassl (1973) also underestimate observed values, but by less than the 'G' values. P, rather than G, values are used in this study as a lower bound to actual atmospheric emission characteristics in the continuum. The temperature coefficients of k, and k„ are +.005 and -.02 ^ l v 2v ( x 10 4 m2g ^atm L) respectively (Grassl 1973 and 1974). 2.1.4 Cloud and Aerosol Formulation Both clouds and aerosols are considered to be homogeneous, plane parallel layers, characterized by a bulk volume absorption coefficient, B^v« Transmissivity is calculated in the standard way: -ln(TR)= 3 A v fdz (2.4) For aerosols, 3 is generally quoted e x p l i c i t l y , and provides a sufficient description of such materials in this context. For clouds, the optical depth T, the liquid water content w, the total water path LWP, the cloud thickness H, and the volume and mass absorption coefficients ( 3 ^ y a n d k^v) are related: T= g. • H (2.5a) Av T= kA • LWP (2.5b) Av k = g. / w (2.6). Av Av 9 TABLE 2.1—Water Vapour Continuum Absorption Parameters For Infrared Radiance Calculation v (cm ) 800 900 1000 1100 B k. .12 .09 .08 .07 lv k 0 19. 10. 10. 12. 2v G k. .012 .004 .0028 .0064 lv k 0 14. 9.0 6.2 4.9 2v P k, .034 .018 .015 .018 lv k„ 14. 9.0 6.2 4.9 2v A l l values ( x 10 m g atm ) B—From Kuhn (after Bignell, 1970) G—From Grassl, 1973 and Paltridge and Piatt, 1976—interpolated P—G, with kn values increased by the quoted measurement error. 10 . is a different constant for each of water and ice clouds (see Table 5.1). The values are computed using equations 2.5, which also give T . H and either w or x are given by the sources and the clouds are assumed homo-geneous: LWP= w • H (2.7) The emissivity e is the vertical emissivity as opposed to the larger (by a factor of 1.66) hemispheric emissivity used in 'dome' flux calculations. Zenith emissivity and optical depth are related: e = 1-e (2.8) Thus, by specifying and the heights of the cloud base and top (expressed as atmospheric pressures), LWP, x and e are a l l determined, and a cloud can be included as an atmospheric layer in equation 2.2. In the RADIANCE model, cloud re f l e c t i v i t y is assumed to be zero, certainly in error at least for cirrus (Piatt, 1973). 2.1.5 Geometry The zenith angle dependence of the radiance i s included in equation 2.2 by the simple secant approximation: N(81) = sec6 • N(6=0) (2.9) Because a l l layers are considered plane parallel and in f i n i t e in the hori-zontal, no modelling of f i n i t e clouds i s performed. Neglected are radiance from cloud sides, and the effects of variable thicknesses in a single cloud (e.g. turrets). The latter effects are certainly not negligible. This can be seen by comparing the emissivities of clouds LOWCB and LOWCD of Table 5.1. The latter has a lower w but appears blacker because i t i s thicker. This experiment observes clouds within 65° of the zenith, so cloud side radiance should be a minor effect. 2.2 Methodology 2.2.1 Constraints There are a number of major constraints on this experiment which limit the scope of observations and conclusions. The f i r s t i s the weather. No data can be gathered during periods of rain or high wind. The former optically interferes with the signal, and both make the equipment inoperable. The second constraint i s the equipment i t s e l f . The narrow band, narrow view radiometer (3.2.2) has two characteristics which limit the experiment. Its threshold radiance i s too high to measure clear sky radiance during much of the year. Thus observation i s limited mainly to summer months. The narrow f i e l d of view implies that in times of order 2 minutes, only a small percentage of the sky can be sampled. The rotating table on which the sky scanning mirror i s mounted yields an effective increment in zenith angle (6) of 5° in the environment (details in chapter 3). The sampling scheme i s a series of concentric hoops over the sky dome. The speed of sampling is somewhat determinable by the experimenter, but limited both by the table's rotating mechanism, and by the compromise between spatial resolution and temporal stationarity. As well, data are recorded on a device with a time constant 20 times that of the radiometer. (Some or a l l of that factor of 20 can be util i z e d in future refinements of this technique.) The rotation speed of the table varies between rotations. This makes automatic data digitizing very d i f f i c u l t , so a time-consuming manual technique i s employed. A third constraint i s the digitizing technique. When averaging the signal over a range of azimuths, some of the signal information, particular-ly regarding short duration extrema, is lost. Averaging the signal over some azimuth increment is preferable to using maximum or minimum signals from that increment^ since the latter bias the information taken from the signal. The fourth constraint i s the lack of local radiosonde information. Although some lower level local temperature and surface humidity infor-mation i s available for some of the experimental days, use of soundings hundreds of kilometres away in a region of major topographic contrast i s inadequate. Because of the uncertainty of upper air conditions at the s i t e , the attempt to correctly model local radiances is almost f u t i l e . Conse-quently the development of radiance discrimination c r i t e r i a depends heavily on off site data. Those results may, therefore, seem somewhat disconnected from the cloud maps themselves. 2.2.2 Map Verification It is desirable to describe how maps w i l l be analyzed, before discussing the details of how they are made. Three verification tech-niques are used here: qualitative inspection, contingency matrix analysis, and comparison of total cloud for a given height class. A l l three involve the comparison of maps with simultaneous all-sky photographs. The third also compares results from the maps and photos at the UBC site (3.1.2) with those of two trained observers in the local area. The qualitative inspection involves examining both maps and photos by eye, as would be done at an operational or research centre upon display of discriminated radiometer (map) information. The point of this analysis is to discern what fraction of the information that a human eye would see at the same location has been transferred to the map. (It i s assumed that cloud observers see the entire sky hemisphere and that their approach i s indistinguishable from that provided by the photographic system (3.2.1)). Key elements in this analysis are shapes, sizes and numbers of cloud elements. Also important i s the treatment of clear sky patches between thick cloud elements. To f a c i l i t a t e this analysis, a series of slides is included to allow comparison with the maps. 13 One must ask i f the maps are 'useful' operationally or for radiation experiments. In operational (day to day meteorological) terms, a map i s useful i f cloud amount, integrated by height class, i s correct to within 2/10 for each class, and i f the general shapes of cloud elements are pre-served. Exact sky position and detailed shapes are not important. Two tenths is chosen as the 'useful' limit because twice that value approxi-mates the range of both scattered (1-5 tenths) and broken (6-9 tenths) cloud as defined by AES MANOBS (1970, p.79). The word twice stands for +: a map estimate of say .2 should be read .2+.2, or (0,.4). Two-tenths is chosen because a criterion of three-tenths i s too lax to ensure that even an estimate at the midpoint of the scattered cloud category is within that category. One-tenth i s too stringent. Consider two estimates of a known .3 stratus coverage. An observer perceives slightly less than a quarter coverage (say .23) and records .2. An infrared map senses .36 stratus and rounds this to .4, .2 different from the observer. A one-tenth usefulness criterion would reject this case. The .13 difference in perceived cover-age represents about one-eighth of the sky. Since cloud data are s t i l l often recorded to this accuracy, i t is f e l t that such differences must be accepted as useful. Therefore one-tenth was rejected and two-tenths accepted as the usefulness criterion. In radiative terms, a map is truthful i f i t is useful and there are no noticeable errors in cloud position. A l l of the above analysis is by nature subjective, so more objective verification c r i t e r i a are sought. One standard method of comparing discriminated information from photographs and infrared maps (and from other sources) is through contin-gency tables (or matrices). A recent i l l u s t r a t i o n is Harris and Barrett's (1978) work with infrared s a t e l l i t e data. The analysis here includes 14 placing a 144 square grid over both maps and photographs and assigning each square a value corresponding to the average cloudiness in the square. Only 103 grid squares are used because confidence in the data decreases radially at large zenith angles. The grid extends to 6=62.5 + 2.5°, the approxi-mate limit of reliable information. The number of map points (representing A8=10° and A<j)=5°) , corresponding to the solid angle in the environment, varies from square to square by up to a factor of ^1 .5 . No weighting of the information is done. Clear sky is assigned class 0 (zero), and high, middle and low clouds to classes 1,2, and 3 respectively. The grids for a map and corresponding photo are analyzed by assigning the photo to the rows, and the map to the columns, of a contingency matrix and comparing the grids square by square. Matrix elements are referred to as ordered pairs (column, row) (e.g. the (0,1) element has a value 8 in the top le f t matrix of Table 5 .6) . The map/photo comparison is perfect i f a l l elements are on the main diagonal, and deteriorates as more elements appear further from i t . For skies containing only high clouds and clear sky, the 4 by 4 matrices reduce to the 2 by 2 case, which can be analyzed using a chi-squared 2 2 (x ) test. In 5.5.2 x data testing for the non-randomness of maps i s presented. The technique has limited value because matrices for most sky conditions are of higher order and contain too many near zero elements for 2 meaningful x results. Therefore, a second analysis is developed by equating effective error with mean square distance from the matrices' mean diagonals. If an element ( i , j ) makes a 'mistake' of order | i - j | , then the effective error, defined by error parameter one (EPp i s : EPL = I ( i - j ) 2 • m (2.10) matrix where m.. is the i j t h matrix element. 15 The objective i s then to set limits on EP^ for various sky types which determine map performance. On a small sample this i s d i f f i c u l t , so much of the EP^ information i s used comparatively. One serious drawback to EP^ i s that i t overestimates the performance of maps of skies dominated by either clear sky or thick low cloud. In each case a mistake can only be made in one direction (e.g. (0,1) but not (0,-1)). Therefore a second error parameter is defined by dividing EP^ by the sum of the non (0,0) and (3,3) elements: EP_ = EP1 / Z m.. (2.11) 1 i * j 1 J i=j^0 or 3 Together EP^ and EP^ give a sense of the ab i l i t y of the maps to represent the photographs' 'true' conditions. Unfortunately, changes in the condition of the sky over 2h to 3>h minutes are not negligible. Anal-yses of pairs of photos taken at that time interval (before and after a single sky scan) using the same matrix and EP techniques, indicate that differences between the photographs may exceed differences between the corresponding map and one of the photographs. Another way to analyze matrix errors i s to examine the direction-a l i t y (or asymmetry) of the map/photo differences. This i s done here by symmetry pairs n/1, n and 1 defined as follows: n = Z m.. 1 = Z m.. (2.12) i > j J K ] J 2 Note that the sums are not weighted by ( i - j ) . n represents the sum of map overestimations, 1 represents the sum of map underestimations, of cloud class. Overestimation implies assigning a higher class—one with a higher radiance. If n'vl in the symmetry pair, then errors are l i k e l y due to cloud position changes. If n and 1 are much different, then some gross d i f f e r -ence in interpretation is occurring between map and photo, usually the 16 misinterpretation of an entire cloud layer due to a shortcoming in the cloud discrimination technique. The third map verification technique involves summing the occur-rences of the various cloud classes in both maps and photographs, and expressing the sky fractions to the nearest tenth. The weighting of the grid imposed on circular maps and photos (which have radius proportional to zenith angle) closely approximates the weighting of a (sky) hemisphere. Hourly cloud data from nearby observers i s compared to the temporally nearest cloud map. An attempt to synchronize radiometer observations with the local cloud observations was defeated because of problems with the recorder and mirror table (3.2.2). Also, short gusty periods prevented usable scans from being performed. However, since the hourly point obser-vations are operationally assumed to represent the hour centred on that observation, time differences of up to 30 minutes between the data sets are not unreasonable. Results of this analysis are expressed using scatter diagrams (Figures 5.9 and 5.14). 2.2.3 Mapping Techniques While verifying maps, one i s l e f t with the question: Are the maps giving a l l the information contained in the radiometer signal? To try to answer this question, two parallel analysis techniques are applied to some of the radiometer signals. One is an objective technique which attempts to mimic 'automatic' data reduction by appropriate processing of the voltage output of the radiometer. The other i s a subjective technique which attempts to extract the maximum possible information from the traces by correcting some of the obvious mistakes of the objective technique. The objective (0) technique follows these guidelines: 1) No thought is allowed during data extraction, even i f an obvious mistake is occurring; 17 2) No after-the-fact map or upper air information can be used; 3) No other simultaneous information can be used with the radiance/voltage trace from the chart recorder, except pre-calculated cloud discrimination radiances; 4) The analysis must be available within at most a few minutes after the radiometer finishes scanning the sky. Guidelines 2 and 3 imply that changes in clear sky radiance from scan to scan cannot be corrected immediately. (When mapping using radiances solely derived from models, guidelines 2 and 3 are sometimes waived to partially compensate for the great distances to the sounding locations.) Guideline 1 implies that even when the recording equipment is 'clipping' signals, the signal must be treated as correct. Guideline 4 rules out complicated schemes to account for time varying local azimuthal variations in clear sky radiance (chapter 4). The subjective (S) analysis ignores guidelines 1 and 4 above. The radiance trace i s analyzed using the same radiances as the corresponding 0 analysis, only i f no change in clear sky radiance (N q) has occurred which would cause mistakes (e.g. large areas of clear sky interpreted as high cloud because of a sudden increase in N ). When such variations do occur, the N o — o values, and the appropriate discrimination radiances, are adjusted accor-dingly. Also, local azimuth changes in N q are eliminated by eye. No results are presented which attempt to compensate for 'clipping' by the chart recorder. This was attempted and found to produce only occasional small changes in the maps. The S analyses provide aids in assessing the performance of some of the 0 maps. Comparisons of error parameters and symmetry of 0 and S maps to the same photo(s) are used. When an 0 map i s s t a t i s t i c a l l y equivalent to the corresponding S map, i t is claimed that the former represents the limit of the technique to record the sky under the given conditions. CHAPTER THREE SITE, INSTRUMENTATION AND DATA ACQUISITION This chapter describes when, where and how the radiance traces were made, and how the input data for the cloud maps were produced. In section 3.1 the s i t e c r i t e r i a and c h a r a c t e r i s t i c s are d e t a i l e d . The photographic and i n f r a r e d radiometer systems are described i n 3.2. Next, i n section 3.3, data reduction i s discussed, e s p e c i a l l y questions of r e s o l u t i o n , error and data from other sources. F i n a l l y , some of the i n s t r u -mental and analysis d e t a i l s of the Werner experiment are given i n 3.4. 3.1 S i t e 3.1.1 C r i t e r i a A l o c a t i o n f or t h i s type of experiment must have a c l e a r view of the sky hemisphere over most azimuths. The l o c a l horizon should be at a zenith angle greater than 75°, the largest usable zenith of the discrim-i n a t i o n system. Nearby hourly cloud observations and 12 hour radiosonde data are each e s s e n t i a l f or c e r t a i n portions of the work. The question of how nearby these observations must be made i s unresolved. The author knows of no studies on t h i s subject p a r t i c u l a r l y , though Werner (1972) used a separation of 30 kilometres for the radiosonde information i n h i s German experiment. Separations f a l l i n g within Oke's (1978) l o c a l scale (100 m to 50 km) would be desirable where topography i s marked. Oke's mesoscale (10 to 200 km) c e r t a i n l y defines an upper l i m i t of separation under a l l surface conditions. In t h i s study the former l i m i t applies. Areas with frequent and/or heavy outbreaks of aerosols that are r e g i o n a l l y a t y p i c a l , 18 19 N O R T H V A N C O U V E R D I S T R I C T W E S T V A N C O U V E R • YHC (DISCOVERY! UBC-V A N 1 — i • 1—* C O Q U I T L A M YVR (AIRPORT)' N 49° N-LANGLEY Scale 1:391 000 0 5 km 10 (Note cloud observer locations YVR and YHC, and heavily urban-ized zone (hatched).) Figure 3.1 Experimental Location (Mesoscale) 20 are unsuitable. F i n a l l y , the s i t e must have access to power and weather-proof storage, as well as be large enough to allow the camera and r a d i o -meter systems to operate simultaneously with minimum mutual interference. 