"Arts, Faculty of"@en . "Geography, Department of"@en . "DSpace"@en . "UBCV"@en . "Okabe, Ian T."@en . "2009-06-11T17:25:24Z"@en . "1995"@en . "Doctor of Philosophy - PhD"@en . "University of British Columbia"@en . "The North American summer monsoon is documented, using precipitation data\r\ntogether with gridded data for outgoing long-wave radiation (OLR), geopotential height\r\nand wind at various levels. The upper level divergence field is diagnosed and compared\r\nwith the precipitation field. A simple wet-dry precipitation index is used to date the\r\nmonsoon onset at stations with daily precipitation data.\r\nThe analysis shows that the monsoon rains advance northward rapidly from late\r\nJune to early July. The monsoon onset is accompanied by the development of a\r\npronounced anticyclone at the jet stream level, by sea-level pressure rises over the\r\nsouthwestern United States, and by decreases in climatological mean rainfall over\r\nadjacent regions of the United States, Mexico and the Caribbean. This coherent pattern\r\nof rainfall changes, that covers much of North and Central America, is shown to be\r\ndynamically consistent with the circulation changes aloft. Hence, the monsoon onset is\r\nembedded within a planetary-scale pattern of circulation changes. The demise of the\r\nmonsoon and the associated upper level anticyclone, which takes place around September\r\nof the year, is more gradual than the onset, and it is accompanied by an increase in\r\nrainfall throughout much of the surrounding region.\r\nThe monsoon exhibits substantial interannual variability with regard to intensity\r\nand onset date."@en . "https://circle.library.ubc.ca/rest/handle/2429/8953?expand=metadata"@en . "3412436 bytes"@en . "application/pdf"@en . "The North American MonsoonbyIan T. OkabeB.Sc. (Honours) University of British Columbia 1982A THESIS SUBMITI\u00E2\u0080\u0099ED IN PARTIAL FULFILMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDoctor of PhilosophyinTHE FACULTY OF GRADUATE STUDIESDepartment of GeographyWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAApril 1995\u00C2\u00A9 Ian T. Okabe, 1995In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of GeographyThe University of British ColumbiaVancouver, CanadaDate L7 /2\u00E2\u0080\u00995DE-6 (2188)11ABSTRACTThe North American summer monsoon is documented, using precipitation datatogether with gridded data for outgoing long-wave radiation (OLR), geopotential heightand wind at various levels. The upper level divergence field is diagnosed and comparedwith the precipitation field. A simple wet-dry precipitation index is used to date themonsoon onset at stations with daily precipitation data.The analysis shows that the monsoon rains advance northward rapidly from lateJune to early July. The monsoon onset is accompanied by the development of apronounced anticyclone at the jet stream level, by sea-level pressure rises over thesouthwestern United States, and by decreases in climatological mean rainfall overadjacent regions of the United States, Mexico and the Caribbean. This coherent patternof rainfall changes, that covers much of North and Central America, is shown to bedynamically consistent with the circulation changes aloft. Hence, the monsoon onset isembedded within a planetary-scale pattern of circulation changes. The demise of themonsoon and the associated upper level anticyclone, which takes place around Septemberof the year, is more gradual than the onset, and it is accompanied by an increase inrainfall throughout much of the surrounding region.The monsoon exhibits substantial interannual variability with regard to intensityand onset date.111TABLE OF CONTENTSAbstract iiTable of Contents iiiList of Tables ivList of Figures vAcknowledgements XChapter 1. Literature Review 11.1 Monsoons 11.2 Arizona or Mexican Monsoon 9Chapter 2. Objectives 192.1 Rationale for the Research 192.2 Objectives of the Research 192.3 Overview of the Thesis 20Chapter 3. Data Sources 223.1 Precipitation 223.2 Upper Air 233.3 Outgoing Long-wave Radiation 28Chapter 4. Precipitation 304.1 Precipitation Regimes 304.2 Wet-Dry Index 43Chapter 5. North American Atmospheric Circulation 575.1 Monthly Composites 575.2 Upper Wind Field 67Chapter 6. North American Monsoon 806.1 Upper-level Divergence 806.2 Divergence and Precipitation 816.3 Precipitation Regimes Revisited 896.4 Outgoing Long-wave Radiation (OLR) 946.5 OLR and Precipitation 1006.6 Interannual Variability 104Chapter 7. Summary and Conclusions 110References 123Appendix 131OLR Variability 1974-1990 131ivLIST OF TABLESTable 1. Mean monthly precipitation (mm) and record length (years) 35Table 2. Precipitation index changes 36VLIST OF FIGURESFigure 1. Global monsoon region as defined by Ramage (after Ramage, 1971) 3Figure 2. NMC octagonal grid 26Figure 3. NMC octagonal grid for North American sector with radiosonde stationsoverplotted 27Figure 4. Location map 34Figure 5a. Monthly precipitation (mm) for Grand Canyon, Arizona 37Figure 5b. Monthly precipitation (mm) for Tempe, Arizona 37Figure 5c. Monthly precipitation (mm) for Guaymas, Mexico 38Figure 5d. Monthly precipitation (mm) for Albuquerque, New Mexico 38Figure 5e. Monthly precipitation (mm) for El Paso, Texas 39Figure 5f. Monthly precipitation (mm) for Chthuahua, Mexico 39Figure 5g. Monthly precipitation (mm) for Winnemucca, Nevada 40Figure 5h. Monthly precipitation (mm) for Logan, Utah 40Figure 5i. Monthly precipitation (mm) for Brownsville, Texas 41Figure 5j. Monthly precipitation (mm) for Caibarien, Cuba 41Figure 5k. Monthly precipitation (mm) for Miami, Florida 42Figure 51. Monthly precipitation (mm) for Mobile, Alabama 42Figure 6a. Wet-Dry index example for Logan, 1971 45Figure 6b. Wet-Dry index example for Tempe, 1971 45Figure 7a. Annual wet-dry index for Logan, 1928-1986 47Figure 7b. Annual precipitation (mm) for Logan, 1928-1986 47Figure 8a. Annual wet-dry index for Tempe, 1926-1984 48Figure 8b. Annual precipitation (mm) for Tempe, 1926-1984 48viFigure 9a. Scatter plot of wet-dry index versus precipitation for Logan, 1928-1986 49Figure 9b. Scatter plot of wet-dry index versus precipitation for Tempe, 1926-1984 49Figure lOa. Average daily wet-dry index for Logan, 1928-1986 51Figure lOb. Average daily wet-dry index for Tempe, 1926-1984 51Figure 1 la. Running sum of average daily wet-dry index for Logan, 1928-1986 52Figure 1 lb. Running sum of average daily wet-dry index for Tempe, 1926-1984 52Figure 12a. Transformed wet-dry index for Logan, 1928-1986 53Figure 12b. Transformed wet-dry index for Tempe, 1926-1984 53Figure 13a. Transformed precipitation for Logan, 1928-1986 55Figure 13b. Transformed precipitation for Tempe, 1926-1984 55Figure 14a. Transformed wet-dry index for Winnemucca, 1928-1986 56Figure 14b. Transformed wet-dry index for Albuquerque, 193 1-1986 56Figure 15a. June composite of mean sea level pressure (mb) with a 1 mb contour interval(bold line indicates the thermal trough axis) 58Figure 15b. July composite of mean sea level pressure (mb) with a 1 mb contour interval(bold line indicates the thermal trough axis) 59Figure 15c. August composite of mean sea level pressure (mb) with a 1 mb contourinterval (bold line indicates the thermal trough axis) 60Figure 15d. September composite of mean sea level pressure (mb) with a 1 mb contourinterval (bold line indicates the thermal trough axis) 61Figure 16a. June composite of 500 mb height (m) with a 30 m contour interval (bold lineindicates the 500 mb ridge axis) 63Figure 16b. July composite of 500 mb height (m) with a 30 m contour interval (bold lineindicates the 500 mb ridge axis) 64Figure 16c. August composite of 500 mb height (m) with a 30 m contour interval (boldline indicates the 500 mb ridge axis) 65Figure 16d. September composite of 500 mb height (m) with a 30 m contour interval(bold line indicates the 500 mb ridge axis) 66viiFigure 17a. June composite of 200 mb height (m) with a 40 m contour interval 68Figure 17b. July composite of 200 mb height (m) with a 40 m contour interval 69Figure 17c. August composite of 200 mb height (m) with a 40 m cOntour interval 70Figure 17d. September composite of 200 mb height (m) with a 40 m contour interval ... 71Figure 18a. June composite of 250 mb wind (26 m s wind vector = 22 mm in length) 72Figure 18b. July composite of 250 mb wind (26 m s\u00E2\u0080\u0099 wind vector = 22 mm in length) 73Figure 18c. August composite of 250 mb wind (26 m s1 wind vector = 22 mm in length)74Figure 18d. September composite of 250 mb wind (26 m s\u00E2\u0080\u0099 wind vector = 22 mm inlength) 75Figure 19. Example of a Tropical Upper Tropospheric Trough (TUTT) 78Figure 20. Weekly averaged positions of the 250 mb upper-level anticyclone 79Figure 21. July minus June divergence (10-8 s1) difference field with a contour intervalof2.0x107s1 83Figure 22. July minus June composite monthly precipitation (locations of stations withmonthly precipitation available for at least 10 years from the WMSSC data set. A \u00E2\u0080\u0098+\u00E2\u0080\u009Cdenotes an increase in precipitation from June to July and locations with a denotes adecrease in precipitation from June to July) 84Figure 23. July minus June composite monthly precipitation overlaid on July minus Junedivergence (10-8 s) difference field with a contour interval of 2.0 x i0 s1 85Figure 24. September minus August composite monthly precipitation overlaid onSeptember minus August divergence(10 s1-)difference field with a contour interval of2.0 x i0\u00E2\u0080\u00997 s1 87Figure 25a. Scatter plot of July minus June monthly precipitation (mm) versus July minusJune divergence (10-8 4) for stations in Mexico, Cuba, Jamaica, and the United Statesup to 42\u00C2\u00B0N latitude 90Figure 25b. Scatter plot of September minus August monthly precipitation (mm) versusSeptember minus August divergence (10-8 s1) for stations in Mexico, Cuba, Jamaica,and the United States up to 42N latitude 90viiiFigure 26. July-June precipitation index for stations in Table 2 overlaid on July minusJune divergence (10 s1) difference field with a contour interval of 2.0 x i0 s 91Figure 27. September-August precipitation index for stations in Table 2 overlaid onSeptember minus August divergence (10-8 s-\u00E2\u0080\u0099) difference field with a contour interval of2.0 x iO-7 s1 92Figure 28a. June composite of OLR (W rn-2)with a contour interval of 10 W m2(contours less than or equal to 240 W rn-2 are solid) 96Figure 28b. July composite of OLR (W rn-2)with a contour interval of 10 W m2(contours less than or equal to 240 W rn-2 are solid) 97Figure 28c. August composite of OLR (W rn-2)with a contour interval of 10 W m2(contours less than or equal to 240 W rn-2 are solid) 98Figure 28d. September composite of OLR (W rn-2)with a contour interval of 10 W rn-2(contours less than or equal to 240 W rn-2 are solid) 99Figure 29. July minus June difference field for OLR (W rn-2)with a contour interval of 5w m-2 (negative contours are dashed) 101Figure 30. September minus August difference field for OLR (W rn-2)with a contourinterval of 5 W rn-2 (negative contours are dashed) 103Figure 31a. OLR index from 1974- 1990 106Figure 31b. Precipitation index from 1974-1988 106Figure 32. Scatterplot of OLR index versus Precipitation index 108Figure 33. Conceptual model of North American monsoon (Thick lines are jetstreamaxes. The strong upper-level anticyclone position is designated by a large H. The weakanticyclone is designated by a small H. Areas of increasingly divergent flow aloft (DIV)and increasingly convergent flow aloft (CON) are also noted) 114Figure 34. Global monsoon region as defined by Ramage (after Ramage, 1971) withapproximate area of North American monsoon added 116Figure Al. July minus June OLR (W m2) difference field for 1974 with a contourinterval of 5 W m2 (negative contours are dashed) 132Figure A2. July minus June OLR (W rn-2) difference field for 1975 with a contourinterval of 5 W rn2 (negative contours are dashed) 133ixFigure A3. July minus June OLR (W rn2) difference field for 1976 with a contourinterval of 5 W m2 (negative contours are dashed) 134Figure A4. July minus June OLR (W rn-2) difference field for 1977 with a contourinterval of 5 W m-2 (negative contours are dashed) 135Figure A5. July minus June OLR (W rn-2) difference field for 1979 with a contourinterval of 5 W rn-2 (negative contours are dashed) 136Figure A6. July minus June OLR (W rn-2) difference field for 1980 with a contourinterval of 5 W rn-2 (negative contours are dashed) 137Figure A7. July minus June OLR (W rn-2) difference field for 1981 with a contourinterval of 5 W rn-2 (negative contours are dashed) 138Figure A8. July minus June OLR (W rn-2) difference field for 1982 with a contourinterval of 5 W m2 (negative contours are dashed) 139Figure A9. July minus June OLR (W rn2) difference field for 1983 with a contourinterval of 5 W rn-2 (negative contours are dashed) 140Figure AlO. July minus June OLR (W rn2) difference field for 1984 with a contourinterval of 5 W rn-2 (negative contours are dashed) 141Figure All. July minus June OLR (W rn-2) difference field for 1985 with a contourinterval of 5 W m2 (negative contours are dashed) 142Figure A12. July minus June OLR (W rn-2) difference field for 1986 with a contourinterval of 5 W rn-2 (negative contours are dashed) 143Figure A13. July minus June OLR (W rn2) difference field for 1987 with a contourinterval of 5 W rn2 (negative contours are dashed) 144Figure A14. July minus June OLR (W rn2) difference field for 1989 with a contourinterval of 5 W rn2 (negative contours are dashed) 145Figure Al5. July minus June OLR (W rn2) difference field for 1990 with a contourinterval of 5 W rn2 (negative contours are dashed) 146xACKNOWLEDGEMENTSThis thesis is the final culmination of several years of graduate work at fourdifferent universities. Without the help of numerous individuals its completion would nothave occurred.To my Ph.D. committee I owe the most. Dr. Gordon McBean, Professor Tim Oke,and Professor Mike Wallace have each played the role of \u00E2\u0080\u009CSupervisor\u00E2\u0080\u009D in my graduatework, and have carefully guided me towards completion. Their dedication and patience iswithout equal.I wish to thank my family for letting me pursue endless years of university studywithout question. Special thanks goes to both my parents for instilling in me theimportance of education.1CHAPTER 1. LITERATURE REVIEW1.1 MONSOONSThe word monsoon is derived from the Arabic mausim, meaning a season, andwas first applied to the seasonal shift in winds that occur over the Arabian Sea (AMS,1959). A consensus on the definition of the monsoon has yet to be reached (Ramage,1971). Most researchers would agree that a monsoon is characterized by two differentseasonal circulations: a surface \u00E2\u0080\u009Cin-flow\u00E2\u0080\u009D into a thermal low during summer; and asurface \u00E2\u0080\u009Cout-flow\u00E2\u0080\u009D from an anticyclone during winter. The circulation arises due todifferential heating between land and ocean.During the summer, uneven heating occurs between adjacent land and oceansurfaces because of the higher thermal capacity of the ocean and its greater ability toredistribute the heat to greater depths by radiation and mixing, and its lower Bowen ratio.The difference in temperature gives rise to a thermally direct circulation similar to themesoscale day-time sea breeze. During winter, the situation is reversed and the land iscooler than the surrounding oceans. This, in turn, creates a circulation similar to thenight-time land breeze. However, the space and time scales of monsoon circulations arelarge enough such that it is affected by the rotation of the planet. The monsoon can bebroken into summer and winter components. Summer (winter) monsoons reach theirhighest intensity near the time of the summer (winter) solstice (Das, 1986).In order to classify a region as having a monsoon climate, Ramage (1971)proposed that it meet several criteria:a) the prevailing wind direction must shift by at least 1 2O between January and July;b) the average frequency of winds from the prevailing direction in January and July2exceeds 40 %;c) the mean resultant winds in at least one of the months (January or July) must exceed 3ms1;andd) fewer than one cyclone-anticyclone alternation occurs every two years, in eithermonth, in a 5 latitude-longitude rectangle.Although the first three conditions have subjective numerical thresholds theyensure that winds are of the right direction and strength in winter and summer. The fourthcriteria emphasizes that a persistent thermal cyclone in summer is replaced by an equallypersistent anticyclone in winter.Using such criteria about one half of the tropics may be defined as having amonsoon climate (Webster, 1987). Ramage\u00E2\u0080\u0099s monsoon area comprises Africa, southernAsia, and northern Australia (Figure 1). This definition does not include precipitation inany manner. Therefore, areas in Africa that receive little summer rainfall would beclassified as a monsoon climate, as would India, which receives over 10 m of rainfall peryear. Hence, most researchers now add rainfall in some way to their definition of amonsoon climate. The most common addition is that the region should have apronounced maximum in precipitation during the summer and a minimum during winter.It is the summer rainfall maximum, and its importance to agriculture, that has createdsuch great research interest in monsoons. The majority of work has concentrated on theIndian, or Asian monsoon, because it is the largest of all monsoons.It is important to note that the beginning of the summer rains does not necessarilycoincide with the change in surface wind direction. Thus, the definition of the monsoononset depends on whether researchers choose the change in wind direction, or the3Figure 1 Global monsoon region as defined by Ramage(after Ramage, 1971)4beginning of the heavy rain, to indicate the start of the monsoon season. Using thedefinition based on rainfall the monsoon begins on the southern tip of India near thebeginning of June, with a standard deviation in arrival time of only one week (Das,1986). The earliest recorded arrival time over the last 100 years was on May 7, 1918 andthe latest on June 22, 1972 (Joseph et at, 1994). It progresses northwestwards until itreaches the northwestern region of India near the beginning of July. Its retreat from Indiais complete by the middle of September.The heavy summer rains are frequently interrupted by periods of light or no rain.These periods are known as \u00E2\u0080\u009Cbreaktt monsoons. Such breaks last about one week. Thedistribution of rainfall varies with topography and the preferred track Of rain-bearingsystems. For instance, the average seasonal monsoon rainfall over northeastern India is1650 mm while the extreme northwest receives less than 300 mm (Das, 1986). Anotherexample is the mountain range along western India known as the Western Ghats (highestelevation 1.5 km). Its orientation is such that it lies nearly perpendicular to thesouthwesterly monsoon winds. Therefore, rainfall can be 5 times greater on the windwardside, than the lee side.To further our knowledge of monsoon climates there have been internationalresearch efforts. The earliest was the International Indian Ocean Expedition during 1963-65, then the Indo-Soviet Experiments of 1973, and 1977 (Das, 1986). The largest projectwas the Monsoon Experiment (MONEX), which began December 1, 1978, and lasted forone year, and coincided with the Global Weather Experiment (WMO, 1982). MONEXhad three components:a) the Asian summer monsoon;b) the Asian winter monsoon; and5c) the West African monsoon.An intense observation period took place from January 5 to March 7, 1979 tocoincide with the winter monsoon. A second phase took place from May 1 to June 30,1979 to cover the beginning of the summer monsoon. A few countries extended theirspecial observational programmes to the end of August to cover the later part of themonsoon. Special observations came from research ships and aircraft, in addition toincreased radar coverage, and real-time GOES satellite information. All data have beenarchived and are available at several international data centres. Most of the researchliterature quoted in this chapter are a direct result of these international efforts.The large scale features of the Asian monsoon are fairly well