RADIATIVE FORCING OF THE SOUTHWEST SUMMER MONSOON (A SATELLITE PERSPECTIVE) By Eric A. Smith Department of Atmospheric Science Colorado State University Fort Collins, Colorado DISSERTATION RADIATIVE FORCING OF THE SOUTHWEST SUMMER MONSOON (A SATELLITE PERSPECTIVE) Submitted by Eric A. Smith Department of Atmospheric Science In partial fulfillment of the requirements for the Degree of Doctor of Philosophy Colorado State University Fort Collins. Colorado Summer 1984 ~ Copyright by Eric A. Smith 1984 All Rights Reserved i COLORADO STATE UNIVERSITY ____ J_u_l~y __ 3~, ____ 19~ WE HEREBY RECOMMEND THAT THE THESIS PREPARED UNDER OUR SUPERVISION BY Eric A. Smith --------------------------------------------------------------- ENT I TLED ___ RAD ___ I...;;...A'--T_I~V_E_F_O"_R"_C'__I'__N'__G____'_O_F_T...;..;H_E__:S_O.....;;.U~T_HWE~S_T_S'__UMME----'-'=-R_M_O.;....N_S ...... O_O_N ____ _ (A SATELLITE PERSPECTIVE) BE ACCEPTED AS FULFILLING IN PART REQUIREMENTS FOR THE DEGREE OF Doctor of Philosophy Committee on Graduate Work Department Head ii ABSTRACT OF DISSERTATION RADIATIVE FORCING OF THE SOUTHWEST SUMMER MONSOON (A SATELLITE PERSPECTIVE) This investigation, presented in four parts (Chapters 2-5), examines the nature of radiative forcing within the Southwest Summer Monsoon. A short introduction is provided in Chapter 1. Chapter 2 explores the problem of converting narrow-field-of-view filtered radiance information, characteristic of operational weather satellite imaging measurements, to estimates of top-of-atmosphere radiation budget parameters. This chapter first provides a discussion of some of the earlier approaches to this problem and their shortcomings. A parameterization approach is then developed which addresses a portioo of the physics which dominates the relationships between spectral radiance and broad band flux. The parameterization is tested on two types of operational weather satellite data sets; the first is derived from the TIROS-N polar orbiting satellite. the second is derived from the GOES-l geosynchronous orbiting satellite. Both of these data sets were prepared in a navigationally-grided and calibrated fashion over a large region encompassing the Southwest Monsoon system in support of the 1979 Summer Monsoon Experiment (SMONEX). The radiation budget estimates are verified against wide-field-of­ view broad band radiation budget parameters measured by the Nimbus-7 Earth Radiation Budget (EBB) instrument. The GOES-1 data are also used in an intercompar1son with the diurnal radiation budget cycle over the 111 Arabian Desert Empty Quarter; the reference cycle is derived from a combination of aircraft (Convair-990) radiometer data and surface radiometer data, in conjunction with theoretical radiative transfer models needed for various transmission calculations. Chapter 2 concludes with a discussion of some of the scientific consequences that have arisen based on the use of radiation budget estimates derived from NOAA operational weather satellite data. Chapter 3 considers exchange processes within the 1979 Southwest Monsoon interpreted with the aid of satellite observations. To set the stage for this analysis, a review of previous research based on satellite measurements is provided. This includes a discussion of the various periodicities that have been identified in terms of monsoon fluctuations. Descriptions of the polar orbiting (TIROS-N) and geosynchronous (GOES-1) weather satellite data sets used in the investigation are provided. Two additional data sets derived from the radiation budget and passive microwave experiments on the Nimbus-7 experimental satellite are used to augment the research. Descriptions of these data sets are also provided. A general discussion of the techniques used to adapt spectral radiance measures obtained from weather satellites, in a radiation budget mode, is also given. First, the large scale radiative forcing of the monsoon is considered. The results are presented in the form of S-day averaged mean fields of the essential radiation budget parameters. These mean fields are used to facilitate a discussion of the phenomonological aspects of the evolution of the monsoon. Results are then presented in the form of time series of the zonal and meridional averages of the radiation budget, over the complete monsoon domain. The domain averaged iv time-latitude and time-longitude sections provide graphic evidence that the principal modulation in monsoon radiative forcing arises from a quasi-long period oscillation related to the systematic propagation of cloud bands out of the equatorial regions and into monsoon latitudes. It is shown that on the large monsoon-domain scale, the quasi-biweekly and shorter period oscillations are in evidence but lack clear organization. It is also shown how 'monsoon breaks' manifest themselves in the meridionally averaged sections of the net radiation parameter. Finally, in terms of large scale forcing, it is shown how in the course of a single month surrounding monsoon onset, the monsoon undergoes a systematic depletion of net radiative convergence into the monsoon, on the order of 5 peta-watts. Next, the contrasts in the monsoon radiation signals are presented, by way of time series, over various oceanic and continental regimes within the monsoon domain. It is pointed out how the shorter period oscillations appear to be much more robust at the local scale. The diurnal cycles over 12 localized regions within the monsoon domain are presented; these are also used to illustrate the large contrasts over the monsoon region on a 1o-day time scale. These contrasts are emphasized by the use of space and time gradients of the diurnal radiation cycle. Finally, the phenomenon of radiative decoupling is discussed. It is pointed out how cloud fields, which are known to 'decorrelate' the short and longwave radiative fluxes at the earth's surface, have a far more general property in which they also 'decorrelate' the top-of­ atmosphere fluxes. v Chapter 4 describes a research program and a methodology designed to investigate the surface energy budget parameters in remote desert and mountain environments. The objectives of this program are twofold: The first is to quantifY the surface energetics of two important heat source regions lying at the periphery of the Southwest Indian Monsoon. i.e. the Arabian Empty Quarter and the Qinghai-Xizang Plateau (Tibetan Plateau). The second is to compare and contrast the surface energetics of the Qinghai-Xizang Plateau with another major plateau system, the North American Plateau. The results of this program are intended to augment our understanding of the controls of desert heat sources (both lowland and highland) on the associated Southwest, East-ASian, and Rocky Mountain Monsoon systems. The characterization of the surface energy budgets is based on a measuring system designed to provide a continuous record of the essential components of surface heat exchange process. The monitoring systems are based on conventional atmospheric and soil sensors configured with microprocessor driven data logging electronics. The system has been designed for both portability and reliability. Sensor components are configured to operate in a virtually automated mode with minimal maintenance. The electronics have been designed to carry out the calibration transformations and most of the statistical processing on-site and to record the output data on conventional cassette tapes for use on simple home and office computer systems. Descriptions of measuring programs carried out in the deserts of Saudi Arabia and in the Rocky Mountain region are discussed. Characteristics of the surface energy budget in the Arabian Empty Quarter and its impact on the Arabian heat low are described. System vi tests carried out in the Rocky Mountain area were designed as feasibility trials in preparation for a new measurement program to be carried out on the Qinghai-Xizang Plateau. Examples of how surface measurements can be incorporated with satellite radiation data for describing bulk atmospheric heat exchange are shown. There are suggestions of how measuring programs similar to those described can be used to quantifY the surface energetics of 'spot coordinates' on the globe to support the problem of developing a global land surface climatology from remote platforms such as proposed by the International Satellite Land Surface Climatology Project. An investigation of the Arabian heat low is given in Chapter S. This research is supported by observations from various satellites, an experimental aircraft and a surface energy budget monitoring station. The observations confirm that the Arabian heat low is neither a radiation sink nor a total energy sink, disputing previous notions of desert heat lows. Top-of-atmosphere radiation budget analysis illustrates the high contrast properties of the radiative exchange fields over the southern Arabian Peninsula, with respect to its surroundings. However, an examination of a four-month time series of net radiative exchange over the Arabian Empty Quarter, indicates the heat low region is in slight radiative excess. Combining these results with estimates of the surface energy budget within the Arabian Empty Quarter and the radiative heating rate estimates of Ackerman and Cox (1982) within the heat low region, provides a relatively complete picture of the energy exchange process of the heat low. A synthesis of these results indicates that the heat low is a total energy source region. It is shown that the surface energy vii budget is in a relatively steady state in the daily mean, but undergoes dramatic diurnal variations and an occasional perturbation due to a moisture driven negative feedback process during intensification. A conceptual structure of the heat low is offered, based on a three layer stratification of the heating mechanisms. The possible role of the Arabian heat low in controlling thermodynamic conditions and forcing baroclinicity in the western Arabian Sea is discussed. It is concluded that the surplus energy properties of the heat low may serve as an important mechanism in controlling moisture transport into the monsoon rainfall regions. Chapter 6 is provided in order to summarize the findings of the four parts discussed above. viii Eric A. Smith Department of Atmospheric