3.1.2 Buchanan Tower Location The f i e l d l o c a t i o n chosen for t h i s experiment i s the roof of the Buchanan Tower, UBC (Latitude 49°20' N., Longitude 123°15' W.). The e l e -vation above sea l e v e l i s taken to be 129 + 8 m (85 + 5 m at the base of the b u i l d i n g , 44 + 5 m for the b u i l d i n g ) . The l o c a t i o n (Figure 3.1) i s 8 to 10 kilometres west of the c e n t r a l business d i s t r i c t of Vancouver and 11 to 12 kilometres northwest of the a i r p o r t . Cloud observations at o f f - s i t e locations Discovery (YHC) and A i r p o r t (YVR) are used. Mountains r i s e to about 1000 metres j u s t north of the d i s t r i c t of North Vancouver and West Vancouver, and to higher elevations further north and northwest. The mountains do not s i g n i f i c a n t l y a f f e c t the horizon at any of the observing loca t i o n s , but do of course contribute considerably to the l o c a l cloud regimes through orographic l i f t i n g . S l i d e 2, among others, i l l u s t r a t e s the view of the sky from the roof. The photos represent changes i n zenith angle l i n e a r l y i n the radius; the e f f e c t s of the buildings and mountains to the east and north are confined to 6>85°. Thus, the UBC l o c a t i o n meets the horizon and cloud observation c r i t e r i a . Power, storage and s i t e s i z e (about 10 by 30 metres) are a l l adequate on the tar and gravel roof. Figure 3.2 shows the roof looking i n the d i r e c t i o n azimuth <j>=60°. The buildings i n that d i r e c t i o n provide a r e l i a b l e check point when or i e n t i n g photographs. Note the two large vents. Empirical tests reveal that t h e i r exhausts have a n e g l i g i b l e e f f e c t on the radiance signals. They are used as a check on o p t i c a l l e v e l l i n g because of their high r e f l e c t i v i t y and known l o c a t i o n . The wooden ramp provides an 21 Figure 3.2 Experimental Location (View of Site) 2 2 adequate footing for the radiometer stand (Figure 3.3) and the camera tripod (Figure 3.4). The chosen location does not satisfy the nearby radiosonde c r i t e r -ion (Figure 3.5). Because of the necessarily intermittent nature of the scanning program, a site close to the experimenter's normal residence was considered essential. In retrospect, performing the experiment at a radiosonde site (such as Port Hardy, B.C., ^  347 km northwest, or Quil-layute, Washington, 180 km southwest) would be preferable, despite possible horizon shortcomings. The chosen site is far enough to the west of the Vancouver urban core that the major urban aerosol and heating effects are confined to a portion of the sky at most. These factors are dealt with in section 4.5. 3.2 Instrumentation and Its Deployment 3.2.1 Sky Photography All-sky photographs were taken using a CANON FTB 35 mm camera. The lens was a CANON Fisheye 7.5 mm f5.6 s.s.c, operating with a sky f i l t e r . This f i l t e r gives the closest approximation to the actual colours per-ceived by the eye of a cloud observer. "This equidistant lens has a f i e l d of view of 180° and provides a circular exposure of 24.143 + .155 mm in diameter on the photographic film...the magnification i s 0.15 x. The focussing i s fixed." (McArthur (1978), pp.28-29) Camera levelling was checked by placing a s p i r i t level on the top of the lens cap in two orthogonal directions. This was done before and after each pair of photographs which accompanied an infrared sky scan. The camera was mounted on a tripod, ^ 8 metres from the infrared system at azimuth 60°. An occulting disc was used to block the solar disc during periods of direct sunlight. This eliminated photo contamination from Figure 3.3 (top) View of Radiometer System Figure 3.4 (bottom) View of Photographic System N N o r t h P a c i f i c Ocean KEY 1 Vancouver (UBC,YVR,YHC) 2 P o r t Hardy (YZT) 3 Q u i l l a y u t e (UIL) 4 Vernon 5 C a m p b e l l R i v e r 6 Vancouver I s l a n d 7 Queen C h a r l o t t e I s l a n d s 8 G u l f o f A l a s k a 9 V i c t o r i a + R a d i o s o n d e S t a t i o n (12 hour) S c a l e 1:9 375 000 Oregon F i g u r e 3.5 E x p e r i m e n t a l L o c a t i o n ( S y n o p t i c Sea 25 internal lens flare due to the sun's rays. Aperture settings of f-8 and f-11 and shutter speeds of 1/125 and 1/250 second were used throughout. The use of the occulting disc eliminated large changes in the amount of light incident on the lens. Standard Kodachrome-II (ASA 64) daylight film with standard commercial processing was used throughout. Occasional cropping of the circular sky image did not affect data quality. Slides were viewed on a Singer Caramate SP viewer which presents a sky image of diameter 0.16 m. An earlier attempt was made to use standard f i e l d of view 35 mm slides to analyze portions of the sky. This proved fruitless because of the transient nature of the sky and the d i f f i c u l t y of quickly and accu-rately orienting the camera in different directions. 3.2.2 Barnes PRT-5 Radiometer System The Radiometer The Barnes PRT-5 consists of an optical unit, an electronics unit, and power and signal cables. It is run on 110 Volt AC power throughout this experiment. The optical unit continuously compares the amount of energy emitted by the sky with that emitted by an internal, reference environment controlled at 45+0.5 °C. The electronics unit processes the radiance comparison into a voltage, related almost linearly to the radiance differ-ence between the sky and the reference (Barnes, 1970). The system can be operated at 3 different response speeds, characterized by the time constant, of the output (5, 50, 500 milliseconds). A shorter time implies a more frequent comparison of sky and reference. The 5 ms range i s used here. The instrument has a f i e l d of view of 2.5 milliradians (or 0.14°) nominal, which represents 7.6+0.2 x 10 ^ of a sky hemisphere. The instrument 26 essentially samples a point at cloud distances: at a distance of 10 km, the viewed area i s approximately a 25 m diameter ci r c l e . The 9.5-11.5 um nominal spectral band pass of the optical unit i s given in Figure 3.6 (from Barnes, 1970.) This curve i s converted to the wavenumber equivalent for radiance modelling purposes. The entire unit can be operated in ambient temperatures of -20 to +40 °C., beyond the climatic extrema for this site. The instrument has three radiance (equivalent temperature) ranges or scales. The lowest, from 2.7 W (-60°C.) to 9.4 W (-10°C), i s used. The signal i s output on a 0-1 volt scale for recording and/or processing. Because the instrument could not be switched to a higher radiance range during a scan when the values from the clouds exceeded 9.4 W, the instru-ment was operated at higher than the listed maximum voltage. Some accur-acy i s lost in these offscale zones, but output saturation does not begin u n t i l the signal exceeds about 1.4 volts, or about 13.1-13.4 W. A small range below 0 volts was also usable, assuming a linear radiance/voltage relationship (Astheimer, pers.commun.). The effective data acquisition limits are, therefore, ^ 2.5 and ^ 13.1 W, a somewhat smaller range than the range of sky radiance sources given in section 2.1. Both the upper and lower instrument threshold radiances are important in this experiment: the upper threshold i s reached several times by clouds, while the lower threshold restricts the effective use of the system to certain parts of the year. The Reflecting System The mirror used in the experiment is aluminum coated by a thin film of sil i c o n monoxide. The entire surface area measures 200 by 250 mm, but only a small area near i t s centre is viewed by the radiometer. The WAVELENGTH (microns) Figure 3.6 Spectral Response of Barnes PRT-5 (after Barnes, 1970) 28 re f l e c t i v i t y (u^) of the mirror was tested at 9.6 um using a tunable CC^ laser and found to be about 0.8. However, empirical tests comparing direct and reflected sky signals on a number of days with clear sky indicated a p.., of closer to 0.97-0.98 for the operating waveband. These tests showed M that the effect of the mirror on the signal was small. The mirror was mounted, during sky scanning experiments, temporarily on the stepping, rotating table designed by Richard Leslie of the UBC Geography Department. The mounting uses 8 strips of tape with glue on one side and a magnet on the other (4 on each of mirror and table). The table begins rotating at an i n i t i a l inclination set by the experimenter, and steps in units of 2% degrees about a horizontal axis. Its f i n a l rotation, at 45° inclination, scans the horizon. A 2% degree inclination change corresponds to a 5 degree change in zenith angle in the sky. Stepping occurs when the mirror and table face the vertical arm of the instrument stand. Data loss is thus limited to one event per rotation. In this experiment the table was started at an inclination of 2.5° and allowed to run out to 45°. (Data for the f i r s t partial rotation was ignored (section During or just after the rotations at 6= 85 or 90°, the second photo for that scan was taken. Rotation time (as determined by elapsed time between sensings of the vertical arm of the radiometer stand) for the table varied continuously between 7.9 and 9.2 seconds. No variation of rotation speed with azimuth occurs at low wind speeds. For winds greater than about 10 ms L, the table acts like a s a i l at the higher zenith angles, and rotation speed varies noticeably, reducing data quality. Rotation speed does vary slightly with zenith angle. Variations for a particular sky scan are typically 0.6-0.8 seconds per rotation. Scan duration can be determined directly from the 29 sum of the rotation times. The Recording System The voltage signal was recorded on a Honeywell Electronik 194 strip chart recorder, connected to the Barnes by standard coaxial cable. The chart recorder was run at 5.08 or 12.7 mm/s on the 2 volt scale. The trace for one rotation was 54 + 6 mm for the slower chart speed and 132 + 16 mm for the faster chart speed. The time constant of the recorder, determined by the step response method, was 0.1 + 0.01 seconds. This time constant determined the response of the entire system. Because the time constant test was performed with a large step change in the voltage, and because of the mechanical nature of the recorder, 0.1 seconds may have underestimated the recorder's capability in following rapid small ampli-tude changes. The Honeywell manual says the frequency response to a 5 Hz, 10% of scan peak to peak sine wave i s down less than 1% of the span. Thus 90% of the sine wave amplitude i s reproduced at 5 Hz. This implies a time constant of 0.11 seconds, the same as that derived from the step response method. Experimental Set Up Figure 3.3 shows the equipment described above, (plus an extra recorder used during some preliminary tests). The Barnes optical unit is attached to the top of the metal stand with four screws and points down into the mirror, which is shown in the figure at a high zenith angle position. The electronics unit is in the lower right, the Honeywell in the lower l e f t . The table has four bolts which f i t into holes on a horizontal metal plate on the instrument stand. It i s levelled using four nuts and periodically checked with a s p i r i t level during a scanning day. The instru-ment stand is levelled by adjusting the four retractable legs. Figure 3.7 shows an e l e v a t i o n view of the m i r r o r , stand and o p t i -c a l u n i t . On the s i d e of the o p t i c a l u n i t are three p r o t r u d i n g pieces of metal which increase the e f f e c t i v e r a d i u s of the o p t i c a l u n i t t o about 0.18 m. At a t a b l e i n c l i n a t i o n of 2.5°, the radiometer 'sees' i t s e l f at some azimuths, t h e r e f o r e the data are unusable. The l a c k of t h i s i n f o r -mation (at a sky z e n i t h angle of 5°) i s not c r i t i c a l because the data at 10° are assumed to provide an adequate near z e n i t h sample. I f t h i s exper-iment i s repeated, the stand's v e r t i c a l arm should be redesigned to avoid t h i s data l o s s . The s i d e of the v e r t i c a l arm f a c i n g the m i r r o r i s painted f l a t b l a c k to i n c rease the radiance c o n t r a s t between i t and the sky. The radiance pulse from the stand on each m i r r o r r o t a t i o n i s used as the t i m i n g pulse f o r the data t r a c e s . (The stand subtends an azimuth angle of l e s s than 5.4° at a l l 6.) The amount of sky radiance data l o s t because of the stand i s determined by t h i s subtended angle and the system time constant: * i = <t» * + <f>/o • \ (3.1) l o s t stand (3 time constants) This works out to about 20° here. 3.3 Data A c q u i s i t i o n and Processing Radiometric data were recorded i n August, 1977 and J u l y , 1978. During the e a r l i e r p e r i o d , a l l - s k y photographs were not taken, so the data could not be mapped and compared (to the photos) as were the data from the l a t e r p e r i o d . However, the August, 1977 measurements of N Q (and modelled N q values f o r the p e r i o d from Port Hardy soundings) form part of the data base f o r the c l e a r sky radiance a n a l y s i s of chapter 4. 3.3.1 Temporal Considerations Because of the high radiance threshold of the Barnes PRT-5 (2.5-2.7 W), sky scans were l i m i t e d to the summer months. The system 31 ~ mounting screws optical unit instrument stand -*• mirror rotating table "1 .27+.02 m J detail omitted •> wood walkway Scale 0 .1 .2 metres Figure 3.7 Radiometer System—Elevation View required on site, real time monitoring. Therefore, data acquisition could not be continuous since only one experimenter was involved. No night data were taken because photographic verification was impossible. It was f e l t that a data set of from four to 8 days, with a range of cloud conditions, would be appropriate within the research context. The month of July, 1978 was selected and scans performed on most days with cloud after July 6. Some attempt was made to coordinate with the UBC SUNSET experiment in eastern Vancouver which was gathering minisonde temperature data. Five days were selected for analysis out of the 7 when scanning took place. The 2 rejected days had insignificant amounts of cloud. For the general synoptic conditions on these days, see Appendix B. On only 1 day, July 26, did precipitation cause an end to scanning. On both July 11 and July 27 data acquisition ended when constant signals from the entire sky were recorded for several consecutive scans. Occasional equipment malfunctions and unexpected emergencies caused the scans to be spaced somewhat irregularly. The time between the start of one scan and the start of the next (the interscan time, t ) was determined partially by the interval between cloud observations and partially by the intrascan time (t^> the time to complete a scan to 0 ). The former i s 1 hour, the latter is about 2-2^ max minutes. Assuming that t should be at least 5-10 times t. (from sampling s l theory), and no less than the time between cloud observations, i t must be in the range 10-60 minutes. Two attempts were made to obtain time series data with t varying between 15 and 60 minutes. The photos taken during one attempt were destroyed in processing. The other series is presented and analyzed in 5.7. The 'simultaneous photographs' were taken either before and after 33 scan (with a time difference of ^ 3 minutes), or during a scan. Because of the need to ensure that the occulting disc covered the sun, and the need to watch the table to ensure correct operation, true simultaneous operation of the systems was impossible. Occasionally the table repeated a particular 9 step, or skipped one. The former provided chances to check sky changes at some 0 over 8 to 10 seconds, while the latter caused the scan to be aborted and rerun. Skipped 9 steps were caused by gusty winds, while repeated 9 steps resulted from electromechanical problems. 3.3.2 Spatial Resolution The zenith angle sampling i s determined by the table step and the radiometer f i e l d of view. Each rotation traces a 0.14° wide ring on an imaginary sky hemisphere. The rings are separated by 5° (AG) centre to centre, so approximately 1/36 of the sky, the ratio of ring width to ring spacing, is sampled. This applies for (9: 7.5° <_ 0 <_ + 2.5°). The more c r i t i c a l resolution in azimuth angle, <j>, is determined by the table rotation speed and the 0.1 second time constant of the system. For a rotation speed of 8.0 seconds, one time constant corresponds to 4.5° in cj>. Assuming that cloud structural changes sensed within three time constants usually cannot be resolved, an azimuth resolution of about 13.5° would be expected. To ensure that the digitizing of the data was not the limiting procedure in the analysis, an azimuth interval for digitizing of 10° was chosen. This <J> interval (Ac))) represents 2 to 2h time constants and is close to the limit of data extraction (1/3 chart square P e r A<t> interval) for the traces done at the slower chart speed. Each AcJ> interval is referred to as a slot. Slots 1 and 36 are ignored in the analysis because of stand contaminat ion. Spatial resolution must be related to the environment being sensed. 34 Consider a cloud at base height (z, ) of 1000 m at 9=65°. A Ad> of 10° D represents 413 m in horizontal extent. That figure increases linearly with z^ (to 2065 m for z^=5000 m), though of course the cloud i s five times as distant horizontally as well. Clearly considerable information about the horizontal extent of small middle and high clouds i s lost. If a recording system as fast as the Barnes could be employed, with a slot equalling the cj) equivalent of four times the new time constant, a factor of 10 improvement in the <j> resolution would be realized, for no change in t^. However the <)> resolution would then be 5 times smaller than the 9 resolution (instead of the present two times larger). A9 would than have to be decreased, increasing t^, which also must be minimized because of stationarity considerations. Recommendations on how to use this factor of 10 w i l l be made under conclusions. 3.3.3 Reduction of Radiometer Data Each data trace i s reduced by applying the discrimination radi-ances (N t ) , converted to chart squares, in order of decreasing magnitude. The average radiance in each slot is mapped directly using a simple FORTRAN program and a Calcomp plotter. The original maps are .229 m diameter circles with different symbols for the different cloud classes: '*' for low cloud; 1 <3>' for middle cloud; '-' for high cloud; and blanks for clear sky. The map compass i s reversed from the normal configuration because the maps are produced (and should be read) looking vertically up into the sky dome. Maps are denoted by a 3 digit number followed by 2 or 3 letters (e.g. Map 112 JE-C i s read experimental day number 112, scan J, d i s c r i -minated using N derived from the clear sky radiance of scan 112 E, using method C for determining N„_ (5.4.3)). If the second letter (E in the 35 example) i s missing, then a l l scans for that day use the same clear sky radiance. On days with higher clear sky radiance, exceeds the Barnes saturation radiance at a lower 9, thereby reducing the amount of usable data. The largest usable 9, 8 m , varies from 40 to 75°, depending on the day and the method used to compute N^ -^ When ^ ^^^0°, the particular values of N2-j are not used since the maps would be too small to compare with the all-sky photos. On two days'data, July 13 and July 27, two separate sets of N , corresponding to clear sky information from d i f f -S L erent times of the day, are used to map selected scans. Reproducibility of the technique is tested on four separate scans. Of the 400 to 500 slots extracted, differences occur in only 10-20 slots. When mapped, and compared to the photos using the grids, differences in only 2 grid squares result. 