known now (Das,1986). Atmospheric circulation in the form of Hadley cells arise in response to majorsources and sinks of heat. When the Sun is overhead of the Tibetan plateau, it acts as amajor heat source, and becomes the location of the ascending branch of a Hadley cell(Koteswaram, 1958). The descending branch is over a large surface anticyclone south ofthe Equator, known as the Mascarene High. In winter the circulation reverses, and theascending branch is found over Indonesia, and surrounding ocean areas. Here, the releaseof latent heat by moist convection is thought to be the major heat source. The descendingbranch during winter is to the north, over the Siberian High.He et al (1987) analyzed First GARP Global Experiment (FGGE) data for an 80day period near the onset of the 1979 summer monsoon, and found descent to the northand west of the area of ascent over the Tibetan plateau. Broccoli and Manabe (1992)found similar ar\u00E2\u0080\u0099\u00C3\u00A8as of descent and ascent around the Tibetan plateau using the GFDLGlobal Climate Model. They hypothesized that drier conditions to the north of theTibetan plateau were due in part to the subsidence created by the Tibetan circulation.6Krishnamurti et al (1989) examined a particularly dry summer, the 1987 Indian monsoon.They found that the upper-level anticyclone shifted eastward allowing for the presence ofa westerly upper wind anomaly into India which brought dry air over the region,inhibiting convection. Under normal circumstances a westerly jetstream is observed northof the upper-level anticyclone while an easterly jetstream is to the south (Soman andKumar, 1993)Regional or synoptic scale monsoon features are less well understood, forexample:a) the rainfall variations within the monsoon season;b) the formation of a low-level monsoon trough along the southern border of theHimalayas, and the thermal low which develops over northwestern India;c) the extension of the monsoon trough into the Bay of Bengal, and why it is a favouredarea for formation of tropical depressions (Das, 1986); andd) the westward propagation of monsoon depressions and equatorial waves.Some evidence suggests that depressions and waves may be just remnants ofwestward moving disturbances that move into the monsoon area, nevertheless they stillinteract with the monsoon. It is this interaction that is not well understood.The sources of moisture for the Asian monsoon have always been clear. Forexample, moisture can come from:a) rain bearing systems like tropical depressions or easterly waves;b) the monsoon trough situated along the northeastern border of India;c) mid-level tropospheric disturbances (near 600 mb) called cyclonic vortices;d) off-shore vortices that develop due to the enhanced topography of the southwesterncoast of India (sometimes the southwest flow does not rise over the mountains but7curves back to form a mesoscale low);e) convective disturbances brought across the Arabian Sea by a low-level jet near the 1.5km level with wind speeds of 15 m s\u00E2\u0080\u0099 (Das, 1986).An example of oceanic influence on the monsoon is the possibility that the Asianmonsoon may be affected by El Ninos. During the 120 year period from 1870-1989 therewere 22 years in which the onset of the monsoon was delayed by more than 8 days. In 16of those years there was also a moderate or strong El Nino (Joseph et al, 1994).Such interannual variability leads to interest in the possibility of monsoonprediction; in particular, the onset of the monsoon, and how much rainfall accompanies it.The focus in work thus far has been to develop regression equations with severalpredictors, among them the geopotential height and temperature over India in April, andthe geopotential height, temperature, and wind at multiple levels over Australia fromJanuary to April (Kung and Sharif, 1980). Using such equations Kung and Sharif (1980)were usually able to predict the start of the monsoon, at the southern tip of India, towithin 3 days for the period 1966-1980. The worst prediction occurred in 1979 when themonsoon was late by ten days. The prediction of the time of onset has been moresuccessful than for rainfall amounts, but work is continuing (Prasad and Singh, 1992;Hastenrath, 1988).The Australian northwest monsoon can be seen as an extension of the Asianwinter monsoon (Joseph et al, 1991). During the northern hemisphere winter thedescending branch of the Hadley cell is over Siberia and the ascending branch is overIndonesia. Thus\u00E2\u0080\u0099 the sections of northern Australia in close proximity to Indonesia aresubject to monsoon rains during their summer season. A dry winter follows, thussatisfying Kendrews (1961) criteria for a monsoon climate.8The African monsoon was also investigated during MONEX . The West Africanmonsoon is the dominate feature, and it is marked with a reversal of surface winds overWest Africa, from northeasterly during the winter season to southwesterly during thesummer. Eastern Africa is also thought to have a monsoon climate but its climate is moreassociated with the movement of the Inter-Tropical Convergence Zone (ITCZ) than anonshore maritime flow (Das, 1986). The ascending branch of the Hadley cell is over theSahel region with the descending branch off the coast of west Africa, which merges withthe southwesterly low-level onshore flow (Das, 1986).General Circulation Models (GCM) could be extremely helpful in understandingthe monsoon. A variety of experiments have already been completed. The mostsuccessful have been sensitivity tests to assess the responses of monsoon characteristicsto variation of several parameters. Changes in sea-surface temperature, snow cover, andsoil albedo have been shown to affect the intensity of the monsoon (Meehl, 1994). Acoupled ocean-atmosphere model would help because the Indian Ocean responds readilyto the shift in monsoon winds. However, the ocean part of such models is still consideredthe weak component in such an endeavour (Meehl, 1989). The removal of the Himalayanmountains in a GCM model indicates that the abrupt northward movement of the westerlyjet would no longer take place. Instead a more gradual northward movement of the jet,and a slower northward progression of the monsoon rains occurred in the model (Hahnand Manabe, 1975). By reducing the elevation of the Tibetan plateau they found thatwithout this elevated heat source the Hadley circulation of the summer monsoon wasmuch weaker. A serious difficulty in modelling the Asian monsoon arises whenrecreating the topography of the Himalayans which reach to the 500 mb level. Fennessy9et al (1994) found that sensitivity to the model\u00E2\u0080\u0099s topography influenced the rainfallsimulation much more than any changes to vegetation, soil moisture, or cloudiness.In general, monsoons are characterized by large anticyclonic circulations in theupper troposphere, which occur in both summer and winter monsoons (Das, 1986). Aneasterly jet is observed to the south of the anticyclone, while a westerly jet is to the north(Chen et al, 1989; Soman and Kumar, 1993). A rapid northward movement of thewesterly jet is common prior to the onset of the summer monsoon. In some years thenorthward movement of the westerly jet is delayed, while in other years the anticyclone isnot as strong. At the surface a strong thermal low (low-level monsoon trough) is presentduring summer monsoons and a strong anticyclone during winter monsoons. This allowsfor the reversal of winds found in monsoon climates. Finally, a predominantly dry winteris followed by a wet summer.1.2 ARIZONA OR MEXICAN MONSOONMonsoons are part of the summer general circulation pattern and occur over mostcontinents (Das, 1986). Ramage\u00E2\u0080\u0099s (1971) definition of monsoon climates does not includeanywhere in North America. Kendrew\u00E2\u0080\u0099s (1961) monsoonal category focuses on the heavysummer rainfall rather than the seasonal shift in winds, and is described as \u00E2\u0080\u009CHeavysummer rainfall and a long dry season centred on winter.\u00E2\u0080\u009D Hence, portions of Mexico andCentral America satisfy Kendrew\u00E2\u0080\u0099s definition. Ramage (1971) does not include MexicoandCentral America as a monsoon region because the weak change in surface winds donot satisfy his criteria. On the other hand Arizona is not included in Kendrew\u00E2\u0080\u0099smonsoonal category, because it does not have a dry winter. However, the term \u00E2\u0080\u009CArizona10monsoon\u00E2\u0080\u009D has become entrenched in the research literature to describe summerprecipitation in Arizona, and its use will be continued here.Because Arizona becomes extremely hot in the summer, the consensus in the1950s was that any moisture advected into the area would rise and condense via low-levelconvergence due to the intense thermal low. This in turn could produce considerableamounts of convective rainfall. Hence, the search for that source of moisture becameparamount. Jurwitz (1953) found that much of Arizona\u00E2\u0080\u0099s summer rainfall is restricted tothe southeast, which he maintained was due to its proximity to a source of moisture, theGulf of Mexico. Using a particular case study, Jurwitz concluded that the trajectory formoist air was directly from the Gulf of Mexico along the Texas-Mexico border at the 700mb level. He also commented that on occasion Arizona can receive considerableprecipitation associated with moisture from the remnants of tropical cyclones as theymake landfall and dissipate over Mexico, if the upper flow is from the southeast. Brysonand Lowry (1955a, 1955b) analyzed several years of data and came to the conclusion thatsoutheasterly flow around an anticyclone, which establishes itself over the middle of theUnited States during the height of summer, brings moisture from the Gulf of Mexico intoArizona.The Bryson and Lowry (l955a, 1955b) hypothesis of moisture transport wasaccepted until Hales (1972, 1974) proposed a completely different one. Hales (1974)noted that flow from the Gulf of Mexico has to cross extensive terrain, mostly above the800 mb level, before it arrives in Arizona. This did not converge with the work of Reitin(1960) who discovered that 50 % of all precipitable water over Phoenix was locatedbelow 800 mb. Therefore, Hales could no longer ascribe to the hypothesis that the Gulf ofMexico was the source region. Based on four case studies, Hales (1972, 1974) found that11the Gulf of California acts as a perfect channel for tropical moisture surges. Under normalconditions, Hales suggested that a balance somehow existed between the hot, dry, thermallow pressure conditions in Arizona, and the cooler, moist, higher pressures across thesouthern part of the Gulf of California. Thus Hales implied there would be little or nowind flow between the higher pressure to the south and the lower pressure to the north.He did not provide the reasoning necessary to understand such a \u00E2\u0080\u009Cbalance\u00E2\u0080\u009D. However, heacknowledged that a local land-sea breeze phenomenon often brings cooler moist air intothe extreme southwest part of Arizona. However, it only extends a few miles inland, asshown by Dodd\u00E2\u0080\u0099s (1965) analysis of dewpoint temperatures. Hales theorized that an\u00E2\u0080\u009Cimbalance\u00E2\u0080\u009D can be triggered by the passage of a tropical cyclone or easterly wave acrossthe southern part of the Gulf of California. The reasoning is as follows. The tropicaldisturbance is associated with considerable convective rainfall. The rain cools the airmassbelow it by evaporation. This cooling causes an increase in surface pressure and induces a\u00E2\u0080\u0098giant sea breeze\u00E2\u0080\u009D towards the thermally-induced lower pressure in Arizona (Brenner,1974). This imbalance initiates a surge of moisture from the tropical air near the southernpart of the Gulf of California, towards Arizona. Two types of surges were noted. Ashallow surge below 1500 m (5000 feet) which typically cools southwestern Arizona by5CC, causing stabilization of the airmass, and thereby, inhibiting thunderstorms. A deepersurge with moisture reaching up to 3600 m (12000 feet) also causes cooling, butaccording to Hales, increases thunderstorm activity by increasing the moisture availablefor convection. Either type of surge can extend as far north as the Arizona-Utah border,and fill the thermal low in less than 24 hours.In a more extensive analysis, Badan-Dangon et al. (1991) used instrumentedaircraft and automatic weather stations to study the three-dimensional structure of the12planetary boundary layer over the Gulf of California. They found that the surges last anaverage of 5 days with southeasterly surface winds of approximately 10 m s within theGulf of California. Their concept of these events differs from that of Hales. Unlike Hales,they did not find sudden wind shifts associated with the leading edge of the pulse, and themoisture transition is not abrupt but lasts over several days. The results of Badan-Dangonet al. (1991) are the more convincing, because of the limited number of cases in theHales\u00E2\u0080\u0099 study.Many have misinterpreted Hales\u00E2\u0080\u0099 study by concluding that the waters of the Gulfof California act as the moisture source for the Arizona monsoon. The surface watertemperature of the Gulf of California is approximately 25\u00C2\u00B0C (Badan-Dangon et al., 1991),and above the ocean is a 200-300 m deep marine layer in which dew points range from18-25\u00C2\u00B0C. Above the marine layer, the air is still relatively moist with dew points rangingfrom 17-21 \u00C2\u00B0C. Given that there is an abundance of moist air in the Gulf of California itseems reasonable that the Gulf itself could provide some of the moisture for the monsoon.However, Hales emphasized that the Gulf of California merely acts as a conduit formoisture advection. He hypothesized the transport of eastern Pacific tropical air from thewaters off western central Mexico to Arizona. Both Hales (1974) and Brenner (1974)conclude that the thermal low over Arizona is filled approximately 24 hours after thesurge begins, and thereafter, a pressure balance exists between the southern Gulf ofCalifornia and Arizona.Several aspects of Hales\u00E2\u0080\u0099 approach remain to be resolved:a) the postulated initial balance between the thermal low and higher pressures to thesouth requires more investigation;b) the forcing mechanism for the surge is uncertain;13c) the duration of the surge is debatable (Hales suggested that after 24 hours the surgeshould no longer exist, but recent observations show it lasts on average for 5 days(Badan-Dangon et al., 1991));d) the role of initially cooler air from the Gulf of California, ahead of the warm moisttropical surge of air is uncertain (the cooler air will inhibit or destroy thunderstormactivity by destroying the thermal forcing. If thermal forcing is the only mechanisminvolved then it will be some time before thunderstorm activity occurs. It is obviousthat the surges occur but an abundance of warm moist air is insufficient to producemonsoon type rains. Considerable vertical motion must be involved);e) the hypothesis does not account for monsoon rainfall in New Mexico, western Texas,southern Nevada, and southern Utah;f) the extreme southwest of Arizona should experience the monsoon rains since itborders the channel for moisture transport but it is eastern Arizona that receives mostof the summer rains and lightning activity (Reap, 1986).Thus the relationship between Gulf of California surges and precipitation in thesouthwestern United States is still unclear (Douglas, 1992a and Douglas et al., 1992).However, despite its shortcomings, Hales\u00E2\u0080\u0099 hypothesis replaced Bryson and Lowry\u00E2\u0080\u0099s(1955a, 1955b) earlier hypothesis, and has remained unchallenged since 1974.After the debate about the source of the monsoon moisture sources was quelled,researchers turned towards more in depth studies of the Arizona monsoon. In 1975, Halesexamined an intense thunderstorm using both radar and satellite imagery. A low-levelmoisture surge occurred the evening before the event so that most of southwesternArizona possessed considerable low-level moisture. However, the thunderstorm formedin the east, along the Arizona-New Mexico border, over the Colorado plateau at an14elevation of over 1500 m (5000 feet). It is unlikely that the Gulf of California moisturesurge reached this far eastward and to this altitude. Although Hales did not mention it,this suggests the possibility of a different source of moisture for this thunderstorm. Thestorm moved southwestward off the plateau into the desert and diminished in strength.The cells intensified again as they encountered the coast mountain ranges east of SanDiego and broke up later as they moved westward over the mountains. Initially, themoisture surge had little if anything to do with the storm, but it probably had an effectwhen the storm re-intensified over the area along the coastal mountains. Even though thiswas a major storm only two stations in Arizona measured precipitation. One measured 80mm (3.1 inches), equivalent to the yearly total for that station. It illustrates both the\u00E2\u0080\u009Cpatchiness\u00E2\u0080\u009D of the rain and the low station density in Arizona.Carleton (1985, 1986) conducted a number of statistical studies relating 3 years ofArizona monsoon rainfall to different atmospheric features, such as the strength of theArizona thermal low, the intensity of the Bermuda anticyclone, height of the 700 mbsurface, eastern Pacific sea surface temperatures, etc. No strong conclusions were reachedbut in later studies he found that Arizona summer rainfall is related to the position of the500 mb anticyclone, whose position lies just to the east of Arizona during the height ofsummer (Carleton, 1987 and Carleton et al., 1990). During wetter Arizona summers the500 mb anticyclone was found to be displaced farther northward. Carleton surmised thatthis allowed more moisture to penetrate from the south. He did not mention whether thismoisture was from Gulf of California surges or transported around the 500 mb high fromthe Gulf of Mexico.Moore et al. (1989) and Adang and Gall (1989) suggest a different way toexamine the Arizona monsoon. During summer the Pacific and Atlantic surface15subtropical highs dominate the surface pressure patterns. The North American continentdivides the two features. Mexico tends to be directly in between the two flows. On thewest is a weak westerly or northwesterly flow, and on the east, an easterly orsoutheasterly flow often prevails. They suggest that this line of confluence be called the\u00E2\u0080\u009CArizona monsoon boundary\u00E2\u0080\u009D. Their analysis suggests it has many of the same features asa mid-latitude front with wind shear causing instability and possible baroclinic waves.This of course assumes that airmasses can transverse the Sierra Madres mountain range.It is also possible that the line of confluence is artificially imposed by the mountains.A series of papers by Howard and Maddox (1988a, 1988b), and Maddox andHoward (1988a, 1988b), and Maddox et al. (1992a) made extensive use of satelliteimagery to document the convective systems that occur over northwestern Mexico duringsummer, and their contribution to the Arizona monsoon. An initial finding shows thatthunderstorm activity shifts from eastern Mexico to western Mexico by the middle of thesummer (Howard and Maddox, 1988b). Their investigation of two case studies ofmesoscale convective systems shows that convective instability is similar in strength tothat associated with severe thunderstorms over the United States (Maddox and Howard,1988a). Despite the strong convection they found no obvious large scale forcing forupward motion (Maddox and Howard, 1988b) and hypothesized that mesoscale processessuch as sea breezes and valley winds may initiate the development of the convectivesystems. Their final paper in the series highlights the typical life cycle and motion ofmesoscale convective systems over Mexico (Howard and Maddox, 1988a). The mainconclusion to be\u00E2\u0080\u0099drawn from their series of papers is that the extent of the convectionover Mexico is clearly evident which hints that the southwest monsoon is much larger inextent than previously thought and appears to extend well into Mexico.16In order to fully explore such aspects of the monsoon a large scale project wasundertaken in the summer of 1990, called the SouthWest Area Monsoon Project, orSWAMP (Meitin et al., 1991), to study:a) central Arizona thunderstorm environments;b) monsoon structures and moisture fluxes; andc) convective systems in Mexico.From July 9 to August 7, 1990, aircraft transits, radar and satellite images,frequent upper air soundings, lightning data, and continuous recording rainfall gaugeswere gathered to form the most extensive observational network ever provided for thearea.The SWAMP findings are summarized in a paper by Douglas et al. (1993). Theyexamined both Mexican and United States rainfall and rawinsonde data, as well assatellite imagery to show that the Arizona monsoon is just the northern extent of a moreprominent Mexican monsoon. Hales (1974) and Brenner (1974) restricted their analysesto Arizona, and therefore, did not see beyond the border, and missed the connection withMexican summer rainfall. Examining monthly precipitation data Douglas et al. (1993)developed a monsoonal index, which is the ratio of July, August, and September rainfalltotals, to the annual mean precipitation. Their monsoonal index shows a maximum inMexico near the southern Gulf of California which continues northward along theMexico coast. It decreases in intensity once it reaches the Mexico-United States border.Douglas et al. (1993) also used satellite imagery to better define the spatial extentof the Mexican monsoon. Using eight years of data (1985-1992) they created colourcomposite maps of the frequency of occurrence of infrared cloud top temperatures colderthan -38\u00C2\u00B0C. This surrogate for deep convection shows that the convection is prominent17over western Mexico in June, but is not as evident over northwestern Mexico until July,thus corroborating their rainfall analysis.Douglas et al. (1993) went on to use upper air data to determine the source ofmoisture for the Mexican monsoon. Rawinsonde data along the northwestern Mexicocoast shows the mid-level flow switching from westerly in June to easterly in July. Thiswould support the earliest hypothesis of advection from the Gulf of Mexico (Bryson andLowry, 1955a, 1955b), but Douglas et al. (1993) is quick to point out that a dew pointanalysis at both the 700 mb and 500 mb levels shows that eastern Mexico has much drierconditions than western Mexico. They suggest that the higher moisture content at upperlevels over western Mexico may be due to ascent associated with the easterly flow overtopography, and/or vertical mixing of boundary level moisture by convective scalemotion. The lack of rawinsonde data makes it difficult to examine the topographicforcing hypothesis. The hypothesis of low level moisture vertically transported byconvection then poses the question of where the low level moisture comes from initially.The Gulf of California could provide the low level source of moisture. Douglas et al.