Science Colorado State University Fort Collins, CO 80523 Summer, 1984 ACKNOWLEDGEMENTS My deepest gratitude must go to Professor Verner Suomi whose vision was foremost in creating the field in which I have labored for eighteen years. The first half of that period was spent under his tutelage and it waa during that time that I gained my appreciation for both the advantages and limitations of meteorological satellite platforms. In the course of the research presented, I was often guided by the many practical lessons I gained from his wise counsel. I also acknowledge my major advisor for this dissertation# Professor Thomas Vonder Haar. The remainder of my Ph.D. Committee was composed of Professors Thomas Brubaker# Thomas McKee and Wayne Schubert. to whom I extend ~ appreciation. There are a number of others who have been of great help to me on this project. Drs. Graeme Stephens and Warren Wiscombe and Professor Steve Cox have been particularly helpful. I am deeply indebted to Mr. John Grafty for his dedication in completing the computational aspects of this project, well beyond the time he was a salaried employee. His assistance haa been absolutely essential. I also express my gratitude to Steve Ackerman, John Davis, Teizi Henmi# Leighton Klein, Dave Loranger# Chris Pasqua, Richard Peek, Dave Randel, Paul Rozenzweig and Charlie Wilkins. They have provided me with both direct assistance and the informal discussions essential to conducting applied research. I extend my warmest regards to Dr. Fawaz Alamy and Professor Marwan Sakkal of King Abdul-Aziz University, Mr. Abdul-Karim Heneidy of the ix Saudi Arabian General Directorate of Meteorology, and Major-Colonel Ali-Hasan El-Naimee, Commander of the Saudi Arabian Southern Force Army Group, whose assistance and hospitality made possible the collection of the Empty Quarter data in a truly formidable but fascinating environment. In addition, these latter acknowledgements would not be complete without mentioning the contributions of Professor Martin Fogle of the University of Arizona and Professor Jeffr~ Eighmy of the Colorado State University Anthropology Department, whose resourcefulness in setting up the Saudi Arabian project, necessary for carrying out desert research, should not go unnoticed. I extend my appreciation to Dr. Robert Fox of SSEC - University of Wisconsin for providing the GOES-l tape recordings, Gregg Hunolt of the National Environmental Satellite Data Information Service for providing the AVHRR data, and the Nimbus-7 ERB Science Team for providing the Nimbus-7 ERB data. I am most grateful to Mrs. Bonnie Grantham, Mrs. Susan Lini and Hiss Loretta Stevens for their excellent care in preparing the manuscript, to Ms. Judy Sorbie who provided her artistry in drafting many of the figures, and to Duane Barnhardt for providing photographic assistance. I cannot conclude this section without expressing my gratitude to my wife, Ms. Karen Greiner, whose encouragement, assistance and advice were instrumental in my efforts to complete this dissertation. This research has been supported under CID-ARMETED Project Contract CSU-SA-KUA-02 and National Science Foundation Grants ATM 78-20375 and ATM 82-00808. Part of the computations were performed at the National x Center for Atmospheric Research~ a division of the National Science Foundation. xi T ABL E OF C ONT ENTS 1.0 INTRODUCTION ••• 1 2.0 THE ESTIMATION OF RADIATION BUDGET PARAMETERS FROM WEATHER SATELLITE SPECTRAL RADIANCE MEASUREMENTS • 4 2.1 Background. • • • • • • • • • •• •• • • 4 2.2 Radiation Budget Data Archives • • • • • 8 2.3 An Ongoing Controversy. • • • •• • • 12 2.4 A Parameterization Approach for Transforming Narrow Band Radiance Measurements to Broad Band Flux Estimates. 20 2.4.1 2.4.2 2.4.3 2.4.4 2.4.5 2.4.6 2.4.7 Geometric Considerations. • . • • • • • Shortwave Spectrum and Bi-Directional Reflectance Normalization. • • • • • Longwave Spectrum and Limb/Path Darkening • Spectral Considerations ••••••••• Radiative Transfer Models • • • • • • • • Smoothed Functional Form Spectral Transformation Mod els • • • • • • • • • • • • • • • • • • • • • Direct Verification of the Spectral Transformation Functions. • • • 2.S Application of the Parameterization to Weather Satellite Data Sets • • • • • • • • • • • • • 2.6 Intercomparison of Weather Satellite Radiation Budget Estimates with Broad Band Measurements • 2.7 2.8 2.6.1 2.6.2 2.6.3 Nimbus-7 Inter comparison. • • • • • • • • • • • • • Aircraft-Surface Radiometer Intercomparison • Magnitude of the Geometric and Spectral Corrections • • • • • • • • • • • Implications for Cloud-Radiation-Climate Studies Using RADBUD Estimates from Weather Satellites Summary. • • • • • • • • • • • • • • 3.0 A MULTI-PLATFORM SATELLITE EXAMINATION OF RADIATIVE FORCING 22 22 29 32 35 41 60 63 68 70 83 89 93 97 WITHIN THE 1979 SOUTHWEST SUMMER MONSOON • . • • • •• 99 3.1 Background •••••••••• 103 3.1.1 Radiation Budget Studies. • 103 xii 3.1.2 3.1.3 3.1.4 3.1.5 Cloud Studies • • • • • • • Long Period Oscillations of the Monsoon • QuaSi-Biweekly Oscillations • • • • • Short Period Oscillations • 3.2 The Satellite Data Sets •••• 3.2.1 3.2.2 3.2.3 The TIROS-N Data Set. The GOES-1 Data Set • The Nimbus-7 ERB/NFOV Data Set. 3.3 Estimation of Radiative Fluxes. 3.3.1 3.3.2 3.3.3 3.3.4 3.3.5 3.3.6 Calibration • • • • • • Geometric Normalization Spectral Transformation • Radiation Budget Parameters • Valida tion. • • Shortcomings. • • • • • • • • 3.4 Idiosyncrasies of the Radiation Budget Calculations. 3.4.1 3.4.2 The Insolation Term • • • • • • Diurnal Variation and Directional Reflectance. • • • • • • • • • 3.5 Large Scale Forcing During the Monsoon Season. 3.6 3.5.1 3.5.2 3.5.3 Five Day Averaged Mean Fields • • • •• •• • • The Time-Latitude and Time-Longitude Sections. R adia tion Blocking. • • • •• ••• • Contrasts in the Radiation Budget. 3.6.1 3.6.2 3.6.3 Oceanic Regimes • • • • • • • • • • Continental Regimes • • • • • • • • Further Remarks on Validation 3.7 Oscillations in Radiative Forcing. 3.8 3.7.1 3.7.2 Regional Periodicities •••• Periodicities in the Zonal Averages • Diurnal Processes. • 3.8.1 3.8.2 Evolution of Dirunal Forcing. Diurnal Gradients in Time and Space xiii 105 107 108 113 115 115 127 148 153 153 1S7 159 160 161 163 163 164 172 180 236 241 252 262 262 267 272 275 279 284 284 286 300 3.9 On Radiative Decoupling •• 3.10 Summary. • •••••• 4.0 INVESTIGATION OF THE SURFACE ENERGY BUDGETS IN REMOTE DESERT 303 308 AND MOUNTAIN mVIR ONMENTS. • • • • • • • • • 311 4.1 Background. • • • • • • • • ••• 4.2 Monitoring Surface Energy Budget Processes From Remote Platforms • • • • • • 4.3 Description of the Present System Configuration. 4.3.1 4.3.2 4.3.3 The Radiation Station • The Tower Station • • • • • • Simultaneous Use of the Two Stations. 4.4 A Measurement Program in the Deserts of Saudi 312 317 321 324 333 337 Arabia • • • • • • • • • • • • • • • • • • 340 4.5 Tests at High Elevation at a Rocky Mountain Site. 365 4.6 A New Experiment on the Qinghai-Xizang Plateau. 383 4.7 Combining Surface Measurements with Satellite Data. 389 4.8 SllmIDary. • • •• •••••••• • •••••• 397 5.0 THE STRUCTURE AND ROLE OF THE ARABIAN HEAT LOW. 5 .1 Background. • • • • 5.2 Design of the Experiment. S.2.1 S.2.2 S.2.3 Data Sources. • • • • • • • Description of the Surface Station. Data Applications • • • 5.3 Energetics of the Arabian Heat Low. S.4 S.5 5.3.1 5.3.2 5.3.3 5.3.4 5.3.5 5.3.6 S.3.7 Large Scale Fields. • • Vertical Structure of the Heat Low. Radiative Properties of the Desert Surface. • Dynamical and Thermodynamical Surface Conditions. • Surface Temperature and Thermal Storage • • • • • • The Surface Energy Budget • The Radiation Budget •••.•.•• Possible Role of the Heat Low. Summary. • 6.0 CONCLUSIONS. 7.0 REFERENCES •• xiv 399 400 402 403 406 413 414 414 422 42S 434 438 4S2 460 464 470 478 491 TABLE Number TABLE 2.1 TABLE 2.2 TABLE 2.3 TABLE 2.4 TABLE 2.5 TABLE 2.6 TABLE 2.7 TABLE 3.1 TABLE 3.2 TABLE 3.3 TABLE 3.4 TABLE 3.5 TABLE 3 .. 6 LIST OF TABLE CAPTIONS Caption Summary of Shortwave RTE Model Physics. Summary of Longwave RTE Model Physics • Coefficients of the NB to BB Parameterization Equations. Minimum and Maximum EBB Temperatures (degrees Centigrade) of the Sinusoidal Temperature Waves Used to Discriminate Between Clear and Cloudy Conditions for Each of the Four Surface Type Categories • .. • • • • • • .. • • • • Components of the Experiments to Validate the Spectral Radiance to Broad Band Flux Transformations • • .. • • • .. • • Nimbus-7:TIROS-N Albedo Intercomparison Statistics (the bias and RMS differences are given in terms of; fractional albedo/percentage wrt mean) ...... . Nimbus-7:TIROS-N Infrared Emittance Intercomparison Statistics (the bias and RMS differences are given in terms of; LW flux/percentage wrt mean) ...... . Band Passes of the 4-Channel TIROS-N AVHRR Instrument. .. • • .. • • • • .. • • • • • • • TIROS-N AVHRR Missing Data Between May 1 and August 30,1979. . . . .........••...