3.3.4 Other Data Reduction Radiosonde information from Port Hardy and Quillayute was extracted manually from the tephigrams archived at the Pacific Weather Centre, Vancouver. The mixing ratio was read directly. Low level temperature soundings (below 2000 m), and surface wet bulb temperatures from Sunset (Figure 3.1) (D. Steyn, pers.commun.) were combined with mixing ratios calculated from dewpoint depressions at Vancouver Airport, to approximate the mixing ratio (q) and temperature in a surface affected layer. In this well-mixed layer, q was assumed constant. The layer was assumed to extend to the base of an elevated inversion, or the l i f t i n g condensation level, whichever was lower. These data were combined with upper layer information from the off site radiosonde soundings in several attempts to 'reconstruct' the local atmosphere. Upper level (50 kPa) and surface charts were used 36 to decide which upper air sounding better approximated the Vancouver conditions. 3.3.5 Error Analysis The errors in this experiment f a l l into 4 categories: 1) probable absolute error (in W) of the measured radiance signal; 2) estimated absolute error (in W) of the modelled discrimination radiances; 3) errors in the digitizing and mapping process; and 4) errors in the grid comparison of maps and photographs. Errors in 1) above can originate with the radiometer, the mirror or the chart recorder. Mirror The mirror is a source of spurious 'extra' radiance given, for a clear sky, by: N • u + B(T„. )• e„ - N (3.2) } J o M Mir M o Assuming a perfect mirror, u + e =1, the 'extra' signal reduces to: V ( B ( TMir> " V ( 3 ' 3 ) ' When the optical unit was alternately pointed at a spot in the sky directly and via the mirror, on various days, the latter signals were between +.3 and -.1 W different from the former. The +.3 occurred on a warm windless day with T., ^ 30°C. (10° greater than the ambient) and N i> 3.0 W. In that M o example e w was calculated to be .019. Values of .015 to .025 for e_. were M M found in most cases. The negative mirror effect could not be explained (except by an abrupt change in radiance from the spot in the sky) since mirror blackbody radiance always exceeded sky radiance, and e^O. The mirror's effect on scanned radiances must be considerably less than .3 W, since when the mirror rotates i t s temperature approaches the ambient and the error term (3.3) decreases. The observed mirror error of 0.2-0.3 W for July 14 is illustrated in Figure 4.6. 37 Radiometer and Recorder The PRT-5 radiometer is accurate, according to Barnes, to + 0.5°C. equivalent temperature. At low radiances ('v 3 W) , this i s + .05 W, while near 9 W, i t rises to + .10 W. In offscale regions below saturation (9.4-o 13.1 W), an error of + .20 W i s assumed, though this may be somewhat low. For the Honeywell, the quoted error i s 0.25% of scan (h, of a chart square) or 0.03 to 0.04 W. Radiance Traces Treating the above errors as relative errors, and using the standard square root of the sum of squares technique, yields the values in Table 3.1. The mirror related error in the low radiance case i s indis-tinguishable from an increase in the clear sky radiance (N q) to an observer who cannot perform a mirror test. Therefore mirror error can be treated as an unwanted source of change in N q . Mirror effects do not appear regularly. If discrimination radiances are computed from measured N q values that implicitly include the effect of the mirror, then the mirror error is eliminated, except for changes in that error with time. These changes should be small (<0.1 W) i f the mirror i s kept well ventilated during sky scanning. Discrimination Radiances The N errors are summarized in Table 3.2. Individual errors are st combined as above. (If computed from N q using Werner's limb variation (method C, section 5.4.3) source 4 error i s reduced.) The largest N ^ ^ error comes in choosing one of the 3 formulations as 'correct'. Source 3 in Table 3.2 i s the estimate of errors made by averaging several modelled situations into a single relationship, plus those inherent in the model. This error can be reduced as more observed cloud data are modelled. Source 5 38 TABLE 3.1—Probable Absolute and Relative Error i n the Radiance Traces at D i f f e r e n t Radiance Values Radiances Small (^ 3 W) Large (^ 9 W) L i m i t i n g (^ 12 W) On scale Errors without mirror e f f e c t 2% (.06 W) 1.2% (.11 W) 1.8% (".22 W) Errors with maxi-mum mirror e f f e c t 10% (.3 W) 2.5% (.23 W) 3.0% (.36 W) TABLE 3.2—Errors i n Discrimination Radiances Source Uncertainty Radiances Affected N 01 N 12 N 23 1. N measurement o 2. Rounding to near-est % square 3. Modelling 4. Limb uncertainty 5. Conversion of modelled to meas-ured radiance scale .11 W .05 W *.3 W v a r i a b l e with .01 W • T o t a l + .12 W + .32 W + .30 W (+ .40 W at high 6) c 39 r e s u l t s from inaccuracies i n converting radiances between the s l i g h t l y d i f f -erent measured and modelled radiance scales. What then are the e f f e c t s of a l l the above errors on the d i s c r i m i -nation process? For the c l e a r sky/high cloud boundary, errors are c h i e f l y from EL, , not the data trace. However, because N R T , - N (.3 W) i s only about 01 01 o three times the probable error i n N ^ , and because of the magnitudes of changes with time (chapter 4), misinterpretations of several tens of g r i d squares can sometimes take place. However, i f the N q i s constantly moni-tored, these errors can be corrected almost immediately. For instance, i f a remote radiometer i s interrogated, a map produced, and the N q values d i s -played simultaneously, the N can be r e c a l c u l a t e d and the map rerun i f  s i g n i f i c a n t changes have occurred. Therefore N Q ^ errors are r e t r i e v a b l e . For high/middle cloud and middle/low cloud boundaries, errors are also more dependent on the N than on the measurements, but are not e a s i l y eliminated. The errors i n the chosen and values add or subtract a few percent from the various map/photo comparison i n d i c e s . The dominant uncertainty i s i n the modelled clouds themselves, since no i n f r a r e d radiance climatology e x i s t s . (This climatology would consist of the means and stan-dard deviations of the i n f r a r e d radiances of d i f f e r e n t cloud types i n a par-t i c u l a r band of the i n f r a r e d . The sample for each cloud type would include measured radiances and radiances modelled from measured l i q u i d water and dimensional v a r i a b l e s . Values at d i f f e r e n t 8 would be needed.) M i s i n t e r -pretation on these boundaries £*vr e n t i r e cloud layers having more or l e s s than the expected radiance i s dealt with i n chapter 5. Errors i n the a c t u a l N (<0.4 W) are small with respect to the range of radiances from the sky. However, only a long term test of the present methodology can determine i f the actual values used here are grossly inaccurate. 40 Other Errors in 3) and 4) above (p. 36) are also d i f f i c u l t to quantify. In 3.3.3 above, digitizing errors are alluded to. If one could digitize at a finer mesh in <{>, more information would be mapped. This cannot be inves-tigated since this experiment i s working at the <$> resolution l i m i t . If a finer mesh of grid squares was used, a better map/photo comparison would be possible, i f one has confidence in photo interpretation at the finer scale. The error which occurs between the discrimination of the traces and the computation of error parameters from contingency matrices i s indistin-guishable from the va r i a b i l i t y of the sky being sensed. 3.4 The Werner Experiment Before presenting results, a brief summary of the experiment of Werner (1972, 1973a,b) is given, along with some of the similarities and differences with the present work. Werner u t i l i z e s a similar theoretical framework, but a far more sophisticated set of equipment. He is able to run his downpointing PRT-5 almost continuously for a year because of a heating and mirror-wiper system which weather-proofs the instrument. The radiometer has a 3° aperture (compared to the present .14°). In 20 rotations (@ - 10 s/rotation) of A0=3° he i s able to sample the entire sky from 0=5.5° to 0=65.5°. No dividing of the signal into A<J> slots i s necessary, as the signal i s contin-uously processed from analogue to d i g i t a l form. The integrated cloud, by height class, is obtained by weighting the signals at each 0 by sin0 /sin0 m; where 0 =64°, the aperture centre line zenith angle for the twentieth max • 6 rotation. Processing i s instantaneous. Werner's calculations indicate that with a response time of .05 seconds (Barnes speed 2) the given rotation speed provides effectively 100 41 separate sky samples for maximum 9. Data loss when switching 6 i s about 30° in <}>. His equipment and sampling/data processing technique are well suited to the experiment. Perhaps the total sampling time of 3 to 3% minutes i s somewhat too long. And there i s no explanation for his not using the Barnes in i t s fastest response position. However, the weakness in that experiment, as in this one, i s the application of the discrimination radiances to the radiometer output. Werner does not temporally vary these values except by season. It i s shown in chapter 4 that significant changes do occur on time scales as short as one hour. He asserts (Werner, 1972) that variations in sub-cloud atmospheric layers only affect the discrimination of high clouds. Results here indicate those layers also affect middle clouds. He also states that cloud radiance depends fundamentally on cloud thickness, through the e versus H relationship of Zdunkowski and Choronenko (1969). The present view seems to indicate the greater importance of LWP, the total water path, but this perception postdates Werner's work. With respect to zenith angle effects, two of the discrimination radiances given in Figure 12 of Werner 1973b (1 and 2, or NQ^ and here) show much less limb brightening than in this present work. Limb brightening from 9=0° to 60° is less than 1 W for both NQJ and This increase with 9 is about the same amount as the analogue. This contradicts the observation in Werner (1972) that limb behaviour varies for clouds of different height classes. The inconsistency may be the result of inferring the limb behaviour of a 9.5-11.5 um PRT-5 from experiments done with a 8-13 um PRT-4 (Figures 4.11, 4.12>Werner 1972). In chapter 6 some comments are made about Werner's results. CHAPTER FOUR RESULTS—CLEAR SKY (BACKGROUND) RADIANCE—N 4.1 Rationale As w i l l be shown in chapter 5, the modelled downwelling radiance at the surface from any cloud varies with the atmosphere in which the cloud i s placed. The dominant atmospheric parameters controlling this variation are the vertical profiles of the humidity mixing ratio, q(z), and the temper-ature, T(z). The radiance from a clear sky, N q , i s primarily a function of q(z) (see 4.5). Its dependence on temperature (and thus season) is mainly indirect, through the relationship between saturation vapour pressure and temperature. N determines the values of the cloud discrimination radiances o (except for ^ . j , under one set of assumptions). Understanding the time variations of N q allows some quantitative assessment of the cloud mapping methods. Changes in N q (both measured and modelled) are examined over 3 time scales: hour to hour, within a day; diurnally within a synoptic regime (of presumably conservative T(z) and q(z)); and between seasons. 4.2 Time Variations of N o 4.2.1 Within Daylight Hours Measured N q data (daylight hours only) from both August, 1977 and July, 1978 (Figures 4.4-4.8) reveal considerable changes on this time scale. On August 10, 1977 a variation of ^ 1 W at 6=50° i s noted over the day. Two changes of ^ .25 W are recorded, each in less than one hour. On August 8 variation over 4 hours reaches 0.4 W. On July 13 and 14, 1978 (Figures 4.5 42 43 and 4.6) similar changes occur over periods between 30 and 75 minutes. There are other similar occurrences. No regular pattern can be discerned in the variations. No sunlight or time of day dependence i s noted; the changes, both increases and decreases, appear unexpectedly. In summary, changes of 0.25 W are not uncommon on this scale, with extreme values on the order of 4 times larger. The former values are greater than experimental errors with no mirror effect (Table 3.1), and comparable to errors with maximum mirror effect. The latter values of N q change exceed experimental errors by a factor of at least 3. 4.2.2 Diurnal and Day to Day Both modelled (Figure 4.1, Table 4.1) and measured (Table 4.2) results show these variations are larger than those occurring in daylight hours. Figure 4.1 shows near zenith modelled N q (from RADIANCE model, Kuhn and Stearns, pers. commun.) at the surface and 100 kPa (z ^ 100-150 m) from atmospheres based on an August 1977 series of routine 12 hour radio-sonde ascents at Port Hardy, B.C. (Figure 3.5). Atmospheres denoted D (day) are for 0000 Z (1700 PDT); those denoted N (night) are for 1200 Z (0500 PDT). There are two pairs of traces, one for each of the B and P continuum parameter sets discussed in 2.1.3. Both sets of parameters are used in the N q analysis since they seem to bracket the measured values. The data are presented at two elevations to show the magnitudes of variations in the near surface layer. For a given atmosphere, B based values exceed P based values by between 1.5 and 2.7 W. Table 4.1 l i s t s 12 hour changes for both B and P series. One third of the B values (and two of 14 P values) exceed 1.0 W. Note the small number of absolute values <_ 0.3 W. If 24 hour changes are examined, l i t t l e reduction occurs in mean change. Table 4.2 shows measured N values for six 44 Continuum Parameters 2.0L 1.6h 1.2h e—extrapolated from 101 kPa i 100 kPa* Surface B * ® ® YZT B ! • p B ® UIL T . -4(model) equiv I -40°C. _* 1 i_ - i 1 1 1 1 1 1 1 1 1 ' ' • -50°C. -60°C. -70°C. -80°C. Dl NI D2 N2 D3 N3 D4 N4 D5 N5 T 00 z 8 Aug. 77 TIME N6 D7 N7 D8 N8 D9 t 00 z 16 Aug. 77 Figure 4.1 Clear Sky Radiance Time Series at Port Hardy (YZT) — August 1977 45 TABLE 4.1—Modelled Radiance Changes Over 12 Hour Intervals (Ordered) Continuum Parameter Type Data from Figure 4.1 Port Hardy August 1977 Surface Values m "2 "IN (W m sr ) -1.4 -1.1 -0.7 -0.6 -0.4 -0.3 (3 occurrences) -0.2 +0.4 +0.9 +1.5 (2 occurrences) +2.1 -0.8 -0.6 -0.5 -0.4 -0.2 (3 occurrences) -0.1 0.0 +0.2 +0.5 +0.8 +1.1 +1.2 TABLE 4.2—Measured Clear Sky Radiance N on Days When Cloud Maps Drawn 1 2 Day Date Radiance (Range) Limb Brightening 6 (W m~2 s r _ 1 ) (W m 2 sr 1) 106 July 11, 1978 *2.50 1.9 107 July 13, 1978 *4.05-4.35 2.6 108 July 14, 1978 *2.90-3.45 2.6 110 July 24, 1978 3.45 2.5 111 July 26, 1978 *4.30 (est.) 3.0 112 July 27, 1978 *3.00 (est.)-3.35 2.5-2.7 —9=10° —For definition, see 4.3.1 *—Figures 4.4-4.8 Day 110 data not displayed 46 days in July 1978. In the f i r s t 3 days (of the same week) changes up to ^ 1.9 W are noted; in the last 3, the changes are ^ 1.3 W . These values are similar to the modelled ones above. It i s d i f f i c u l t to separate diurnal from synoptic regime variations with these data. One way to- attempt this i s to examine the vertical profile of N as a time series (Figure 4.2: Note reduced abscissa). Since N o o depends strongly on q(z), and since changes in q above the mixed layer often accompany synoptic changes, variations in N q at 50 or 70 kPa would appear to be a good indicator of synoptic change. However, the noticeable changes from N3 to D4 and N6 to D7 (both at altitude and at the surface) and D5 to N5 (at the surface) do not accompany any documented synoptic changes. Though several low pressure disturbances passed through the general area (Appendix B), the Figure 4.1 data represent one slightly modified synoptic regime (clear, with occasional cloudy intervals, some fog, and a general warming over time). The ranges for N q surface variation for the whole time series (2.4 W and 1.4 W in the B and P cases respectively) thus represent approximate upper bounds for radiance changes at this time scale. 4.2.3 Seasonal Figure 4.3 summarizes measured and modelled N results for the months ° o June to November, including 2 summer periods. For data at 9=60°, a range of 5.5 W is noted. Smallest values occur in late f a l l (November) with maxima in the two summers as expected. (The November model results are indirectly confirmed by hand measurements with the Barnes which show 'on-scale' readings (Nq> 2.7 W) only for 9> 75-80°.) Two conclusions can be drawn from these data. F i r s t , the changes between June and August are partially diurnally/synoptically related, and partially due to increases, over time, in temperature and saturation vapour 47 August 1977 Port Hardy ^ 02 ' 1 ' 1 1 1 1 1 1 1 1 1 1 ' - 0 2 NI D2 N2 D3 N3 D4 N4 D5 N5 (D6) N6 D7 N7 D8 N8 12 Z 12 Z 8 August TIME 15 August Figure 4.2 Clear Sky Radiance Profile Time Series 48 Measured 10 Modelled ( B parameters) J July 1977 N November 1977 # October 1977 u w I O •,a A A X October 11, 1977 October 17, 1977 August 1977 July 1978 July 14, 1977 June 1-6, 1978 10 20 30 40 50 60 70 ZENITH ANGLE (9) (DEGREES) minimum on-scale radiance for Barnes PRT-5 80 Figure 4.3 Seasonal Changes in Modelled and Measured Clear Sky Radiance Versus Zenith Angle 49 pressure of a deep layer from the surface to the mid-troposphere: 5.5 W i s too great a range to be explained by only diurnal/synoptic changes. In the spring and f a l l , changes in radiance from warm Pacific Maritime air to cold Arctic Continental air may occasionally approach this value. The second conclusion is that for much of the year, at this location, a radio-meter with a much lower threshold radiance (on the order of 0.5 W in the 9.5 to 11.5 um band) is needed to evaluate N q (and, by inference, to map clouds using a technique dependent on N q ) . If a wider band is used to increase radiance signal, larger absolute variations in N occur. How o such variations would compare to cloud signals is a question for other investigators. 4.2.4 Implications The sizes of the above variations indicate that N q must be computed (or observed) at least once a day, to bring N q uncertainty to between 0.3 and 1.0 W (within a factor of 3 to 10 of the measurement error ^  + 0.1 W— (Table 3.2)). Observation of the variable several times a day is recom-mended (5.5). Figure 5.5 shows that Werner (1973a) also considers seasonal variations important in this type of analysis. The seasonal differences noted in the present study are much larger than Werner's values of 1.0-1.5 W (for NQ^, which is assumed to closely follow N Q ) . 4.3 Angular Dependence of N u o This section deals primarily with zenith angle (6) dependence of N Q . Azimuth (cf>) dependence, observed at high 0 in some cases, i s discussed in 4.5.5 and 5.5. When discriminating clouds using radiances based on N q , the limb (9) behaviour of N i s fundamental. Uncertainties in limb variation o must be combined with those in the near zenith N q values when assessing map results. 