(1993) suggest that local evaporation from the Gulf of California may or may not play animportant role but acknowledge that Hales\u00E2\u0080\u0099 (1974) hypothesis that the Gulf of Californiaacts as a conduit for moisture transport from the tropical Pacific ocean likely plays animportant role. However, as discussed previously, Hales (1974) hypothesis still has anumber of deficiencies.Although Douglas et al. (1993) did not provide an answer to the moisture sourceof the monsoon, \u00E2\u0080\u0098they did identify that the Arizona monsoon has a larger spatial extentwith the strongest changes occurring in northwestern Mexico. Given their evidence they18suggest that the Arizona monsoon should be more appropriately named the Mexicanmonsoon.A second project occurred in the summer of 1993 called the ExperimentoMeteorologico del Verano (EMVER) (Douglas, 1993) to supplement the spatial densityof upper air observations obtained in the SWAMP project, especially over Arizona.The most important discovery thus far to arise from these projects was the exactnature of the trajectory of low level flow in the Gulf of California during the monsoonseason (Douglas et al, 1991, 1992 and Douglas, l992a, 1993). A distinct southeasterlyflow was evident up the Gulf of California. Surface winds near 9 m s\u00E2\u0080\u0099 were apparentover the Gulf but just inland along the coast winds were nearly calm. This indicates thatin past studies the southeasterly flow may have been missed by the network of coastalweather stations. EMVER balloon observations also noted a strong diurnal change in thewinds. In the early morning hours the flow was parallel to the coast but by afternoonshifted to an onshore and upsiope flow, presumably due to a strong sea breezedevelopment.EMVER observations also captured many \u00E2\u0080\u009CGulf surges\u00E2\u0080\u009D of air up the Gulf ofCalifornia which are critical to Hales (1974) moisture source hypothesis. The twoprojects also showed the variability of the monsoon. The observation period of theSWAMP project was unusually wet while the EMVER project was characterized byseveral extended dry spells. Analysis of the extensive data sets from both the SWAMPand EMVER projects is continuing and there is considerable hope that they will answermany questions about the mesoscale aspects of the monsoon.19CHAPTER 2. OBJECTIVES2.1 RATIONALE FOR THE RESEARCHIn his monographtThe Climates of the Continents\u00E2\u0080\u009D, Kendrew (1922) describedvarious precipitation regimes\u00E2\u0080\u0099 that exist in North America, but offered little, or noreasons as to why they existed. Since that time several researchers have quantified thevarious regimes using different statistical techniques, but there is still little if any workdone to show why they occur. This thesis was originally undertaken to better understandthe summer precipitation regimes that exist in North America, in particular, the initialfocus was on the Arizona (or Southwest) monsoon. The research expanded beyondArizona, as the spatial extent of the monsoon became evident.2.2 OBJECTIVES OF THE RESEARCHPeople living in Arizona and New Mexico are quite familiar with the dangerouslightning storms, strong winds, heavy rains, and flash floods that come with the summermonsoon. Atmospheric scientists are also familiar with the events, but a completeunderstanding of the monsoon is still not in hand (Douglas et al., 1993). Much of the pastresearch on the Southwest monsoon has focussed on either, case studies of singularevents, or mesoscale aspects of the phenomenon (Douglas, 1992b, 1992c). This thesis is aclimate diagnostic study of the summer monsoon. The results may be of interest toclimate modellers, since the monsoons are an important part of the global circulation.Theprimary goal is to provide other researchers and forecasters with a large scale view of the\u00E2\u0080\u0098The character of the seasonal distribution of rainfall at any place (AMS, 1959)20phenomenon, with a view to its better prediction in the future. In order to accomplish thisgoal the following specific objectives are undertaken:a) to document the changes that occur in summer precipitation (spatial changes are wellknown but temporal changes require further investigation);b) to examine changes in atmospheric circulation that coincide with the changes insurface precipitation (circulation from the surface to the top of the troposphere will beinvestigated);c) relate the changes that occur in atmospheric circulation to changes in surfaceprecipitation;d) examine independent evidence (outgoing long-wave radiation which is a surrogate forconvective precipitation) to corroborate the relationship between changes inatmospheric circulation and changes in surface precipitation;e) re-examine the spatial scale of the monsoon; andt) examine the interannual variability of the monsoon.2.3 OVERVIEW OF THE THESISThe thesis starts with a review of past literature related to the phenomenon. Anexamination of monthly precipitation in North America identifies different precipitationregimes, including the summer monsoon. A simple index is used to explore the rapidchanges in precipitation associated with the monsoon. Standard meteorological fields arethen analyzed, in an attempt to dynamically interpret the summer monsoon. Divergencecalculations show that upper-level anticyclones are related to areas of convective activity.Outgoing Long-wave Radiation (OLR) is used to corroborate the relationship. The year-21to-year variability of the monsoon is explored with OLR difference fields. Finally,conclusions and implications of the research are presented.I22CHAPTER 3. DATA SOURCES3.1 PRECIPITATIONMonthly precipitation data used in this study are from the World Monthly SurfaceStation Climatology (WMSSC) and the United States Climate Division data set, bothavailable from the National Climatic Data Center (NCDC), in Asheville, North Carolina.The WMSSC contains monthly precipitation time series for over 3000 stationsworld wide. NCDC provide quality control by scanning and correcting for severalhundred gross errors related to incorrect data entry. Extreme values beyond 5 standarddeviations from the long period monthly mean are also inspected. Those that NCDCbelieve to be obviously the result of publication error are set to \u00E2\u0080\u0098missing\u00E2\u0080\u009D.In addition to the NCDC quality control the following procedures wereimplemented for the data used in this study. Monthly precipitation data were first filteredto find missing values. An objective check was conducted to compare average annualprecipitation to individual anomalously high monthly precipitation. Finally each year, forevery station used in this study, was inspected visually by means of a time series ofmonthly precipitation to highlight possible incorrect values. No additional suspect datawere found using these additional techniques.The United States Climate Division data set was developed by the NationalClimate Data Center (NCDC). Each individual state was separated into 5 to 10homogeneous climate divisions, and all available stations within each area were weightedequally and averaged, to produce the mean monthly precipitation for each division. Thesize of the division depends on the spatial homogeneity of the climate within a state. Theborders of the climate divisions are based on station availability, topography, and23differences in monthly station data. The borders changed slightly in 1965 from theiroriginal definition in 1951. Obviously, this creates a problem when attempting to evaluatetrends in the time series. The changes also create problems in the early part of the record,when some divisions had only one or two stations. However, it is one of the mostcomprehensive monthly precipitation data sets for the United States and is essentiallycontinuous from 1895 to 1988. Only climate divisions in Arizona and New Mexico wereexamined in this study.Daily precipitation data for Albuquerque, Logan, Tempe, and Winnemucca arefrom the Daily Climate Summary, available from NCDC.The accuracy of rainfall measurements is hard to determine. The treatment oftrace values as zero in the monthly totals can lead to an underestimate of precipitation, ascan evaporation from rain gauges. On the other hand, the possibility of rain splashing intothe gauge can lead to an overestimate of rainfall. Groisman and Legates (1994) estimate arain gauge measurement error of 5% during summer over the United States.3.2 UPPER AIRUpper air data used in this study were taken from the National MeteorologicalCenter\u00E2\u0080\u0099s (NMC) Northern Hemisphere octagonal grid data set. Gridded fields of monthlymean sea level pressure, 500 mb height, 200 mb height, and daily 250 mb zonal andmeridional wind components were used. Mean sea level pressure and 500 mb height7 fields have the most extensive past coverage starting in January 1946. Obviously, overthe decades there have been considerable changes both in the availability of observations,and the objective analysis schemes used to assimilate those observations.24A lack of surface observations in the early part of the record, and a serious lack ofupper air observations led to the practice of \u00E2\u0080\u009Cbogusing\u00E2\u0080\u009D data. Skilled meteorologicalanalysts constructed \u00E2\u0080\u009Csubjective analyses\u00E2\u0080\u009D based on the available data to estimate thepressure or height in data sparse regions. As the number of observations increased overthe years, and as remotely sensed data became available in the late 1960s, the practice of\u00E2\u0080\u009Cbogusing\u00E2\u0080\u009D was reduced. Hence, the early part of the pressure and height data sets are ofquestionable accuracy. As time progressed the data sets improved in quality. As toexactly when the data sets improved in quality enough to justify accepting them withoutquestion is not known, and some may argue that the time has yet to come.Objective analysis schemes changed considerably over the period of the data sets.One of the first schemes used at the National Weather Analysis Center (the forerunner tothe NMC) employed a least squares fitting method to infer grid point values fromsurrounding observations. In practice, the method employed a minimum of tenobservations (Dey, 1989), which meant it performed poorly in data sparse areas. Thetechnique also failed to preserve temporal continuity in analyses, since each analysis wasconstructed without prior knowledge of preceding data. This often led to substantialerrors in the analysis in data sparse regions. Tn 1958, NMC changed their objectiveanalysis scheme to one developed by George Cressman (Cressman, 1959). It corrected fortemporal continuity by using NMC\u00E2\u0080\u0099s previous 12 hour forecast field, valid for the time ofthe analysis, as the first guess field for the analysis. Grid point values were determined byweighting observations around the grid point. Errors in observations could be readilydetected when they varied considerably from the first guess field. Data sparse areas couldbe analyzed much more accurately using objective rather than subjective means. TheCressman scheme lasted until 1974 when it was supplanted for a short period by the25Hough analysis. This new scheme incorporated vertical consistency by using spectralobjective techniques. Tn 1978 the analysis scheme was changed to a technique employingmultivariate statistical methods. This technique remained in place for the duration of theavailable data set, however, there were continual improvements to the basic scheme. Infact, Trenberth and Olson (1988) have documented over 40 changes to the NMCobjective analysis scheme from 1978 to 1987.The NMC gridded data fields are not without error, and the objective schemeshave changed over the period of record, but they remain one of the most error-freeanalyses available for use in climate and meteorological studies. The larger concern withthe NMC analyses is that the archive is incomplete. Of particular concern to this study is13.4% of the data fields are missing during the summers of 1982 and 1983 (Trenberthand Olson, 1988).NMC archives the analyses in global, hemispheric, and octagonal formats. Theoctagonal grid is a specific grid designed by NMC so that a 47 by 51 array of evenlyspaced grids can be overlaid on a Northern Hemisphere poiar stereographic secantprojection. The four corners of the array were cut off to reduce the size of the data set,and yet maintain as much Northern Hemisphere coverage as possible. Figure 2 shows theoctagonal grid overlaid on a polar stereographic secant projection. Each dot represents agrid point. It should be noted that the octagonal u and v wind components are not inrelation to compass directions, but to the x and y coordinate system of the polar7 stereographic projection. Figure 3 is an expanded view of North America. Note the lackof data in the southwest corner of the picture. Overlaid on Figure 3 are the locations ofthe upper air sites on which the grid data is based. The weather ships are no longer inoperation.t\u00E2\u0080\u0099)27..----rk.->... /... .\\,_/Figure 3 NMC octagonal grid for North American sectorwith radiosonde stations overplotted283.3 OUTGOING LONG- WAVE RADIATIONMeasurements of Outgoing Long-wave Radiation (OLR) from polar-orbitingNOAA satellites started in June 1974 (Gruber and Krueger, 1984). Each time a satellitefailed, another satellite was sent up to continue the data collection. In spite of thedifferent radiometer characteristics a twice-daily global OLR data set has been derived.The satellite infrared radiometer measures in a narrow-band window, typically from 10.5to 12.5 jim. This radiance is converted to an equivalent blackbody window temperature,Tb, by means of a regression equation that relates Tb to 99 different atmospheres withvarying vertical moisture and temperature profiles. Because several different satelliteshave collected the data in the past, another regression equation is used to convert all thedifferent satellite Tb data to the same flux temperature, Tf. Outgoing long-wave radiationcan then be estimated using the Tf data via aT4,where a is the Stefan-Boltzmannconstant. Given that the actual infrared window has changed over the years from 10.5 -12.5 jim, to 10.5 - 11.5 jim, and then to 11.5- 12.5 jim, it is somewhat surprising to findthe data are fairly consistent, with expected errors of only 11 W rn-2 (Gruber andKrueger, 1984). Another problem exists with the OLR data set. Each time a differentsatellite was launched, it had a slightly different Equator crossing time, which introducesa diurnal bias since the satellite was observing the same areas at different times of theday. The simplest way to reduce the error associated with different Equator crossingtimes is to average the twice-daily data. This reduces the error to at most 7 W rn-2(Gruber and Krueger, 1984). A correction has already been applied to the OLR data toaccount for the different equator cross times (Gruber and Krueger, 1984).29Outgoing Long-wave Radiation (OLR) flux data are available from the NationalOceanic and Atmospheric Administration (NOAA). It is a twice-daily global data set withresolution of 2.5\u00C2\u00B0 latitude by 2.5\u00C2\u00B0 longitude.30CHAPTER 4. PRECIPITATION4.1 PRECIPITATION REGIMESIn \u00E2\u0080\u009CThe Climates of the Continents\u00E2\u0080\u009D, Kendrew (1922) was one of the first todescribe the different precipitation regimes in North America (excluding Mexico).Examination of monthly and daily precipitation over the United States shows a peak inrainfall during summer over much of the country (Epstein and Barnston, 1988). Thissummer rainfall is extremely important for agricultural purposes (Fritsch et at., 1986). Instudying daily precipitation events in summer that were greater than 12.7 mm (0.5 inch)in depth, Heideman and Fritsch (1988) found that 80 % of all warm season precipitationis convective in nature, being indirectly or directly associated with thunderstorms. Thiscontrasts with winter precipitation, when maxima correspond to preferred storm tracks ofsynoptic cyclones. Topography can play a role in enhancing precipitation in both seasons.Kendrew (1922) categorized precipitation in the southwest United States as theArizona type; a winter maximum due to cyclonic activity, and a summer maximum due tolocal heating causing convective rainfall. Preceding the summer rains is an almostrainless June. The southeastern United States was labelled the Gulf type distinguished bya summer precipitation maximum later than the Arizona type, and overall greater rainfallthroughout the year. Markham (1970) combined both the Arizona and Florida areas, andcalled it the August tropical regime. Markham assumed it was the strength of the tradewinds, and height of the trade inversion, that was the main factor for the tropical rains.The higher the inversion, the greater the strength of the convection. Texas was notincluded in the tropical regime since its maximum in precipitation is delayed until31September. Markham (1970) reasoned that occasional torrential rainfall from hurricanestended to influence the Texas statistics.Harmonic analysis of rainfall (Horn et at., 1957 and Horn and Bryson, 1960) wasused to determine the annual march of precipitation in the United States. These resultsshow that the region of summer rainfall in Arizona and New Mexico also extends intosouthern Nevada and Utah. They