• GOES-l VISSR Data Tabulation For June. 1979 (x-full sector; s-short scan; m-missing sector) • Estimated Biases and Uncertainties in the TIROS-N AVHRR RADBUD Estimates. . .. . . .. .. .. . . . . Estimated Biases and Uncertainties in the GOES-l VISSR RADBUD Estimates. .. . . . . . Color Enhancement Table for Figure 48 .. .. .. .. xv 37 40 45 66 71 77 81 117 117 132 160 160 237 TABLE Number TABLE 3.7 TABLE 3.8 TABLE 3.9 TABLE 3.10 TABLE 3.11 TABLE 3.12 TABLE 3.13 TABLE 3.14 Caption Summary Statistics of SO Strips of Zonal Mean Albedo Over the Monsoon Domain Longitudes (300 -1000 E) • • • • • • • • • • • • • • • • • Summary Statistics of SO Strips of Zonal Mean Emitted Infrared Flux Over the Monsoon Domain o 0 Longitudes (30 -100 E) •••••••••••• Names and Location Coordinates of Five Regions within the Monsoon Domain Used to Study Radiation Budget Contrasts. • • • • • • • • • •• • • • Statistical Intercomparison Between the 79-Day Time Series of Nimbus-7 and TIROS-N RADBUD Parameters for Two Oceanic and Three Continental Regions • • Quality Index of the Nimbus-7 Versus TIROS-N Periodogram Intercomparisons. • • • • • • • • • Periodicities in the Net Radiation Term at the Regional Scale. • • • • • • • • • . • • Periodicities in the Net Radiation Term for SO Zonal Averages. • • • Three-Way Correlation Tabulation Between the AVHRR Retrievals of the Time Series of SO Zonal and 255 256 263 273 276 280 285 Meridional Averages • • • • • • • • • • • • • • • •• 306 xvi Figure Number Fig. 1.1. Fig. 2.1a. Fig. 2.1b. Fig. 2.1c. Fig. 2.2. Fig. 2.3. Fig. 2.4. Fig. 2.S. Fig. 2.6a. Fig. 2.6b. LIST OF FIGURE CAPTIONS Caption The Southwest Monsoon domain and its associated meteorological components [from Johnson and Houze (1984)] • • • • • • • • • • • • • • • . • The diurnal radiative structure of the tropical troposphere under suppressed and convectively enhanced cloud conditions. The top-of-atmosphere terms are based on a RADBUD parameterization applied to SMS-1 geosynchronous satellite measurements. The surface terms are based on GATE B-scale shipboard radiometers. The insets in the lower left corners represent ordinate magnifications of the net longwave radiation term to illustrate its diurnal or semi- diurnal characteristics • The same as Fig. 2.1a for stratus cloud and broken convective cloud conditions • • • • • • • • • Schematic of GATE tropospheric bulk radiative heat budget methodology and RADBUD notational key ••••• Conceptual diagram of the geometric-spectral parameterization scheme • • • • • • • • • • • Examples of bi-directional reflectance normalization models for ocean. desert. Himalayan range, and cloud. Limb darkening parameterization given in terms of measured window temperature (T ebb ) and the satelli te zeni th angle (a ) • • • • • • • • • • s Schematic illustration of the SW and LW radiative transfer model structure. Shortwave NB to BB relationship for cloud free water surface. Top part shows the statistical averages of the raw model calculations. Bottom part shows the smoothed functional form • • • • Same as Fig. 2.6a except for cloud covered water surface • • • • • • • • • • • • • • • • • • • • • xvii 3 16 17 18 21 28 31 38 46 47 Figure Number Fig. 2. 7a. Fig. 2. 7b. Fig. 2. Sa. Fig. 2.Sb. Fig. 2.9. Fig. 2.10a. Fig. 2.10b. Fig. 2.11. Fig. 2.12a. Fig. 2.12b. Caption Shortwave NB to BB relationship for cloud free desert surface. Top part utilizes PW as the ordinate with individual curves respresenting different ZA's. Bottom part reverses the roles of PW and ZA • • • Same as Fig. 2.7a except for cloud covered desert surface Same as Fig. 2.6a for cloud free semi-arid surface •• Same as Fig. 2.6b for cloud covered semi-arid surface. Note that in these two graphs the ordinate is PW. • • Shortwave NB to BB relationship for Himalayas. The top part is for a cloud-free surface; the bottom part is for a cloud-covered surface • • • • • • • • Longwave NB to BB relationship for clear case. These are smooth analytic representations for TM = 24SoK and 2S00 K • • • • • • • • • • • • • • • • • Continuation of Fig. 2.l0a for TM 2600 K • • • • • • • • • • • • Longwave NB to BB relationship for cloudy case. These are raw statistical composites for TM 22SoK and 23SoK . • • • . . • . • . • • . . • . . Longwave NB to BB relationship for cloudy cases. These are smooth functional representations for TM = 2lSoK and 22SoK. • • • • • • • • • • . . . . Continuation of Fig. 2.12a for TM and 24SoK • • • • • • • • = 23SoK Fig. 2.12c. Continuation of Fig.2.l2a for TM = 25SoK. Fig. 2.13a. Comparison of theoretically synthesized infrared NB to BE relationships in a clear atmosphere to observa­ tionally based relationships derived from satellite and aircraft measurements taken over tropical atmospheres • • • • • • • • • • Fig. 2.13b. Same as Fig. 2.13a for a cloudy atmosphere. Fig. 2.14a. Fig. 2.14b. Portrayal of the global surface type map based on the atlas of Edlin and Huxley (1973) Enlargement of the monsoon region • • • xviii 48 49 so 51 52 S3 54 S5 56 57 58 61 62 64 65 Figure Number Fig. 2.15. Fig" 2.16" Caption Schematic illustration of the application of the parameteriz ation procedur e over a 0.5 by o. S degr ee sub-grid domain containing an ensemble of cloud top heights . . . . . . . . . . . . . . . . . . . . . Nimbus-7 ERB-WFOV grid format over the Southwest Monsoon region. " Fig. 2.17. Schematic illustration of the weighting scheme used in the spatial averaging of the weather satellite RADBUD estimates to bring them into geographic spatial 67 73 correspondence with the Nimbus-7 ERB-WFOV targets 74 Fig. 2.18. Intercomparison between Nimbus-7 ERB-WFOV albedo measurements and TIROS-N AVHRR (channel 1) albedo estimates during May and June, 1979 over the Southwest Monsoon region. • • " • • • " • • • • • 7 S Fig. 2.19" Intercomparison between Nimbus-7 ERB-WFOV Daytime emitted IR measurements and TIROS-N AVHRR (channel 4) daytime emitted IR estimates during May and June, 1979 over the Southwest Monsoon region. 78 Fig" 2.20. Same as Fig. 2.19 for nighttime IR" " • • 79 Fig. 2.21. Same as Fig. 2.19 for combined day-night time IR. 80 Fig. 2.22. Schematic illustration of a multiple platform design for intercomparison experiments incorporating operational weather satellites, experimental satellites, aircraft. and surface stations. 84 Fig. 2.23. Comparison of GOES-l diurnal limb-corrected EBBT window temperatures with top-of-atmosphere EBBT model calculations incorporating actual surface temperature measurements to specify the lower boundary conditions. • • " • • " • • • • • " • • • • • " " •• 86 Fig. 2.24. Comparison of GOES-l diurnal IR emittance estimates with top-or-atmosphere IR flux model calculations incorporating the surface data to specify the lower boundary conditions. The graph also includes the upward and downward surface IR irradiance measurements as well as the model calculated downward surface irradiance • " • • " " • • • " • •• 87 xix Figure Number Fig. 2.25. Fig. 2.26. Fig. 2.27. Fig. 3.1. Fig. 3.2. Fig. 3.3a. Fig. 3.3b. Fig. 3.3c. Fig. 3.3d. Fig. 3.4. Fig. 3.5. Caption Comparison of GOES-l diurnal albedo estimates with top-of-atmosphere albedo model calculations incorporating the surface pyranometer data to specifY the lower reflectance boundary conditions. The graph also includes the measured surface albedo term, a theoretical clear sky transmittance term. an atmospheric transmittance corrected for aerosol extinction, and an effective atmospheric absorptance term. • • • • • • • • • • • • • • • Comparison of SMS-l daily averaged RADBUD estimates to estimates without geometric corrections (NO BDR; NO LD) and without both geometric and spectral corrections (NO BDRBB. NO LDBB). The RADBUD parameters represent GATE-Phase 3. B-scale averages of 0.5 by 0.5 degree retrievals. • • • • Similar to Fig. 2.26 except in this case the RADBUD time series represent diurnal averages as opposed to daily averages. The left portion of the figure represents supressed (cloud free) conditions; the right portion represents enhanced (cloud covered) conditions ••••••.••••• Degiction of the TIROS-N analysis region (Eq-3SoN; 30 E-l00oE). The numbers indicate the background type (blank-water; I-desert; 2-semi-arid; 3-Tibetan Plateau). • •• • ..•• Schematic illustration of the multi-orbital swath TIROS-N mapping scheme. • • • • • • • • • • TIROS-N AVHRR Channel 1 Arabian sector (daytime) on May 23, 1979 • Same as Fig. 3.3a for Indian sector TIROS-N AVHRR Channel 4 Indian sector (daytime) on May 23. 1979. • •••••• Same as Fig. 3.3c for Indian sector. Albedo (top) and shortwave absorbed flux (bottom) fields derived from TIROS-N AVHRR imagery on June 18. 1979. . . . .............. . Daytime (top) and nighttime (bottom) flux equivalent temperature fields derived from TIROS-N AVHRR imagery on June 18, 1979 ••••••••••••••• xx 88 90 91 118 119 120 121 122 123 125 126 Figure Number Fig. 3.6. Fig. 3.7a. Fig. 3. 7b. Fig. 3.70. Fig. 3. 7d. Fig. 3.7e. Fig. 3. 7f. Fig. 3.7g. Fig. 3.8. Fig. 3.9. Caption Degiction of the GOES-1 analysis region (300 S-40oN; 30 E-1000 E). The numbered regions are named below the map. These regions correspond to the geographic divisions used in the diurnal analysis The 2:00 GMT VIS (top) and IR (bottom) GOES-1 sectors for June 19, 1979 ••• Same as Fig. 3.7a for 5:00 GMT. Same as Fig. 3.7a for 8:00 GMT. Same as Fig. 3.7a for 11:00 GMT • Same as Fig. 3.7a for 14:00 GMT. The 17:00 and 20:00 GMT GOES-1 IR sectors for June 19, 1979. • • • •• • ••• Same as Fig. 3.7f for 23:00 GMT • Schematic illustration of the mapping scheme used for the GOES-1 processing Portrayal of the landmarks and coordinates used for the GOES-l navigation analysis [image is May 4, 1979 (day 124), 8:00 GMT] A. Socotra/Ra's Shu'ab 120 32'30"N, 53 0 19'00"E BI Madagascar/Nosy Lava 140 34'00"S, 47 0 36'30''E C. Sri Lanka Strait/Ramesvaram- 90 19'00"N, 790 18'30"E D. Bahrain/Ra's Al Barr 250 48'00"N, 500 35'00"E 129 134 135 136 137 138 139 140 142 EI KuwaitI Jazirat Faylakah - 29°24 '00 liN, 48°23 '30"E 143 Fig. 3.10a. Unmapped GOES-1 VIS image on May 1, 1979. 144 Fig. 3.10b. Same as Fig. 3.10a for the mapped version •• 145 Fig. 3111a. GOES-1 VISSR mapped Arabian sector (VIS channel) on May I. 1979 (=10:30 LT) • • • • •• ••••• 146 Fig. 3.11b. TIROS-N AVHRR mapped Arabian sector (VIS channel) on May 1, 1979 (=15:00 LT) ••• I • • • • • • . • . 147 xxi Figure Number Caption Fig. 3.12. Intercomparison between GOES-1 VISSR and TIROS-N AVHRR EBB temperatures for 4 days in June. An individual point represents the means for a 10 x1° box; the comparison regions extends from Eq-35 0 N; 55 0 -650 E. The GOES-1 sectors are at 11:00 GMT which corresponds closely to the 15:00 LT TIROS-N pass at the 600 E longitude • • • • • • • • • • • • • • • • • Fig. 3.13. Three month (June-July-August, 1979) global mean of the infrared emitted flux field derived from the Nimbus-7 ERB-NFOV measurements. • Fig. 3.14a. Same as Fig. 3.13 for December, 1979. Fig. 3.14b. Same as Fig. 3.13 for March, 1979 • Fig. 3.15. Fig. 3.16. Fig. 3.17. Fig. 3.18. Fig. 3.19. Fig. 3.20. Fig. 3.21. GOES-1 VIS geosynchronous satellite sector over the Arabian Sea; date is June 3, 1979; image time is 7:40 GMT. The aircraft mission track overlaid on the satellite image represents the third Arabian Sea mission flown out of Bombay by the NASA Convair-990 during SMONEX. . • • • • • The top photo is the instantaneous downward flux (K~) at TOA corresponding to the AVHRR composite image on June 18, 1979. The bottom photo is the derived instantaneous noontime total net flux field (Q*) based on the K~ source term • Five day averaged TOA solar irradiance field based on the instantaneous K~ calculations for the June 16-20 daytime compOsite AVHRR maps. The contour interval is 80 W·m • • • • • • • • . • • • • The meridionally averaged K~ time series based on the S-day average TOA solar irradiance map shown in Fig. 3.17 ••••••••••••••• The meridionally averaged K~ time series based on removing solar time dependence in the irradiance calculation • • • • • • • • • • • • • • • • • • • • • The zonal averaged K~ time series associated with solar time dependence (top) and no solar time dependence (bot tom). • • • • • • • . The emitted infrared flux fields for June 18, 1979. The top photo corresponds to daytime; the bottom photo corresponds to nighttime •.