50 4.3.1 Modelled N Results—July 1978 o Figures 4.4-4.8 show both measured and modelled N q values for the 5 days in July, 1978 on which cloud mapping i s done. Ignore for the moment the values of the radiances plotted. Consider only the increase in N q with 9, or limb brightening—6, between 9=8° and 9=62°: 6 = N (62) - N (8) (4.1) o o 6 i s uniformly larger for B models than for P models. B values are from 3.0 to 3.5 W, while P values are from 1.5 to 2.0 W. For the August, 1977 soundings at Port Hardy, the 6 ranges are (2.6,3.9 W) for B cases and (1.4,2.6 W) for P cases. Higher 6 values correspond to higher N q values because 6 is really a measure of the increased path length through the atmosphere's emitters at higher 9. On days of higher N q there i s more emitting material, and the increase in path length has a greater absolute effect. If the radiometer i s situated higher in the atmosphere, modelling indicates that <5 w i l l decrease. The 6 decrease between the surface (p=101.5-102 kPa) and 100 kPa i s small (the differences in N at 9=62° are within ' - o 0.1 W of those at 0=8°, as shown in Figures 4.4-4.8). However, extra-polation of surface results to elevations more than say 500 metres i s not recommended. The model employed here can be used to consider elevated conditions. Substantial additional limb brightening occurs for 9 in the range (60°,75°)(Figure 4.5). Because the model assumes a plane parallel atmos-phere, no attempt is made to examine N q values at 9> 75°. (The cutoff angle i s chosen arbitrarily to coincide with the maximum usable cloud map angle during the measurement periods.) When choosing atmospheres in which to model clouds, i t i s important to decide which continuum approximation 51 July 11, 1978 Day 106 0 Measured Data (^ 0700 PDT) * Model 0500 PDT at Port Hardy modified with a local (Vancouver) surface layer B —Bignell-Kuhn continuum parameters P —Revised Piatt continuum parameters ® Model 1700 PDT, modified synoptically 1 Radiance value at 100 kPa 0 I I I L. 1 1 1 -J • 0 10 20 30 40 50 60 70 80 ZENITH ANGLE (9) (DEGREES) Figure 4.4 Clear Sky Radiance Versus Zenith Angle—Day 106 July 13, 1978 Day 107 Figure 4.5 Clear Sky Radiance Versus Zenith Angle—Day 107 53 July 14, 1978 Day 108 0 I i i i _ _ i 1 1 -J ' 0 10 20 30 40 50 60 70 80 ZENITH ANGLE (6) (DEGREES) Figure 4.6 Clear Sky Radiance Versus Zenith Angle—Day 108 July 26, 1978 Day 111 Figure 4.7 Clear Sky Radiance Versus Zenith Angle—Day 111 July 27, 1978 Day 112 O Measured data 1035 PDT ^ Measured data 1210-1350 PDT • Measured data 1120-1123 PDT X Model 1700 PDT (YZT) <S) Model f0500 PDT (YZT) L 1700 PDT 26 JULY (UIL) 0 10 20 30 40 50 60 70 80 ZENITH ANGLE (9) (DEGREES) Figure 4.8 Clear Sky Radiance Versus Zenith Angle—Day 56 best represents the atmosphere. The d i f f e r e n c e i n limb behaviour (as i n d i c a t e d by 5) between B and P cases i s at l e a s t as important as the d i f f e r e n c e i n a c t u a l c l e a r sky radiance values. 4.3.2 Measured N q R e s u l t s — J u l y 1978 The same f i g u r e s (4.4-4.8) show measured N q values f o r cloud mapping days. Measured 6 values are i n Table 4.2. There i s a general c o n f i r m a t i o n that higher 6 values accompany higher N q v a l u e s , but the data set i s too small to d e r i v e a meaningful q u a n t i t a t i v e r e l a t i o n s h i p (see Figure 4.9). The range of measured <5 values l i e s between the ranges of B and P models. The continuum emission c h a r a c t e r i s t i c s of the a c t u a l atmosphere are t h e r e f o r e bracketed by those of the two models. 4.4 Comparison of Modelled and Measured Values R e s u l t s f o r the 5 days (Figures 4.4-4.8) i n d i c a t e that P models c o n s i s t e n t l y underestimate measured values w h i l e B models overestimate on 3 days and estimate w e l l ( w i t h i n ^ 0.5 W) on the other two. The magnitudes of B overestimations are, on average, l e s s than those of P underestimations. The c o n c l u s i o n i s that the B models are the b e t t e r estimators of the observed atmospheres. Therefore, when modelling clouds on p a r t i c u l a r days (e.g. f o r d e r i v a t i o n of i n 5.4.3) B models only are used. When d e r i v i n g r e l a t i o n -ships expected to apply to any day, weighted averages of the B and P models are used. Determination of a 'true' or 'best estimate' continuum parameter set f o r t h i s IR band must be a p r i o r i t y f o r f u t u r e work. The attempt to model the c l e a r sky radiance at the measurement s i t e to the accuracy necessary f o r d e r i v i n g usable N t o — 0.1-0.3 W) i s not s u c c e s s f u l . A l l of the maps presented ( i n chapter 5) use N q values derived from measurements e a r l y on that mapping day. Modelled N q values are not used because, f o r the 5 days i n question, there are no soundings at Figure 4.9 Measured Limb Brightening Parameter (6) Versus Measured Near Zenith Clear Sky Radiance (N ) 58 0500 PDT (or 1700 PDT the previous day) which are complete, and which give modelled N 's close to the measured values. The two cases where the B models o do estimate well include one sounding which occurred at the end of a mapping day, and one where the humidity data for 90 to 75 kPa i s missing, and therefore estimated. When N q values from these two modelling attempts (x's on Figures 4.5 and 4.7) are used for mapping, the resulting maps are indistinguishable from those using the measured N Q'S, despite the dif f e r -ences in limb behaviour noted in 4.3. (The good agreement of N q derived from the estimated humidity data with the measured N q values i s s t r i c t l y coincidence.) Future mapping and modelling should include on site sounding of the whole atmosphere at a 12 hour interval, plus more frequent sampling of the lowest 2000-3000 metres. 4.5 Sensitivity Analysis of to Atmospheric Factors This section shows how changes in N q are related to changes in atmospheric variables. The fundamental variable is water vapour. 4.5.1 Water Vapour Mixing Ratio (g) To test the N Q—q relationship, q 2 i s defined as the mean mixing ratio from 50 to 100 kPa, sampled at 5 kPa intervals. Model atmospheres (B parameters) yield the following relationship: N Q = C l • ? 2 + c 2 (4.2) cl = (1 + .05) x 10~3 (W nf 2 sr" 1) c 2 = -0.40 +0.15 ( " ) (see Figure 4.10). For P parameters, Cj^ and c 2 are about k the B parameter _ _3 -1 * values. Values of q 2 of between 3 and 6 (x 10 kg kg ) correspond to the summer data used to derive (4.2). If appropriate upper air data are Assumed units for a l l subsequent q values. 59 'u 4 CM I a 3 X August 1977 • July 1978 / /^•eye f i t to y August 1977 data eye f i t to July 1978 data N Q= .99 q 2 - .49 ( W.m 2 sr l) — - 2 - 1 N = 1.04 q, - .24 ( W m sr ) ^-(1.15,.60) November 1977 3.0 4.0 5.0 6.0 - -1 -3 (Mean q in layer q 2 (kg kg ) x 10 5 0 t Q 1 0 Q ^ Figure 4.10 Dependence of Modelled N q on q 2 60 available, without computer f a c i l i t i e s , i t is recommended that (4.2) be used to provide a f i r s t order estimate of N q. This value should be replaced by modelled, or preferably measured, N q values when available. (Equation 4.2 correctly predicts N q for a November model atmosphere with q^-l.lS, therefore i t may be useful seasonally, as well as diurnally in summer. Twelve hour variations in q^ of between 0.3 and 2.0 occurred in the August, 1977 data.) 4.5.2 Temperature (T) An analysis similar to that of 4.5.1 yields no significant relation-ship between N q and (defined analogously to q^). See Figure 4.11. On a seasonal basis, the relationship between temperature and vapour pressure does cause lower values of N q to accompany the lower values of (4.2.3). However, diurnally no estimate of N q changes with T^  can be made. More data from different seasons and locations are needed to test this tentative conclusion that there i s a lack of correlation between N q and temperature. 4.5.3 Ozone This experiment's active sensing band includes part of the 1043 cm 1 ozone band. Therefore, i t i s appropriate to examine N q sensitivity to changes in ozone distribution and amount. The model employed throughout this analysis has a constant ozone mixing ratio profile (Table A.5). To test for ozone sensitivity, two extreme profiles representing maximum and minimum ozone mixing ratio at a l l levels are used. These profiles (Table A. 5) are synthetically derived from the sample extrema of 7 separate profiles from different authors and latitudes (Figure 4.12). Above 20 kPa, a wide range of values i s represented. Presumably this range brackets the climatological variations at the experimental site. The maximum values below 20 kPa (tropospheric) are adapted from Johnson (1979), which 61 6 r u w 4 C M 1 s 3 X A u g u s t 1977 • J u l y 1978 9=8V x X X x X T 0 N 2 o ( - 1 2 . 6 , 0 . 6 ) November 1977 10 T 2 ( " C . ) 15 F i g u r e 4 . 1 1 Dependence o f M o d e l l e d N q on 62 100 © x D Kuhn (Radiance) Roewe Mid-latitude Winter Roewe Tropical AFCRL Panama 20/3/63 AFCRL Thule,Greenland 27/2/63 Walshaw and Goody W.Germany 2/8/51 Walshaw and Goody " 11/7/51 2.5 5.0 7.5 10.0 12.5 15.0 V — ' W - J .1 17.5, 25.0 OZONE MIXING RATIO (x 10~ ) Figure 4.12 Ozone Mixing Ratio Versus Atmospheric Pressure 63 examines stratospheric ozone penetrations to near surface levels. (Highest observed values at 3 levels are used to approximate values at the 3 nearest levels in the model.) The authors conclude that "many ( i f not most) low pressure troughs regularly inject large quantities of ozone into the tropo-sphere over wide regions as they move eastward across the United States." (Ibid., p. 361). The sensitivity analysis indicates that ozone changes above 20 kPa result in N q changes of + < 0.02 W. Tropospheric ozone gives additional changes of + 0.13/- 0.01 W. While both negative changes are negligible, for this analysis, the positive change in the troposphere is marginally significant. The 'routinely observed' change in N q because of a tropo-spheric ozone intrusion is l i k e l y to be significantly less than 0.13 W, which i s within the instrument error of this experiment. Horizontal mixing (at altitude) between intrusion and observation w i l l lower concentrations. Also, the values used in the analysis are maxima for each observing ele-vation. Spatial averages are certainly lower, perhaps by a factor of 2 or 3 judging from Johnson's profiles. Thus, some N q variations of ^ 0.05 - 0.1 W may be attributed to ozone. It i s unlikely that the majority of the N Q variation noted above (4.2.1 and 4.2.2) is due to ozone. It is assumed that the objections of Walshaw and Goody (1956) to the use of a mean pressure to characterize a vertically extensive emitting layer are answered by the multiple layer nature of the model. 4.5.4 Nascent Cloud During the change from atmospheric water vapour to mainly liquid water or ice clouds, transitory invisible sub-micron pre-cloud droplets are formed (Rogers, 1976). These droplets, generally not pure water, have highly variable radiative properties (Deirmendjian, 1969). Such droplets 64 may e x i s t i n a reg ion of the sky perce ived as c l e a r from the ground. Because t h e i r r a d i a t i v e behaviour i s presumed to be Intermediate between that of gaseous and l i q u i d water, these smal l d r o p l e t s , when they occur , cause an increase i n perce ived c l e a r sky rad iance . I t i s not p o s s i b l e to tes t the s e n s i t i v i t y of N q to t h i s v a r i a b l e us ing the present model. 4 .5 .5 Urban E f f e c t s T h i s experiment takes p lace a few ki lometres west of a major urban c e n t r e . What are the e f f e c t s on N q , i f any, of urban r e l a t e d changes i n temperature, humidity and aerosols? I f s i g n i f i c a n t , are these e f f e c t s present over the e n t i r e sky, as viewed from the research s i t e , or are they r e s t r i c t e d to some ranges of 9 and <f>? Azimuth variations in N Q ( for 8 greater than about 4 0 ° ) of from 0.1 to 0.7 W are observed on some days, with l a r g e r v a r i a t i o n s o c c u r r i n g for ( 6 0 ° <9< 7 5 ° ) ( F i g u r e 4 .13) . u <° 1.5 C M I e l.0|_ ~ 0.5 o z 0.0 10 Data for 1614 PDT 8 August 1977 20 3fl 40 50 ZENITH ANGLE (9) (DEGREES) 60 70 80 Figure 4.13 Azimuthal Change i n N Versus Z e n i t h Angle 65 Because the higher variations are at higher 9 , the cause must be surface related, since at higher 9 a larger portion of the optical path i s below the top of the surface affected layer. To investigate temperature and humidity effects, low level urban and 'rural' soundings (below 75 kPa) from St. Louis (Spangler and Dirks, 1974) are combined with upper layer information from Port Hardy, B.C. The N q data from four such a r t i f i c i a l soundings at dawn and near noon are shown in Figure 4.14. The 'rural' stations are 20 to 30 kilometres from the Arch in central St. Louis. There are no significant urban/rural differences in average mixing ratio or temperature at any level at either dawn or noon. At both times urban/rural radiances are negligible ( < 0.1 W). More examples of urban/rural comparisons are needed to confirm those observations. _3 Between near dawn and near noon (^  5 hours) an average drying of 1.3 x 10 kg kg 1 and a warming of 2-4 °C. occurs in the layers below 80 kPa. The effect on urban N q values i s a decrease of from 0.5 W (P model) to 0.8 W (B model). Because the q and T changes do not represent extreme conditions, the indicated significant diurnal radiance changes may not be an isolated occurrence. This N q decrease due to daytime drying may explain some of the magnitude changes described in 4.2.1 and 4.2.2. Not enough vertical infor-mation is available to verify that contention with the present data. The main conclusion i s therefore that azimuthal variations of N cannot be o explained by T and q variations. Aerosols are the f i n a l 'clear sky' radiator to be examined. Data from Ackerman .et-al. (1976) for 'light' and 'average' aerosols are modelled as a 5 kPa (429 m) layer based at 100 kPa. Ackerman considers the 'light' aerosol to be essentially a no aerosol condition. Here i t is considered a lower limiting case. The 'average' aerosol (for southern California) i s Synthetic Atmosphere 75 kPa and higher. Lower levels from Spangler and Dirks (1974). 66 u CO CM I e 3 U — Arch (urban St. Louis) 0700 CDT R 1 — Troy (rural) 0640 CDT U 2 — Arch 1230 CDT R 2 — B e l l e v i l l e (rural) 1130 CDT A l l data for 12 August 1971 Elevations in kPa 6=0° B Parameters P Parameters Surface 100 95 90 70 Surface 100 95 90 .2 R, R. R. Figure 4.14 Synthetic St. Louis Area Clear Sky Radiances (refer to 4.5.5 for explanation of synthetic) 67 taken as the upper limiting case. The increases in N q from clear to 'with aerosol' cases f a l l in the range 0.3-0.6 W for 5 kPa of the light aerosol (increasing with 9 from 8° to 75°): increases for the average aerosol are 1.0-1.5 W. It is assumed that the radiometer in this experiment views unpol-luted maritime air to the west and anthropogenically altered a i r to the east and north-east (including the urban/regional plume which generally tracks east of the city centre (hatched area in Figure 3.1) and the indus-t r i a l areas further east). The <j> variations in N q noted above may be due to aerosols contained in the latter a i r , but no local data exist to support the contention. The calculated aerosol related changes in N q are generally larger than the observed changes in N q with <J>. This may be due to inapprop-riately high aerosol extinction or layer thickness values being used in the model. It also could be caused by the radiometer viewing only a fraction of the thickness of the aerosol layer. This w i l l in fact occur unless the urban plume i s directly over the radiometer at UBC, a rare event which re-quires down valley flow (east to west) to exceed up valley flow (west to east) over a 24 hour period. The sea to land and up valley (daytime) circulation is generally stronger and more pronounced than the down valley and land to sea (nocturnal) circulation on days when aerosol concentrations are l i k e l y to be high (those dominated by weak anti-cyclonic circulation and generally clear or partly cloudy skies). If one assumes that urban related aerosols are generated primarily during daylight hours, then the former c i r -culation takes the plume further to the east during the day than the latter returns i t to the west at night. The upper (return) portions of these circulations (Hay and Oke, 1976) are ignored in the above analysis. If part of the plume does pass directly over the radiometer, time variations of 68 near zenith N q may occur as air columns with markedly different aerosol concentrations are sensed by the radiometer. Contrasts between continental and maritime aerosol loading of air masses are not considered. 4.5.6 Elevation, Near the Surface N variation between 100 kPa and the surface (Figures 4.4-4.8) i s o 0.2-0.4 W near the zenith and 0.3-0.6 W at 6=62°. Surface pressures vary from 101.5 to 102.0 kPa, yielding vertical N q gradients of 0.1 to 0.4 W/kPa. Because surface pressure varies significantly from Vancouver to Port Hardy, because of the elevation of the radiometer (see 3.1.2), and because of the importance of the surrounding surface in determining sources of radiance, i t is f e l t that the differences between surface and 100 kPa modelled radiances represent a f i r s t order estimate of the data extrapolation over large horizontal distances. The actual atmospheric pressure at the site need only be measured when a simultaneous on site sounding i s performed. 4.6 Summary In summary, this chapter shows that clear sky radiance (Nq) depends primarily upon water vapour amount and distribution, and that the limb brightening of that radiance depends on the N q value. Temperature, though important for cloud radiances, appears in i t s e l f to be relatively unim-portant in determining the radiance of a clear sky. Ozone and urban effects, except for aerosols, have transient negligible effects on N q over a l l time scales examined. Urban aerosols in relatively high concentrations can cause azimuthal increases in N in the direction of the city. Such o variations can be restricted to 9 > 6 by situating the radiometer far max enough upwind of the main aerosol source. (Upwind refers to the predomi-nant wind direction during the synoptic regime(s) conducive to high aerosol concentrations. The upwind distance in this case i s about 15 km.) 69 If the polluted layer i s over the radiometer, systematic increases in N q and 6 would be expected. Inhomogeneities in the aerosol concentration may cause variations in N q of the same order as those observed during daytime periods in this experiment. Predicting N q from upper air soundings requires the radiometer and the sounding station to be much closer than the 150-350 kilometres in this experiment. In practice, equation 4.2, using measured values of N Q for some day i , and estimates of q^ on days i and i+1, may be a better predictor of N q than modelling based on distant soundings. This concept has not been tested. CHAPTER FIVE RESULTS—CLOUDY SKIES In this chapter the clear sky radiance results of chapter 4 are combined with modelling results described in 5.1-5.3 to produce equations for discrimination radiances (N ) in 5.4. The latter are used to make the st cloud maps of different sky types (5.5, 5.6) and the map series of 5.7. In 5.8 a brief evaluation i s made of overall map performance. 5.1 Context for Cloud Modelling The forms of the relationships among clear sky radiance, zenith angle and discrimination radiances are not obvious and have not been previously stated ex p l i c i t l y by other workers. In 5.1-5.4 the derivation of one set of such relationships i s given. 