also indicated another summer maximum over thepanhandle of Florida. A more elegant way of displaying the harmonics was used by Hsuand Wallace (1976) employing vectors. The length of the vector indicated the amplitudeof the harmonic, and the direction indicated the time of the maximum amplitude inprecipitation. Essentially, the same summer maxima as those of Horn and Bryson (1960)were found.Walsh et at. (1982) used a rotated factor analysis to determine areas of spatiallycoherent monthly precipitation. They found large areas of coherence to be related tocyclone tracks during winter, but during summer the spatial coherence broke down due tothe predominance of convective precipitation. Englehart and Douglas (1985) found thatprecipitation frequency has more spatial coherence than precipitation amount. Usingprecipitation frequency they found a spatially coherent precipitation regime duringsummer over Arizona and western New Mexico. Horn et al. (1957) used a simplerapproach of intermonthly precipitation changes to delineate different precipitationregimes. Positive and negative changes in precipitation amounts for successive monthswere analyzed. The southwest monsoon was easily identified by a sharp increase inrainfall from June to July, accompanied by drying in adjacent areas. Horn et at. (1957)hinted at the existence of some synoptic circulation patterns that affect all the areassimultaneously, but offered no suggestions.32Even the Baja Peninsula of Mexico exhibits unusual precipitation patterns.Hastings and Turner (1965) found that the precipitation patterns of the northern andsouthern ends of the Baja are completely different. Over the extreme northwest corner ofthe peninsula, close to San Diego, rainfall is a maximum during winter and rare in thesummer. On the other hand, the southern tip has at least twice the rainfall of thenorthwest area, but it comes during summer and fall. There is a secondary maximum inwinter, but it is considerably weaker, and there is no rain in the south during the spring.Research by Ives (1949) and Wallen (1955) led to a description of the rainfalldistribution in Mexico. In general annual rainfall is a maximum in southern Mexico, withthe northwest or Sonoran state being the driest year round. Another maximum existsalong the western slopes of the Sierra Madres. Portig (1965) described annual rainfall inCentral America ranging from 1 m at inland sheltered areas to 6.5 m in exposed coastalareas. All stations possess rainfall maxima in the summer season, but as in Mexico manystations exhibit a bimodal distribution, with rainfall peaks occurring in June andSeptember with relatively drier conditions prevailing in July and August. In Mexico thisphenomenon is referred to as the \u00E2\u0080\u0098la canicula\u00E2\u0080\u009D or August drought, while many in CentralAmerica refer to it as \u00E2\u0080\u0098el veranillo\u00E2\u0080\u009D. Portig (1965) contends that in Central America thisminimum in July and August may be related to the movement of the intertropicalconvergence zone, and in Mexico by changes in the intensity of the Bermuda anticyclone.He later acknowledges (Portig, 1976) that the strength of the anticyclone and theassociated increase in subsidence is not in phase with the precipitation, so the reasoning isnot completely satisfactory.33Mosino and Garcia (1974) suggest other mechanisms. Common hypotheses forthe dry spell invariably relate it to the variation of the Sun\u00E2\u0080\u0099s declination, and its effect oninsolation and heating. The maximum in June occurs in southern Mexico when the Sun\u00E2\u0080\u0099soverhead position moves northward to the Tropic of Cancer, and the second maximum inSeptember occurs as the Sun\u00E2\u0080\u0099s overhead position moves southward. However, this wouldnot account for the dry spell near the Tropic of Cancer over northeastern Mexico, sincethe Sun is directly overhead only once. More importantly, as Mosino and Garcia (1974)point out, the rainfall peaks in June and September do not occur on the same dates everyyear as would be the case if it was dependent on the Sun\u00E2\u0080\u0099s zenith angle. They propose thatthe dry spell is related to changes in the atmospheric circulation. In mid summer an upperair trough develops over the Atlantic and extends southwestward over Florida and intoCentral America. Thus trade wind disturbances, such as easterly waves and tropicalcyclones tend to get caught by the upper trough and recurve northeastwards along theeastern edge of the trough. This has the effect of cutting moisture off from the Gulf ofMexico, Mexico and Central America. However, this does not explain why the dry spellis restricted to the eastern side of Mexico, since easterly waves would cross Mexico to thePacific ocean. Cavazos and Hastenrath (1990) postulate that the rainfall maximum inSeptember in northeastern Mexico is due to tropical storm activity.The annual cycle of precipitation (mm) based on monthly data for various stationsover the southern United States, Mexico, and Cuba (for locations see Figure 4) is shownin Figures 5a to 51. Table 1 gives the mean monthly precipitation (mm) and the recordlength (years) for each station. Monthly data for Logan and Tempe were derived fromdaily precipitation data. A total of twelve sites were chosen. Site selection was based onthe availability of the data in the World Monthly Surface Station Climatology (WMSSC),34Figure 4 Location map35with the requirement of a minimum record length of 25 years (except for Caibarien whichonly possessed 10 years of data). The first six stations represent locations where theprecipitation increases from June to July. The remaining six stations have contrastingsummer precipitation regimes.Station Years Jan Feb Mar May Jun Jul p Oct Nov DecGrand Canyon 49 33 39 36 25 16 12 39 56 4.0 26 21 40Tempe 59 20 20 22 9 5 3 20 31 20 14 14 26Guaymas 29 18 6 5 1 2 1 46 71 28 17 8 18Albuquerque 57 10 10 12 13 16 17 32 38 24 22 11 12El Paso 109 11 10 8 6 8 17 42 39 33 21 11 14Chthuahua 63 7 7 5 5 14 37 88 87 74 23 10 9Winnemucca 104 24 21 21 20 22 19 6 7 10 17 20 25Logan 59 39 39 49 55 48 36 15 22 33 43 4.0 41Brownsville 37 35 37 14 38 66 70 42 68 141 ?TCaibarien 10 35 26 30 58 193 205 124 151 250 250 85 53Miami 37 52 56 60 81 164 225 154 180 210 166 74 49Mobile 124 123 128 156 121 109 139 180 165 128 82 100 129Table 1. Mean monthly precipitation (mm) and recordlength (years).Before discussing the precipitation climatologies in Figure 5 an index isintroduced to help evaluate the changes in rainfall between successive months. A simpleindex of the change in rainfall from June to July and from August to September,normalized by the two month total precipitation amount (xlOO) is calculated to show theprecipitation changes in adjacent months. The indices given in Table 2 are as follows:(JULY- JUNE)/(JTJLY + JUNE) x 100(SEPTEMBER - AUGUST)/(SEPTEMBER + AUGUST) x 10036The index ranges from -100 to + 100. If June has zero precipitation then the indexwill be + 100. If July is zero then the index will be -100. Therefore, the index representsthe relative change from month to month.Station Jul-Jun (index) Sep-Aug (index)Grand Canyon 53 -17Tempe 74 -22Guaymas 96 -43Albuquerque 31-23El Paso 42 -8Chihuahua 41 -8Winnemucca -52 18Logan-41 20Brownsville -25 35Caibarien-25 25Miami-19 8Mobile 13 -13Table 2. Precipitation index changesThe Grand Canyon, Tempe2and Guaymas (Figures 5a, 5b, and 5c respectively),all exhibit southwest monsoon signatures, characterized by fairly dry Junes followed byvery wet Julys. The July-June indices range from a low of 53 at the northern location to ahigh of 96 at the southern site of Guaymas. Note that the above three stations have amajor peak in precipitation during summer, and a secondary peak of precipitation duringwinter.Albuquerque, El Paso, and Chthuahua (Figures 5d, 5e, and 5f respectively) exhibitweaker southwest monsoon signatures, with indices ranging from 31 to 42. These stationsdo not exhibit a secondary winter precipitation peak.2Tempe, AZ is a suburb of Phoenix,AZ and was chosen instead of Phoenix because its precipitation recordremained unbroken by station relocations.37BISS60504030201004030-20-10-MONTHJan Feb Mar Apr May Jun JuL Aug Sep Oct Nov DecMONTHFigure 5 a) Monthly precipitation (mm) for Grand Canyon, Arizona (top),b) Monthly precipitation (mm) for Tempe, Arizona (bottom)38Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMONTH8060402004030-BB20-10-\u00E2\u0080\u0094 I I I I I I I I IJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMONTHFigure 5 c) Monthly precipitation (mm) for Guaymas, Mexico (top),d) Monthly precipitation (mm) for Albuquerque, New Mexico (bottom)3950-40302O,H, H, II I IJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMONTH100 -75-BI50-25-_____I II IIHJ IH IJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMONTHFigure 5 e) Monthly precipitation (mm) for El Paso, Texas (top),f) Monthly precipitation (mm) for Chihuahua, Mexico (bottom)SS30MONTH\u00E2\u0080\u0094- iSep Oct Nov DecFigure 5 g) Monthly precipitation (mm) for Winnemucca, Nevada (top),h) Monthly precipitation (mm) for Logan, Utah (bottom)40SS25-2015-10-5-nI I- - I F-- IJan Feb Mar Apr May Jun Jul Aug6050403020100Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMONTH41150 -____________________________j100-Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMONTH300-250 -200 -SS150-g 100-II I I I m I IJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMONTHFigure 5 i) Monthly precipitation (mm) for Brownsville, Texas (top),j) Monthly precipitation (mm) for Caibarien, Cuba (bottom)42250-200 -150-100-50-__U - I I I 1 1 1 T T T iJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMONTH200-150-ES100-50 -0- rn rn rn rn- rn rn i rn- rn rn rnJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMONTHFigure 5 k) Monthly precipitation (mm) for Miami, Florida (top),I) Monthly precipitation (mm) for Mobile, Alabama (bottom)43Winnemucca and Logan (Figures 5g and 5h, respectively) are examples of thedrying that takes place to the north of the southwest monsoon in July. Both sitesexperience a considerable drop in precipitation from June to July, while their counterpartsto the south simultaneously experience a large increase in precipitation.A different summer precipitation regime occurs around the Gulf of Mexico. Inparticular, Brownsville and Caibarien (Figures 5i and 5j, respectively) have two rainyseasons interrupted by a relatively drier mid-summer period. Caibarien shows this best,with rainfall maxima in May and June, as well as September and October, and a relativelydrier July and August. The drying in August at Brownsville is not as pronounced as inJuly so its pattern is not as clear as that at Caibarien.The southeastern United States exhibits a mixture of precipitation regimes. Miami(Figure 5k) resembles Caibarien with a drier mid-summer period, while just to thenorthwest, Mobile (Figure 51) experiences a single summer precipitation peak, that ismore reminiscent of a weak monsoon signature.Winnemucca and Logan (Figures 5g and 5h, respectively) begin to dry at thesame time as Tempe and Albuquerque (Figures 5b and 5d, respectively) begin theirmonsoon rains. In the next section this is investigated further using daily data.4.2 WET-DRY INDEXResearchers normally study precipitation as total amounts of precipitable waterwhether in daily, monthly, or yearly formats. The use of total amounts has two importantlimitations. One is that trace amounts of precipitation are neglected. This isinconsequential if one is interested solely in the total amount of precipitation. However, if44one is concerned with the percentage of days on which precipitation occurs, then traceamounts have to be accounted for, in some way. Another limitation is that totals are oftenbiased towards large events. For example, heavy convective rains that occur duringsummer severe weather have been known to produce total rainfall amounts in one daythat match or exceed the long-term monthly average for the location.To deal with the biases introduced by both large precipitation events and tracemeasurements a new method of studying daily precipitation is developed here. The firststep is to recode the total daily precipitation amounts into a more simplistic measure. Allprecipitation events, whether large or trace, are recoded as \u00E2\u0080\u009C1 \u00E2\u0080\u009Cs. Then all days withoutprecipitation are recoded as \u00E2\u0080\u009C-l\u00E2\u0080\u0099s. All missing data are recoded as zeros. Thus a runningsum of the recoded data, the wet-dry index, increases during wet periods and decreasesduring dry periods. Extended periods of missing data are easily visible graphically sincethe \u00E2\u0080\u009Cwet-dry index\u00E2\u0080\u009D remains constant.Figure 6 shows an example of two such running sums for 1971. Figure 6a shows arelatively wet period during winter and spring in Logan, and a relatively dry periodduring summer and fall. Tempe on the other hand (Figure 6b) is generally dry throughoutthis year except during the summer when some indication of the monsoon appears. Notethat the rains did not fall every day during the summer of 1971 but were interspersed withdry days.The final index number at the end of the year is a measure of the year\u00E2\u0080\u0099s dryness orwetness. For instance, an index of -365 would be a year of 365 continuous dry days.Conversely, +365 would indicate a year of 365 continuous wet days. The index can alsobe directly converted into the number of dry or wet days in the year as follows:4520-150-200-250-300YEAR-DAYFigure 6 a) Wet-Dry index example for Logan, 1971 (top),b) Wet-Dry index example for Tempe, 1971 (bottom)0-20 --40 --60- I\u00E2\u0080\u00A2 I I0 50 100 150 200 250 300 350-50YEAR-DAY46WET = (365 + INDEX)/2DRY = (365 - INDEX)12Of course, the total number of days (365 in the above example) must be adjustedfor Leap Years as well as missing data.Figure 7a shows the time series of the annual wet-dry index for Logan from 1928to 1986 and Figure 7b is a time series of annual total precipitation for the same timeperiod. Note that even though the period beyond the mid 1960s shows a relatively wetterperiod than earlier, this is only reflected in the total annual precipitation in the 1980s.Hence the number of wet days increased in the latter part of the record but this was notnecessarily accompanied by increases in rainfall amounts. Figure 8a is the time series ofthe annual wet-dry index for Tempe from 1926 to 1984, and Figure 8b is a time series ofannual total precipitation for the same time period. Tempe does not show the discrepancyin the latter part of the record shown for Logan. Figure 9 gives an idea of how well thewet-dry index is correlated with precipitation. A correlation coefficient of 0.74 (95%confidence limit from 0.59 to 0.84) for Logan means that approximately 54 % of thevariance is explained. Tempe has a slightly larger correlation of 0.79 (95% confidencelimit from 0.67 to 0.87), and 62% of the variance is explained.A 59-year average of the wet-dry index for each day of the year for Logan isshown in Figure lOa. (Note that almost all days have negative wet-dry index values. Thisindicates that on average there are more dry than wet records for any given calendar day).What is notable is a break in the wet-dry index as one goes from a very wet to a very dryperiod in the year, in a relatively short period of time. Thus the average break occurssometime in the middle of June (year day, YD, 160-170). A gradual return to wetterconditions occurs in the later part of the year. It is interesting to note that the wet-dry47Inn/1J\A/1\\1/\th1//ut\1930 1940 1950 1960 1970 1980YEAR1000-800-600-c)400-I I I I I1930 1940 1950 1960 1970 1980YEARFigure 7 a) Annual wet-dry index for Logan, 1928-1986 (top),b) Annual precipitation (mm) for Logan, 1928-1986 (bottom)48-225 --250 --275 -350-300--325 -1930 1940 1950 1960 1970 1980YEAR500-400-300-200 -100 -0- I I1930 1940 1950 1960 1970 1980YEARFigure 8 a) Annual wet-dry index for Tempe, 1926-1984 (top),b) Annual precipitation (mm) for Tempe, 1926-1984 (bottom)49-300200 400 600 800PRECIP (mm)1000-200-225-250-275-300-325-350Figure 9 a) Scatter plot of wet-dry index versus precipitation for Logan, 1928-1986 (top),b) Scatter plot of wet-dry index versus precipitation for Tempe, 1926-1984 (bottom)PRECIP (mm)50050index does not reach the same level in the last two months of the year as in the first twomonths of the year, even though the monthly precipitation chart (Figure 5h) shows thatthe precipitation amounts to be almost equal. This indicates that there are fewer days inNovember and December which have precipitation, but they must have on average ahigher daily precipitation total to produce monthly totals on par with January andFebruary. Thus frequency of precipitation could be used as another technique todistinguish between precipitation regimes. Figure lOb is the 59-year average of the wet-dry index for each day for Tempe. Here one goes from a very dry period of the yearwhich lasts to around YD 175 to a very wet period by YD 195.In order to remove some of the noise in the signal a running sum was applied tothe wet-dry index. Figure 1 la is the running sum of the average daily wet-dry index forLogan, and Figure 1 lb for Tempe. This display makes it easier to find the break in slopefor Logan, but the change at Tempe is now harder to see. The robustness of the break ofslope for both Logan and Tempe were tested using two standard techniques. The firstcompares the odd years to the even years, and the second compares the first half of therecord with the latter half. Since the breaks persist (not shown), it seems that both Loganand Tempe exhibit different temporal summer precipitation regimes. Logan goes fromwet to dry in the summer and Tempe goes from dry to wet.To better emphasize the break point the time series of the indices were adjusted soas to have a mean of zero. To accomplish this the final running sum index value at theend of the year was divided by 365. This average value is then subtracted from the indexgiven for each day (i.e., before a running sum is done). Now a running sum of thisadjusted data should force the index to end at zero. Figure 12a shows the transformedindex for Logan with a more pronounced break point near YD 165.The transformed index51-0.25000Figure 10 a) Average daily wet-dry index for Logan, 1928-1986 (top),b) Average daily wet-dry index for Tempe, 1926-1984 (bottom)0 50 100 150 200 250 300 350YEAR-DAYYEAR-DAY520-500-100-150IFigure 11 a) Running sum of average daily wet-dry index for Logan, 1928-1986 (top),b) Running sum of average daily wet-dry index for Tempe, 1926-1984 (bottom)0 50 100 150 200 250 300 350YEAR-DAYYEAR-DAY5330-20-1:\u00E2\u0080\u009410\u00E2\u0080\u0094 I I I I I0 50 100 150 200 250 300 350YEAR-DAYI10 50 100 150 200 250 300 350YEAR-DAYFigure 12 a) Transformed wet-dry index for Logan, 1928-1986 (top),b) Transformed wet-dry index for Tempe, 1926-1984 (bottom)54also identifies the more gradual return to a wetter climate that begins after YD 300.Tempe (Figure 12b) shows the onset of the monsoon beginning around YD 190. Thishappens to correspond to the first week of July, which is why the monsoon signal showsup so strongly in the monthly data set. The change in Logan occurs in the middle of June,which gives the appearance in the monthly data that the drying occurs more gradually.A similar procedure is undertaken for precipitation totals, and the finaltransformed precipitation is shown in Figure 13. Both transformed precipitation curvesshow the same breaks as the transformed wet-dry index curves. This indicates that neitherLogan or Tempe was unduly biased by trace precipitation records or large precipitationevents.Two more stations are examined to show that Logan and Tempe are not unique tothe region. The transformed wet-dry index curves are shown for Winnemucca andAlbuquerque (Figures 14a and 14b, respectively). Winnemucca is almost identical toLogan with the break occurring near YD 165. Albuquerque has a much earlier break thanTempe and occurs near YD 170.In the next Chapter the changes in atmospheric circulation that accompany therapid onset of the monsoon rains are examined.552.5-2-c)1-C0.5-0--0.5-- I0 50 100 150 200 250 300 350YEAR-DAY0 50 100 150 200 250 300 350YEAR-DAYFigure 13 a) Transformed precipitation for Logan, 1928-1986 (top),b) Transformed precipitation for Tempe, 1926-1984 (bottom)563020\u00E2\u0080\u00A2rn._I_,\u00E2\u0080\u0094 I I I I I0 50 100 150 200 250 300 350YEAR-DAY20 -10--10-\u00E2\u0080\u0094 I I I I I0 50 100 150 200 250 300 350YEAR-DAYFigure 14 a) Transfonned wet-dry index for Winnemucca, 1928-1986 (top),b) Transformed wet-dry index for Albuquerque, 1931-1986 (bottom)57CHAPTER 5. NORTH AMERICAN ATMOSPHERIC CIRCULATION5.1 MONTHLY COMPOSITESGeneral circulation studies of the Northern Hemisphere summer have beenrelatively rare. Almost all of the work has concentrated on winter circulation patternswhen gradients are at their strongest and more exciting changes in weather and climateoccur. Summer circulation patterns are fairly flat with little gradient in pressure, height,or temperature, hence the lack of research interest in the season.An exception is a paper by White (1982), who produced an observational study ofNorthern Hemisphere summers (June to September inclusive composited from 1966 to1977). Composited 200 mb height fields showed a zonal flow pattern over North Americaextending from the arctic to 30N latitude. South of 30N the gradient was weak and theonly feature was an upper-level high centred just south of the Baja peninsula. Such upper-level highs are characteristic of all monsoon circulations around the globe.Because of the lack of summertime observational studies the first step in thepresent study was to examine the standard monthly meteorological fields. Monthly fieldswere readily available for Mean Sea Level Pressure (MSLP), and 500 millibar height(500 mb) data from January 1946 to June 1989. Monthly composites for both werecalculated from 1946 to 1988, inclusive.Since pressure and height gradients are relatively weak during the summer overNorth America, non-standard contour intervals were used. Figure 15 shows the MSLPcomposites for June, July, August, and September (hereafter referred to as JJAS) using a1 mb contour interval. The thermal trough line is shown in bold. The Pacific and Atlanticsurface highs show approximately the same position and strength throughout the summer58Figure 15a June composite of mean sea level pressure (mb) with a 1 mbcontour interval (bold line indicates the thermal trough axis)59Figure 15b July composite of mean sea level pressure (mb) with a 1 mbcontour interval (bold line indicates the thermal trough axis)60\ \-A\u00E2\u0080\u0099 -prjj: L/ 1012312-/ 10121-Figure 15c August composite of mean sea level pressure (mb) with a 1 mbContour interval (bold line indicates the thermal trough axis)61Figure 15d September composite of mean sea level pressure (mb) with a 1 mbcontour interval (bold line indicates the thermal trough axis)62months. In between the highs, a thermal trough is evident in Mexico and thesouthwestern United States. The thermal trough weakens by 1 mb from June to July andmoves slightly westward through the summer. Pressures increase over eastern Mexico,New Mexico, and western Texas by over 2 mb in July. An increase in pressure isopposite to what one would expect with the arrival of the summer rains in July. Thepreferred location of the thermal trough and change in strength is corroborated byRowson and Colucci (1992).Figure 16 shows the 500 mb height composites for JJAS using a 30 metre interval.The 500 mb ridge line is shown in bold. By focussing on the ridge line that lies parallel tothe latitude circle, we can see that the ridge moves northward over North America ratherquickly, moving from 25\u00C2\u00B0N latitude to 35N from June to July. The position remainsrelatively stable from July to August, but retreats to 25\u00C2\u00B0N in September.Evidence of this upper-level anticyclone dates as far back as 1921 when balloonlaunches from San Francisco revealed a southerly wind aloft and not the easterlies thatwould be expected in a land-sea breeze circulation (Reed, 1933). Even then forecastersbelieved that the tropical moisture affecting the southwest was being advectednorthwestwards from the Gulf of Mexico around the anticyclone that they discovered at4000 m (Reed, 1937). Bryson and Lowry (1955a, 1955b) came to the same conclusionsome time later.Since the JJAS sequence of 500 mb height composites changed much more thanMSLP did, the next step was to investigate higher levels in the atmosphere. Monthlyfields of 200 mb heights were not readily available, but were constructed from twice daily200 mb height data available from 1962 to 1989. All available data were averaged toproduce monthly composites for JJAS. Figure 17 shows the 200 mb height JJAS63/ \\u00E2\u0080\u00A2,. )jcj ee Th \/ i\u00E2\u0080\u0099% I /1_______Figure 16a June composite of 500 mb height (m) with a 30 mcontour interval (bold line indicates the 500 mb ridge axis)64- \u00E2\u0080\u00A2\ \ \\\\ kx\-\-k\\u00E2\u0080\u0098\\ \ I \f-S \ \ )\u00C3\u00A7 \ /J f% \ \ fr\u00E2\u0080\u00A2..1--kii \ \ )c\u00E2\u0080\u0099 \>\u00E2\u0080\u0098-Z I ZI . 1 1 / \u00E2\u0080\u00A2L.- \ / ,% \ \\ 4) -\u00E2\u0080\u009817ZI--. / -_____\)_,__Figure 16b July composite of 500 mb height (m) with a 30 mcontour interval (bold line indicates the 500 mb ridge axis)651 I. \ \u00E2\u0080\u0098\ % \u00E2\u0080\u0098, \u00E2\u0080\u0098 j L_ : N. N. N. \S \:\u00E2\u0080\u0098\ \ \\u00E2\u0080\u00981 \ \ \ \ \/\u00C3\u00A7_\__ 5412 1 r--i-\u00E2\u0080\u0094-L\ \\u00E2\u0080\u009C*-\u00E2\u0080\u0098\ \ \ % \ \ \ -cS \ \-\ / 1 \ v-A-/ ) \u00E2\u0080\u0098N 4\ \ \ \ \\\\\ \u00C3\u00A7<\\u00C3\u00A7Q rAJN \\f ! V) j\ \u00E2\u0080\u0098 \u00E2\u0080\u0098 \ \ \ \2% \ / \u00E2\u0080\u0098O-(- A I \u00E2\u0080\u009C. I \u00E2\u0080\u009C I t-1.\\u00E2\u0080\u0094 \u00E2\u0080\u00A2:/ S61.\--/ 5880Figure 16c August composite of 500 mb height (m) with a 30 mcontour interval (bold line indicates the 500 mb ridge axis)66\u00E2\u0080\u0098C(5\u00E2\u0080\u0094c-i 5297i7_\\ \44)\u00E2\u0080\u0098\u00E2\u0080\u0098- 1\u00E2\u0080\u0099IIII\//I1_______Figure 16d September composite of 500 mb height (m) with a 30 mcontour interval (bold line indicates the 500 mb ridge axis)67sequence using a 40 metre contour interval. In June there is an indication of an upperridge building over North America. By July a closed upper high of considerable strengthforms over northwestern Mexico. The high remains quasi-stationary during August, andby September it has weakened and retreated southward. Thus an upper-level anticyclonesimilar to that found in the Asian, Australian, and African monsoons is found over NorthAmerica as well.5.2 UPPER WIND FIELDThe height field gradient revealed by the summer composites is quite weak overthe southern portions of North America. Therefore, to better emphasize the upper-levelcirculation, the 250 mb wind fields were examined. Twice-daily 250 mb level u and vwind component data were available from May 1965 to June 1989. The JJAS sequence ofmonthly composites based on daily data is shown in Figure 18. A 25.7 m s (50 knot)wind is equivalent to a wind vector length of 22 mm.In June anticyclonic flow is centred near 2ON over western Mexico. Magana andYanai (1991) refer to the upper-level anticyclone over Central America as the Mexicananticyclone. They believe it is \u00E2\u0080\u009Cmaintained by divergence associated with convectionover Central America\u00E2\u0080\u009D. The mean position of the polar jetstream cuts across thecontinent near 50N. A weaker subtropical jetstream tracks over the northern sections ofthe Baja peninsula, and gradually merges with the polar jetstream over the continent. ByJuly the upper-level anticyclone is much farther north, centred around the MexicoArizona-New Mexico border. There is also clear evidence of an eastward extension of theanticyclonic circulation toward northern Florida. The polar jetstream is still in the samerelative position as in June, but now the subtropical jetstream has moved fartherI Co Co tJC\u00E2\u0080\u009D0 C C.,o\u00E2\u0080\u0094.C z70/ 1 I I I 1\u00E2\u0080\u0099 ). I I I --\u00E2\u0080\u0094.t\u00E2\u0080\u0099 \ \ \ \ \\i/I kc1\ LY I \u00E2\u0080\u0098 I L\_-kf\ \ \ \ \_-\u00E2\u0080\u009Ci i 60s_Jl \ \u00E2\u0080\u0098\ \u00E2\u0080\u0098\ Vt---Jk-ii1\ \\%\ } !J/ \ \\\\\\N ,$)Q/,N\Zc2/ /\_1\u00E2\u0080\u0099/_______Figure 17c August composite of 200 mb height (m)with a 40 m contour interval71c2/___ii___Figure 17d September composite of 200 mb height (m)with a 40 m contour intervaL72\u00E2\u0080\u009C \u00E2\u0080\u0098i.- \u00E2\u0080\u0098\ \ \-- \ ,X\u00E2\u0080\u0099//>( \ \ \ \\\u00E2\u0080\u00984 / NN\ \ \ \ si\u00E2\u0080\u009DN\\ \ \ S___-. -. \u00E2\u0080\u0094s-------------__\u00E2\u0080\u0098\_ \u00E2\u0080\u0094 \u00E2\u0080\u0094..-\u00E2\u0080\u0098 \u00E2\u0080\u00A2\u00E2\u0080\u00A2/----N\u00E2\u0080\u0094b \u00E2\u0080\u0094 - \u00E2\u0080\u0094.\u00E2\u0080\u00A2 c/ - \u00E2\u0080\u0094 -\u00E2\u0080\u0094.:*-,.\u00E2\u0080\u0098 / \ N, \u00E2\u0080\u00A2\u00E2\u0080\u00A2*. \u00E2\u0080\u00A2\u00E2\u0080\u00A2_ \u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2% \u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2 ,\u00E2\u0080\u00A2_\u00E2\u0080\u00A2\u00E2\u0080\u00A2 \u00E2\u0080\u00A2\u00E2\u0080\u00A2_\u00E2\u0080\u00A2 \u00E2\u0080\u00A2,j\u00E2\u0080\u00A2- _\u00E2\u0080\u00A2_\u00E2\u0080\u00A2\u00E2\u0080\u0094\u00E2\u0080\u0098*.\u00E2\u0080\u0098\u00E2\u0080\u0094-,. \u00E2\u0080\u0094 \u00E2\u0080\u0094,_\u00E2\u0080\u0094.$\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u0094 _____v7 -.\u00E2\u0080\u0098 \u00E2\u0080\u009C-\u00E2\u0080\u0098\u00C3\u00A7 \u00E2\u0080\u0094 \u00E2\u0080\u0094\u00E2\u0080\u009C (,S.\u00E2\u0080\u0094 \u00E2\u0080\u00A2\u00E2\u0080\u0094 \u00E2\u0080\u00A2- \u00E2\u0080\u00A2 \u00E2\u0080\u009C -..\u00E2\u0080\u0094 N. -... \u00E2\u0080\u0094. \u00E2\u0080\u0094 \u00E2\u0080\u0094. \u00E2\u0080\u0094.Figure 1 8a June composite of 250 mb wind(26 m wind vector = 22 mm in length)t\u00E2\u0080\u0099JcrS. o II t\u00E2\u0080\u0099J -t\u00E2\u0080\u0099JT1000 I-I II0t\u00E2\u0080\u0099.)<\u00E2\u0080\u0094.UiCcrt\u00E2\u0080\u0099)CCI-tII C C C-76north and merges with the polar jetstream over the northwestern United States. The maincirculation cell drifts slightly southward in August, and by September it has retreatedsouthward to 25N along western Mexico. The extension towards the east is now harderto discern. The subtropical jetstream has returned to its mean position over the northernBaja peninsula.South of the upper-level ridge there is often an area of cyclonic shear, referred toas the Tropical Upper Tropospheric Trough (TUTT) (Whitfield and Lyons, 1992). ATUTT is a relatively narrow cyclonic shear zone, elongated southwest to northeast, thatappears in the climatological mean summer circulation pattern of the upper troposphere.An example is given for the mean 250 mb-level wind field for July (Figure 19). Vorticitymaxima at the 200 mb-level propagate along the TUTT axis. The TUTT or vorticitymaximum is not usually reflected in any feature at the surface. The TUTf ischaracterized by anomalous sinking motion near the centre due to upper-levelconvergence, and weak rising motion along the eastern flank which can produce somerainfall (Whitfield and Lyons, 1992). The rainfall associated with TUTTs isovershadowed by the monsoon rains, but TUrfs can be very important in day-to-dayconvective events.To better examine the motion of the upper-level anticyclone, weekly 250 mb-levelwinds were composited out of twice daily data. The individual sequence of maps are notshown but the seasonal evolution of the estimated central position of the main Mexicanhigh is shown in Figure 20. The number beside each point is the week that wascomposited. The northward motion of the upper-level high is extremely fast during thefirst 7 weeks. The high then tends to meander during the height of summer (weeks 7-10).When the high begins to retreat it appears to stall (weeks 12-14) for some time before77heading farther south at a somewhat slower pace than it had originally advanced. Sincethe northward advance is more rapid than its retreat, the motion is asymmetric.T1 C I0079/ / IN / / / WEEK DATEN / / 1 May 28 - Jun 03/ / 2 Jun 04 \u00E2\u0080\u0094 Jun 10N / / -.. 3 Jun 11 - Jun 17N / / 4 Jun 18 \u00E2\u0080\u0094 Jun 24/ 10/ / 5 Jun 25 \u00E2\u0080\u0094 Jul 019 / j 6 Jul 02 \u00E2\u0080\u0094 Jul 0811\u00E2\u0080\u0099 7 8 / j 7 Jul 09 \u00E2\u0080\u0094 Jul 15/ / 8 Jul 16 \u00E2\u0080\u0094 Jul 22/ j 9 Jul 23 - Jul 296/ / j 10 Ju130\u00E2\u0080\u0094AugOSj 11 Aug 06 - Aug 12/ 2 14 / I 12 Aug 13 \u00E2\u0080\u0094 Aug 1913 Aug 20 - Aug 26/ 14 Aug 27 Sep 02/ / 15 Sep 03 \u00E2\u0080\u0094 Sep 09N / 16 / 1 16 Sep 10 - Sep 164 17 / I 17 Sep 17 \u00E2\u0080\u0094 Sep 23/ / 18 Sep 24 - Sep 30/ 18 / 19 Oct 01 \u00E2\u0080\u0094 Oct 07/ / / 20 Oct 08 \u00E2\u0080\u0094 Oct 14/--.- / I 21 Oct 15 - Oct 21/ -. I 22 Oct 22 \u00E2\u0080\u0094 Oct 281 2/ 0/ / 21---111 22Figure 20 Weekly averaged positions of the250 mb upper-level anticyclone80CHAPTER 6. NORTH AMERICAN MONSOON6.1 UPPER-LEVEL DiVERGENCEGiven that a thermal surface trough occurs throughout the summer months overMexico and the southwestern United States, there should also be low-level convergenceand upward vertical motion. If enough moisture is available, the ascending motion wouldlikely lead to the development of convective clouds. In addition, upper-level divergenceis associated with mid-tropospheric ascent.Krishnamurti (1971) found east-west upper-level circulation patterns across thetropics while examining upper-level data for the summer of 1967. The strongest upper-level divergence centres are located over western Mexico and southeast Asia. Thecorresponding areas of upper-level convergence of the thermally-direct circulation areover the Atlantic and eastern Pacific Oceans around 30N. He discovered that theintensity of the east-west upper-level circulation is comparable to the Hadley-typecirculation, and that it is distinct from the Walker circulation which is known to be farthersouth along the Equator.The relationship between upper-level anticyclones and convective precipitationhas been noted in a number of studies. Leary (1979) and Leary and Houze (1979) showedhow the development of a single tropical cloud cluster (100 km diameter) was related toan upper-level (200 mb) anticyclone in the large-scale flow. A cloud cluster formed in anarea of weak upper-level divergence. Low-level convergence in the cluster eventuallyincreased until a closed surface cyclonic circulation formed. Finally, upper-leveldivergence increased as the cloud cluster reached maturity. McBride and Gray (1980)81found that upper-level divergence was evident in 87 cloud clusters they examined in boththe Pacific and Atlantic Oceans.In the case study of a severe thunderstorm by Hales (1975) there was a definite300 mb anticyclone located directly over the southwestern United States, but he onlyexamined the low-level convergence zones. Two squall lines in a particular Arizonamonsoon season were examined by Smith and Gall (1989) and upper-level anticyclonicflow was clearly found in both cases. Unfortunately, upper-level divergence calculationswere only available for one case, thus it was only suggested that upper-level divergencecould have played a role in the development of the squall line.While studying mesoscale convective complexes (MCC) Fritsch and Maddox(1981 a, 1981 b) found that all ten cases exhibited a strong anticyclonic flow perturbationat 200 mb. Unsure as to whether the anticyclonic flow was there prior to the developmentof the MCC or whether the MCC developed its own upper-level anticyclone, they ran anumerical experiment which showed that an upper-level anticyclone developed after thecomplex formed. However, they still regard it as a \u00E2\u0080\u0098chicken and egg\u00E2\u0080\u009D question.6.2 DIVERGENCE AND PRECIPITA HONIn May 1986 NMC changed the way they conducted their analyses, resulting inimproved tropical upper-troposphere divergence fields (Mo and Rasmusson, 1993; Dey,1989). Hence pre-1986 upper-level divergence calculations are not as good as those inlater years. However, the interest here is in the changes that take place in the uppercirculation, and this should be less sensitive to alterations in the analysis scheme. Thechanges are most abrupt between June and July, and less so between August and82September. Hence, difference fields between adjacent summer months were calculatedfor divergence fields. Hopefully, this differencing will reduce the biases introduced by thenew analysis scheme. Figure 21 shows the smoothed July minus June upper-leveldivergence field. The upper-level flow becomes increasingly divergent aloft overnorthwestern Mexico, while smaller increases occur just to the west of Florida. There isalso an increasingly divergent flow aloft over Lake Superior, which may be anomaloussince it is based on a single grid point. However, no glaring enor is evident in the dataset. The flow becomes increasingly convergent aloft over the northwestern United States,and from the Great Plains States to the Gulf of Mexico, as well as over Cuba andJamaica.Similar difference calculations were done for adjacent monthly precipitation datafrom the WMSSC data set. Figure 22 shows whether a particular station increases inprecipitation from June to July (indicated by a \u00E2\u0080\u009C+\u00E2\u0080\u009C symbol), or decreases in precipitationfrom June to July (indicated by a \u00E2\u0080\u009C-\u00E2\u0080\u009C symbol). Generally, there is an increase inprecipitation over western Mexico, and northward into Arizona, New Mexico, Colorado,and southern Utah and Nevada. The eastern United States also experiences an increase inprecipitation from June to July, while precipitation decreases in the rest of the UnitedStates, eastern Mexico, Cuba, and Jamaica.The relationship between changes in upper-level divergence and changes insurface precipitation becomes clearer when the two fields are superimposed (Figure 23).Over the southern United States and Mexico, wherever there is an increasingly divergentflow aloft from June to July, we observe a corresponding increase in precipitation.Conversely, wherever there is an increasingly convergent flow aloft from June to July, weobserve a corresponding decrease in precipitation. For example, there is an increase in83Figure 21 July minus June divergence (10-8 s\u00E2\u0080\u0099) difference fieldwith a contour interval of 2.0 x i0 s84Figure 22 July minus June composite monthly precipitation (locations of stationswith monthly precipitation available for at least 10 years from the WMSSC data set.A \u00E2\u0080\u009C+\u00E2\u0080\u009C denotes an increase in precipitation from June to July and locations witha \u00E2\u0080\u009C-\u00E2\u0080\u009C denotes a decrease in precipitation from June to July)85Figure 23 July minus June composite monthly precipitation overlaid onJuly minus June divergence (1o-8 -4) difference fieldwith a contour inten\u00E2\u0080\u0099al of 2.0 x i086precipitation through the southeastern United States down into Florida, until we reachthe southern tip, where precipitation actually decreases in July. This decrease extendssoutheastward into Cuba and Jamaica and corresponds well with an increasinglyconvergent flow aloft. A decrease in July precipitation is evident throughout the GreatPlains which corresponds to an area of increasingly convergent flow aloft. Through muchof western Mexico and northward into Arizona and New Mexico, there is an increase inprecipitation, which corresponds to an area of increasingly divergent flow aloft. Therelationship does not hold as well through Texas and eastern Mexico where the area ofincreasingly convergent flow aloft is to the east over the Gulf of Mexico. An increase inprecipitation over the northeastern United States does not correspond to the increasinglyconvergent flow aloft, which may be related to precipitation mechanisms other thanconvection.Figure 24 shows September minus August divergence superimposed onSeptember minus August precipitation changes. The dominant features in this pattern aresimilar to those in the July minus June pattern, but of reversed polarity. The upper-levelflow becomes increasingly divergent over the extreme northwestern United States andfrom the Great Plains States to the Gulf of Mexico, as well as Cuba and Jamaica.Conversely, the flow becomes increasingly convergent aloft over Arizona, New Mexico,western Mexico, and the eastern