••.••••••• xxii 149 152 154 155 165 166 167 169 170 171 173 Figure Number Caption ~ Fig. 3.22. Directional reflectance functions for a water surface based on satellite, aircraft and surface observations and a theoretical plane surface Fresnel calculation. The sources are Ellis (1978), Griggs and Margraff (1967), Jacobowitz, et ale (1976), Minnis and Harrison (1984), Payne (1972), Raschke (1969), and Raschke, et ale (1973). 175 Fig. 3.23. Directional reflectance data for snow based on satellite, aircraft and surface observations. The sources are Bartman (1967), Dirmhirn and Eaton (1975), Griggs and Margraff (1967), Kondratyev (1969), Korff, et ale (1974), Paltridge and Platt (1976), Robinson (1958), Salmonson (1968), and Salmonson and Marlatt (1968). • • • • • • 176 Fig. 3.24. Directional reflectance data for ice based on surface observations of Korff, et ale (1974) and the reference book of Kondratyev (1969) • • 177 Fig. 3.25. Directional reflectance functions for cloud based on satellite data [the Raschke, et~. function incorporates aircraft and satellite data]. The sources are Ellis (1978), Raschke, et ale (1973), and Ruff, et ale (1968) • • • • • • • . . • • • 178 Fig. 3.26a. The May 1-6 (pentad 1) mean fields of shortwave ab­ sorbed radiation (K*) and total net radiation (Q*) over the TIROS-N SMONEX domain (Eq-300 N; 30o-100o E) The isopleths are in W.m-2 based on a 100 w.m-~ contour interval. • • • • • • • • • • • • • • • 182 Fig. 3.26b. The May 1-6 (pentad 1) mean fields of daytime emitted longwave radiation [Lf(day)] and nighttime emitted longwave radiation (Lf (n1§ht)] over the TIROS-N SMONEX domain (Eq-300 N; 30 -1000 E). The isopleths are in W.m-2 based on a 20 W.m-2 contour interval. • • • • • • • • • • • Fig. 3.27a. Same as Fig. 3.26a for May 7-11 (pentad 2). Fig. 3.27b. Same as Fig. 3.26b for May 7-11 (pentad 2). . . . Fig. 3.28a. Same as Fig. 3.26a for May 12-16 (pentad 3) Fig. 3.28b. Same as Fig. 3.26b for May 12-16 (pentad 3) • Fig. 3.29a. Same as Fig. 3.26a for May 17-21 (pentad 4) • xxiii 183 184 185 186 187 188 Figure Number Caption Fig. 3.29b. Same as Fig. 3.26b for May 17-21 (pentad 4) 189 Fig. 3.30a. Same as Fig. 3.26a for May 22-26 (pentad S) • 190 Fig. 3.30b. Same as Fig. 3.26b for May 22-26 (pentad S) 191 Fig. 3.31a. Same as Fig. 3.26a for May 27-31 (pentad 6) 192 Fig. 3.31b. Same as Fig. 3.26b for May 27-31 (pentad 6) 193 Fig.3.32a. Same as Fig. 3.26a for June 1-S (pentad 7). 194 Fig. 3.32b. Same as Fig. 3.26b for June 1-S (pentad 7) •• 195 Fig. 3.33a. Same as Fig. 3.26a for June 6-10 (pentad 8) • 196 Fig. 3.33b. Same as Fig. 3.26b for June 6-10 (pentad 8) 197 Fig. 3.34a. Same as Fig. 3.26a for June 11-15 (pentad 9). 198 Fig. 3.34b. Same as Fig. 3.26b for June II-IS (pentad 9) •• 199 Fig. 3.35a. Same as Fig. 3.26a for June 16-20 (pentad 10) 200 Fig. 3.35b. Same as Fig. 3.26b for June 16-20 (pentad 10) • 201 Fig. 3.36a. Same as Fig. 3.26a for June 21-25 (pentad 11) 202 Fig. 3.36b. Same as Fig. 3.26b for June 21-25 (pentad 11) 203 Fig. 3.37a. Same as Fig. 3.26a for June 26-30 (pentad 12) • 204 Fig. 3.37b. Same as Fig. 3.26b for June 26-30 (pentad 12) • 20S Fig. 3.38a. Same as Fig. 3.26a for July 1-5 (pentad 13) 206 Fig. 3.38b. Same as Fig. 3.26b for July 1-S (pentad 13) 207 Fig. 3.39a. Same as Fig. 3.26a for July 6-10 (pentad 14). 208 Fig. 3.39b. Same as Fig. 3.26b for July 6-10 (pentad 14) •• 209 Fig. 3.40a. Same as Fig. 3.26a for July 11-15 (pentad 15) • 210 Fig. 3.40b. Same as Fig. 3.26b for July 11-15 (pentad 15) • 211 Fig. 3.41a. Same as Fig. 3.26a for July 16-20 (pentad 16) 212 Fig. 3.41b. Same as Fig. 3.26b for July 16-20 (pentad 16) 213 xxiv Figure Number Caption Page Fig. 3.42a. Same as Fig. 3.26a for July 21-25 (pentad 17) . 214 Fig. 3.42b. Same as Fig. 3.26b for July 21-25 (pentad 17) . . . . 215 Fig. 3 .. 43a. Same as Fig. 3.26a for July 26-30 ( pentad 18) .. 216 Fig. 3. 43b. Same as Fig. 3.26b for July 26-30 (pentad 18) .. 217 Fig. 3.44a. Same as Fig. 3.26a for July 31-August 5 ( pentad 19) · 218 Fig. 3 .. 44.b Same as Fig. 3.26b for July 31-August (pentad 19) 219 Fig. 3.45a. Same as Fig. 3.26a for August 19-24 (pentad 23) 220 Fig. 3.45b. Same as Fig. 3.26b for August 19-24 (pentad 23) .. 221 Fig. 3.46a. Same as Fig. 3.26a for August 25-30 ( pentad 24) 222 Fig. 3.46b. Same as Fig. 3.26b for August 25-30 (pentad 24) .. 223 Fig. 3.47a. Mean monthly average of daily integrated net flux (0·) for May, 1979 based on ERB/NFOV data • 224 Fig. 3.47b. Same as Fig. 3.47a for June, 1979 • · .. · · 225 Fig. 3 .. 48a. Mean six-day average of daily integrated net flux (0·) from the ERB/NFOV measurements for May 7-12 (top-period 1) and May 13-18 (bottom-period 2) over the entire Eastern Hemisphere • · · · · · · · .. · . · 226 Fig. 3.48b. Same as Fig. 3.48a for May 19-24 (period 3) and May 24-30 (period 4) . . . · .. · · · · 227 Fig. 3.480. Same as Fig. 3.48a for May 31-June 5 (period 5) and June 5-10 (period 6) . . . .. · .. .. .. 228 Fig. 3.48d. Same as Fig. 3.48a for June 12-17 (period 7) and June 18-23 (period 8) .. · · · · .. · .. · 229 Fig .. 3.48e. Same as Fig. 3.48a for June 24-29 (period 9) and June 30-July 5 (period 10) · · · · · . .. .. .. 230 Fig. 3. 48f. Same as Fig. 3.48a for July 6-11 (period 11) and July 12-17 (period 12) .. · .. · .. · . .. 231 Fig. 3.48g. Same as Fig. 3.48a for July 18-23 (period 13) and July 24-29 (period 14) .. · · .. 232 Fig. 3.48h. Same as Fig. 3.48a for July 30-August 4 (period 15) and August 5-10 (period 16) . . .. · · · · · · . 233 xxv Figure Number Fig. 3.48i. Fig. 3.48j. Fig. 3.49a. Fig. 3.49b. Fig. 3.S0a. Fig. 3.S0b. Fig. 3.51. Fig. 3.52. Fig.3.S3a. Fig. 3.S3b. Fig. 3.54 • Caption Same as Fig. 3.48a for August 11-16 (period 17) and August 17-22 (period 18) •••• Same as Fig. 3.48a for August 23-28 (period 19) and August 29-September 3 (period 20) • • • • • Zonal average time series of albedo (0.5 degree resolution) over the TIROS-N SMONEX domain. • • Same as Fig. 3.49a for daytime flux equivalent temperature [EBBTt{day)] ••••••••••• Meridional average time series of albedo (0.5 degree resolution) over the TIROS-N SMONEX domain. • • Same as Fig. 3.S0a for daytime flux equivalent temperature [EBBTt(day)] •••••••••••• Pattern of solar zonal forcing to the SW Monsoon radiation budget. • • • • • • •• • ••.• Time-latitude section of noontime net radiative convergence (0.) into the monsoon domain. The contour interval 1s 100 Wt m-2 ••••.• Same as Fig. 3.52 for K. (top) and Albedo (bottom). Contour intervals are 100 Wem-2 and 1~ respectively ••••••••••••••• Same as Fig. 3.S3a for Lt(~~y) and Lt(night). Contour interval is 20 Wem for both diagrams •• Time-longitude section of noontime net radiative convergence (0.) into the m~2soon domain. The contour interval is 100 W·m •••••••••. Fig. 3.SSa. Same as Fig. 3.54 for K. (lef~~ and Albedo (right). Contour intervals are 100 Wtm and 10% respectively. • • Fig. 3.SSb. Same as Fig. 3.SSa for Lt(~~y) and Lt{night). Contour interval is 20 W-m for both diagrams. Fig. 3.56. Normalized cumulative frequency distribution functions for the SW Monsoon domain (Eq-30oN; 30o-l00o E) of the net radiative exchange term for the pentad periods during May, 1979 (the period represents a 6-day average) •••••• xxvi (Q.) first 234 235 242 243 244 245 246 247 249 250 251 253 254 257 Figure Number Fig. 3.57. Fig. 3.58. Caption Same as Fig. 3.56 for the six periods between May 27-31 and June 21-25. • ••••••• Same as Fig. 3.56 for the six periods between June 15-21 and July 16-20 •••••••• Fig. 3.59. Same as Fig. 3.56 for the six periods from July 16-20 and August 25-30 (there is a 13 day missing 258 259 data gap from August 6-18). • • • • • • • • 260 Fig. 3.60. Time series from May to August, 1979 of radiation budget parameters over the Arabian Sea based on ERB/NFOV measurEments • • • • • 264 Fig. 3.61. Same as Fig. 3.60 for the Bay of Bengal 265 Fig. 3.62. Same as Fig. 3.60 for the Indian Sub-continent. 268 Fig. 3.63. Same as Fig. 3.60 for the Tibetan Plateau. • • 269 Fig. 3.64. Same as Fig. 3.60 for the Arabian Empty Quarter. 270 Fig. 3.65. Intercomparison between the periodograms of Lt (day) for the79-day time series from Nimbus-7 and TIROS-N over the Bay of Bengal. • • . . • . • 277 Fig. 3.66a. Periodograms of 79-day (May 19-August 5) time series of tr. derived from Nimbus-7 for grid square over the Arabian Sea (left) and the Bay of Bengal (right) • • • • • • • • • • • • • • • • • • • • 281 Fig. 3.66b. Same as Fig. 3.66a for the Indian Sub-continent and the Tibetan Plateau • • • • • • • • • • • • 282 Fig. 3.66c. Same as Fig. 3.66a for the Arabian Empty Quarter. 283 Fig. 3.67. Fig. 3.68. Fig. 3.69. Fig. 3.70. Fig. 3.71. Fig. 3.72. Diurnal radiation budget over Arabian Sea based on GOES-1 measurements for three 10-day periods in June •• Same as Fig. 3.67 for Bay of Bengal Same as Fig. 3.67 for Indian Sub-continent Same as Fig. 3.67 for Western Tibetan Plateau. Same as Fig. 3.67 for Eastern Tibetan Plateau. Same as Fig. 3.67 for Arabian Peninsula ••••• xxvii 287 288 289 290 291 292 Figure Number Fig. 3.73. Fig. 3.74. Caption Same as Fig. 3.67 for East Africa ••• Same as Fig. 3.67 for Northern Deserts 293 294 Fig. 3.75. Same as Fig. 3.67 for South East Asia. 295 Fig. 3.76. Same as Fig. 3.67 for Indian Ocean (SoN-lOoS) • 296 Fig. 3.77. Same as Fig. 3.67 for Indian Ocean (lOoS-200S). 297 Fig. 3.78. Same as Fig. 3.67 for Indian Ocean (200S-300S). 298 Fig. 3.79. Temporal gradient of diurnal C* over Indian Sub- continent • • •• •••• 301 Fig. 3.80. Same as Fig. 3.79 for Arabian Sea. 302 Fig. 3.81. Spatial gradient of diurnal Q* between Indian Sub- conti nent and Arabian Sea • • • • • • • • • •• 304 Fig. 4.1. Schematic illustration of the Radiation Station. 