5.1.1 Methodology To derive N — N Q — 9 relationships, clouds which approximate border-line conditions between cloud classes are treated at 3 values of 9. The clouds were placed in two atmospheres which exhibit extremes of N q (Figure 4.1). Downwelling radiances at both cloudbase (N ,) and surface (N ) were zb s examined under the two sets of continuum parameters (2.1.3), using the model described briefly in 5.1.2. The N G values were generalized to deter-mine the form and values for and N ^ J the relationships were then applied to particular days. was approached similarly, yielding method A values (Figures 5.1 and 5.2). These values were thought to be suspect since, when used on Day 111, very thick low clouds were interpreted as middle clouds. So a second method (B) was tried. (B on cloud maps in 5.5-5.7 refers to method B, not to Bignell continuum parameters.) For method B, 70 71 0 10 20 30 40 50 60 70 80 ZENITH ANGLE ( 6 ) (DEGREES) Figure 5.1 Discrimination Radiances Versus Zenith Angle—Day 107 72 0 10 20 30 40 50 60 70 80 ZENITH ANGLE (9) (DEGREES) Figure 5.2 Discrimination Radiances Versus Zenith Angle—Day 112 73 borderline low/middle clouds were modelled in atmospheres which best approximated the measured clear sky radiance for each mapping day. The atmospheres were taken from modified and unmodified soundings from Port Hardy and Quillayute (Figure 3.5) up to 24 hours before and 6 hours after sky scans were performed. The sounding which appeared from the synoptic charts to be the most representative one occurring before 0600 on the day in question was generally used. When not available, a 1700 sounding for the day was used. The results of this analysis for a variety of scanning ( J ) and non-scanning days are plotted on Figure 5.3. The independent variable is T, the average temperature in the atmospheric layer 70-90 kPa, the range of base heights of borderline low/middle clouds. The linear relationship indicated is not used in the present analysis because of the lack of on site T measurements, but i s suggested as a useful tool for future work. i s also weakly dependent on q (average mixing ratio in the layer 80-100 kPa). — -3 -1 — In the absence of q changes of more than 2 x 10 kg kg , T variations greater than about 1.6 K cause changes which significantly affect the map analysis. T changes over 12 hour intervals for the series in Figure 4.1 range from 0.1 to 2.7 K, with 80% less than 1.4 K. If T can be measured or estimated within 12 hours of a scan, the above evidence indicates that deriving from T is a good f i r s t approximation i f the more detailed infor-mation needed to model i s not available. Differences between method B results and the N^^ to T relationship are +.05, -.10, +.35, +.40 and +.50 W, for the 5 measurement days. Though some mapping differences result from changes of that magnitude, the simpli-city of the —T technique recommends i t . (With a sufficiently large data set at any location, the — T — q relationship, or one involving different Figure 5.3 N.„ Versus T (Modelled Atmospheres) 75 T and q variables, might be better determined.) 5.1.2 Modelled Clouds Table 5.1 shows the clouds used in the modelling. They are treated as homogeneous ($^v> T, q constant with height) i n f i n i t e horizontal layers with bases and tops at some pressure level. (For conversion of pressures to altitudes in metres, see Table A.1.) Cloud T and q are equal to the mean of the ambient values at cloud base and top, except when the cloud thickness exceeds 5 kPa. In such cases, T and q for each 5 kPa of cloud equal those of the clear atmosphere for the same layer. Note that the mass absorption coefficient for ice clouds is approximate (note 4, Table 5.1, after Paltridge and Piatt (1976)). Therefore ice cloud inputs and results are less certain than those for water clouds. Cloud MEDCD represents altostra-tus, while other middle clouds are altocumulus (after Piatt and Bartusek (1974)). The low clouds are stratocumulus (Piatt, 1976), while the high clouds represent a range of cirrus (Piatt, 1973). 5.2 Modelled Cloudbase Radiances (N , ) —zb 5.2.1 Low and Marginal Low/Middle Clouds Table 5.2 (top) illustrates N , values for three such clouds, under zb a variety of conditions. Note the negligible ( < 0.1 W) difference with choice of continuum parameters. Limb brightening, 6(N i s from 2 to 3 W. Values of are higher for the atmosphere (SD7) with the higher N q. There is l i t t l e relationship between N , and N . The N , difference is primarily r zb o zb due to differences in cloud temperatures of from 3 to 7 K between the soundings (Tables A:3 and A.4). Vertical temperature variations account for the variation of N ^  with z, . Table 5.3 illustrates that for cloud LOWCC, the lower based case b (zK=95 kPa) has a higher radiance with slightly more limb brightening. 76 TABLE 5.1—Modelled Clouds 12 3 Cloud ' w 6 T e H W = H-w (g m J) A V (km) -2, (km-1) (§ m > LOWCA .24 19. .67 LOWCB . 18 14. 2.1 LOWCC5 .05 4.0 1.3 LOWCD .13 10. 3.3 MEDCA5 . 10 8.0 1.6 MEDCB5 .005 .40 .24 MEDCC I 4 ' 5 .044 .80 1.0 MEDCD .014 1.1 2.0 HIA I 5 .011 .20 .05: HIB I 5 .019 .35 .32 HIC I 5 .028 .50 .99 .49 (.27) .035 8.4 .88 (.83) .15 27. .74 (.84) .33 17. .96 (.96) .33 43. .80 .20 20. .21 .60 3.0 .65 1.3 (57.) .86 1.8 25. .054 .28 (3.0) .28 .93 (18.) .63 2.0 (55.) 1 Low clouds are based from 75 to 95 kPa, middle clouds from 55 to 70 kPa and high clouds from 25 to 45 kPa. 2 LOWCC base of 75 kPa = L2 HIA base of 40 kPa = Hi MEDCA " " 70 kPa = Ml HIB " " 40 kPa = H3 MEDCB " " 70 kPa = M2 HIC " " 40 kPa = H2 MEDCC " " 60 kPa = M3 3 From relationship shown on Figure 7 of Piatt (1976). The indiv-idual cloud e's measured by Piatt are shown ( ). 4 2 - 1 2 - 1 k^=0.08 m g for water clouds; 0.018 m g for ice clouds, I. A l l clouds not designated I are water clouds. 5 Indicates cloud used to derive discrimination radiances. 77 TABLE 5.2—Surface and Cloudbase Radiances for Marginal Low/Middle Cloud Cases -2 -1 Cloudbase Radiance (N , ) (W m sr ) Z D Continuum 0 (°) Sounding SN2 Sounding SD7 Parameters L 2 M l M 3 L 2 M l M 3 8 9.81 9.89 6.43 10.47 10.80 7.36 P 35 10.64 10.58 7.10 11.34 11.53 8.12 62 12.41 11.87 8.81 13.16 12.88 10.04 8 9.86 9.91 6.44 10.58 10.84 7.37 B 35 10.68 10.60 7.12 11.43 11.57 8.14 62 12.43 11.87 8.82 13.20 12.90 10.04 Surface Radiance (N ) (W m sr ) 8 10.24 10.32 7.19 11.33 11.63 8.80 p 35 11.06 11.01 7.95 12.19 12.36 9.68 62 12.78 12.35 9.89 13.97 13.79 11.91 8 10.73 10.68 8.03 12.18 12.41 10.15 B 35 11.53 11.38 8.84 12.99 13.11 11.07 62 13.15 12.73 10.91 14.58 14.46 13.29 -2 - i Clear Sky Radiance (NQ) (W m sr ) 8 1.44 2.84 P 35 1.72 3.37 62 2.88 5.41 8 2.97 5.41 B 35 3.51 6.28 62 5.57 9.29 78 TABLE 5.3—Cloudbase and Surface Downwelling Radiances as Functions of Cloudbase Height and Zenith Angle -2 -1 -2 -1 Continuum z b(kPa) N z b (W m sr ) N g (W m sr ) Parameters 6=8 6=35 6=62 6=8 6=35 6=62 Cloud LOWCC p 75 95 9.81 11.3 10.6 12.3 12.4 14.3 10.2 11.5 11.1 12.4 12.8 14.3 B 75 95 9.86 11.6 10.7 12.5 12.4 14.4 10.7 11.9 11.5 12.7 13.2 14.3 0=0 Cloud LOWCA2 0=0 P 85 7.08 7.99 B 85 7.47 Cloud LOWCD2 9.13 P 85 13.5 13.7 B 85 13.6 13.8 Using model atmosphere SN2 (Appendix A) Using model atmosphere SDl (Appendix A) 79 For this lower cloud, at 6=8°, a significant difference (0.3 W) between the B and P cases occurs because of the effect of the atmosphere behind the cloud. This effect is more pronounced on days with higher q; Table 5.3 i s for a very dry atmosphere above 90 kPa (Table A.2). This background effect decreases for clouds based higher in the atmosphere, since there i s less water vapour above (behind) them. For extremely optically thick clouds, such as MEDCD and LOWCD, this effect is minimized because N , approaches the zb blackbody radiance of the mean cloud temperature. 5.2.2 High and Marginal High/Middle Clouds Table 5.4 (top) shows N , values for two high clouds and middle zb cloud M2. The limb behaviour of the latter is similar to the middle clouds in Table 5.2, described above. Both high clouds show significantly lower 6(N^) absolutely, but roughly the same in relative terms, ^(N^) / 1^(8°). N , is independent of continuum parameters, as in 5.2.1. Variation between zb soundings is small ( < 0.2 W) for the high clouds, but considerable for cloud M2, especially at higher zenith angles. A l l variations in in these cases are small compared to the variations introduced by the inter-vening atmosphere (that between the cloudbase and a ground observer). 5.3 Modelled Surface Radiances (N ) —s 5.3.1 Low and Marginal Low/Middle Clouds Tables 5.2 (middle) and 5.3 indicate surface radiances for clouds of this type. N g for optically thick LOWCD i s affected only slightly by the sub-cloud layers, while optically thin LOWCA has N g-N z b of from 1 to 2 W, depending on the continuum parameters (Table 5.4). The effect for LOWCC, with intermediate optical properties, f a l l s between those extremes. The 9 dependence of the intervening atmosphere is small. At higher 9 , N g-N z b decreases. In Table 5.2 cloud M3 shows the largest N -N because i t i s 80 TABLE 5.4—Surface and Cloudbase Radiances for High and Marginal High/Middle Cloud Cases -2 -1 Cloudbase Radiance (N , ) (W m sr ) zb Continuum 9 (°) Sounding SN2 Sounding SD7 Parameters Hi H3 M2 HI H3 M2 8 0.39 1.84 2.69 0.41 1.94 3.07 p 35 0.47 2.15 3.18 0.49 2.27 3.61 62 0.79 3.28 4.99 0.84 3.44 5.65 8 0.39 1.85 2.78 0.42 1.94 3.26 B 35 0.47 2.16 3.27 0.50 2.27 3.83 62 0.80 3.28 5.12 0.85 3.45 5.95 Surface Radiance ( N ) (W m sr ) 8 1.77 3.08 3.81 3. 16 4.40 5.24 p 35 2.11 3.60 4.45 3.73 5.12 6.08 62 3.48 5.47 6.79 5.93 7.63 8.96 8 3.26 4.41 5.06 5.66 6.65 7.32 B 35 3.84 5.11 5.85 6.56 7.61 8.34 62 6.02 7.54 8.55 9.61 10.65 11.47 -2 -1. Clear Sky Radiance (N ) o (W m sr ) 8 1.44 2.84 P 35 1.72 3.37 62 2.88 5.41 8 2.97 5.41 B 35 3.51 6.28 62 5.57 9.29 81 highest in the atmosphere. N G variation with continuum parameters Is. important. Differences of up to 1.4 W occur in some cases, with the more absorbing B case showing the higher radiances. These differences are a key error source in e s t i -mating ^22' Limb behaviour i s almost unaffected by the sub-cloud atmos-phere and by B/P differences in most cases. 5.3.2 High and Marginal High/Middle Clouds . Table 5.4 shows that for these clouds the surface radiance (N ) i s s often more dependent on the intervening atmosphere radiance than on the cloudbase radiance. For cloud HI, which represents marginally thin cirrus, the N G values are merely the clear sky (N Q) values incremented by about 10%. This behaviour, plus empirical considerations, leads to the N^^ formulation in 5.4.1. Clouds H3 and M2 are more substantial. H3 is close to an average cirrus (Piatt 1973). M2 i s an optically thinner altocumulus placed lower in the atmosphere. Both clouds exhibit more limb brightening at the surface than the clear sky radiance for the same sounding (Table 5.4). This feature is incorporated in the derivation of Because of their strong depend-ence on N q, the N G values also vary substantially with continuum parameter and sounding. Absolute differences are larger than for the lower clouds of Table 5.2; these differences represent large percentages of the absolute values. The large radiance overlap between- thick cirrus and thin altocumulus (and presumably altostratus) is the key source of modelling error in the values. 5.4 Determination of Discrimination Radiances 5.4.1 —Clear Sky/High Cloud Boundary Radiance The relationship used i s : 82 N01 ~ N o + 0 , 3 0 ( W ) ( 5 , 1 ) ' with a l l 0 dependence i m p l i c i t l y included i n N^. This r e l a t i o n s h i p closely p a r a l l e l s the surface radiance for cloud HI under B parameters. Some increase i n the constant i n (5.1) with 0 may be indicated i n Table 5.4, e s p e c i a l l y under P parameters. Because of the uncertainty i n the i c e cloud absorption r e l a t i o n s h i p (5.1.2), such a refinement i n equation 5.1 was rejected. S l i g h t overestimation of high cloud at high 0 may be expected, mainly on days with low N Q . 5.4.2 N 1 2 — High / Middle Cloud Boundary Radiance The r e l a t i o n s h i p used i s : N 1 9 = N + c ' (5.2) 1 2 o , where Q = 2.15 + 0.01-(0-8°) W : 0 <_ 35° (5.3a) 5 = 2.42 + 0.013-(0-35°) W : 9 _> 35° (5.3b) Equation 5.2 i s derived by combining the N g-N Q values for clouds H3, M2 (from Table 5.4) and H2 (Figure 5.4) as follows: for each cloud, the four values of r, at some 0 were combined i n a 1:3:3:1 weighted average (for SN2—P, SN2—B, SD7—P, and SD7—B values r e s p e c t i v e l y ) . Then at each 0 the clouds' values were combined i n a second weighted average 2:2:1 (H3:M2:H2). Both weightings were a r b i t r a r y , but designed to emphasize median conditions. This procedure yielded three numbers at 8=8, 35 and 62° which were then l i n e a r l y i nterpolated to give equations 5.3. Error i n the above procedure was estimated by changing the weightings i n the averaging processes. An error i n t, of + 0.2-0.3 W was found. This error dominates the equation. It must \ day for any p a r t i c u l a r cloud, large misinterpretations may r e s u l t . However, be stressed N 1 2( 0 ) i s highly averaged. On any p a r t i c u l a r 8 3 84 by including the 9 dependence through both C, and N q , and including a linear variation with N , i t is f e l t that such errors are minimized. o 5.4.3 —Middle/Low Cloud Boundary Radiance A number of approaches are used to derive values for N 2 3. One set of values (method B on Figures 5.1 and 5.2) is derived by modelling cloud L2 under B parameters in the sounding which best approximates for each of the measurement days (including after the fact soundings). Cloud Ml is used as a check. This method i s empirical, s t a t i s t i c a l l y untenable and violates objective analysis criterion 2 (2.2.3). Never-theless i t was used because a l l available information was needed to relate T, q and N^. N 2 3 represents clouds of substantial emissivity .7 or .8). As cloud temperatures increase, emitted radiances increase accordingly. If clouds from some range of elevations can be misinterpreted by a temperature representing those layers should be used to set N,^ . T, used here, is the unweighted mean of the temperature at 5 kPa intervals from 70 to 90 kPa. N also affects N„„. It can either be included e x p l i c i t l y , o 15 8 i v i n g : N 2 3 = N 2 3 ( T , N q ) (5.4), or i t can be replaced by some q (the unweighted mean of q in the sub-cloud layer), q i s a surrogate for N q . It ignores the radiant effect of water vapour above the cloudbase, but is used because i t should be easily e s t i -mated from the surface or synoptic charts (or verified with minisondes). Equation 5.4 changes then to the form: „ „ ,— —. ,.. c. N 2 3 = N23(T,q) (5.5). For the 5 measurement days, q for modelled atmospheres range from 6.1 to -3 -1 7.4 ( x 10 kg kg ). For the August 1977 series the range i s 5.2 to 9.0. The corresponding T's are 4.5 to 15.9 and 10.5 to 16.7, a l l in °C. A 85 — -3 sensitivity analysis indicates a q change of 1 x 10 i s roughly equal to a cloud temperature change of 1 K. Though this i s only an order of magni-tude relationship, the above q and T indicate that changes in the latter dominate. Therefore, to a f i r s t approximation, q can be ignored in deriving ^23 ^ u t ^ o r a best estimate must be included). The eye f i t line on Figure 5.3 indicates: N 2 3 = 8.6 + 2.0(f) W (5.6). This relationship has not been tested, and should not be used without rigorous on site testing and reformulation. It does however cover data from different seasons (and q's from 2.0 to 9.0). The scatter i s presum-ably due to q since the below line excursion of the 'A' point at T= 14° C. corresponds to a low q value. No attempt i s made at a formulation of (5.4). Two attempts at relating N 2 3 to N q alone (method A) are shown on Figure 5.4 as lines 1 and 2, and as case A (from line 1) on Figures 5.1 and 5.2. The equation of line 1 (method A) i s : = 8 > 3 2 + Q w ( 5 > 7 ) > zj o This relationship yields values which agree well with the method B values on 3 out of 4 days tested, in the zenith, but the limb behaviour is quite different. For Day 106, methods A and B agree to within 0.1 W at 0=10°, but differ by 1.0 W at 6=65°. For Day 111 the corresponding values are 0.5 and 0.8 W. In the analysis of cloud scans (5.6.2), some attempt is made to determine the differences in map interpretation due to these variations, particularly at 6 ^ 50°. The limb brightening 6(N 2 3) in method A i s 1.0-1.8 W, in method B 2.0-2.5 W, and from Werner (Figure 5.5) about 1.0-1.5 W. It appears that the model, using both sets of continuum parameters, is overestimating S(N,>3). 12 r- A summer 11 10 Q N 2 3 winter • a ta ~ 7 u CO C N I a w 5 co ® » 8> ® <3 m N 1 2 summer <S N Q 1 summer • N^ 2 winter N Q 1 winter 10 20 3 0 4 0 5 0 6 0 70 ZENITH ANGLE (9) (DEGREES) 8 6 Figure 5.5 Werner Discrimination Radiances Versus Zenith Angle (a f t e r Werner 1973a). 87 Therefore a hybrid method, C, i s also tested. Method C uses method B zenith values and limb shape, but scales 6 down to about Werner's values. Because of the broad range of values and 6's from the three methods, the systematic overestimation i n method B, from using only B continuum para-meters, i s ignored. On Day 111 some thick altocumulus (z^ ^ 3500 m) were observed emitting more radiance 1.5 W more) than several d i f f e r e n t low clouds. This e f f e c t i s not v i s i b l e on the maps for Day 111 (Figure 5 . 13 ) because exceeds a l l cloud radiances on that day. 5 .5 Maps for Skies Dominated by High (Cirrus) Clouds The cloud maps are divided into 2 classes: those dominated by high clouds (HIGH); and those dominated by low (and low plus middle) clouds (LOW). This section ( 5 . 5 ) deals with HIGH scans. 5 . 5 . 1 E f f e c t of N Variations o Maps of t h i s type generally have large areas of clear sky. The <f> v a r i a t i o n s of N q cause some cl e a r sky areas to be interpreted as high cloud. This type of m i s i n t e r p r e t a t i o n i s mainly r e s t r i c t e d to the high 9 0> 5 5 ° ) regions of objective maps. (When one can see the radiance/voltage trace and correct for the v a r i a t i o n s s u b j e c t i v e l y , the problem e s s e n t i a l l y disappears.) The r e s u l t i s extra ( 1 , 0 ) contingency matrix elements ( 2 . 2 . 2 ) and increased p o s i t i v e asymmetry for that scan. In Table 5 .5 note p a r t i c -u l a r l y maps 107 BB, CB and DB where n > 1 s i g n i f i c a n t l y i n 4 of the 5 cases. N q temporal v a r i a t i o n s of 0 .2 to 0 .6 W can sometimes t o t a l l y i n v a l -idate HIGH type maps. Scan 107 I was mapped using two sets of N q values (107 B < 107 J) d i f f e r i n g by between 0 .2 and 0 .7 W depending on 9. Table 5 . 6 and Figure 5 .6 show a s u b s t a n t i a l increase i n the amount of clear sky mapped i n the higher N case (107 I J ) . The largest changes are near the 88 TABLE 5.5—Cloud Map Statistics for Objective and Subjective Analyses for Predominantly High Cloud Scans Map EP^ EP^ Symmetry1 Objective Subjective 107 BB2 15: 19 .65:.63 12/3:9/10 107 CB 18:19 .75:.79 13/5:14/5 107 DB 48 .70 41/7 107 IB 42 .61 22/20 107 IJ 36 .69 5/29 108 BA 27,254'5 .61..584 7/20,6/19 108 E*A 19:32 .56:.76 7/12:12/20 108 FA 19:44 .41:.76 10/9:23/21 108 GA 30:44 .49:.62 16/14:20/24 108 HA 20:29 .38:.53 2/18:8/21 108 HH 13:28 .25:.48 2/11:11/17 107 BBS 10:10 .56:.43 7/3:2/8 107 CBS 11:12 .65:.71 6/5:7/5 107 DBS 27 .56 20/7 107 IBS 29 .54 7/22 108 BAS 4 19,15 .44,.37 6/13,4/11 108 E*AS 22:27 .61:.69 10/12:12/15 108 FAS 18:47 .33:.76 12/6:24/23 108 GAS 22:35 .41:.56 11/11:14/21 108 HAS 20:32 .36:.54 5/15:12/20 See 2.2.2 for a description of these measures of map/photo agreement. See 3.3.3 for an explanation of map notation. x:y is read "x is the comparison of the indicated map with a photo taken just before the scan, y is for the map versus the photo just after. If only one number is given, the photo is during the scan." Same photograph, before scan, 1 f-stop difference. The f i r s t photograph is denoted 108 BI, the second 108 B2. 