United States. Northern California does showincreasingly convergent flow aloft, which is inconsistent with the increase in precipitationover that region. However, it should be noted that by September convective precipitationis no longer the dominant type of precipitation at the latitude of northern California.Synoptic weather systems coming from the Pacific become more dominant. Increases inprecipitation are apparent in the northwestern United States, Texas and eastern Mexico,87Figure 24 September minus August composite monthly precipitation overlaidon September minus August divergence (10-8 s1) difference fieldwith a contour interval of 2.0 x l0 s88as well as Cuba and Jamaica. Decreases in precipitation are found in northwesternMexico, Arizona, New Mexico, sections of the Great Plains, and the eastern UnitedStates.Once again, the relationship between changes in upper-level divergence andchanges in the surface precipitation is again clear (Figure 24). Over the southern UnitedStates and Mexico wherever the upper-level flow becomes increasingly divergent(convergent) aloft from August to September we observe a corresponding increase(decrease) in September precipitation. For example, there is a decrease in precipitationthrough the southeastern United States down into Florida, until we reach the southern tip,where precipitation increases in September. The increase extends southeastward intoCuba and Jamaica and corresponds well with the upper atmosphere becomingincreasingly divergent aloft. An increase in September precipitation is evident in Texas,and eastern Mexico, which corresponds to an area of increasingly divergent flow aloft.Through much of western Mexico and northward into Arizona and New Mexico, there isa decrease in precipitation, which corresponds to an area of increasingly convergent flowaloft.The relationship is no longer valid in some sections of the northern United States.The weakness of the relationship can probably be atthbuted to the effect of synoptic scaleweather systems that begin to affect the northern United States in September.A simple linear regression analysis is employed to show the scatter of July minusJune divergence, to July minus June precipitation. Upper-level divergence values areinterpolated from the raw grid point data at every precipitation site in Jamaica, Cuba,Mexico, and the United States up to a latitude of 42\u00C2\u00B0N, which coincides with the borderof northern California, Nevada, and Utah. Figure 25a shows the July-June scatterplot of89153 data points with the best fit line as shown. The correlation coefficient (r) is 0.57(95% confidence limits from 0.45 to 0.67). The coefficient of determination (r2) is 0.33,indicating that 33% of the variance can be explained, which shows that there is a fairlystrong relationship between changes in divergence aloft and changes in surfaceprecipitation. Similarly, September minus August data (Figure 25b) gives a correlationcoefficient of 0.45 (95% confidence limits from 0.32 to 0.57) and a coefficient ofdetermination of 0.20. The lower value was expected because the retreat of the monsoondoes not regularly occur at the end of August but sometimes occurs as late as the end ofSeptember.Horel et al. (1989) used Outgoing Long-wave Radiation (OLR) to study theannual cycle of convection during 1980-87 over the Amazon basin, and found that theposition of the Bolivian high at 200 mb correlated well with inferred convection duringthe rainy season. Their results gives credence to similar correlations found here overMexico, the southern United States, and the northern Caribbean.6.3 PRECIPITATION REGIMES REVISITEDBuilding on this notion of the relationship between changes of surfaceprecipitation and those in upper-level circulation, it is appropriate to re-examine thesummer precipitation regimes given in Section 4.1.Figure 26 shows the onset index (from Table 2) overlaid on the July-June changein upper-level divergence. Figure 27 shows the demise index (from Table 2) overlaid onthe September-August change in upper-level divergence.The monsoon is surprisingly apparent in the southeastern United States, althoughHsu and Wallace (1976) did suggest that \u00E2\u0080\u009CSmall areas of late summer maximum90150100500-50\u00E2\u0080\u00A2100-150JULY-JUNE DIV (E-08) (uS)200SS250200S! 15010050Cl)0-50-150SEPT-AUG DIV (E-08) (uS)Figure 25 a) Scatter plot of July minus June monthly precipitation (mm)versus July minus June divergence (10-8 s1) for stations in Mexico, Cuba,Jamaica, and the United States up to 42\u00C2\u00B0N latitude (top)b) Scatter plot of September minus August monthly precipitation (mm)versus September minus August divergence (10-8 s) for stations in Mexico,Cuba, Jamaica, and the United States up to 42\u00C2\u00B0N latitude (bottom)-100 -50 0 50 10091Figure 26 July-June precipitation index for stations in Table 2overlaid on July minus June divergence (10-8 s1-) differencefield with a contour interval of 2.0 x i07 a-192Figure 27 September-August precipitation index for stations in Table 2overlaid on September minus August divergence (10-8 s4) differencefield with a contour interval of 2.0 x l0 s193[precipitation] over Florida are probably monsoonal in character\u00E2\u0080\u009D. Only small changesin precipitation are evident, but they are consistent with the changes in divergence aloft.Most researchers agree that the convective rainfall in Florida is associated with the low-level convergence zones associated with sea breezes. However, Burpee (1979) shows thatthe strength of convergence of the sea breezes has surprisingly little correlation with theamount of rainfall. Mobile has one of the lowest July-June precipitation indices (seeTable 2) because the upper-level ridging to the east is considerably weaker over the Gulfcoast states.Much of the U.S. Midwest shows precipitation peaks in June and September, withslightly lower precipitation amounts in July and August. Keables (1989) concluded thatthe month of June is wet in the Midwest because the 700 mb flow was more oftensouthwesterly, providing moisture advection from the Gulf of Mexico. As summerprogresses a 700 mb ridge builds over British Columbia turning the flow morenorthwesterly, hence cutting off the supply of moisture to the region. Drier Julys andAugusts result. As the end of summer approaches, the 700 mb ridge to the westdiminishes somewhat allowing a more southwesterly flow pattern again, which leads towetter Septembers. Our analysis of the 700 mb flow patterns (not shown) indicate a directwesterly flow over the US Midwest in June and September, while the flow turns slightlynorthwestwards (approximately 300\u00C2\u00B0) during July and August. The mean 700 mb flow isnever southwesterly. During the months of July and August the upper-level flow becomesmore convergent aloft than in June and September, which coincides nicely with thebimodal precipitation pattern Keables (1989) was examining. Similarly, the relative \u00E2\u0080\u009Cdry\u00E2\u0080\u009Dspell in July and August through Texas and eastern Mexico, is consistent with the upper-94level flow becoming increasingly convergent aloft. In southern Mexico, the relationshipis not as clear.To the north of the monsoon, considerable drying takes place as the monsoonrains begin, in early July. The effect is strongest in northern Nevada and Utah butstretches northward to the Pacific northwest. Lowry (1956) first suggested that theatmospheric circulation links Oregon\u00E2\u0080\u0099s drying trend in early July to the monsoon to thesouth.6.4 OUTGOING LONG-WAVE RADIATION (OLR)Outgoing long-wave radiation has been used as a proxy measurement forconvection over the tropics for some time (e.g. Bess et al., 1989 and Maddox et al.,1992b). Any OLR values less than or equal to 240 W rn-2 are regarded as beingassociated with cold convective cloud tops (Arkin et al, 1989), and therefore, convectiverainfall. Assuming blackbody conditions, 240W m2 is equivalent to 255 K (-18CC).Lifting a saturated air parcel with a temperature of 20CC, from the 1000 mb level, thecloud top must ascend to over 6 km (20000 feet) to cool to -18CC in a standardatmosphere. Therefore, if cloud tops are over 6 km, the clouds are considered to beconvective. Of course, this calculation is for a single measurement of a cloud top. In thecase of an individual OLR grid point, which is averaged over a 2.5 degree latitude by 2.5degree longitude box, cold convective cloud tops would be averaged with wannersurroundings and produce significantly warmer average temperatures than -l8C. Hence,the actual convective cloud top heights are more likely considerably colder than -18CCand much higher than 6 km. A considerable amount of research has been done tocalculate exactly what average OLR grid value corresponds to cold convective cloud tops95that are presumably raining underneath. The current consensus is that this thresholdcorresponds to 240 W rn-2. There are still errors associated with using OLR as a proxyfor convective precipitation. The largest such error is the influence of cirrus clouds,which can have very low OLR values yet not necessarily be associated with convectiveprecipitation.Figure 28 shows composite OLR for JJAS using a contour interval of 10 W rn-2.Here the solid lines indicate areas of convective activity. It must be emphasized here thatover the extreme northern United States the ground temperatures are cool enough to biasthe OLR values. Hence, it is not appropriate to use OLR as a proxy for convectiveprecipitation under such circumstances. Since we are concerned only with the summermonths the distinction between cool ground and cold cloud tops is not that difficult tomake. In other seasons, it would be more difficult and some kind of parameter for cloudcover would have to be incorporated. The OLR values are also monthly averages, henceareas which do not have daily convective clouds will not reach the threshold value of 240Wm2.Convection in June is evident over northwestern South America, Central America,and westward along the intertropical convergence zone. By June convection has alsostarted its northward march into southern Mexico. The high (warm) OLR values overnorthwestern Mexico are an indication of both clear skies and the intense surface thennalheat low already evident in the area. By July rapid change has occurred and convectionreaches well into northwestern Mexico, close to the Arizona and New Mexico borders.Note that the 240 W m2 threshold does not reach into the southeastern United States,which indicates that convective cloud tops do not occur daily over the region. There isalso a slight northward extension of the intertropical convergence zone over the easternz1 ICC)oC) (t C C t\u00E2\u0080\u0099.)\u00E2\u0080\u009CCt1 t%)00C c.-c.\u00E2\u0080\u0099cCC\u00E2\u0080\u0098\u00E2\u0080\u0094 0 \u00E2\u0080\u0098cirj III o C C0000 C,, CDCCDCDCC\u00E2\u0080\u0099JC\u00E2\u0080\u0099 CD C -cCCDCDC\u00E2\u0080\u0094\u00E2\u0080\u0098-.CD C100Pacific. The convective rainfall is at full intensity during the month of July. The patternis similar in August. However, by September the area of convection is retreatingsouthward. Note again that the onset and demise of the convective rainfall is notsymmetrical, that is, the progression is quicker than the retreat. Negri et al (1993) foundsimilar OLR distribution patterns while examining the rainfall climatology over Mexico.6.5 OLR AND PRECIPITATIONOLR data are used to corroborate the relationship between upper-level divergenceand precipitation. Figure 29 is July minus June OLR with a contour interval of 5 W m2.Negative values are indicative of cooler cloud tops, hence the more negative the values,the more convective July is compared with June. Conversely, positive areas indicate lessconvective activity in July than June. Obviously, there is a considerable decrease in OLR,and hence increase in convective activity, over western Mexico and into Arizona andNew Mexico. There is also a small decrease in OLR, and hence increase in convection,along the northern Gulf of Mexico as far east as Florida. Areas showing a decrease(increase) in convection (OLR) include Central America eastward to Cuba. Other areasshowing strong decreases (increases) in convection (OLR) are centred over the U.S.Midwest southward into Texas, as well as from Nevada northeastwards into Montana.When compared with the divergence and precipitation map for the same period (Figure23) the general pattern is almost identical. The increase in convection inferred from thedecrease in OLR over northwestern Mexico matches well with an increasingly divergentflow aloft and an increase in precipitation, and the position of the upper-level anticyclone.The extension of the upper-level anticyclone eastward into Florida corresponds to anincreasingly divergent flow aloft, increased (decreased) convection (OLR) and101zzzzzzq- \u00E2\u0080\u0094 - - -.-Figure 29 July minus June difference field for OLR (W m2)with a contour interval of 5 W m2 (negative contours are dashed)\u00E2\u0080\u0094ViiiV A102precipitation. Similarly, areas of decreased (increased) convection (OLR) andprecipitation compare well with regions of increasingly convergent flow aloft.Figure 30 is September minus August OLR and is essentially a reversal, albeit aweaker change than July minus June. Now convection (OLR) decreases (increases) overwestern Mexico, Arizona, New Mexico, and the northern Gulf of Mexico region. Whileconvection (OLR) increases (decreases) slightly over Texas, Central America, and areassouth of Florida. There are at least four reasons for the weaker September minus Augustchanges. Firstly, convection is indirectly being inferred from the OLR, and therelationship is not perfect. For example in September the cooler ground temperatures inthe northern United States can begin to bias the OLR data. Secondly, the onset anddemise of the monsoon is asymmetric. Recall that the monsoon onset is more rapid thanits demise, hence changes in precipitation are less dramatic. Thirdly, the demise does notfall nicely at the transition periods between the two months of August and September,whereas the onset does start near the beginning of July. Finally, the increased effect ofsynoptic weather systems over the northern sections of the United States duringSeptember will tend to mask out convective precipitation.Mo and Rasmusson (1993) looked at OLR, divergence, and precipitation in adifferent manner. Instead of comparing precipitation with divergence, they compared theOLR and divergence fields for the summer of 1987. They found that OLR departuresfrom the mean had a spatial correlation coefficient of 0.8 with 200 mb divergence. Theyalso found that OLR is a poor proxy for divergence in areas of weak convection or areasof subsidence. That is, areas of 200 mb convergence are poorly related to OLRdepartures. This means that our comparison of precipitation and divergence would alsofail in areas of upper-level convergence.103Figure 30 September minus August difference field for OLR (W m2)with a contour interval of 5 W m2 (negative contours are dashed)1046.6 INTERANNUAL VARIABILITYThe largest increases in OLR from June to July occur over northwestern Mexico.Monthly OLR means are calculated from the twice daily OLR data, for each year ofavailable data. Fifteen years of data from 1974 to 1990 were used (1978 and 1988 weremissing).The onset of the monsoon brings the most noticeable change in convectiveactivity. The 15 years of July minus June OLR difference fields are shown in AppendixA. Many years closely resemble the average July-June field shown in Figure 29. Othersshow the monsoon did not reach as far northward as Arizona and New Mexico. Someshow an increase (decrease) in convection (OLR) across the northern Gulf of Mexico,while others do not. One year in particular (1979) showed a weak Mexican monsoon witha stronger inferred increase in convection over the Gulf of Mexico.Because the spatial variability of the monsoon makes it difficult to estimate itsstrength each year, a simple objective index is developed. The largest change in OLRbetween June and July almost always occurs over northwestern Mexico, as in the averagefield. The OLR grid is based on 2.5 by 2.5 degree latitude-longitude boxes, and 12 out ofthe 15 year maximum OLR differences occurred at the same grid point (one was just tothe left of that grid point and two were to the right). Thus, the maximum value of thosethree particular grid points (points A,B ,C see Figure 4) is used to create an index toexamine the temporal variability of the monsoon. That is, the maximum July-June OLRdifference normalized by the July+June OLR for either point A,B, or C is the simpleobjective OLR index.105The mean OLR difference over the 15 year period was -48.0 W rn-2, with astandard deviation of 19.0 W m2. This indicates considerable variability, and shows thatthe onset of the monsoon is not always a strong event. Gadgil et at. (1992) found that theOLR values from NOAA-SR (1974-78) were consistently higher than those from NOAA7 (1982 and onward) and could not be accounted for by an increase in convection sincerainfall data did not appear to corroborate the lower values. Thus, any long term trendanalysis of OLR data is subject to criticism. However, we are looking at differenceamounts between adjacent months, which should be less sensitive to the effects ofsystematic error.Figure 3 la shows the OLR index for increased convective activity versus year.The more negative the number the stronger the onset of the monsoon. The 1970s stillappear to have had stronger monsoons than the 1980s. This is consistent with Gadgil etal.