322 Fig. 4.2. Schematic illustration of the Tower Station. • • 323 Fig. 4.3a. A Radiation Station deployed near the CSU Flight Facility •• • • • •• ••.• • • . • 325 Fig. 4.3b. Hand held soil probes (thermistors and soil blocks) and a rainguage situated some meters away from the main frame of the Radiation Station • • • • • •• 326 Fig. 4.4. A close up photograph of a CSI CR-7. The programming panel is on the right; the analog interface, A-D converter and I/O modules are exposed to illustrate the wiring configuration. A cassette tape recorder (not shown) is mounted on a plate on the left front side. The cable mounts are seen at the left on the outside of a fiberglass enclosure which includes a front panel that completely seals the electronics. • • • • • • •• 328 Fig. 4.Sa. A Tower Station deployed near the CSU Flight Facility. The foothills to the northwest of Fort Collins are seen in the background • • • • 335 Fig. 4.5b. The author and Professor Reiter are shown reprogramming the eddy flux data logger on the Tower Station • • • • • • • • • • • • • • • • • xxviii 336 Figure Number Caption Fig. 4.6. Wind, temperature, humidity, and soil parameters taken from May 6 to May 10, 1981 in the mountain city of Taif, located southeast of Mecca. on the western escarpment of the Arabian Peninsula • • • Fig. 4.7. Subsurface soil temperatures at six depths taken from May 18 to May 28, 1981 on top of a Barchane sand dune immediately to the west of Jeddah • • Fig. 4.8a. Preliminary measurements taken near the village of Sharouwrah, within the Arabian Empty Quarter, from June 1 to 5. 1981. The wind, temperature, humidity, and soil temperature measurements are plotted. At the top right of this figure is a phase-amplitude inset of the subsurface temperature waves • • • • • • Fig. 4.8b. The radiometer outputs in uncalibrated units (millivolts). The pyrgeometer signals are not yet adjusted for thermal drifts in the instrument and are thus not proportional to irradiance • • • • Fig. 4.9a. The top photograph shows the author activating the data loggers on the eve of the first Sharouwrah test (June 1, 1981). following a sand storm. The electronics were placed in a footlocker which was then partially buried in the sand and covered with a solar blanket to minimize solar heating damage to the electronics. The bottom photograph shows Professor Sakkal cleaning the upward radiometer domes the following morning. • Fig. 4.9b. Dune scenes from the Empty Quarter (north of Sharouwrah). In the bottom photograph, the project technician (Mr. Salah) is seen walking toward the measurement station. • • • • • • • • Fig. 4.10. METEOSAT image (VIS channel) of Arabian Peninsula indicating the four Saudi Arabian monitoring sites: 1) Jeddah, 2) Taif, 3) Najran. 4) Sharouwrah. The satellite image was kindly provided by Prof. Fig. 4.11. G. E. Hunt. . . .. • •...••..••• Half hourly data traces of air temperature and relative humidity during June, 1981 at the Sharouwrah site • • • • • • . • • • • • • • • • xxix 343 344 345 346 347 348 349 351 Figure Number Fig. 4.12. Fig. 4.13. Fig. 4.14. Fig. 4.15. Caption Half hourly radiative flux signals during June, 1981 at the Sharouwrah site. The definition of symbols is as follows: K~, Kt, K* are the downward, upward, and net solar fluxes L., Lt, L* are the associated infrared terms; finally Q* is the total net radiative flux. • • • • • • • • .• • • • • • The visible (VIS), near-infrared (NIR), and total solar (TOT) directional reflectance signals for June, 1981 at the Sharouwrah site ••••• Diurnal averages of the seven terms of the surface radiation budget (left) and the three-dimensional reflectance terms (right) • • • • • • • • • • • • • • Diurnal averages of the three principal components of the surface energy budget during June, 1981 at Sharouwrah. The definition of symbols is as follows: C* is the net radiative flux. S is the subsurface storage term, and SH is the bulk sensible heat term • • • • • • • • • • Fig. 4.16a. Summer (left part) and winter (right part) diurnal averages of directional radiative fluxes at Taif, Saudi Arabia. The one standard-deviation lines are 352 353 355 358 plotted along with the means. • • . • • • • • • 359 Fig. 4.16b. Same as Fig. 4.16a for the net radiative fluxes. 360 Fig. 4.16c. Same as Fig. 4.16a for the surface energy budget terms. The standard deviation lines are not plotted • • • • • • • • • • • • • • • • • • • 361 Fig. 4.17a. Midwinter (December-January) diurnal averages of directional radiative fluxes (left part) and suface energy budget components (right part) at Najran, Saudi Arabia. The one standard deviation lines are given for the directional radiative fluxes. • • 362 Fig. 4.17b. Same as Fig. 4.17a for late winter (February) . 363 Fig. 4.17c. Fig. 4.18. Same as Fig. 4.17a for summer (July) ••••• A TIROS-N image (channel 1:0.55-0.90 ~m) of the Indian Sub-continent and the Tibetan Plateau. The proposed monitoring site near Lhasa is indicated 364 with black arrows • • • • • • • • • • • • • • • • •• 367 xxx Figure Number Caption Fig. 4.19. Photographic illustration of the Pingree Park mountain site. The view is to the south; the Mummy Range is seen in the background. The Radiation Station is seen at the right; the Tower Station at th e 1 ef t . . . . . . . . • . . . . . . . . . Fig. 4.20. Time series of Radiation Station parameters from September 30 to October 6, 1983 at Pingree Park, CO • Fig. 4.21. Mean temperature runs from the 'Sampling Interval Test' held at Pingree Park on October 14, 1983 •••• Fig. 4.22. The same as Fig. 4.21 for mean relative humidity. Fig. 4.23. The same as Fig. 4.21 for mean vertical velocity ••• Fig. 4.24. The same as Fig. 4.21 for the correlation between vertical velocity and temperature • • • • • • • • Fig. 4.25. Mean temperature runs from the 'Integration Time Test' held at Pingree Park on October 27, 1983 •• Fig. 4.26. The same as Fig. 4.25 for mean vertical velocity. Fig. 4.27. The same as Fig. 4.25 for standard deviation of t anper atur e • • • •• •••••• • • • • • Fig. 4.28. The same as Fig. 4.25 for standard deviation of vertical velocity • . • . • • • • Fig. 4.29. The same as Fig. 4.25 for the correlation between vertical velocity and temperature • • • • • • • • Fig. 4.30. Diurnal averages of surface directional radiative fluxes (top) and surface net radiative fluxes (bottom) in the Gobi desert, April 8-14, 1984 • • Fig. 4.31. Same as Fig. 4.30 for surface directional reflectance and upward/downward equivalent black body Fig. 4.32. t emperatur es. ••• • • • • • • • • • • Photograph of the Himalayan Range near Srinagar, India taken from the NASA Convair-990 on June 11, 1979. . . . . . . . . . • . . . . . . . . . . . . xxxi 368 370 372 373 374 375 377 378 379 380 381 384 385 387 Figure Number Caption Page Fig. 4.33. Examples of GOES-1 (top) and TIROS-N (bottom) weather satellite imagery. The GOES-1 image is a VIS channel sector, taken at 11:00 LT on June 10, 1979. The flight track of a CV-990 differential heating mission (flown that day out of Dhahran, Saudi Arabia) is seen superimposed over the image. The TIROS-N image is a near-IR sector (channel 2:0.7-1.02 ~m) taken at 15:00 LT on June 20, 1979 • • • • • • • 391 Fig. 4.34. Net radiation map of the monsoon region for the period May 19-24, 1979 derived from the Nimbus-7 ERB-NFOV data. The color scheme is given below the figure. • • • • • • • • • • •• • • • • 392 Fig. 4.35. Time series of radiation budget terms (albedo, daytime and nighttime infrared emittance, net flux) over the Saudi Arabian Empty Quarter from May through August, 1979. • • • • • • • • • • 394 Fig. 4.36. Same as Fig. 4.35 for a region centered near the Tibetan city of Lhasa • • • • • 395 Fig. 4.37. Examples of TIROS-N channel 1 and GOES-l VIS imagery taken on the same day (May 1, 1979) at different times. The TIROS-N image is at approximately 15:00 LT, whereas the GOES-1 image is at approximately 12:30 LT. • •••• • • • • • • •• 396 Fig. 5.1. Schematic illustration of the observational platform network . . . . · . . . . . · · · · · · · · · · · Fig. S.2a. Mission map of the May 6, 1979 differential heating mission . · · · · · · · · Fig. S.2b. Same as Fig. S.2a for the May 9, 1979 regional energy budget mission. · · · · · · · · Fig. S.2c. Same as Fig. 5.2a for the May 14, 1979 differential hea ting mission · · · · · · · · · · Fig. 5.3. A visible METEOSAT image of the Arabian Peninsula with the four surface measurement sites indicated · · · 404 407 408 409 (1-Jeddah; 2-Taif; 3-Najran; 4-Sharouwrah). • • • 410 Fig. 5.4. Schematic illustration of the Sharouwrah surface energy budget station • • 412 xxxii Figure Number Fig. S. Sa. Fig. S.Sb. Fig. S.6a. Fig. S.6b. Fig. 5.7. Fig. S.8. Fig. S.9a. Fig. S.9b: Fig. S.10. Fig. 5.11. Fig. 5.12. Fig. 5.13. Caption Five day averaged instantaneous albedo fields (referenced to local noon) over the Southwest Monsoon region during 1979. Top of figure covers the June 6-10 period; bottom of figure covers the June 11-15 period. • • • • • • • • • • • • • • • • • • Same as Fig. S.Sa for the June 16-20 period (top) and June 21-25 period (bottom) ••••• Five day averaged emitted longwave flux fields (referenced to local noon) over the Southwest Monsoon region during 1979. Top of figure covers the June 6-10 period; bottom of figure covers the June 11-15 period • • • • • • • • • Same as Fig. 5.6a for the June 16-20 period (top) and June 21-25 period (bottom) •••••• Surface pressure field representative of mean July conditions over Southwest Monsoon region [from van de Boogaard (1977)] .•••••••••••• Streamline field at 8S0 mb representative of mean July conditions over Southwest Monsoon region [from Ramage and Raman (1972)] ••••••••••••• Five day averaged top-of-atmosphere net radiation fields (referenced to the Southwest Monsoon region during figure covers the June 6-10 period; covers the June II-IS period. • • • instantaneous net local noon) over 1979. Top of bottom of figure Same as Fig. 5.9a for the June 16-20 period (top) and the June 2i-2S period (bottom) •.••• Thermodynamic structure of the Arabian heat low based on the May 9, 10 and 12 Empty Quarter missions. Daytime and nighttime vertical motion profiles over the Arabian heat low region based on the kinematic analyses of Blake, et ale (1983) •••••••• Time series (15 minute sampling) of upward and downward solar fluxes (VIS, NIR, total) for the June, 1981 period at Sharouwrah, Saudi Arabia • • • • Same as Fig. 5.12 for longwave fluxes (total infrared spectrum) • • •• •• • • • • • • • • • • • • • • xxxiii 415 416 418 419 420 421 423 424 426 427 428 429 Figure Number Fig. 5.14. Fig. 5.15. Fig .. 