89 TABLE 5.6—Contingency M a t r i c e s f o r Selected Maps of Skies Dominated by High Clouds Map 107 IB Map 107 I J Map 107 IBS 0 1 2 0 1 2 0 1 2 Photo 0 34 21 0 51 4 0 49 6 0 107 I 1 8 24 . 1 16 16 1 11 21 1 2 0 12 3 1 12 2 0 11 4 Map 108 BA Map 108 BAS 0 1 0 1 0 59 7 60 6 0 Photo 1 20 17 13 24 1 108 BI 0 60 6 62 4 0 Photo 1 19 18 11 26 1 108 B2 Map 108 FA Map 108 FAS 0 1 0 1 0 57 10 55 12 0 Photo 1 9 27 6 30 1 108 F l 0 45 23 41 24 0 Photo 1 21 14 23 15 1 108 F2 90 5 a) MRP 1071B JULY 13/78 1 6 : 2 7 - 3 0 BUCH TOWER s b) MAP 107IBS JULY 13/78 1 6 : 2 7 - 3 0 BUCH TOWER Figure 5.6 KEY - HIGH <!> MIDDLE * LOW s MAP 1071J JULY 13/78 1 6 : 2 7 - 3 0 BUCH TOWER Figure 5.6 92 horizon, but substantial changes occur as well in other parts of the sky. The maps are significantly different, since one predicts 3/10 high cloud, the other 5-6/10. Map 108 HH (Table 5.5) has lower error parameters and more symmetry than 108 HA, which uses N q values about 0.25 W higher. However 108 HH misinterprets additional clear sky near the horizon as high cloud because of N q azimuth variation. In this case the maps differ insignif-icantly, with one predicting 4/10 cirrus, the other 3-4/10. Thus, using the clear sky radiance from the f i r s t scan of the day (that has sufficient clear sky) to define may result in serious map errors. If N q i s taken from calculations using off site upper a i r infor-mation, larger errors usually result. On Day 108 use of the B model indic-ated on Figure 4.6 results in unacceptable errors (EP^'s > 1, asymmetries > 40). Modelled N 's should be avoided for HIGH skies; this is usually o not a problem since such days usually have enough clear sky to use measured values. 5.5.2 Case Studies This section examines scans 108 B, 108 F and 107 I (Figures 5.6-5.8; slides 1-4; Tables 5.5-5.7)? 108 B Maps 108 BA and 108 BAS (Figure 5.7) show a sky with a small amount of cirrus in the southeast. Slide 1 (photo 108 BI) reveals that both maps have accurately resolved the major cirrus features present. (Slide 1 is the f i r s t of 2 photos taken 10 seconds apart, with the camera changed by one f-stop.) Individual hooked tufts cannot be resolved, because of system resolution. Comparing 0 and S maps, the latter eliminates the spurious When examining slides note that the viewing horizontal corresponds to the bisector of the pie shaped 'blind wedge' on the maps. Figure 5 . 7 94 s a) MAP 108FR J U L Y 1 4 / 7 8 1 5 : 0 6 - 0 9 BUCH TOWER s b) MRP 108FAS J U L Y 1 4 / 7 8 1 5 : 0 6 - 0 9 BUCH TOWER Figure 5.8 95 2 1 TABLE 5.7—x Statistics for Selected Objectively Analyzed Maps 2 Map X Statistic Map i s Significant at .05 Yes or No 2 (The x • -i n c for 1 degree of freedom = 3.84.) 106 F-B 6.0 Y 106 F-C 12.2 Y 106 I-A 9.4 Y 106 I-B 1.8 N 106 I-C 15.4 Y 107 DB 3.5 N 107 DBS 19.9 Y 108 BA 16.6:20.83 Y:Y 108 E*A 25.5:3.2 Y:N 108 FA 36.7:.37 Y:N 108 GA 17.4:2.1 Y:N 108 HA 42.2:19.5 Y:Y 108 HH 59.2:20.8 Y:Y 112 DB-A2 — : 1.15 -:N Plus subjective map 107 DBS. A l l Day 106 and 112 matrices have been grouped slightly. Photo convention as for Table 5.5. 96 cirrus at high 8, but indicates too much cirrus in the northwest. Table 5.6 shows that the contingency matrices in the 0 cases are significantly poorer than in the S cases. In the former the value of the (1,1) element i s less than the sum of the (0,1) and (1,0) elements, while in the latter the inequa-l i t y is reversed. In this case both the 0 and S matrices are s t a t i s t i -2 cally non-random at the 0.01 level in a x test (Table 5.7), with the S matrices more non-random. Comparisons of the maps with photos 108 BI (slide 1) and 108 B2 il l u s t r a t e the 'noise' in the statistics associated with the photographic technique. In Table 5.5 note changes in EP^ of 2 to 4 and in EP^ of .03 to .07. Since the photos are equally representative of the sky at the time, this magnitude of change in the EP's i s insignificant. The EP's can be considered significantly different between maps 108 BA and 108 BAS, where differences of 8 to 10 and .15 to .20 are noted in EP^ and EP^ respectively. Thus 108 BA is not performing at the limit of the technique. Tables 5.5 and 5.7 indicate that map/photo comparisons with EP^'s > .65 and EP^'s > 35-40 are s t a t i s t i c a l l y random (not significant) 2 2 according to the x test. The x test indicates that the analysis for map 2 108 BA ijs significant. Thus, i f a map is significant in the x test, this does not imply that the map i s doing as well as the method allows, but does indicate reasonable performance. 108 F Maps 108 FA and 108 FAS (Figure 5.8) and photos 108 Fl and 108 F2 (slides 2 and 3) i l l u s t r a t e the problem of interpreting cloud fields which change significantly over t^, the intrascan time. In Table 5.5 the EP^ data indicates uniformly lower values for map comparisons to 'before-scan-photos' than for comparisons to photos taken after the same scans. EP^ data are less 97 d r a m a t i c , b u t t h e same t r e n d i s i n d i c a t e d f o r most maps. T a b l e 5.7 i n d i c a t e s t h a t 3 maps w h i c h a r e s i g n i f i c a n t w i t h r e s p e c t t o a ' b e f o r e ' photo a r e i n s i g n i f i c a n t w i t h r e s p e c t t o t h e ' a f t e r ' one. The maps and p h o t o s f o r 108 F show s i g n i f i c a n t movement o f t h e c i r r u s e l ements i n t h e l o w e r h a l f o f t h e photo and i n t h e n o r t h . The 46 22 c o n t i n g e n c y s t a t i s t i c s f o r t h e two p h o t o s a r e 21 14 , EP^=43, EP2=0.75. These i n d i c a t e p o o r e r agreement between t h e p h o t o s t h a n between t h e maps and photo 108 F l . The c l o u d m o t i o n s , e s p e c i a l l y n e a r t h e z e n i t h , cause t h e s t a t i s t i c a l d e g r a d a t i o n o f t h e map c o m p a r i s o n t o t h e ' a f t e r - s c a n -photo ( s l i d e 3 ) ' . That a s s e r t i o n i s s u p p o r t e d by t h e symmetry i n f o r m a t i o n . I n a l l 4 (0 and S) c o m p a r i s o n s , t o b o t h p h o t o s , e r r o r s a r e c l o s e t o s y m m e t r i c a l . But i n t h e ' a f t e r ' p a i r ( u s i n g photo 108 F 2 ) , b o t h numbers a r e h i g h e r , i n d i c a t i n g i n c r e a s e d (0,1) e r r o r s a r e j u s t as l i k e l y as i n -c r e a s e d (1,0) e r r o r s . T h i s same c o n c l u s i o n can be drawn f r o m 108 GA and somewhat from 108 E*A and 108 HA. F o r t h o s e 4 s c a n s , no s i g n i f i c a n t s t a t i s -t i c a l improvement i s n o t e d between 0 and S maps, i n d i c a t i n g t h a t t h e 0 maps a r e p e r f o r m i n g a t t h e l i m i t o f t h e t e c h n i q u e . Under t h e c r i t e r i a o f 2.2.2 a l l t h e 0 maps i n T a b l e 5.5 appear t o be ' u s e f u l ' , b u t o n l y a few a r e ' t r u t h f u l ' . The m i d - t r o p o s p h e r i c w i n d s ( v^Q) d u r i n g t h e s e s c a n s were between about 10 and 15 m s ^. C a l c u l a t i o n s i n d i c a t e t h a t t h e a n g u l a r d i s p l a c e m e n t of a c i r r u s element between p h o t o s w o u l d be 10 t o 25° (assuming c l o u d s move w i t h t h e f l o w ) d e p e n d i n g on 6, z^ and v ( z ) between 50 kPa and z^. D i s p l a c e -ments o f t h a t o r d e r a r e o b s e r v e d on s l i d e s 2 and 3. The s t a t e d V^Q v a l u e s a r e thought t o be l o w e r t h a n t h e c l i m a t i c median f o r t h e s e c l o u d c o n d i t i o n s i n t h i s r e g i o n . A c t u a l d i s p l a c e m e n t s may s u b s t a n t i a l l y e xceed t h o s e o b s e r v e d h e r e . 98 Summarizing, to verify a HIGH sky map, photos taken before or during the scan must be used. Further, a map loses i t s 'truthfulness' almost as soon as the scan ends, and i t s 'usefulness' in about 8-15 minutes (the time for cloud elements to move from near zenith to near horizon positions). (This was confirmed visually during the experiment, but i s not photograph-i c a l l y presented here.) If the intrascan time were reduced to say half of the present value, the truthfulness of the map would increase at the end of the scan, but would decay just as rapidly after the scan. 107 I Scan 107 I (Figure 5.6, slide 4 of photo 107 I) i s discriminated into two 0 and one S maps. 0 map 107 IB uses N Q'S from the start of the recording day, while 107 IJ uses those near the time of the scan. The latter i s more symmetrical, and as noted in 5.5.1, has significantly less high cloud. EP's are negligibly different. Table 5.6 shows 107 IJ with more (1,0) and less (0,1) elements than 107 IB. This indicates that, while coincident N o's reduce map errors from N q temporal changes, they may increase errors due to N q changes with <j>. This occurs because, by more closely approximating the actual (presumed constant at each 0 ) N q, more of the spuriously assigned N q values are interpreted as high cloud. Map 107 IJ i s slightly better than map 107 IB (subjectively judged) as a representation of photo 107 I. o Comparing both of the 0 maps to map 107 IBS, Table 5.5 shows that the latter is significantly better, s t a t i s t i c a l l y , than 107 IB and slightly better than 107 IJ. However, comparing the actual maps, 107 IJ appears better than 107 IBS. The statistics clearly offer only a rough guide to performance. Map 107 IB is not performing as well as the technique can, but 107 IJ, which uses clear sky information obtained after the start of 99 the scanning day, performing at the technique l i m i t . In summary, these case studies indicate that c i r r u s skies can be adequately mapped using t h i s technique despite considerable movement of the clouds. Daytime changes i n N q (0.2-0.3 W) s i g n i f i c a n t l y change a map pattern, but not the s t a t i s t i c s used to evaluate the map. However, because daytime changes i n N q can be larger than the change on Day 107, under c i r r u s conditions, one must adjust d i s c r i m i n a t i o n radiances N^^ and throughout a scanning day. 5.5.3 Comparisons of Cloud Integrated by Class (High, Middle, Low) Figures 5.9 show comparisons of class integrated cloud for Days 107 and 108. Numbers beside symbols indi c a t e the r e p l i c a t i o n of that symbol at the given l o c a t i o n (e.g. x^ implies 3 class 1 p o i n t s ) . The 4 diagrams i , i i , i i i and i v compare the i n f r a r e d maps and sky photos with the observers at Vancouver Ai r p o r t (YVR) ( i and i i ) and Vancouver Discovery (YHC) ( i i i and i v ) . The graphs (vi) compare the observers' r e s u l t s ( i n d i c a t i n g mesoscale v a r i a t i o n s ) , while the graphs (v) compare the map and photo r e s u l t s , which help to decide map usefulness. Only data for 9 <_ 65° were used, the same range as i n the s t a t i s t i c a l analyses of 5.5.2. A l l data are expressed as a f r a c t i o n of the sky covered. The dashed p a r a l -l e l l i n e s i n d i c a t e the l i m i t s of usefulness defined i n 2.2.2. The purpose here i s to determine i f the measured cloud f i e l d i s s p a t i a l l y coherent over a scale of 10-20 kilometres. That information i s then used i n the assess-ment of the performance of the mapping technique. Figure 5.9a (Day 107) shows that the observers and the photo data agree (within useful l i m i t s ) i n a l l cases. The IR maps are overestimating high cloud and underestimating c l e a r sky and middle cloud. This i s due to misinterpretation by the map of altocumulus as high cloud. The photographic OOT TOT 102 evidence (slide 4) supports this explanation. Throughout the day the observers reported 1/10 cumulus or towering cumulus which was not detected by the maps or photos. That cumulus most li k e l y occurred near the horizon in a region of the sky not used in the map and photo analysis and also not visible at UBC. For Day 107 the IR maps performed poorly twice and ade-quately twice. Other data for the day (not shown) indicate that assessment may be too harsh, but those cases involved skies with very l i t t l e cloud. The results for Day 108 (Figure 5.9b) show the maps doing consider-ably better. Only one 'map-related-point' f a l l s outside the usefulness limits, indicating the maps are representing areal conditions well. The photographs do the same. On this day any of the 3 recording locations would be a representative 'regional' station, with only occasional occurrences of variations between locations of more than 2/10 for any cloud class. That observation i s confirmed by the other scans for the day (not shown). In summary, on HIGH sky days, though mesoscale cloud var i a b i l i t y i s l i k e l y to be small, the infrared maps may seriously misinterpret those skies. 5.6 Maps for Skies Dominated by Low and Middle Cloud Skies of this type present at least 3 situations not encountered in section 5.5. One or more extensive cloud layers may exist in the sky. Does the infrared technique interpret each layer uniformly at different zenith angles? Small low and middle cloud elements, and small openings in exten-sive cloud layers, may be present. How well can the technique resolve these features? Thirdly, an extensive thick low cloud may be present in the sky and the infrared map shows i t a l l as middle cloud. Is such a map worthless or not? Those 3 situations occurred on Days 106, 111 and 112. Examples are 103 treated in 5.6.3. Some attempt is made in this section to evaluate methods A, B and C for computing discrimination radiances. Thus the s t a t i s t i c a l summary of 5.6 (Table 5.8) contains up to four maps from some scans. No subjective (S) analyses are performed for scans of these skies because the influence of clear sky radiance in the mapping is small (5.6.1). 5.6.1 Map Variation With Changes in Clear Sky Radiance (NQ) Variations with N q are much less important for these LOW skies than for the HIGH skies in 5.5. Primarily this i s because discrimination radiance N22 i s most important here. varies more slowly with N q changes both in time and zenith angle than do N^ ai*d NQ^. Tests of this variation over the daytime scale could only be done on Day 112, since on the other LOW sky days, only one reasonably clear sky was scanned. (There Is. significant map variation over the day to day time scale. For example, i f the N for Day 111 are used on Day 112, a l l low cloud in Figures 5.10 and 5.11 becomes middle cloud.) From Figure 4.8 one can see 3 measured clear radiances which differ by up to 0.4 W. An increase occurred from 1035 PDT (scan 112 B) to 1120 PDT (scan 112 D), followed by a decrease to 1210 PDT (scan 112 E) and constant values u n t i l at least 1500 PDT. Infrared maps for scans 112 D and 112 E are made using method A and different sets of values of N Q. Consider maps 112 EB-A, 112 ED-A and 112 EE-A. There are no significant differences in the EP^s and EP^s (Table 5.8), except higher values (154, 1.75) for 112 ED-A versus the second photo and some change in asymmetry. Integrated cloud values differ by less than one tenth in each cloud class. These 3 maps represent: the one made with the f i r s t N Q'S of the day (112 EB-A)(Figure 5.10); the one made with the most recent N Q before the scan being discriminated (112 ED-A); and the one made with the N for that scan (112 EE-A). One expects the last to be the best of the 3, 104 TABLE 5.8:—Cloud Map Statistics for Predominantly Low and Middle Cloud Scans, Objective Analyses Only Map E P 1 EP 2 Symmetry 111 c-c 1 64:55 .62: .53 1/45:6/40 111 D-C 70:73 .70: .72 9/46:15/49 111 E-C 93:94 .90: .91 1/62:1/72 111 G-C 90:85 .87: .83 2/85:11/71 106 C-A 60 .61 29/22 106 C-B 47 .48 14/27 106 C-C 2 5 9 .60 28/22 106 E-A 47:28 .47: .28 1/46:13/15 106 E-B 49:30 .48: .29 0/49:12/18 106 E-C 48:29 .48: .29 2/46:14/15 106 F-A 33 .69 7/26 106 F-B 50 .72 3/47 106 F-C 32 .67 6/26 106 I-A 33 .69 7/26 106 I-B 55 .76 5/50 106 I-C 28 .64 6/22 112 BB-A 85:493 1.73 1.11 8/34:3/28 112 BB-B 87:51 1.71 1.11 8/36:3/30 112 BB-C 83:47 1.77 1.12 8/32:3/26 112 DB-A 63:71 1.26 1.45 24/12:7/34 112 DB-C 67:69 1.34 1.47 25/12:7/32 112 DD-A 96:100 1.88 2.00 24/16:7/40 112 EB-A 80:129 .90 1.42 14/37:20/41 112 EB-C 77:132 .89 1.45 14/34:23/41 112 ED-A 78:154 .94 1.75 7/45:15/49 112 EE-A 77:133 .87 1.46 13/38:19/42 112 FE-C4 58:88 .92 .1.33 6/31:11/32 112 HE-A 41:35 1.03 .88 9/17:7/16 112 HE-C 42:38 1.05 :.95 10/17:7/16 112 JE-A 55 .89 8/32 112 JE-B 67 .88 8/47 112 JE-C 43 .83 8/23 For a l l Day 111 maps, method C and method A using clear sky o scan 111 D are identical. Photo taken 2 minutes before start of scan. Second photo taken during scan, not after. Equivalent to 112 FE-A in EP^ + .02 in EP 2 < 105 5 a) MRP 1 1 2 E B - R J U L Y 2 7 / 7 8 1 2 : 1 0 - 1 3 BUCH TOWER s b) MRP 112ED -R J U L Y 2 7 / 7 8 1 2 : 1 0 - 1 3 BUCH TOWER Figure 5.10 106 a) mP 112DB-R JULY 27/78 1 1 : 2 0 - 2 3 BUCH TOWER s b) MRP 112DD-R JULY 27/78 1 1 : 2 0 - 2 3 BUCH TOWER Figure 5.11 107 but in this case i t is negligibly so. Thus, on the basis of this single analysis, correcting this day for clear sky changes i s unnecessary. The maps for scan 112 D indicate a drawback to such correction procedures. Using the above expectation, map 112 DD-A (Figure 5.11) should be s t a t i s t i c a l l y better than map 112 DB-A. The reverse is true. The reason for this i s that while the former map represents the clear sky patches present as clear sky, cloud motion over times of order t^ change the position of those clear patches. Thus the contingency matrices for map 112 DD-A (Table 5.9) record more (0,3) elements (clear on the map, low cloud on the photo) than do the matrices for map 112 DB-A. The latter misinterprets the clear sky as high cloud and reports these grid squares as (1,3) elements. Thus a small number (11) of grid changes (which however do not change integrated cloud amount by even 1/10 for any cloud class) has a large effect on the error parameters. The implications are that in such skies, a map's usefulness may be quite unrelated to i t s error parameters, but i t s truthfulness i s . In other words cloud patterns are much better preserved than cloud locations. This pattern preservation is illustrated by a comparison of Figure 5.11 with photos 112 Dl and 112 D2 (slides 5 and 6) and Figure 5.10 with photos 112 El and 112 E2 (slides 7 and 8). 