\u00E2\u0080\u0099s (1992) systematic error hypothesis, even though OLR difference values wereemployed.In order to verify that the OLR index is an appropriate measure of the monsoon,comparison is made to monthly precipitation. Unfortunately, the area of the index gridpoints has no precipitation station. The closest station from the WMSSC data set isGuaymas, Mexico which is 20 south of the points A, B, and C. To the north is Arizonaand New Mexico, whose closest stations are approximately 20 to the north of points A, B,and C. Therefore, to give an estimate of the precipitation in the area of points A, B, and Can average was taken of Guaymas, and the southeastern Arizona and southwestern NewMexico climate data divisions (each climate data division is an average of all availablestations within each division). Figure 3 lb is a plot of the combined average precipitationindex (July - June)/(July + June).106H LU-0.1-C-0.15-\u00E2\u0080\u00940.2\u00E2\u0080\u00941974 1976 1978 1980 1982 1984 1986 1988 1990YEAR1-\u00E2\u0080\u0094--0.75-0.5-0.25-U- \u00E2\u0080\u0094 9- \u00E2\u0080\u0094 \u00E2\u0080\u0094 9- \u00E2\u0080\u0094 9-\u00E2\u0080\u0094 9- \u00E2\u0080\u0094 9- 9-1974 1976 1978 1980 1982 1984 1986 1988 1990YEARFigure 31 a) OLR index from 1974-1990 (top),b) Precipitation index from 1974-1988 (bottom)107The scatter plot (Figure 32) gives a correlation coefficient (r) of -0.70 (95%confidence limits range widely from a low of -0.24 to a high of -0.90 because of thelimited sample size). There are only 13 years of overlapping data, so any conclusions aretenuous at best. Nevertheless, agreement between the two indices gives support to thesuggestion that the OLR index is an appropriate measure of the monsoon.The weakest OLR index occurs in 1987, which corresponds to the lowestprecipitation index. The second weakest OLR index is in 1984 which corresponds to thesecond lowest precipitation index. The next two weakest OLR indices are 1979 and 1986.Precipitation indices for these two years are fairly low but do not correspond to the sameranking as the OLR index. The four weakest OLR indices in order of increasing strengthare 1987, 1984, 1986, and 1979.The monsoon season in 1984 is a special case in that the monsoon started a monthearlier than normal, with the upper-level anticyclone already into northwestern Mexico inJune. Thus the July-June indices were artificially low since they were the differencesbetween two wet months rather than a dry to wet transition. Hence, 1984 was not a weakmonsoon year, but started a month earlier instead.Examination of the remaining years (1979, 1986, 1987) reveals a commonfeature. Each year shows the upper-level anticyclone had not yet reached northwesternMexico by July, and in fact was still weak and centred over the southern Baja peninsula.In both 1986 and 1987 the upper-level high managed to move farther northward byAugust, and hence the monsoon was delayed somewhat. However in 1979, the upperlevel high never reached farther north than the middle of the Baja peninsula so themonsoon did not reach into the southwestern United States. Reyes and Cadet (1986,1988)108C0PREC INDEXFigure 32 Scatterplot of OLR index versus Precipitation index109investigated water vapour fluxes during the summer of 1979, in an attempt to determinethe moisture sources for the southwest monsoon. Unfortunately, this turned out to be oneof the weakest monsoons of the period which is probably why their findings were not assubstantial as they had hoped.110CHAPTER 7. SUMMARY AND CONCLUSIONSMonsoon circulations are found throughout much of the extratropics. Using ascheme based on surface wind criteria, Ramage (1971) classified Africa, southern Asia,and northern Australia as having a monsoon climate. An atmospheric circulation patternthat coincides with all monsoons is a large upper tropospheric anticyclone (Das, 1986). Awesterly jetstream is commonly observed to the north of the anticyclone, while aneasterly jetstream is to the south (Chen et at, 1989; Soman and Kumar, 1993). Duringwinter a strong surface anticyclone establishes itself over continental regions. Thiscreates an \u00E2\u0080\u009Coffshoret\u00E2\u0080\u0099wind component and dry conditions generally prevail throughoutthe region. In summer the flow reverses as the surface anticyclone is replaced by a strongthermally induced low pressure system. The \u00E2\u0080\u009Conshore\u00E2\u0080\u009D wind component often coincideswith the onset of heavy summer rainfall in the region.Ramage\u00E2\u0080\u0099s (1971) classification scheme excludes any monsoon climates in NorthAmerica because there is no distinct seasonal shift in surface winds. However, Kendrew(1961) categorizes portions of Mexico and Central America as having a monsoon climatebased on a summer rainfall maximum followed by a dry winter. Arizona is not includedas a monsoon climate because it does not have a dry winter. However, the term \u00E2\u0080\u009CArizonaor Mexican monsoon\u00E2\u0080\u009D has become entrenched in the research literature to describe thesummer precipitation in the southwestern United States and northwestern Mexico.The Arizona or Mexican monsoon has been extensively studied in the past. Earlyresearch efforts focussed on moisture sources for the heavy summer rains. However, thereare still many questions that remain to be resolved with the current hypothesis ofmoisture transport up the Gulf of California (Hales, 1974). Case studies of singular events111or mesoscale aspects of the phenomena comprise the bulk of the more recent researchefforts. Thus much of the research has essentially been done in isolation from the largerscale atmospheric flow. As a result there are considerable gaps in our understanding ofthe relationship between the monsoon and the larger scale atmospheric circulation. Thisthesis attempts to address those gaps in understanding.In Chapter 4 an investigation of monthly precipitation data showed severalsummer precipitation regimes throughout the southern United States and Mexico. Overthe southwestern United States and northwestern Mexico rainfall increases sharply fromJune to July. A less abrupt increase takes place over the southeastern United States. Thereverse is true over the northern United States Rocky Mountains, Texas, eastern Mexico,and the northern Caribbean, where rainfall decreases from June to July. Thesesimultaneous but reverse changes in precipitation hint at the possibility of a larger scaleconnection between these summer precipitation regimes.To highlight the temporal change in precipitation a few stations in the westernUnited States were analyzed in more depth using daily precipitation data. Mexican dailyprecipitation data were not available for use in this study. A simple wet-dry index wasdeveloped and applied to graphically show the abrupt changes that occur in summerprecipitation. The heavy summer rains begin in earnest over the southwestern UnitedStates near the end of June or beginning of July. To the north over the northern UnitedStates Rocky Mountains strong drying takes place near the middle of June.The question of where the moisture for the heavy summer rains initially comesfrom is difficult to answer. Preliminary analysis of water vapour from satellite imagerysuggests there are multiple potential moisture sources. One such possible moisture sourcecould be the northern edge of the ITCZ, which moves northward during summer. Waliser112and Gautier\u00E2\u0080\u0099s (1993) study of a satellite derived climatology of the 1TCZ for a 17 yearperiod shows that the large-scale deep convection along the ITCZ widens during July,with the northern edge reaching southern Mexico. Satellite imagery shows tropical watervapour plumes (well-defined boundaries of middle to upper-level moisture) extendingfrom the northern edge of the ITCZ through Mexico and into the United States on aregular basis during the summer monsoon (Thiao et at, 1993). This large scale source ofmoisture hints at the possibility that local scale sources of moisture in Arizona orsurrounding areas may be less important than previously thought.In Chapter 5 the changes in large scale atmospheric circulation that accompanythe onset and demise of the monsoon are examined. The summer rainfall so characteristicof other monsoons, does not coincide with the presence of the deepest surface thermaltrough. In fact, the deepest thermal low occurs in June over Arizona a month in advanceof the monsoon rains. The arrival of the monsoon rains is accompanied by a weak rise inmean sea level pressure (MSLP) over the affected area. This is contrary to what isnormally expected with convective rainfall. It also suggests the possibility that othermechanisms, other than local convection, might be involved.An examination of available National Meteorological Center upper level datashows an upper-level anticyclone which corresponds to surface precipitation as in othermonsoons around the world. Weekly composites of upper-level wind data show the rapidprogression of the upper-level anticyclone from southern Mexico during the beginning ofJune to the southwestern United States-northwestern Mexico border by the middle ofJuly. The anticyclone meanders near the border for the remainder of July. In thebeginning of August the anticyclone starts its southward track only to stall in the laterhalf of August just southeast of Guaymas. The retreat begins again in the beginning of113September but at a slower pace than its initial advance through Mexico. Since thenorthward advance is more rapid than its retreat, the motion is asymmetric.Analyses of upper-level divergence in Chapter 6 show that changes inprecipitation are consistent with changes in divergence aloft. Over northwestern Mexicoand the southwestern United States upper-level divergence corresponds well with thesummer monsoon signature. Increased precipitation from June to July corresponds to anincreasingly divergent flow aloft. As precipitation decreases from August to Septemberan increasingly convergent flow aloft is evident. Similarly over the northern United StatesRocky Mountains decreased precipitation from June to July corresponds to increasinglyconvergent flow aloft. The reverse is true from August to September, as precipitationincreases over the Rockies correspond to increasingly divergent flow aloft. The bimodalpeak in precipitation in the summer half of the year, around the Gulf of Mexico, alsomatches the increasingly divergentlconvergent patterns aloft.Using OLR as an independent data set, it is confirmed that convective rainfallcorrelates well with upper-level divergence. Precipitation, OLR, and upper leveldivergence information demonstrate a remarkable spatial correlation across the southernUnited States, Mexico, and the northern Caribbean.These analyses show a great deal of consistency between the large scaleatmospheric circulation and the different summer precipitation regimes in the area. Aconceptual model of the monsoon at its height in July shows this in Figure 33. An upperlevel anticyclone is well established over northwestern Mexico, with a ridge extendingeastward to a weaker anticyclone over the southeastern United States. The subtropical jethas moved farther north where it converges with the polar jet over the northern114Figure 33 Conceptual model of the North American monsoon (Thick lines are jetstreamaxes. The strong upper-level anticyclone position is designated by a large H. The weakanticyclone is designated by a small H. Areas of increasingly divergent flowaloft (DIV) and increasingly convergent flow aloft (CON) are also noted)115United States. From June to July areas of increasingly divergent/convergent flow aloftmatch changes in precipitation (OLR) at the surface.This thesis was originally undertaken to better understand the summerprecipitation regimes that exist in North America, in particular, the Arizona or Mexicanmonsoon. Its primary goal was to provide other researchers and forecasters with a largescale view of the phenomenon, in hopes that this broadened context would provide abasis for improved prediction of summertime rainfall anomalies over North America. It isnow clear that the onset of the monsoon is part of a much larger pattern of circulationchanges that occur rather abruptly in late June to early July. Precipitation, OLR and upperair data present a mutually consistent picture of these changes, which extend over most ofthe southern United States, Mexico, and the northern Caribbean. The rapid northwarddevelopment of the upper level anticyclone coincides with the abrupt changes found inthe precipitation and OLR data. The upper-level anticyclone provides a much betterindicator for the onset and demise of the monsoon than the previously used 500 mb ridgelines, or the strength and position of the thermal low (Carleton, 1985,1986, 1987).Forecasters would be well advised to monitor this atmospheric feature in order to helpforecast the monsoon onset.The monsoon can no longer be viewed as a meso- to synoptic scale phenomena: itis an integral part of the summertime circulation pattern over North America. Hence itwould be better to rename it the North American monsoon. A re-evaluation of thepresently accepted precipitation climatology of the area should also be undertaken. Thesummer precipitation regimes that heretofore have been treated as separate, should betreated as connected entities. The view that summer monsoons exist only in Asia, Africa,and Australia also needs re-examination. Figure 34 shows Ramage\u00E2\u0080\u0099s (1971) monsoon116Figure 34 Global monsoon region as defined by Ramage (after Ramage, 1971) withapproximate area of North American monsoon added117region as defined by his wind criteria with the approximate extent of the North Americanmonsoon as defined by precipitation and OLR data. The northern and western boundariescoincide with the approximate limits of the monsoon rains over the southwestern UnitedStates. The eastern boundary is more artificial since precipitation data are not analyzedover the Atlantic Ocean. The southern boundary is undefined because the differentialheating between land and ocean masses decreases as the land area diminishes.Many of the remaining questions concerning the North American monsoon relateto the following questions:1)Is there another unexplored source of moisture for the monsoon?2) Why is the onset of the monsoon so rapid?3) How does the monsoon affect surrounding areas of North and CentralAmerica?4) What controls the interannual variability of the monsoon?A cursory examination of hourly satellite imagery during several North Americansummer monsoons has led the author to believe that some of the water vapour thatcondenses over the southwestern United States and northwestern Mexico comes from thesouthwestern Gulf of Mexico through the Rio B alsas river basin (south of Mexico city)then northwestward through the Rio Grande river basin to the west coast of Mexico.From there the convection remains close to the western slopes of the Sierra MadreOccidental mountain range all the way north to the Mexico-United States border nearArizona and New Mexico. This southerly flow may be related to the gulf surges or low-level southerly jetstream found in the Gulf of California and surrounding areas during thesummer monsoon (Douglas, 1992a).118This hypothesis could be investigated using 850 mb gridded wind data to examinethe air flow around the region. However, the scale of the river basins is small enough thatboth the observation network and the objective analysis scheme that converted the data togridded values might not be able to resolve it. It might be possible to use satellite derivedwinds from cloud motion to see if the flow across this basin at the top of the planetaryboundary layer is strong enough to account for the moisture flux that supplies themonsoon and if it tends to be particularly strong during periods of heavy monsoonrainfall. If there were not enough trade wind cumulus clouds to support estimates of thewind field in this region, (subject to the availability of resources) constant densityballoons could be used to track the air flow from the pathway to northwestern Mexico.Another question deals with the rapid onset of the North American monsoon.More work is needed to better document the onset date of the monsoon as a function ofgeographical location. Weekly rather than monthly composites of infrared satelliteimagery would better define the onset. Higher resolution OLR data are now available tobetter define the mesoscale structure of the monsoon. Eventually high quality dailyprecipitation data will be available for Mexico to help provide a better temporal sequencefor the monsoon.Since the solar forcing changes only slightly from June to July it is surprising thatnorthwestern Mexico and the southwestern United States go from their driest to theirwettest seasons in less than a month. How does the atmosphere change so quickly?One possible avenue to explore is the atmosphere\u00E2\u0080\u0099s ability to move from onesteady state equilibrium to another in a short period of time. For example, the upper flowover western North America during wintertime can be meridional for weeks at a time andthen, within a matter of a few days, change to zonal and remain that way for weeks,119before switching back to a more meridional flow. Perhaps the monsoon circulation andnon-monsoon circulation are another example of such a phenomena. A diagnostic studyof gridded upper air data could determine whether such abrupt changes in circulation arecommon. However, a more theoretically-oriented general circulation modelling studywould be necessary in order to begin to understand why such rapid changes occur in theatmosphere.Another interesting possibility is to investigate the effect of boundary conditionson the climate (Meehl, 1994): Extreme heating of the surface can occur only after the soilmoisture has been largely depleted. Thus the differential heating between land and oceanmasses can increase quickly, once the solar energy that was used to melt mountain snowsand evaporate soil moisture during the spring becomes available to heat the parchedsurface during early summer. Reduced water availability will also cause changes invegetation which, in turn, affect the albedo of the area. An analysis of observational datacould be undertaken to see if relationships exist between certain anomalous boundaryconditions and the monsoon. if any relationships are indicated then a climate model couldbe used to determine how much of a role the absence or presence of a particular boundarycondition plays in the rapid onset of the monsoon.When the monsoon begins over northwestern Mexico and the southwesternUnited States an almost simultaneous drying occurs both to the north and to the east. Amore comprehensive analysis of daily precipitation data can better define the strength andextent of this negative correlation. The simultaneous occurrence of flooding in theMississippi river basin and a relatively weak monsoon in the summer of 1993 is areflection of such a negative correlation. A more comprehensive diagnostic study could120verify whether strong (weak) monsoons are consistently associated with suppressed(enhanced) summer convection in surrounding areas.Several processes could contribute to the observed negative correlation describedin the previous paragraph. It is conceivable that the monsoon takes most of the availablemoisture in the region, leaving less for the surrounding areas, or that it inducessubsidence over neighbouring areas thus creating an inversion and reducing convection.General circulation modelling studies would be best to explore whether these hypothesesbear merit.A preliminary examination of the OLR data shows that there is considerableinterannual variability of the North American monsoon. Both the timing and intensity ofthe onset of the monsoon vary. The causes underlying the interannual variability of theNorth American monsoon need to be investigated. Since the monsoon is believed tooccur in response to the differential heating of the land and adjacent oceans it seemsreasonable that the variability of either land and ocean thermal characteristics could giverise to monsoon variability (Joseph et al, 1991, 1994).Much like the heating of the Tibetan plateau is thought to play a major role in theAsian monsoon, the Great Basin of the United States may play a role in the NorthAmerican summer monsoon (Tang and Reiter, 1984). There are several factors toexamine in determining the thermal characteristics of the land. Since the amount of snowcoverage varies from year to year, the eventual melting of all the snow takes a differentamount of time each year. Once the energy is no longer needed to melt the snow it can gotowards evaporating soil moisture. Once the soil is dry the energy can go towards heatingthe surface and overlying air, enabling the heating to occur at a faster rate. Hence, bothsnow cover and soil moisture may be important factors in determining whether a weak or121strong monsoon will occur. Hence, the preceding seasons may be very important tomonsoon variability. A colder than normal winter with more snow could conceivably leadto a weaker monsoon, or an anomalously warm, dry spring may lead to a strongmonsoon. Even vegetation and albedo changes could affect differential heating in thesummer.Changes in sea-surface temperature (SST) in either the eastern Pacific and theGulf of Mexico could give rise to changes in differential heating. On a local scale warmerSST in the Gulf of California would increase evaporation, thereby increasing theavailable moisture for the monsoon, while colder SST could lead to a stronger sea breezecirculation which could enhance the vertical motion along the western flanks of the SierraMadre Occidental mountain range during the day.Large scale changes to SST in the tropical Pacific associated with ENSO havebeen shown to be related to the Asian monsoon (Joseph et al, 1994). Hence, there existsthe possibility that the North American monsoon may feel the effects of SST anomaliesin the eastern equatorial Pacific.During El Nino years the ITCZ tends to be south of its climatological position.Previously, the ITCZ was proposed as a source of moisture for the North Americanmonsoon, since tropical water vapour plumes regularly extend from the northern edge ofthe 1TCZ through Mexico and into the United States during the summer (Thiao et al,1993). Hence, the strength and position of the ITCZ could also play a role in the NorthAmerican monsoon.Diagnostic studies could be performed to see if relationships exist between any ofthe above mentioned factors and the summer monsoon. If significant relationships are122found, climate modelling experiments would have to be employed in order to try tounderstand the underlying linkages.It is conceivable that both local and remote factors could contribute to thevariability of the North American monsoon. Since the monsoon is such an importantcomponent of the summer climate for much of North America, progress in understandingthe factors that influence it could yield major benefits in terms of improved summertimeclimate