5 .. 16 .. Fig .. 5.17. Fig. 5.18. Fig .. 5.19 .. Fig .. 5.20. Fig. 5.21. Fig. 5.22. Fig. 5.23. Fig. 5.24. Caption Diurnal surface radiation budget for June. 1981 at Sharouwrah, Saudi Arabia. The shortwave upward, downward and net terms are indicated by Kf , K~ , K*; the longwave upward, downward and net terms are indicated by Lt , L~ , L*; total net is indicated by Q*. • • Surface directional reflectance in the visible (VIS), near-infrared (NIR) and total solar (TOT) spectrums for June, 1981 at the Sharouwrah site ...... . Daily averaged surface reflectance in the visible, near-infrared and total solar spectrums for the June, 1981 period at the Sharouwrah si tee • • • .. .. • • .. • Diurnal averages of temperature and relative humidity during June, 1981 at Sharouwrah. Saudi Arabia • • Same as Fig .. 5.17 for wind direction. The thick dashed line is a smoothed representation. • Same as Fig. 5.17 for wind speed [AVE (V)], wind magnitude [AVE «u2 + v2)1/2)] and standard deviation deviation of wind speed .. • .. .. .. .. .. .. • .. • .. • • • Time series (15 minute sampling) of wind speed and magnitude during June, 1981 at Sharouwrah • • • Time series (15 minute sampling) of sub-surface temperatures (2, 20. 35 cm), air temperature. relative humidity, wind speed and magnitude, and wind direction from May 30 to June 1, 1981 at Sharouwrah, Saudi Arabia. The diagram in the upper right hand corner provides the amplitude (oC) and phase (local time) as a function of depth of the sub-surface thermal waves.. The small table in the upper left hand corner gives the peak amplitudes at each measuring depth (June, 1981) ......... .. Diurnal averages of sub-surface sand temperatures at three depths (2, 20, 35 cm) at the Sharouwrah site (June, 1981). .. • .. • • • • • • • • .. ..... Daily averaged soil temperatures at the three measurement depths at Sharouwrah (June, 1981) Time series (15 minute sampling) of soil temperatures at the three measurement depths at Sharouwrah (June. 432 433 435 436 437 439 440 441 443 444 1981) . . . . . . . . . .. . . • . . . . . . . . . .. 447 xxxiv Figure Number Fig. 5.25. Fig. 5.26. Fig. 5.27. Fig. 5.28. Fig. 5.29. Fig. 5.30. Fig. 5.31. Fig. 5.32. Fig. 5.33. Fig. 5.34. Fig. 5.35. Caption Comparison between the measured sub-surface sand temperatures (solid line) and the calulations from the analytic formulation (dotted line) ••••• Scattergram of the diurnal amplitude of surface temperature versus bulk Richardson number for the June, 1981 Sharouwrah data. • • • • • • • • • Time series (15 minute sampling) of the principal surface energy budget terms [sensible heat (SH); storage (S); net radiation (0*)] at Sharouwrah (June, 1981). • • • • •••• Diurnal averages of the three principal terms in the surface energy budget at Sharouwrah (June, 1981). Daily averaged surface energy budget process at Sharouwrah, Saudi Arabia during June, 1981. Time series (15 minute sampling) of the air temperature and relative humidity at Sharouwrah during June, 1981 • . • • • . • • . • • . • • • Diurnal processes of shortwave radiation exchange within the Arabian heat low representative of June conditions. • • • . . . . . . . . . . . . Daily time series of the radiation budget parameters over the Arabian heat low from May through August, 1981. These data were obtained from the Nimbus-7 ERB Narrow-Field-Of-View radiometers ••••••••• Net radiation field derived from the Nimbus 7 ERB-NFOV radiometers over the E~~tern Hemisphere from May 19-24, 1979 [KEY (W" m ): dark red to light red ranges from +140 to +80; dark brown to light yellow ranges from +80 to -10; white to grey ranges from -10 to -20; light green to dark green ranges from -20 to -140; light blue to dark blue ranges from -140 to -200]. •• ••••• Conceptual three layer structure of the Arabian heat low • • • Schematic illustration of the possible role of Arabian heat low in supporting the maintenance of Southwest Monsoon rainfall • • . • . • • • • xxxv 451 453 455 456 457 459 461 463 465 466 469 1.0 INTRODUCTION Much of what is now known about the terrestrial radiation budget process is based on satellite observations at the seasonal, annual and multi-annual time scales and at the large space scales out to the zonal average. Previous satellite investigations have mainly focused on the 'global radiation budget' and the required atmospheric and oceanic transports needed to balance the zonal radiation gradients imparted by the net radiative convergence term. This investigation is an attempt to concentrate within these scales and to investigate a feature of the global radiation budget, i.e., the 'Southwest Summer Monsoon System', which manifests itself as a significant anomaly in terms of mean top-of-atmosphere radiation exchange. The overall study is designed around the use of weather satellite data to reveal features of the monsoon that cannot readily be described by traditional wide-field-of-view radiation budget instruments. In so doing, there are various fundamental problems that arise. The foremost is the development and testing of a methodology which is consistent with utilizing weather satellite spectral radiance information in a 'radiation budget mode'. Chapter 2 of this investigation specifically addresses this problem. Based on the methodology and parameterizations developed in Chapter 2, Chapter 3 then goes on to examine the radiative exchange processes within the 1979 Southwest Monsoon. This monsoon was the subject of an 2 extensive observational program under the auspices of The 1979 Summer Monsoon Experiment (SMONEX), a component part of the global scale First GARP Global Experiment (FGGE) of 1979. Figure 1.1, provided by Johnson and Houze (1984), is used to illustrate the Southwest Summer Monsoon domain and to highlight what are now considered to be the foremost elements of the 'Southwest Summer Monsoon System'. In an attempt to focus in more detail on various key radiative elements of the monsoon, Chapter 4 proceeds to develop a strategy for monitoring monsoon heat source regions, specifically the Arabian heat low and the Tibetan elevated heat source. The monitoring approach incorporates the use of satellite. aircraft, and surface platforms which are used to describe the external boundary conditions essential in depicting the radiative structure of an atmospheric heat source feature. Finally, Chapter S explores in detail the Arabian heat low, based on the monitoring strategy outlined in Chapter 4. This chapter of the investigation is concerned specifically with the energetic structure and potential role of the Arabian heat low within the larger scale Southwest Monsoon. SAUDI ARABIAN HEAT LOW INDIAN OCEAN TIBETAN PlATEAU ~ HEAT SOURCE Fig. 1.1. The Southwest Monsoon domain and its associated meteorological components [from Johnson and Houze (1984)]. w 2.0 THE ESTIMATION OF RADIATION BUDGET PARAMETERS FROM WEATHER SATELLITE SPECTRAL MEASUREMENTS This investigation focuses on a historical problem in satellite meteorology concerned with the use of spectral radiance measurements, obtained from operational weather satellites. to estimate flux exchange across the upper boundary of the earth's atmosphere. This problem has its roots in the very early period of radiation budget science, during the time when satellite estimates of the global radiation budget could only be obtained from a somewhat patchwork collection of satellite radiation measurements taken over a variety of spectral and angular configurations. The early results of Vonder Haar (1968) and Vonder Haar and Suomi (1969. 1971) gave testimony to the fact that the earth's radiation budget could be monitored quite well with non-idealized sensors. Since the advent of the modern sun-synchronous polar orbiter and geosynchronous orbiter operational weather satellite programs (1970 and 1974 respectively), there has been a continuous interest in using data from these systems for radiation budget research. 2.1 Background There is an established literature addressing problems in radiation exchange, climate controls and cloud-circulation-radiation feedback which incorporates radiation budget estimates from both polar orbiting and geosynchronous meteorological satellites. These data have been applied to a variety of scientific problems. For example, the investigations of Heddinghaus and Krueger (1981), Ohring and Gruber (1983), Saunders and Hunt (1983), Short and Cahalan (1983), and Virji, s et ale (1982) have all focused on using weather satellite data. in a radiation budget mode, to map either global or regional scale radiation budgets or the correlative variables of cloudiness and rainfall. Short and Wallace (1980) have extended the use of longwave radiation budget estimates, derived from sun-synchronous polar orbiter satellite data, to examine 12 hour difference in cloudiness. Smith (1980b) has investigated the diurnal modulation of the radiation budget by cloud systems, using radiation budget estimates derived from geosynchronous satellite data. In the attempt to assess and validate a climate model of the 3- dimensional GCM type (in this case the ECMWF model), Geleyn, et ale (1982) have incorporated radiation budget estimates from operational polar orbiters. Along these lines, Ohring and Adler (1978) have employed the same data to assess the accuracy of a zonally averaged energy balance climate model (EBCH), incorporating simple dynamics, in Simulating the zonally averaged emitted infrared radiation budget term. Simmonds and Chidzey (1982) have also used these types of data to develop longwave emittance parameterizations for EBCM's. There have been a number of studies designed to determine the role of clouds on the radiation budget, i.e. the cloud sensitivity type studies motivated by the earlier works of Schneider (1972), Budyko (1974), and Cess (1976). The results of Cess and Ramanathan (1978), Cess, et ale (1982), Hartman and Short (1980), Ohring and Clapp (1980), Ohring, et ale (1981), and Potter, et ale (1981) have attempted to come to grips with the debate over cloud 'reciprocity' and the space and time