5.6.2 Map Variation With Changes in Discrimination Method Table 5.8 indicates that differences among maps using methods A, B and C are generally small. Differences between comparisons with before-and after-scan photographs are generally larger. For Day 111, method B could not be used because i t s N^^ values exceeded the radiometer saturation value at 9 < 60°. Maps from methods A and C are identical for that day. For Day 106, methods A and C maps show EP^ values agreeing to within 5, EP0 values to within .05 and symmetry pairs to within + 4. Method B maps TABLE 5.9—Contingency Matrices for Selected Maps of Skies Dominated by Low and Middle Clouds Map 112 DB-A Map 112 DD-A 0 1 2 3 0 1 2 3 0 0 0 4 0 0 0 4 0 0 Photo 1 0 6 14 1 0 6 14 1 1 Photo 112 DI 2 0 2 8 5 0 5 5 5 2 112 DI 3 0 4 6 53 4 3 4 52 3 0 0 0 1 0 0 0 1 0 0 Photo 1 0 1 1 1 0 1 1 1 1 Photo 112 D2 2 0 1 7 4 0 6 2 4 2 112 D2 3 0 8 25 54 4 5 25 53 3 Map 112 EB-A Map 112 ED-A 0 1 2 3 0 1 2 3 0 9 5 2 0 15 1 0 0 0 Photo 1 15 24 7 0 22 18 6 0 1 Photo 112 El 2 1 8 14 0 2 8 13 0 2 112 El 3 1 4 8 5 1 4 8 5 3 0 9 6 0 0 12 3 0 0 0 Photo 1 11 18 12 1 18 13 10 1 1 Photo 112 E2 2 2 7 12 1 5 5 11 1 2 112 E2 3 4 9 8 3 6 8 7 3 3 109 TABLE 5.9—(continued) Map 111 D-C 0 1 2 3 0 3 2 0 0 1 4 21 7 0 Photo 2 0 24 24 0 111 Dl 3 0 5 13 0 0 1 5 0 0 1 6 22 10 0 Photo 2 0 22 16 0 111 D2 3 0 3 18 0 Map 111 G-C 0 1 2 3 0 0 1 0 0 1 1 2 1 0 Photo 2 1 9 14 0 111 Gl 3 0 2 72 0 0 0 3 0 0 1 2 5 8 0 Photo 2 0 5 16 0 111 G2 3 0 1 63 0 Map 106 F-A Map 106 F-B Map 106 F-C 1 2 3 1 2 3 1 2 3 1 0 2 0 0 2 0 0 2 0 Photo 2 0 15 5 0 19 1 0 16 4 Photo 106 F 3 0 26 55 0 47 34 0 26 55 106 F 110 have lower EP's in 1 case, greater EP's in 2 cases and similar values in one case. Changes are of order 20 in E P ^ , with almost a l l increases in 1 of the symmetry pair. Table 5.9 and Figure 5.12 for map 106 F-B show large increases in middle cloud, especially near the horizon, compared to the other methods. Photo 106 F (slide 9) shows that the cloud thickens towards the horizon, so method B i s certainly in error. The N 2 3 values are increasing too quickly with 6, causing low clouds at high 6 to be inter-preted as middle cloud. Note that method B maps have uniformly larger 1 (and smaller n) values in their symmetry pairs. This i s a consequence of 2 the higher N 2 3 values at most 9. Table 5.7 shows that x s t a t i s t i c s for maps 106 F-B and 106 I-B are worse than those for other methods. In the latter case the change of method causes the map to lose significance at the 0.05 level. On Day 112, differences among the 3 methods are small for a l l scans except 112 D (discussed in 5.6.1) and 112 J (method B for scans 112 F and 112 H are not shown). For scan 112 J, method C i s marginally better (lower E P p than method A, which i s better than method B. The differences occur mostly at high 9 where large portions of the radiometer signal approximate N23-To conclude, method B i s demonstrably worse than the other methods for several scans (low cloud i s misinterpreted as middle cloud). The other 2 methods are equivalent in a l l but one case. Since method A uses only N Q , when the latter is available early in the day, that method should be used. If N q i s not available, i t can be estimated from Figure 4.10 i f q 2 i s known. Method A could then be used (for a l l three N ). If a more exact know-st ledge of the temperature and humidity profile is available, method C could be used for N o 0 (with the same values as method A for N n 1 and N^). Exten-I l l s a ) MflP 106F-C JULY 11/78 0 8 : 2 6 - 2 9 BUCH TOWER-b) MflP 106F -B JULY 11/78 08=26-29 BUCH TOWER Figure 5.12 112 sive testing of a l l 3 methods at different locations i s suggested, with emphasis on A and C. 5.6.3 Case Studies In the cases below, and the others in Table 5.8, no systematic differences are found in map agreement with photos taken before a scan and those taken after. This i s different from the HIGH sky data in Table 5.5. 106 F Comparing map 106 F-C (Figure 5.12a) and photo 106 F (slide 9) one sees that the textured stratocumulus cloud i s well represented by the map. The break to the southwest is partially obscured by the blind sector. The break shows on the map as the The clouds near the opening appear to be mixed stratocumulus and altocumulus, so the large area of middle cloud in the west of the map i s partly true and partly in error. Table 5.9 shows that 70% of the grid squares are correctly interpreted and Table 5.7 shows 2 the map is significant in the x test. Table 5.8 shows EP^ =32 and EP2=.67. Map 106 F-A is negligibly different. These results indicate the map is approximating the photo well. Similar conclusions can be made for the other Day 106 scans, except perhaps 106 C where the cloud type in the photo i s d i f f i c u l t to assess. Map 106 F-B (Figure 5.12b) i s a somewhat poorer representation of the sky, since one should conclude from the map that a layer of mixed middle and low cloud i s present (with the latter in the east and north). Table 5.9 confirms that only about % of the grid is correctly interpreted. For a layered cloud situation, this level of performance is only barely adequate. 111 G Map 111 G-C (Figure 5.13b) and photo 111 Gl (slide 10) show that a thick, low cloud is being interpreted by the map as a middle cloud. Clearly a) MAP 111D - C JULY 26/78 08=52-55 BUCH TOWER s b) MAP 111G-C JULY 26/78 10:04-07 BUCH TOWER Figure 5.13 114 the value of N^ is t o ° high for a l l 9; the mapping method f a i l s badly for this scan. Table 5.9 confirms the failure. Note the large values for map=2, photo=3. About the only encouraging feature of the map i s the way the edge of the cloud bank i s shown. There appears to be no explanation for the gross error in N 2 3 o n this day. If_ such an error can be recognized early on a measurement day, perhaps some use can be made of the map by assuming a l l middle cloud is low cloud. U l D Map 111 D-C (Figure 5.13a) and photo 111 Dl (slide 11) show a sky one hour earlier than scan 111 G. Both scans occur within the 2 hours preceding the onset of local thunderstorms. Given that the map i s incapable of correctly displaying low cloud, this map is surprisingly informative. The cirriform clouds to the east and the cirrostratus and altostratus to the north and west are well represented. As well, the openings in the fracto-and altocumulus show up well. Table 5.9 indicates that about % the grid i s correctly interpreted by the map, while Table 5.8 shows EP.^70, EP2=.70 and significant asymmetry (1 > n). Despite this apparently poor s t a t i s t i c a l summary, in the opinion of this investigator this map has value for a synoptic user. However, i t i s certainly not a truthful representation of the sky. The map also shows an anomaly in that openings in clouds the size of the one to the northeast in the photo cannot be resolved, while those similar to the ones in the east and southeast can be resolved. 112 D Maps 112 DB-A and 112 DD-A (Figure 5.11) and photos 112 Dl and 112 D2 (slides 5 and 6) i l l u s t r a t e a broken fairweather cumulus sky. The difference between the photographs taken about 3 minutes apart indicates a rapidly changing sky. Comparing the photos (using the map/photo comparison 115 technique) yields EP's of 136 and 2.83 and symmetry of 9/37. The photos agree more poorly with each other than each does with the 3 maps (Table 5.8). Thus one would expect large differences between the map comparisons to the 'before' and 'after' photos. Table 5.8 indicates this i s not the case, except with the symmetry where n and 1 change by about 20. Table 5.9 indicates that much of the increased low cloud in photo 112 D2 (versus 112 Dl) i s misinterpreted as middle cloud, but that errors from the wispy cumulus in the north are eliminated. Map 112 DD-A resolves openings and 2 cloud edges better than map 112 DB-A. A x test (Table 5.7) on the latter shows i t is not significant at 0.05. This is an underestimation of the map's value. 112 E Maps 112 EB-A and 112 ED-A (Figure 5.10) and photos 112 El and 112 E2 (slides 7 and 8) show a scattered f a i r weather cumulus sky. The 2 photos appear quite similar, yet the statistics comparing them are EP's of 106 and 1.22 and symmetry of 26/33. Table 5.8 indicates the maps agree with photo 112 El better than the photos agree. Agreements between maps and photo 112 E2 are worst. In Table 5.9, note the higher values in the lower l e f t for matrices using the second (as opposed to the f i r s t ) photo. These increases show that cloud motion over time t^ has caused the bulk of the increases in the error parameters. The maps for scans 112 E and 112 D satisfy the usefulness criterion of cloud shape preservation. 5.6.4 Comparisons of Cloud Integrated by Class (High, Middle, Low) The data for this section are summarized in Figures 5.14a, b and c for Days 106, 111 and 112 respectively. Notation is similar to Figure 5.9, with letters beside symbols indicating the discrimination method (A,B,C) 9 1 T LU 811 119 used. The purpose i s the same as for section 5.5.3. On Day 106, large mesoscale cloud variations are apparent. More than half the low and middle cloud observations differ by more than 2/10 between the observers. Under such conditions, comparison of class inte-grated cloud results from UBC (maps and photos) and the observers i s not a valid method to test map accuracy. The maps and photos agree well, except for one B map, showing the maps are representing well the on site conditions. A 21 minute stationary test was performed by orienting the mirror to a near-zenith position and recording the radiometer signal. The signal was discriminated and integrated using time as a surrogate for sky position. The results differed by up to 3/10 (for the 3 cloud classes) with those of a map made from a scan 6 to 9 minutes after the stationary test ended. Another such test (for 18 minutes at 0=35°, (j>=300°) yielded results which agreed to within 3/10 with the subsequent map (made from a scan 4 to 7 minutes after the end of the stationary test). These isolated tests do not contain enough information to compare them systematically to the maps. On Day 111 (Figure 5.14b) there is less mesoscale variation than on Day 106. The maps and photos both agree f a i r l y well with the YVR observer, but more poorly with the YHC one. Map/photo comparison i s quite good, considering the error in ( which prevented the maps from recording low cloud. A near-zenith stationary test agreed to within 1/10 for a l l cloud classes with a subsequent scan. The results for Day 112 (Figure 5.14c) again show considerable mesoscale variation. The map/photo agreement is excellent ( a l l but one map are within the + .2 useful limits) while agreement between the UBC data and the observers is tenuous at best, with numerous outlying points. The 120 one stationary test differs by up to 4/10 per cloud class with the subsequent map. From this section one can conclude that maps are generally useful in describing local conditions. There are some provisos. If gross discrimination errors are made, the local conditions may not be well represented. (Usefulness also depends on cloud shape. In the small sample taken here, the maps appear to follow shapes well, si£ class integrated  cloud becomes the main usefulness criterion.) Stationary tests are of only limited use in describing class integrated cloud. 5.6.5 'Maximum' Low Cloud Radiance When low clouds were observed in this experiment, there appeared to be a maximum observed radiance value (on a particular day) which was independent of cloud thickness, as perceived from the surface. This varied weakly with 0, with less than 1 W of limb brightening observed from the zenith to 9=75°. On Day 106 (July 11, 1978) by 1430 PDT, radiances were the blackbody equivalents of between 4 and 8 °C. The cloud bases observed at YVR were 550-600 m for towering cumulus and 1050 and 1800 m for the two layers of stratocumulus. YHC reported stratocumulus at 1400 m, with cumulus at 1000 m. The temperatures at between 85 and 90 kPa 1000-1500 at Port Hardy were in the range 5-8 °C. and 7-11 °C. for the 0500 and 1700 PDT soundings respectively. The locally adapted atmosphere (using the technique of 3.3.4) had temperatures of 4-11 °C. A l l of the actual and estimated temperatures approximate those inferred from the cloud radiances using the blackbody assumption. These 'limiting' clouds do confirm that assumption in the observing wavelength band. Removing the effects of instrument saturation only improves the comparison. This 121 observation is confirmed on July 27, 1978 (Day 112) when cloudbase (900 m) temperatures and equivalent blackbody temperature agreed at about 9 + 1 °C. 5.7 A Time Series of Maps If infrared maps are used for operational purposes, they w i l l certainly be made regularly at some interscan time, t (3.3.1). Deter-mining t from instrument and environmental considerations i s the purpose of this section. This whole experiment was not primarily designed to test the information content of such a series of maps. However, on Day 112 the 6 maps below provide an opportunity to examine a time series to identify some of i t s characteristics. The 6 maps used are, sequentially, those in Figures 15a, 11a, 10a, 15b, 15c and 15d for maps 112 BB-C, 112 DB-A, 112 EB-A, 112 FE-C, 112 HE-C and 112 JE-C. Photographs corresponding to the maps are on slides 12, 6, 8, 13, 14 and 15. The times between the maps are 47, 50, 33, 67 and 43 minutes respectively. These are not constant, yet they f a l l between about 30 and 60 minutes, the latter the most common time interval for synoptic network cloud observations. Examining the map series, one can conclude l i t t l e about what occurred between the scans, except perhaps that a low cloud dominated the eastern sky between 1240 and 1350 PDT (Figures 15b and 15c). This was confirmed by scan 112 G (not shown) at 1340. Also, for a l l scans except 112 E, low clouds dominate the sky. The structure present in Figures 10a and 15b might suggest cumulus clouds, though the evidence from other maps would not confirm that observation. In short, the series of maps adds l i t t l e to the information that can be inferred from each map separately. The photo series shows l i t t l e more information in the f i r s t 3 slides, but the last 3 (slides 13-15) indicate a layer of stratocumulus or 122 5 b) MflP 112FE-C JULY 27/78 1 2 : 4 3 - 4 6 BUCH TOWER Figure 5.15 123 KEY - HIGH <!> MIDDLE * LOW ¥ * ****** * * * * *A * - * V „<* J* J" * X X «> ' * X X * X S a) MRP 112HE-C JULY 27/78 1 3 : 5 0 - 5 3 BUCH TOWER Figure 5.15 124 altocumulus approaching from the northwest. Clearly, the photo series can provide a continuity which the map series cannot. The maps' lack of information can be explained using cloudbase velocity (v.), maximum usable zenith angle in the map (0 ), the cloudbase J b max height (z, ) and the cloud transit time (t_) (defined as the time for a cloud D i to follow an apparently straight path through the zenith from 9 a t some ej) to 0 at <J) + 180°). In general: max - r i b t_ = 2(z /v.)(tan 9 ) (5.8) , T b b max or for 6 =65°: max t T = 4.28(zb/vb) (5.9) . Assuming the cloud moves with the mean wind, for v, *\> 4 m s \ z, ^ 1500 m, b b t^ i s about 27 minutes. Since at any time v^ typically increases with z^ (cirrus at z^ ^ 10 km regularly experience > 25 m s S the values of t^ w i l l not vary by much more than a factor of 2 for different cloud layers on the same day. If the steering flow speed (v^Q) an<i the surface wind speed (v ) are obtainable from charts or measurements, the v^'s for various cloud layers can be estimated. On Day 112, V,-Q was 10-15 m s ^ west to southwest, v was 1-4 m s * easterly then westerly, and z, for the cumulus s b was 600-1150 m and for the altocumulus 2300-3000 m. Logarithmically inter-polating the velocity from the surface (10 metres) to 50 kPa (for a f i r s t approximation) yields v^'s o r 9-10 m s * for the cumulus and ^ 11 m s ' for the altocumulus. These yield t ' s of ^ 8 and % 16 minutes respectively. If one assumes that t should be no less than 5 times the intra-s sampling (intrascan) time t^, and no more than the transit time t^,, i t i s clear that neither the altocumulus nor the cumulus clouds can be adequately tracked by the map or photo series presented. Five times t^ i s about 10-12% 125 minutes while t^, i s about 12 minutes. A much more rapid sampling (smaller t g ) i s needed on such a day to generate a map series which can be inter-preted as a whole. This sort of computation could be done for a number of synoptic conditions to determine the optimal t in each case. In conclusion, short cloud transit times across the map area require interscan times generally less than 20 minutes. Therefore, to avoid a ratio t / t , of less than 10, a considerable decrease in t, i s s i 1 suggested. This conclusion i s supported by evidence from the repeated 9 data (3.3.1). Up to 5 of the 34 slots at 6=10° changed over 8 seconds in two separate tests, while 2 of 34 slots changed at 9=65° in another. These results, plus those indicating rapid cloud position changes over 3 minutes ( and, support the recommendation to reduce t^. (<() reso-lution w i l l be decreased unless radiometer/recording system response is also improved.) 5.8 Overall Performance of the Maps In order to judge the maps as a group, the contingency matrices for the 23 scans represented in Tables 5.5 and 5.8 were summed. Only the best 2 matrix for each scan was used. A x test was performed on the resulting matrix: 0 1 2 3 0 541 136 7 0 1 140 298 68 1 2 4 115 315 40 3 1 47 265 391 2 The x s t a t i s t i c (2369 to the nearest integer) far exceeded the c r i t i c a l 2 value. (For 9 degrees of freedom at 0.005, x • • ,=23.6.) Though this test ° ^ c r i t i c a l 2 would be expected to yield an affirmative result, the large x value ind i -cates that the method i s overwhelmingly significant. The majority of the 2 X value does not come from the 6 cells (elements) most off the main 126 diagonal, thus the small number of cases in these cells does not degrade the s t a t i s t i c . No such analysis could be done for particular days, because the number of observations in each c e l l was generally too few. CHAPTER SIX CONCLUSIONS This study demonstrates that significant useful information i s contained in cloud maps produced by the infrared radiance contrast tech-nique in summer at Vancouver, B.C., Canada (sections 5.5-5.8). The technique i s inferior to manual observation because no cloud type infor-mation is provided by i t and cloud height information i s in only 3 classes Maps do describe the angular distribution of clouds in the sky hemisphere (which observers do not describe). The maps are poorer at this than (instantaneous) all-sky photographs. The principal external (non-cloud) variables which affect the quality of maps produced are the amount and vertical distribution of water vapour in the atmosphere. The water vapour mixing ratio, q, averaged over the lower half of the atmosphere, can be used to estimate the clear sky radiance (NQ) i f the latter cannot be measured, or modelled directly from nearby radiosonde information. Changes in q and N q over time scales from hours to seasons adversely affect maps of predominantly cirrus skies. Map of skies dominated by lower clouds are insensitive to daytime changes in N but can be adversely affected by changes over days and longer periods. Other natural atmospheric variables insignificantly affect the maps, though high concentrations of atmospheric aerosols may degrade maps of cirrus skies by being spuriously interpreted as cirrus cloud at certain azimuths. 