prediction.123REFERENCESAdang, T.C., and R.L. Gall, 1989: Structure and dynamics of the Arizona monsoonboundary. Monthly Weather Review, 117, 1423-1438.American Meteorological Society, 1959: Glossary ofMeteorology. AmericanMeteorological Society, Boston.Arkin, P.A., V.R. Krishna Rao, and R.R. Keilcar, 1989: Large scale precipitation andoutgoing long-wave radiation from INSAT- lB during the 1986 southwestmonsoon season. Journal of Climatology, 2, 6 19-628.Badan-Dangon, A., C.E. Dorman, M.A. Merrifield and C.D. Winant, 1991: The loweratmosphere over the Gulf of California. Journal of Geophysical Research, 96, C9,16877-16896.Bess, T.D., G.L. Smith, and T.P. Charlock, 1989: Ten-year monthly data set of outgoinglong-wave radiation from Nimbus-6 and Nimbus-7 satellites. Bulletin of theAmerican Meteorological Society, 70, 480-489.Brenner, I.S., 1974: A surge of maritime tropical air - Gulf of California to thesouthwestern United States. Monthly Weather Review, 102, 375-389.Broccoli, A.J., and S. Manabe, 1992: The effects of orography on midlatitude NorthernHemisphere dry climates. Journal of Climate, 5, 1181-1201.Bryson, R.A., and W.P. Lowry, 1955a: Synoptic climatology of the Arizona summerprecipitation singularity. Bulletin of the American Meteorological Society, 36,329-339.Bryson, R.A., and W.P. Lowry, 1955b: Synoptic climatology of the Arizona summermonsoon. Scientific Report No. 1, USAF Contract AF19(604)-992, Department ofMeteorology, University of Wisconsin, Madison, 24 pp.Burpee, R.W., 1979: Peninsula-scale convergence in the south Florida sea breeze.Monthly Weather Review, 107, 852-860.Carleton, A.M., 1985: Synoptic and satellite aspects of the southwestern U.S. summermonsoon. International Journal of Climatology, 5, 3 89-402.Carleton, A.M., 1986: Synoptic-dynamic character of bursts and breaks in the south-westU.S. summer precipitation singularity. International Journal of Climatology, 6,605-623.124Carleton, A.M., 1987: Summer circulation climate of the American southwest, 1945-1984. Annals of the Association of American Geographers, 77, 619-634.Carleton, A.M., D.A. Carpenter and P.J.Weser, 1990: Mechanisms of interannualvariability of the southwest United States summer rainfall maximum. Journal ofClimate, 3, 999-1015.Cavazos, T., and S. Hastenrath, 1990: Convection and rainfall over Mexico and theirmodulation by the southern oscillation. International Journal of Climatology, 10,377-386.Chen, T-C., M-C. Yen and H.Van Loon, 1989: The effect of the divergent circulation onsome aspects of the 1978/79 southern hemisphere monsoon. Journal of Climate, 2,1270-1288.Cressman, G., 1959: An operational objective analysis system. Monthly Weather Review,87, 367-374.Das, P.K., 1986: Monsoons. World Meteorological Organization, WMO No. 613, l55pp.Dey, C.H., 1989: The evolution of objective analysis methodology at the NationalMeteorological Center. Weather and Forecasting, 4, 297-312.Dodd, A.V., 1965: Dew point distribution in the contiguous United States. MonthlyWeather Review, 93, 113-122.Douglas, M.W., R.A. Maddox and K.W. Howard, 1991: The summer monsoon overnorthwestern Mexico: results from the southwest area monsoon project.Submitted to V Congreso Nacional de Meteorologia, October 2-4, 1991, Cd.Juarez, Chihuahua, Mexico.Douglas, M.W., R.A. Maddox, K.W. Howard, and S. Reyes, 1993: The Mexicanmonsoon. Journal of Climate, 6, 1665-1677.Douglas, M.W., C. Watts and S. Reyes, 1992: Gulf of California surges and associatedrainfall during the southwest North American monsoon. Preprints, Symposium onWeather Forecasting, January 5-10, 1992, Atlanta, Georgia, USA, 27-30.Douglas, M.W., 1992a: Structure and fluctuations of the summertime low-level jet overthe Gulf of California and environs. Preprints, Fifth Conference on MesoscaleProcesses, January 5-10, 1992, Atlanta, Georgia, USA.Douglas, M.W., l992b: Structure and dynamics of two monsoon depressions. Part I:observed structure. Monthly Weather Review, 120, 1524-1547.Douglas, M.W., 1992c: Structure and dynamics of two monsoon depressions. Part II:vorticity and heat budgets. Monthly Weather Review, 120, 1548-1564.125Douglas, M.W., 1993: Northern Mexico\u00E2\u0080\u0099s Monsoon. Preprints, Sixth Mexican NationalMeteorological Conference and Second Iberoamerican Congress on AtmosphericEnvironment, October 27-29, 1993, Mexico City, Mexico.Englehart, P.J., and A.V. Douglas, 1985: A statistical analysis of precipitation frequencyin the conterminous United States, including comparisons with precipitationtotals. Journal of Climate and Applied Meteorology, 24, 350-362.Epstein, E.S., and A.G. Barnston, 1988: A precipitation climatology of five-day periods.NOAA Technical Report 41, U.S. Department of Commerce, 162 pp.Fennessy, M.J., J.L. Kinter III, B. Kirtman, L. Marx, S. Nigam, E. Schneider, J. Shukia,D. Straus, A. Vernekar, Y. Xue, and J. Zhou, 1994: The simulated Indianmonsoon: a GCM sensitivity study. Journal of Climate, 7, 33-43.Fritsch, J.M., and R.A. Maddox, 1981a: Convectively driven mesoscale weather systemsaloft. Part I: observations. Journal of Applied Meteorology, 20, 9-19.Fntsch, J.M., and R.A. Maddox, 1981b: Convectively driven mesoscale weather systemsaloft. Part II: numerical simulations. Journal of Applied Meteorology, 20, 20-26.Fritsch, J.M., R.J. Kane and C.R. Chelius, 1986: The contribution of mesoscaleconvective weather systems to the warm season precipitation in the United States.Journal of Climate and Applied Meteorology, 25, 1333-1345.Gadgil, S., A. Guruprasad and J. Srinivasan, 1992: Systematic bias in the NOAAoutgoing long-wave radiation data set? Journal of Climate, 5, 867-875.Groisman, P.Y., and D.R. Legates, 1994: The accuracy of United States precipitationdata. Bulletin of the American Meteorological Society, 75, 2 15-227.Gruber, A., and A.F. Krueger, 1984: The status of the NOAA outgoing long-waveradiation data set. Bulletin of the American Meteorological Society, 65, 95 8-962.Hahn, D.G., and S. Manabe, 1975: The role of mountains in the south Asian monsooncirculation. Journal of the Atmospheric Sciences, 32, 1515-1541.Hales, J.E., 1972: Surges of maritime tropical air northward over the Gulf of California.Monthly Weather Review, 100, 298-306.Hales, J.E., 1974: Southwestern United States summer monsoon source - Gulf of Mexicoor Pacific Ocean? Journal of Applied Meteorology, 13, 33 1-342.Hales, J.E., 1975: A severe southwest desert thunderstorm: 19 August 1973. MonthlyWeather Review, 103, 344-351.126Hastenrath, S., 1988: Prediction of Indian monsoon rainfall: further exploration. Journalof Climate, 1, 298-304.Hastings, J.R., and R.M. Turner, 1965: Seasonal precipitation regimes in Baja California,Mexico. Geografiska Annaler, 47A, 204-223.He, H., J.W. McGinnis, Z. Song, and M. Yanai, 1987: Onset of the summer monsoon in1979 and the effect of the Tibetan plateau. Monthly Weather Review, 115, 1966-1995.Heideman, K.F., and J.M. Fritsch, 1988: Forcing mechanisms and other characteristics ofsignificant summertime precipitation. Weather and Forecasting, 3, 115-131.Horn, L.H., R.A. Bryson, and W. P. Lowry, 1957: An objective precipitation climatologyof the United States. Scientific Report No. 6, USAF Contract AF19(604)-992,Department of Meteorology, University of Wisconsin, Madison, 33 pp.Horn, L.H., and R.A. Bryson, 1960: Harmonic analysis of the annual march ofprecipitation over the United States. Annals of the Association of AmericanGeographers, 50, 157-171.Horel, J.D., A.N. Hahmann and J.E. Geisler, 1989: An investigation of the annual cycleof convective activity over the tropical Americas. Journal of Climate, 2, 1388-1403.Howard, K.W., and R.A. Maddox, 1988a: Mexican mesoscale convective systems - asatellite perspective. Preprints, III InterAmerican and Mexican Congress ofMeteorology, November 14-18, 1988, Mexico City, Mexico, 404-408.Howard, K.W., and R.A. Maddox, 1988b: A satellite-based climatology of warm seasonthunderstorms over Mexico. Preprints, III InterAmerican and Mexican Congressof Meteorology, November 14-18, 1988, Mexico City, Mexico, 414-418.Hsu, C-P.F., and J.M. Wallace, 1976: The global distribution of the annual and semiannual cycles in precipitation. Monthly Weather Review, 104, 1093-1101.Ives, R.L., 1949: Climate of the Sonoran desert region. Annals of the Association ofAmerican Geographers, 39, 143-187.Joseph, P.V., B. Liebmann, and H.H. Hendon, 1991: Interannual variability of theAustralian summer monsoon onset: possible influence of Indian summer monsoonand El Nino. Journal of Climate, 4, 529-538.Joseph, P.V., J.K. Eischeid, and R.J. Pyle, 1994: Interannual variability of the onset of theIndian monsoon and its association with atmospheric features, El Nino, and seasurface temperature anomalies. Journal of Climate, 7, 8 1-105.127Jurwitz, L.R., 1953: Arizona\u00E2\u0080\u0099s two-season rainfall pattern. Weatherwise, 6, 96-99.Keables, M.J., 1989: A synoptic climatology of the bimodal precipitation distribution inthe upper Midwest. Journal of Climate, 2, 1289-1294.Kendrew, W.G., 1922: The Climates of the Continents. 1st Edition, Clarendon Press,Oxford, 286-298.Kendrew, W.G., 1961: The Climates of the Continents. 5th Edition, Clarendon Press,Oxford.Koteswaram, P., 1958: The easterly jet stream in the tropics. Tellus, 10, 43-57.Krishnamurti, T.N., 1971: Tropical east-west circulations during the northern summer.Journal of the Atmospheric Sciences, 28, 1342-1347.Krishnamurti, T.N., H.S. Bedi, and M.Subramaniam, 1989: The summer monsoonof1987. Journal of Climate, 2, 321-340.Kung, E.C., and T.A. Sharif, 1980: Regression forecasting of the onset of the Indiansummer monsoon with antecedent upper-air conditions. Journal of AppliedMeteorology, 19, 370-380.Leary, C.A., 1979: Behaviour of the wind field in the vicinity of a cloud cluster in theintertropical convergence zone. Journal of the Atmospheric Sciences, 36, 631-639.Leary, C.A., and R.A. Houze, 1979: The structure and evolution of convection in atropical cloud cluster. Monthly Weather Review, 36, 437-457.Lowry, W.P., 1956: Notes on the ending of Oregon\u00E2\u0080\u0099s Rainy season. Bulletin of theAmerican Meteorological Society, 37, 426-428.Maddox, R.A, and K.W. Howard, 1988a: Mexican mesoscale convective systems - twocase examples. Preprints, Ill InterAmerican and Mexican Congress ofMeteorology, November 14-18, 1988, Mexico City, Mexico, 89-93.Maddox, R.A, and K.W. Howard, 1988b: Mexican mesoscale convective systems - largescale environmental conditions. Preprints, III InterAmerican and MexicanCongress of Meteorology, November 14-18, 1988, Mexico City, Mexico, 395-399.Maddox, R.A., M.W. Douglas and K.W. Howard, 1991: Mesoscale precipitation systemsover southwestern North America: a warm season overview. Preprints,International Conference on Mesoscale Meteorology and TAMEX, December 3-6,1991, Taiwan, 393-402.128Maddox, R.A, K.W. Howard and A.J. Negri, 1992a: Analysis of GOES infraredconvective cloud-top temperatures for extended periods: an overview of the 1990warm season for a subtropical region. Preprints, Sixth Conference on SatelliteMeteorology and Oceanography, January 5-10, 1992, Atlanta, Georgia, USA,205-208.Maddox, R.A, K.W. Howard and A.J. Negri, 1992b: Wide-area, long-term satellite rainestimation using GOES high resolution imagery. Preprints, Sixth Conference onSatellite Meteorology and Oceanography, January 5-10, 1992, Atlanta, Georgia,USA, 260-263.Magana, V., and M. Yanai, 1991: Tropical-midlatitude interaction on the time scale of 30to 60 days during the northern summer of 1979. Journal of Climate, 4, 180-201.Markham, C.G., 1970: Seasonality of precipitation in the United States. Annals of theAssociation of American Geographers, 60, 593-597.McBride, J.L., and W.M. Gray, 1980: Mass divergence in tropical weather systems.Quarterly Journal of the Royal Meteorological Society, 106, 501-538.Meehi, G.A., 1989: The coupled ocean-atmosphere modeling problem in the tropicalpacific and Asian monsoon regions. Journal of Climate, 2, 1146-1163.Meehl, G.A., 1994: Influence of the land surface in the Asian summer monsoon: externalconditions versus internal feedbacks. Journal of Climate, 7, 1033-1049.Meitin, J.G., K.W. Howard and R.A. Maddox, 1991: Southwest area monsoon project.Daily operations summary. available from the National Severe StormsLaboratory, Boulder, CO, USA, 80303.Mo, K., and E.M. Rasmusson, 1993: The 200-mb climatological vorticity budget during1986-1989 as revealed by NIvIC analyses. Journal of Climate, 6, 577-594.Moore, T.J., R.L. Gall and T.C. Adang, 1989: Disturbances along the Arizona monsoonboundary. Monthly Weather Review, 117, 932-94 1.Mosino, P.A., and E. Garcia, 1974: The climate of Mexico. World Survey of Climatology,Volume 2, Climates ofNorth America, edited by R.A. Bryson and F.K Hare,Elsevier, New York, 373-404.Negri, A.J., R.F. Adler, R.A. Maddox, K.W. Howard, and P.R. Keehn, 1993: A regionalrainfall climatology over Mexico and the southwest United States derived frompassive microwave and geosynchronous data. Journal of Climate, 6, 2144-2161.Portig, W.H., 1965: Central American rainfall. Geographical Review, 55, 68-90.129Portig, W.H., 1976: The climate of Central America. World Survey of Climatology,Volume 12, Climates of Central and South America, edited by W. Schwerdtfeger,Elsevier, New York, 405-47 8.Prasad, K.D., and S.V. Singh, 1992: Possibility of predicting Indian monsoon rainfall onreduced spatial and temporal scales. Journal of Climate, 5, 1357-136 1.Ramage, C.S., 1971: Monsoon Meteorology. Academic Press, New York, 296 pp.Reap, R.M., 1986: Evaluation of cloud-to-ground lightning data from the western UnitedStates for the 1983-84 summer seasons. Journal of Climate and AppliedMeteorology, 25, 785-799.Reed, T.R., 1933: The North American high level anticyclone. Monthly Weather Review,.61, 321-325.Reed, T.R., 1937: Further observations on the North American high level anticyclone.Bulletin of the American Meteorological Society, 18, 297-298.Reitin, C.H., 1960: Distribution of precipitable water vapor over the continental UnitedStates. Bulletin of the American Meteorological Society, 41, 79-87.Reyes, S., and D.L. Cadet, 1986: Atmospheric water vapour and surface flow patternsover the tropical Americas during May-August 1979. Monthly Weather Review,114, 582-593.Reyes, S., and D.L. Cadet, 1988: The southwest branch of the North American monsoonduring summer 1979. Monthly Weather Review, 116, 1175-1187.Rowson, D.R., and S.J. Colucci, 1992: Synoptic climatology of thermal low-pressuresystems over south-western North America. International Journal of Climatology,12, 529-545.Smith, W.P., and R.L. Gall, 1989: Tropical squall lines of the Arizona monsoon. MonthlyWeatherReview, 117, 1553-1569.Soman, M.K., and K.K. Kumar, 1993: Space-time evolution of meteorological featuresassociated with the onset of Indian summer monsoon. Monthly Weather Review,121, 1177-1194.Tang, M., and E.R. Reiter, 1984: Plateau monsoons of the northern hemisphere: acomparison between North America and Tibet. Monthly Weather Review, 112,6 17-637.Thiao, W., R.A. Scofield, and J. Robinson, 1993: The relationship between water vaporplumes and extreme rainfall events during the summer season. NOAA TechnicalReport NESDIS 67, 69pp.130Trenberth, K.E., and J.G. Olson, 1988: Evaluation of NMC global analyses, 1979-1987.National Center for Atmospheric Research Technical Note, NCARJTN-299+STR,Boulder, Colorado, 82pp.Waliser, D.E., and C. Gautier, 1993: A satellite-derived climatology of the ITCZ. Journalof Climate, 6, 2162-2174.Wallen, C.C., 1955: Some characteristics of precipitation in Mexico. GeografiskaAnnaler, 37, 51-85.Walsh, J.E., and M.B. Richman and D.W. Allen, 1982: Spatial coherence of monthlyprecipitation in the United States. Monthly Weather Review, 110, 272-286.Webster, P.J., 1987: The elementary monsoon. Monsoons, edited by J.S. Fern and P.L.Stephens, John Wiley & Sons, New York, 3-33.White, G.H., 1982: An observational study of the northern hemisphere extratropicalsummertime general circulation. Journal of the Atmospheric Sciences, 39, 24-40.Whitfield, M.B., and S.W. Lyons, 1992: An upper-tropospheric low over Texas duringsummer. Weather and Forecasting, 7, 89-106.WMO, 1982: International Conference on the Scientific Results of the MonsoonExperiment. Denpasar, Bali, Indonesia, October 26-30, 1981, WorldMeteorological Organization.131APPENDIXOLR VARIABILITY 1974-1990The following maps of OLR have undergone \u00E2\u0080\u009Csmoothing\u00E2\u0080\u009D hence the centralvalues are not as high as the OLR index, which were obtained from the raw grid pointdata.132Figure Al July minus June OLR (W m2)difference field for 1974with a contour interval of 5 W m2 (negative contours are dashed)I1I-.0C.,cMC..0 CC)CD0 0.t1.sI-.0> CDCD 00CD CDSQ135--i-\u00E2\u0080\u0098E\\u00E2\u0080\u0098 \u00E2\u0080\u0098tr\ : \u00E2\u0080\u009C ( \u00E2\u0080\u0098 1\ \u00E2\u0080\u0098 \u00E2\u0080\u0098 ; :-\ * s___,I,. \u00E2\u0080\u0094I \u00E2\u0080\u0098 * /I \ \u00E2\u0080\u0098- ,\u00E2\u0080\u0098 ,*\u00E2\u0080\u0098 \u00E2\u0080\u009C \u00E2\u0080\u009C \u00E2\u0080\u0098* %\u00E2\u0080\u0098**\u00E2\u0080\u0098.__.___)I /I\u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094\u00E2\u0080\u009420 g \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 S--\u00E2\u0080\u0094--\u00E2\u0080\u0094-\u00E2\u0080\u0094\u00E2\u0080\u0094-\u00E2\u0080\u0094\u00E2\u0080\u0094\u00E2\u0080\u0094. \u00E2\u0080\u0098\u00E2\u0080\u0094-\u00E2\u0080\u0094-\u00E2\u0080\u0094sE;-E:Z,Figure A4 July minus June OLR (W m2)difference field for 1977with a contour interval of 5 W m2 (negative contours are dashed)136) ) \u00E2\u0080\u0098Z\u00E2\u0080\u0099--/ / Ij - -- - - -- --- - - -- - - - - -- -) -Figure A5 July minus June OLR (W m2)difference field for 1979with a contour interval of 5 W m2 (negative contours are dashed)137Figure A6 July minus June OLR (W m2)difference field for 1980with a contour interval of 5 W m2 (negative contours are dashed)I-.OC)CDCD\D00139- -,r\u00E2\u0080\u0094I-) I W7\u00E2\u0080\u00997fl \u00E2\u0080\u0098\I)i \u00E2\u0080\u0098 -\u00E2\u0080\u0094 - - \u00E2\u0080\u0094 - \u00E2\u0080\u0094- - \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 - \u00E2\u0080\u0094 - -/--_- - - - - \u00E2\u0080\u0094, \u00E2\u0080\u0098I\u00E2\u0080\u0099 - -\u00E2\u0080\u0098 \u00E2\u0080\u0098\u00E2\u0080\u0094 - -\u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094\u00E2\u0080\u0094, _,\u00E2\u0080\u0094\u00E2\u0080\u0094\u00E2\u0080\u0094. %_%d1\u00E2\u0080\u0094__.....- -------, ,\u00E2\u0080\u0098 \ I : 1\u00E2\u0080\u0098\u00E2\u0080\u0098Jt\u00E2\u0080\u0099/\u00E2\u0080\u0094 \u00E2\u0080\u0094\u00E2\u0080\u0094 I S \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 - I 5 S% \u00E2\u0080\u0094\u00E2\u0080\u0094 S___Figure A8 July minus June OLR (W m2) difference field for 1982with a contour interval of 5 W m2 (negative contours are dashed)0>, I- Ui.C00czo CCD\u00E2\u0080\u00980C-GO141I 1 \ I \u00E2\u0080\u0098 -\u00E2\u0080\u0098 / I I 1 \u00E2\u0080\u0094-) ?\u00E2\u0080\u0098 \31L\u00E2\u0080\u0099 1 ) 0 J J /,2r:-S\u00E2\u0080\u0094 \u00E2\u0080\u0094 S\u00E2\u0080\u0094 \u00E2\u0080\u0094\u00E2\u0080\u0094____i --J\u00E2\u0080\u0094-. \u00E2\u0080\u0098I,\u00E2\u0080\u0099 -Figure AlO July minus June OLR (W m2) difference field for 1984with a contour interval of 5 W m2 (negative contours are dashed)142Figure All July minus June OLR (W m2)difference field for 1985with a contour interval of 5 W rn-2 (negative contours are dashed)-/___...._____,1\u00E2\u0080\u0099\u00E2\u0080\u0099_i /\S\u00E2\u0080\u0098 \u00E2\u0080\u0098S\u00E2\u0080\u00985\u00E2\u0080\u0099, 55I - I04:\u00E2\u0080\u0099\u00E2\u0080\u0098 ,- \u00E2\u0080\u00945\u00E2\u0080\u0099143\:*I:Figure A12 July minus June OLR (W m2)difference field for 1986with a contour interval of 5 W m2 (negative contours are dashed)II UiCD CD-\u00E2\u0080\u00A2 CD \u00E2\u0080\u0098\u00E2\u0080\u0094CD CD 1jD CD CD c.1454\:1Figure A14 July minus June OLR (W m2)difference field for 1989with a contour interval of 5 W m2 (negative contours are dashed)crCDC)>CI\u00E2\u0080\u0094 UioI\u00E2\u0080\u0094CD o UiCDCDCDi-\u00E2\u0080\u0099CC)0C)CD."@en . "Thesis/Dissertation"@en . "North America"@en . "1995-05"@en . "10.14288/1.0088894"@en . "eng"@en . "Geography"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "The North American monsoon"@en . "Text"@en . "http://hdl.handle.net/2429/8953"@en .