scales associated with the cancellation effect. These studies are intended to provide the essential 'climate Signal' needed to interpret 6 the purely artificial cloud sensitivity experiments carried out in GeM environments; see e.g. Meleshko and Wetherald (1981). Schneider. et ale (1978). Shukla and Sud (1981). and Wetherald and Manabe (1980). Finally. in a fairly new branch of radiation budget science, various investigators have utilized radiation budget data obtained from weather satellites to try to quantifY the relationships between radiation budget parameters and measures of the general circulation. Jensenius. et ale (1978) have attempted estimating outgoing longwave radiation (OLR) at the top-of-atmosphere based on predictor variables diagnosed from the National Meteorological Center's Limited Fine Mesh Forecast Model. OLR estimates have been used by Liebmann and Hartmann (1982). Lyons (1981), and Weickmann (1983) to study various elements of the general circulation. Lau and Chan (1983a. 1983b) have used OLR estimates to investigate atmospheric teleconnections while Murakami (1980a. 1980b) has studied oscillations in the Winter Monsoon based on OLR estimates. The advantages of using weather satellite radiation budget (RADBUD) estimates over the traditional wide-field-of-view measurements are threefold: 1) the spatial resolution is much higher; 2) global coverage is near real-time, continuous, and the data is readily available; 3) and in the case of geosynchronous satellite data, diurnal processes can be examined. Thus, there is an ongoing need for parameterizat10ns which can be used to overcome the main disadvantage in using weather satellite measurements for proxy radiation budget estimates; i.e •• the fact that the measurements represent narrow-band (NB), narrow-field-of-view 7 (NFOV) radiances rather than the preferred broad-band (BB), wide-field­ of-view (WFOV) fluxes associated with hemispheric irradiances. The problem of transforming spectral radiance measures [both shortwave (SW) VIS and longwave (LW) IR terms] obtained from weather satellites to broad band flux estimates has. to date, been treated rather crudely. The foremost approach taken by various investigators has been the use of simplified quaSi-linear regression equations relating the narrow band parameters (independent variables) to the broad band parameters (dependent variables) [see e.g. Abel and Gruber (1979). Gruber, et ale (1983), Gube (1980, 1982a, 1982b», Ohring, et ale (1984), Smith and Yonder Haar (1980a), and Wydick (1983)]. This approach as applied to the NOAA scanning radiometer (SR) data (see Gruber and Winston, 1978) has led to various ambiguities particularly evident in the studies of Cess, et ale (1982) and Simmonds and Chidzey (1982). Although the regreSSion approach is indicative of the first order differences between the NB-NFOV and the BB-WFOV quantities, it does not address the essential physics leading to these differences. nor does it account for a variety of non-linear processes inherent to the transformation problem. For brevity, the problem can be divided into geometriC considerations (SW anisotropy and LW limb darkening), and spectral considerations (SW and LW narrow to broad band transformations), taking place in either clear or cloudy atmospheres. In the following investigation there are four basiC objectives. The first is to develop a thorough but practical parameterization procedure for transforming weather satellite measurements to radiation flux estimates, an approach which is intended to overcome many of the fallacies associated with the linear or quasi-linear type statistical 8 regression techniques. Second, the parameterization is tested on polar orbiting [TIROS-N/Advanced Very High Resolution Radiometer (AVHRR)] and geosynchronous [GOES-l/Visible-Infrared Spin Scan Radiometer (VISSR)] satellite data taken over the Southwest Monsoon region. Third, the accuracy of the radiation budget estimates are verified and assessed by intercomparison with a combination of satellite WFOV data (from the Nimbus 7 Earth Radiation Budget Experiment), aircraft radiometer data (from the NASA Convair-990). surface radiometer data (from a Saudi Arabian Desert Empty Quarter Site), and various theoretical calculations needed in the intercomparison analysis. Finally there is an attempt to resolve some of the controversy that has arisen associated with the use of weather satellite type radiation budget data. 2.2 Radiation Budget Data Archives The historical archives of earth radiation budget data consist of two basic types of products; those derived from broad band (BB), wide­ field of view (WFOV) flate plate radiometers specifically designed for global radiation budget monitoring, and those derived from narrow-band (NB), narrow-field of view (NFOV) collimated radiometers designed for pictorial weather system monitoring. The principal BB-WFOV data have been derived from a series of radiation budget experiments flown on the Nimbus-2, Nimbus-3, Nimbus-6, and Nimbus-7 NASA experimental satellites. All of these experiments incorporated broad band radiation detectors (the MRIR instruments on Nimbus-2 and 3 utilized four spectral infrared channels to integrate up to the total terrestrial spectrum). The principal NB-NFOV data have been derived from the scanning radiometers (SR) and advanced very high resolution radiometers (AVHRR) flown on the NOAA operational sun-synchronous polar orbiting satellite series. There 9 have also been sporadic attempts to retrieve RADBUD parameters from geosynchronous satellite imaging radiometers, particularly those flown on the U.S. GOES vehicles and the European METEOSAT vehicles. The basic designs and descriptions of the series of instruments flown on the Nimbus satellites can be found in the reports of Nimbus Project (1965), McCulloch (1969), W. L. Smith, et al. (1975, 1977) and Jacobowitz, et al. (1978). The further reports of ERB Science Team (1984) and Soule (1983) are helpful in reviewing the details of the Earth Radiation Budget (ERE) data and the sensors flown on the most recent earth radiation budget experiments, i.e. those of Nimbus-6 and Nimbus-7. The BB data sets themselves, and the scientific results of these experiments, have been described elsewhere. The results of Vander Haar (1968) and Vonder Haar and Suomi (1969, 1971), which suggested the earth was somewhat warmer and darker than suggested by earlier 'non-satellite' studies [see London and Sasamori (1971)], were essentially confirmed by the investigations of Raschke and Bandeen (1970), Raschke, et al. (1971), Raschke, et al. (1973), and Vander Haar, et al. (1972), based on the Nimbus-2 and Nimbus-3 data. The combined data sets from the earlier Vander Haar (1968) study were then combined with the Nimbus-2 and Nimbus-3 data by Vander Haar and Ellis (1974) in order to perform a series of studies. These later investigations considered the monthly, seasonal, and annual zonal averaged radiation budget climatology [Ellis and Vonder Haar (1976)], the required oceanic transport [Cort and Vonder Haar (1976)], the interannual variability of the global heat balance [Ellis, et al. 10 (1978)], and the effects of cloudiness on the global radiation budget [Ellis (1978)]. The Nimbus-6 experiment promoted a completely new generation of earth radiation budget studies. With the addition of a scanner which can be used to partially decompose the angular components of the radiative flux term [see Stowe (1983)], and a renewed interest in the mathematics of basic radiation budget principals, a series of theoretical papers came out of the university and NASA laboratories. The studies of Campbell and Vonder Haar (1978), Bess, et al. (1980, 1981), Green (1981, 1983), G. L. Smith and Green (1981), and G. L. Smith and Bess (1983), focused on the problem of enhancing the spatial resolution of w1de-field-of-view (WFOV) measurements through deconvolution techniques. King and Curran (1980) explored theoretically the possible differences in interpreting the albedo field due to anistropy. In applied studies, Jacobowitz, et al. (1979) reproduced almost exactly the earlier findings of the 'sixties' satellites with an 18 month Nimbus-6 satellite data set [approximately 31' global albedo and a 2540 K terrestrial equilibrium temperature]. Campbell and Vonder Haar (1980a) provided an updated climatological survey using a two year Nimbus-6 data set, and in a following report [Campbell and Vonder Haar (1980b)], went on to investigate the source of year-to-year differences using this data set (July 1975 to June 1977). Next, Campbell (1980) studied the transport problem on an annual scale, in terms of both the required equator to pole exchange and the more subtle east to west exchanges. In addition, the divergence of energy flux was divided into an oceanic component and a continental component; these results depicted 11 the annual cycle of winter inflow and summer outflow associated with continental radiative energy exchange. Stephens. et al. (1981) synthesized the Nimbus-6 period by directing their attention to the geographical distribution of the annual variability of the net flux after the removal of the semi-annual and annual cycles forced by the periodic variation in solar declination and distance. This investigation attempted to identifY the 'global action spots' principally responsible for inter-annual changes in the radiation budget climate. The regions identified were the equatorial western Pacific. the stratocumulus regime off the west coast of South America, the Sahara Desert. and notably. the Southwest-Asian Monsoon. The most recent measurements of the earth radiation budget now available from the Nimbus-7 ERB experiment have just begun to be scrutinized. Smith and Vonder Haar (1983) have examined the fluctuations in the Southwest Monsoon based on 6-day averaged fields of net radiation through the 1979 monsoon season. Randel (1983) has completed a study of the space-time variations in the global radiation budget (based on a one-year Nimbus-7 data set) with some emphasis on the hemispheric contrasts. This study also looks into the differences in the resultant albedo field based on whether wide-field-of-view or narrow-field-of-view (scanner) data were used. Arking and Vemury (1983) have suggested that the probable cause of these differences is the application of angular correction models to the scanner measurements. Based on the two most recent earth radiation budget experiments, broadband WFOV data have been available continuously since the Nimbus-6 launch (June, 1975) and Nimbus-7 launch (November, 1978), with the exception of a 4.5 month gap from July, 1978 to mid-November 1978. The 12 Earth Radiation Budget Experiment (ERBE) planned to start in the mid- 1980's is expected to continue the broadband WFOV time series on into the late 1980's; see Hall and Barkstrom (1981). The availability of continuous records of 'radiation budget estimates' based on operational weather satellites is confined to the Scanning Radiometer (SR) and Advanced High Resolution Radiometer (AVHRR) data sets of Winston, et ale (1979) and Gruber, et ale (1983). These data have been used extensively by a number of investigators as discussed in the preceeding section. Descriptions of the processing of these data are found in Gruber (1978), Gruber and Winston (1978), Gruber, et ale (1983), and Ohring and Gruber (1983). Background information on the SR and the newer AVHRR radiometers can be found in the reports of Fortuna and Hambrick (1979), Lauritson, et ale (1979), and Schwalb (1978). These so called NOAA data have been available almost continuously since June, 1974 with the exception of a ten month gap from March, 1978 to December, 1978 due to the changeover from the NOAA SR type instruments to the TIROS-N AVHRR type instruments. 