6.1 Map Quality 'Useful' maps, as defined in 2.2.2, are obtained on 4 of the 5 127 128 mapping days. These include a broad range of sky conditions. On the other day, some serious misinterpretation of low cloud as middle cloud occurs. No specific fact i s offered to explain this behaviour. Perhaps the clear sky radiance was changing so rapidly that the values were not represen-tative only one hour later. Maps made using methods A and C are of compar-able quality, while those from method B are frequently significantly poorer in quality, especially near the horizon. The principal constraint on map quality is the non-stationarity of the cloud patterns under most sky conditions. The maps preserve cloud shape far better than cloud position, especially in the cases of discrete cirrus and cumulus elements. It is recommended that i f this experiment i s replicated, the intrascan time (tj) be reduced to the order of 1.0 to 1.5 minutes or less. Though this may increase data loss on each sky scanning revolution, the decreased importance of cloud movement should s t i l l improve agreement between the maps and the actual sky conditions. If possible, verification photos should be taken before, during and after the scan using an automatic timer. In order to improve resolution, a d i g i t a l or analogue magnetic tape recording system with a time constant smaller than that of the Barnes radiometer is suggested. Display of map data at a mesh of 5° in <f> would be a definite improvement requiring no change in 9 sampling. A finer <j> mesh requires a finer 9 mesh, which increases t^. With the present radiometer, major simultaneous improvements in both map resolution and accurate cloud positioning are impossible. A more serious question concerns the maps' spatial representa-tiveness. Section 5.7 suggests that scanning every 12 to 20 minutes would be needed to provide a series of maps containing any sort of continuity. A reduction of t, to less than 1.2 minutes would allow an interscan time 129 (t ) of 12 minutes to satisfy the sampling criterion t / t 1 > 10. An s s J. upper bound on t can be estimated from surface and mid-tropospheric winds as demonstrated in 5.7. Such a map series, and the individual maps that compose i t , would be synoptically useful as point measurements but not as mesoscale repre-sentations of the cloud f i e l d at a particular time. Also, no conclusions can be drawn about how such a mapping system would perform over an extended time period. 6.2 Validity of Technique Assumptions The blackbody cloud assumption i s confirmed for thick low clouds, but i s rejected for cirrus. No attempt is made to test the blackbody assumption for middle clouds because of a lack of data. The assumption of Werner (1973a,b) that seasonally constant discrimination radiances are valid is rejected. His criterion for judging maps (total cloud integrated for a l l cloud classes) i s too lax as demonstrated by Figure 5.14. In that figure the class 0 data (for clear sky) equal one minus the total cloud as a sky fraction. Large errors in particular cloud classes are hidden by only considering the overall total. The assumption of 3 usable cloud classes i s conditionally accepted, since the maps are shown to be useful. However, expressing a continuum cf radiance sources as 3 discrete cloud classes (and clear sky) i s fundament-all y unsound i f i t is based solely on radiance, not meteorological concepts. Multichannel techniques providing more discrimination options and/or more cloud classes seem the proper next step. The assumption of year round operation is rejected, at least at this and higher latitudes, with the present radiometer configuration. The lack of sufficient clear sky signal (to register on the Barnes) through much 130 of the year at many zenith angles implies that the present technique w i l l not work with the equipment tested during those periods. (Some information on low clouds may be obtainable since their radiances are higher and less dependent on N q than those for higher clouds.) If the radiometer i s changed by increasing the infrared channel width, then a new analysis of the clear sky and discrimination radiances must be performed. The maps produced may not have the same characteristics as those presented here. The only certain way to extend the technique further over the globe and throughout the year i s to find a radiometer with a much lower threshold radiance and which can operate outside of a laboratory. The assumption (by Werner 1973a) that limb brightening (6) is constant for discrimination radiances N ^ ^ , a n d N ^ ^ i s incorrect (3.4). His values for ^(^-j) are used in method C of this work because his two reports (1972,1973a) agree on those data. Total integrated cloud was Werner's primary objective, while cloud maps (by height class) were the objective here. Concern with the clear sky radiance i s high in this study because thin cirrus/clear sky discrimination is important in distinguishing cloud shapes (especially i f the clouds are mainly cirriform and high level altocumulus). Mistakes with that discrimination may account for a great deal of the scatter in Figure 13 of Werner (1973b) where he compares his infrared derived cloud totals with those of nearby trained observers. Finally, the use of subjective map analysis, described in 2.2.3, provided l i t t l e extra information, took a great deal of time and was d i f f i -cult to reproduce. It should not be used. 6.3 Suggestions for Future Work The use of all-sky photographs to compare with the maps was an important aspect of this study, since standard narrow view photographs cannot give the same overall sense of the cloud cover. During future work with this technique, or an electronically based equivalent, a l l sky-photography should be continued. As suggested in chapter 4, more research into the actual emission of the atmospheric infrared continuum (and i t s relationship to water vapour) i s needed to resolve the discrepancy among the sets of continuum parameters in Table 2.1. There is the question of a context for this study. For those who must implement a remote cloud detection system at some location, this study answers some of the scientific and design questions. At the moment the Barnes PRT-5 appears to be the only commercially available radiometer sturdy enough to do the job. Engineering problems, such as the construction of a weather-proof mirror rotation system remain, though Werner (1973a) has pointed the way to the solution of some of them. It would appear, though, that u n t i l a lower threshold instrument becomes available, manual observers may s t i l l be good value for the money for cloud observation purposes. This mapping methodology should not, however, be put aside but rather tested under experimental conditions (using methods A, B and C employed here) at locations with more atmospheric water vapour year round. This w i l l test the equations for N in section 5.4 and the relationships among temperature, humidity and radiance in Figures 4.9, 4.10 and 5.3 to see i f they apply more generally. Additionally, other wave band radiometers should be tested similarly to see i f a different choice can alleviate the threshold radiance problem. Numerically modelling the radiative transfer in the operating wave bands of such radiometers would be recommended. For work on the receipt of hemispheric long- and short-wave radiation at a point on the ground, maps made from this technique w i l l have to be compared to both sky photos, and downwelling fluxes averaged over the intrascan time of about 1 to 2 minutes. Slower response flux radiometers should not be used in such comparisons. 133 REFERENCES Ackerman, T.P., K.-N. Liou and CB. Leovy, 1976: Infrared Radiative Transfer in Polluted Atmospheres. J_. Appl. Meteor., 15, 28-35. Air Force Cambridge Research Laboratory, 1964: Ozonesonde Observations Over North America—Volume 1. Laboratory Report 64-30 (_1) , Cambridge, Mass., U.S.A. Atmospheric Environment Service (AES) (formerly Canada Department of Transport—Meteorological Branch), 1970: MANOBS (Manual of standard procedures for surface weather observing and reporting). 6th edition, Toronto, Canada, 157-163. Barnes Engineering Company, 1970: Instruction Manual for Precision Radiation Thermometer Model PRT-5. Report No. BEC 06611270, Stamford, Connecticut, U.S.A. Grassl, H., 1973: Separation of Atmospheric Absorbers in the 8-13 um Region. Contributions to Atmospheric Physics, 46, 75-88. Grassl, H., 1974: Influence of Different Absorbers in the Window Region on Radiative Cooling (and on Surface Temperature Determination). Contributions to Atmospheric Physics, 47, 1-13. Harris, R. and E.C. Barrett, 1978: Towards an Objective Nephanalysis. J. Appl. Meteor., 17, 1258-1266. Hay, J.E. and T.R. Oke, 1976: The Climate of Vancouver. B_.C_. Geographical  Series, No. 23, Tantalus Research, Vancouver, Canada, 48 pp. Johnson, W.B., 1979: Measurement of Stratospheric Ozone Penetrations into the Lower Troposphere. Preprints AMS 4th Conference on Atmospheric.Turbulence, Diffusion and Air Pollution, Reno, Nevada, U.S.A., January, 1979. McArthur, L.J.B., 1978: The Spatial Distribution of Diffuse Solar Radiation in the Sky Hemisphere. M. ^ c_. Thesis (unpublished), Department of Geography, UBC, Vancouver, Canada, 137 pp. Oke, T.R., 1978: Boundary Layer Climates, Methuen, London, England, 372 pp. Paltridge, G.W. and C.M.R. Piatt, 1976: Radiative Processes in Meteor-ology and Climatology. Developments in Atmospheric Science, No. 5, Elsevier, New York, 318 pp. Piatt, C.M.R., 1973: Lidar and Radiometric Observations of Cirrus Clouds. J. Atmos. Sci., 30, 1191-1204. 134 Piatt, C.M.R., 1976: Infrared Absorption and Liquid Water Content in Stratocumulus Clouds. Quart. j;. R. Meteorol. Soc., 102, 553-562. Piatt, C.M.R. and K. Bartusek, 1974: Structure and Optical Properties of Some Middle Level Clouds, j;. Atmos. Sci., 31, 1079-1088. Roewe, D. and K.-N. Liou, 1978: Influence of Cirrus Clouds on the Infrared Cooling Rate in the Troposphere and Lower Stratosphere. J. Appl. Meteor., 17, 92-106. Rogers, R.R., 1976: A Short Course in Cloud Physics. International  Series in Natural Philosophy, Vol. 84, Pergamon, Toronto, Canada, 227 pp. Spangler, T.C. and R.A. Dirks, 1974: Mesoscale Variations of the Urban Mixing Height. Bound.-Layer Meteorol., 6, 423-441. Walshaw, CD. and R.M. Goody, 1956: An Experimental Investigation of the 9.6 um Band of Ozone in the Solar Spectrum. Quart. J^ . R. Meteorol. Soc. , 82, 177-186. Werner, C., 1972: Wolkentemperatur- und Wolkenhohenmessung mit Radiometer und Lidar. Forschungsbericht aus der Wehrtechnik BMVg-FBWT 72-17, DOKZENTBw, Bonn, W. Germany, 92 pp. Werner, C., 1973a: Gerat zur Autonatischen Bestimmung des Wolkenbedeck-ungsgrades. Forschungsbericht aus der Wehrtechnik BMVg-FBWT 73-13, DOKZENTBw, Bonn, W. Germany, 66 pp. Werner, C. , 1973b: Automatic Cloud Cover Indicator System. J^ . Appl. Meteor., 13, 1394-1400. Zdunkowski, W.G. and I. Choronenko, 1969: Incomplete Blackness of Clouds in the Infrared Spectrum. Contributions to Atmospheric  Physics, 42, 206-223. Program RADIANCE may be obtained from Dr. P.M. Kuhn and Ms. L. Stearns at NOAA-ERL-ARL, Boulder, Colorado, 80302 U.S.A. 135 APPENDIX A ADDITIONAL TABULATED INFORMATION FOR INTERPRETATION OF MODELLED RESULTS TABLE A.1 TABLE A.2 Pressure/Height Conversion P (kPa) z (m) Pressure/Temperature/Humidity Data for Sounding SDl: Port Hardy, B.C. (YZT) 00 Z 8 August, 1977. P (kPa) T (°C.) 100 U l 95 540 90 988 85 1457 80 1949 75 2466 70 3012 65 3591 60 4206 55 4865 50 5574 45 6344 40 7185 35 8117 30 9164 EC/AES Tephigram from document 7488/2 of ICAO standard atmosphere 2nd edit. 1964. q ( x 10 kg kg l) 101.7 17.0 8.8 100.0 17.0 9.0 95.0 17.0 8.7 90.0 15.5 9.1 85.0 13.0 6.5 80.0 11.0 5.0 75.0 9.0 3.2 70.0 5.0 2.8 65.0 1.0 2.3 60.0 -3.5 2.8 55.0 -8.0 1.5 50.0 -12.5 0.8 45.0 -17.5 0.5 40.0 -24.5 0.7 35.0 -31.5 0.6 30.0 -38.5 0.4 25.0 -45.5 0.2 20.0 -55.0 0.1 15.0 -55.0 0.05 Clear Sky Radiance (N ) -2 -1 ° ( W m sr ) (B parameters) N 8 3.9 35 4.4 62 6.1 136 TABLE A.3 Pressure/Temperature/Humidity Data for Sounding SN2: Port Hardy, B.C. (YZT) 12 Z 9 August, 1977. P (kPa) T (°C.) q ( x 10 _ 3 kg kg 1) 102.0 11.5 8.8 100.0 12.0 8.8 95.0 15.0 9.0 90.0 20.5 3.2 85.0 18.0 2.9 80.0 14.7 2.4 75.0 10.3 1.9 70.0 5.8 1.6 65.0 1.0 1.2 60.0 -4.0 0.92 55.0 -8.0 0.73 50.0 -12.7 0.55 45.0 -18.3 0.45 40.0 -23.7 0.35 35.0 -31.0 0.50 30.0 -40.0 0.15 25.0 -49.5 0.05 20.0 -61.0 0.02 TABLE A.4 Pressure/Temperature/Humidity Data for Sounding SD7: Port Hardy B.C. (YZT) 00 Z 14 August, 1977. P (kPa) T (°C.) q ( x 10"3 kg kg"1) 100.8 20.0 10.0 100.0 20.0 10.2 95.0 19.0 11.7 90.0 22.0 9.1 85.0 20.5 8.0 80.0 17.0 6.2 75.0 13.0 5.0 70.0 10.0 3.8 65.0 7.2 2.2 60.0 3.4 1.75 55.0 -2.0 1.4 50.0 -8.0 1.0 45.0 -14.5 0.76 40.0 -21.2 0.55 35.0 -29.5 0.20 30.0 -38.2 0.10 25.0 -48.5 0.05 20.0 -57.0 0.02 137 A. 5—Ozone Mixing Ratio Profiles (for section 4.5, .3) Pressure Ozone Mixing Ratio ( x 106) (kPa) 1 2 3 A B C 105.0 0 0 0 101.3 0.039 0.10 0.0014 61.5 0.074 0.29 0.001 50.6 0.084 0.33 0.001 35.6 0.090 0.50 0.001 19.A 0.15 1.0 0.10 14.5 0.18 3.7 0.10 10.3 0.43 5.0 0.10 5.50 3.0 6.0 0.70 2.90 7.3 9.3 7.0 2.37 10.4 10.4 7.4 1.60 11.5 15.0 7.0 1.30 11.8 17.0 6.0 0.88 11.0 17.0 9.0 0.48 9.0 18.0 9.0 0.28 7.0 20.0 7.0 0.06 5.0 20.0 5.0 0.02 3.7 20.0 3.0 0.01 0.89 20.0 0.5 Standard profile used in a l l modelling except ozone sensitivity. 'Maximum' ozone case—for section 4.5.3. 'Minimum' ozone case—for section 4.5.3. Approximation to zero. 138 APPENDIX B SYNOPTIC CONDITONS B. 1 August 8-JL5, 1977 The weather for this period was controlled by an upper air (50 kPa) high centred a few hundred kilometres west of Vancouver Island (Figure 3.5). A number of weak low pressure disturbances passed through the area of southwestern B.C. Pressure gradients, temporally and spatially, were low. The pressure extrema for the 3 stations, Vancouver, Port Hardy and Quillayute for the period were 100.5 and 102.2 kPa, while temperature ranges were (15, 27), (10, 21) and (12, 29) °C respectively. Surface dewpoint depressions varied from 0 to 10 °C , indicating large changes in relative humidity throughout the period. Mixing ratios also varied substantially. Winds at 50 kPa (v^Q)» after the f i r s t 24 hours, were less than 15 m s L, usually in the range 2 to 10 m s L. There was often an effective 'decoupling' of the air at the three stations, both because of low wind velocities and local (sea breeze) circulations. Under these conditions, estimating upper air conditions at Vancouver from those at Port Hardy of Quillayute became very d i f f i c u l t . Sky conditions were generally clear, with occasional cloud accom-panying or preceding the weak lows. Port Hardy often had morning fog, which may have affected the mixing ratios recorded below 90 kPa. This was not a serious drawback to the use of the soundings for clear sky analysis, since the radiance trends present at the surface were usually observed at 90 kPa as well. B.2 July Jl-14 and 24-27, 1978 Since cloud mapping was done in this period, more details are 139 given than in B.1. On July 11 (Day 106) a cold low over central Alberta was assoc-iated with cloud over most of southern B.C. at 0500 PDT (Figure 3.5). The upstream flow showed a high centred about 1500 km to the southwest of Vancouver and a fast moving low, with attendant fronts in advance of i t , the same distance to the west. Winds at 50 kPa were ^ 17 m s ' north-westerly at Quillayute and ^ 7 m s * north-northwesterly at Port Hardy. For the next 48 hours they remained < 10 i s ^ from the north or northwest for both stations. Satellite cloud photos indicated a band of middle or high cloud extending west from the low centre to the coast. Offshore was an area of low cloud. At 0800 cloud observations from Vancouver to Campbell River indicated overcast or broken thick or double layered altocumulus. Pressure gradients were low, air temperatures 10-15 °C and dewpoint depressions 1-3 °C. L i t t l e change was reported at 1100, with the low cloud to middle cloud boundary situated near the coast. Three hours later low clouds were reported as far inland as Aldergrove (60 kilometres east of the Vancouver Airport) and by 1700 the low clouds covered the entire mainland coast. During the day the cold low moved off rapidly to the east, implying extensive drying in the lower atmosphere. During the night of July 11-12, the low approaching the coast began occluding and f i l l i n g in the Gulf of Alaska. The next day the low intensified, bringing clouds as far south as Campbell River during the daylight hours. By 2000 PDT mixed cirrus and f a i r weather cumulus were reported at Vancouver Airport. The next night the outer coast (of Vancouver Island) had rain, while the inner coast was clear with occasional low cloud. 140 In the Vancouver area, July 13 (Day 107) began clear and cool (T ^  14 °C ) with light southwest winds. Northwestward of Campbell River there was a large mass of cloud, moving rapidly southward. By noon satellite photographs indicated high cloud north from near Vancouver to the Gulf of Alaska. The edge of the thicker cloud mass was just south of Campbell River. This thicker mass moved southeastward u n t i l , by 1700, i t was photographed 30 to 50 kilometres northwest of Vancouver. This cloud appears on Figures 5.6. (The temperature that day rose to 25 °C at 1400 with the dewpoint rising from 11 °C at 0500 to 14 °C at 1400.) Through the night and the next day, July 14 (Day 108), fog and occasional light rain were prevalent throughout the area, except for Vancouver and the immediate v i c i n i t y . Upper level winds began to shift to the southwest and increase to 10 to 15 m s L. Local surface winds remained light and variable. Clouds near Vancouver were cirriform, with some cumulus reported on the horizon. The pressure situation was weak and confused. For the next 13 days the high ridge to the southwest moved in to dominate the Vancouver area. Minimal cloud was reported u n t i l July 24. On July 24 a thin stratus deck approached Vancouver from the south but dissipated before covering the city for more than a few minutes. The next day was clear, but by late afternoon the high was weakening as a cold front approached from the northwest. This front stalled over the Queen Charlotte Islands during the night. At the same time a band of deep convective clouds was moving rapidly northward through Oregon and Washington. Upper level winds for the period were moderately strong (10 to 25 m s"1) from the south and west. By 0200 July 26 Portland, 500 km south of Vancouver, was reporting altocumulus and cirrus, while some cumulus had penetrated into Washington state. By 0500 these cumulus 141 were in south-central B.C. No cumulonimbus had yet been reported at Vancouver, and Quillayute, the upper air station to the south, was reporting low and middle level coastal cloud and fog. By 0800 cumulonimbus was reported at Vancouver Airport, Victoria and other local areas. The Day 111 cloud observations occurred among the chaotic skies of the next 3 hours. By 1100, cumulonimbi were general in the area. Thunderstorms occurred from about 1100 u n t i l about 1400 with the rest of the day cloudy and cool. The thunderstorms moved north of Vancouver but did not reach Port Hardy. July 27 (Day 112) was a day of weak pressure gradients, signi-ficant diurnal heating and light winds. Fair weather cumulus dominated the skies, with some clearing evident west of Vancouver on the sa t e l l i t e photo at 1045 PDT. 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items