2.3 An Ongoing Controversy The two types of data discussed above have been used for a variety of scientific purposes as indicated in the Section 2.1. In the course of these studies a pattern has been established in which the results and inferences of similar type investigations are often in dispute due to the type of data set used in the analysis. It goes without saying, of course, that scientific conclusions should not be based on the inaccuracies and biases inherent in the input data, nevertheless, the history of atmospheric science has been plagued with this problem, 13 particularly on the global scale. Satellite radiation budget science is no more immune from data set inconsistency than temperature profile science is immune from disparity between balloon and satellite temperature-moisture retrievals. However, there is one key aspect to the radiation budget controversy that should be kept in mind. It must be assumed at the outset that the BB-WFOV measurements are far more representative of broad band top-or-atmosphere fluxes than the NB-NFOV measurements, Simply due to the selective absorption properties of the atmosphere (the gas, aerosol, and clouds constituents) and the oft-time anisotropic nature of the atmospheric radiation field due to both atmospheriC constituents and the lower surface. The WFOV instruments are for the most part designed to make direct measurements of the desired radiation budget parameters. The errors inherent in a measurement are essentially those due to residual calibration problems which are presumably very small. On the other hand a NFOV measurement represents only a tiny slice of the overall geometric-spectral domain that constitutes a shortwave or longwave flux or irradiance measure. Thus there are large gaps (even after instrument calibration) between weather satellite radiance measures and top-of atmosphere flux estimates. It is in trying to fill these gaps that one would expect errors to arise and, in fact. it is the shortcomings in filling these gaps that have led to the various pitfalls radiation budget science has often found itself mired in. An important example of a discrepancy between data sets is discussed in the study of Cess, et ale (1982) in which they examined the cloud-radiation feedback problem in conjunction with a pair of WFOV data 14 sets and a NFOV data set. Their results indicated that the now familiar cloud sensitivity factor (0) given by: where & = aK./aF - aL,/aF c c = Absorbed Shortwave Radiation = OutgOing Longwave Radiation = Fractional Cloud Cover (2.1) and averaged on a global scale leads to a negative value(cloud albedo dominates) if NB-NFOV data are used. but leads to a positive value (cloud greenhouse effect dominates) if BB-WFOV data are used. This is a ludicrous contradiction and bodes ill for the future of climate modeling if it is not resolved. A second outstanding example of a data set discrepancy is found in a study by Simmonds and Chidzey (1982) in which they conclude that on a global-seasonal basis the NB-NFOV NOAA radiation budget data leads to a much improved parameterization over BB-WFOV Nimbus data in estimating longwave radiative emittance from surface temperature and cloud cover. This method of parameterization, of course, is part of the bedrock of the energy balance type climate models. There is an obvious explanation for this contradiction which is discussed in Section 2.7. A more subtle pitfall in radiation budget science is illustrated in a study by Geleyn, et al. (1982). They attempted to resolve inadequacies in the radiation component of an operational version of the European Center's Medium Range Forecast Model. They tested two radiation algorithms by intercomparing the zonal annual averages of planetary albedo and IR emittance with the 1976 NOAA RADBUD data composited in the same fashion. There is wisdom in this approach up to some point, however, as pOinted out in Ohring and Gruber(1983) there are 15 still significant inconsistencies remaining in the zonal annual averages between the NOAA NB-NFOV data and the NIMBUS BB-WFOV data. Furthermore. zonal-annual averages serve to obscure the even larger inconsistencies on the more localized space and time scales. Until these inconsistencies are resolved it is questionable as to how much tuning should be done to the radiation algorithms in weather forecasting or GeM models based on weather satellite RADBUD parameters. A final example is based on this author's investigation (Smith. 1980b) of the contrast in the diurnal radiative structure between a suppressed tropical atmosphere. an atmosphere containing stratus cloud. an atmosphere containing moderate broken convection. and a heavily convectively enhanced atmosphere (see Fig. 2.1). This investigation was carried out over the 1974 GATE B-Scale Array in the tropical eastern Atlantic. Bulk tropospheric radiation budget estimates were derived by applying some fairly simple geometriC and spectral transformation schemes to SMS-l geosynchronous satellite radiance measurements and combining these estimates with ship board radiometer flux measurements; see part c of Fig. 2.1 for a schematic illustration of the method (surface flux parameterizations were not employed for any of these results). The results for the four types of cloud regimes are illustrated in parts a and b of Fig. 2.1. Whereas there are some obvious and dramatic differences in the magnitudes and diurnal asymmetrics of the various radiative terms indicative of the real atmosphere (note specifically the net atmospheric radiation terms which have been stipled), there are also some questionable results. A close examination of the solar radiation convergence under convectively enhanced conditions, for example. BULK ATMOSPHERIC RADIATIVE HEATING BUDGETS IH GAT[ REGION DERIVED FROM SMS-l SATELLITE AND SURFACE (SHIP) RADIOMETFRS B-SCALE DIURNAL BUDGET B-SCALE DIURNAL BUDGET SUPPRESSED CONDITIONS ENHANCED CONVECTION PHASE 3 CLOUD CLASSIFICATION PHASE 3 CLOUD CLASSIFICATION 1200r 0 10 1200 ·A- r " O·A- O·s -- I \ 9 I \ Q·s --1000 ~ 0. 0 ••••• I \ 8 1000 Q·o ••••• K·A -- .. \ . . K"A -- / -" I . 0 \ 7 8001- K· S -- . . I \ I . 0 \ 800 K·S --. 0 I \ K· 0 - -- f . . , 6 K· 0 - --. 0 I ••• ·0. \ I . . , 6001- Lo A - . 0 0 \ 5 600 Lo A - I . 0 \ I . 0 0 L· s ----- I : : \ Los ----- / . . \ 4 I : . 400~ L· o -- I : . \ .. \ 3 ... 400 L·O -- I : . \ >. 0 f' 200 2 ~ f' 200 ~ E I 0 E ~ ~ 0 o w 0 x ~ x :) -I a::: :) ...J ...J LL_200 -2 ~ LL -200 -3 ~ w -400 - -4 I -5 -GOO -I -6 -800 ~A' •• I~ ) __ '._ \.-..... ~I 1-7 -800 - 1-8 -ICX)()' 21-24 -ICX)() 9-12 12-15 15-18 18-21 0-3 3-6 6-9 9-12 12-15 15-18 18-21 0-3 3-6 6-9 DIURNAL TIME PERIOD DIURNAL TIME PERIOD 10 9 B 7 5 4 3 ... >. 0 2 ~ ~ 0 o w ~ -I a::: -2 ~ -3 ti w -4 I -5 -6 -7 -8 21-24 Fig. 2.1a. The diurnal radiative structure of the tropical troposphere under suppressed and convectively enhanced cloud conditions. The top-of-atmosphere terms are based on a RADBUD parameterization applied to SMS-l geosynchronous satellite measurements. The surface terms are based on GATE B-scale shipboard radiometers. The insets in the lower left corners represent ordinate magnifications of the net longwave radiation term to illustrate its diurnal or semi-diurnal characteristics. .... 0'\ B-SCALE DIURNAL BUDGE T B-SCALE DIURNAL BUDGET STRATUS CLOUD CONDITIONS BROKEN CONVECTION PHASE 3 CLOUD CLASSIFICATION ALL PHASE CLOUD CLASSIFICATION 1200r a 10 1200 10 °A- QOA- aos -- 9 aos -- 9 1000 ~ Qoo ••••• I , 8 1000 o· 0 ••••• , 8 I \ / KOA -- \ KOA -- / \ \ 7 \ 7 8001- Kos -- ... \ 800 Kos -- \ , . . I •••• -. K- 0 - -- I . ° 6 K· 0 - -- . \ 6 . . . I . . \ I • . \ 600~ L· A - I . • I 5 600 L· A -- . 5 , . . \ I : 0. \ Los ----- . . . L· s ----- I : .. \ ' . . 4 4 400~ Loo-- '. ~.\ L· o -- . .. \ J • •• \ 3 .. 400 I : 3 "j" >- I : ~ 0 I • 0 2 "'0 f' 200 I . 2 '0 ~ 200 ~ ~ E D E I 0 .... ~ ~ ...,J a a w a a w x ~ x ~ ::::> -I a: ::::> ~ -I 0:: ....J ....J lL.._200 lL.._200 -2 ~ -2 ~ -3 ~ \\ -3 ~ -400 1-1701- '\. " If I w -400 '\ 1/ w , .. -4 I ~ \ ..... {7 -4 I -600~ " ------ ~-5 -600 -5 -6 - "- -6 -800 l-z~ I I I I I I J-7 -800 - I I I I I I • I 0-3 Hi 6-9 9-12 ~I~ &-e 0,3 3-6 6-9 9-12 Il-I!\ IH5 &21 21,24 -7 -8 -8 -K)()() I I -ICIOO 0-3 3-6 6-9 9-12 12-15 15-18 18-21 21-24 0-3 3-6 6-9 9-12 12-15 15-18 18-21 21-24 DIURNAL TIME PERIOD DIURNAL TIME PERIOD Fig. 2.lb. The same as Fig. 20la for stratus cloud and broken convective cloud conditions. 18 BULK ATMOSPHERIC BUDGETS RADIATION SATELLITE - SHIP SATELLITE- PARAMETERIZATION -1 r--1 ~ I 1 ___ It ___ -----0---- t, I ... t I ..-----------....... ( PARAMETERIZATION) --- S ~-:-.- ... -.--:-:--.. -:-;-:-.. :~ NOTATION As • SURFACE ALBEDO K~s,Kts,K· s • SURFACE SHORTWAVE UP, DOWN, NET Ao • TOP ALBEDO Kto,Kto,K*o • TOP SHORTWAVE UP,DOWN,NET K * A • ATMOSPHERIC SHORTWAVE NET Lts,Lt s,L· s • SURFACE LONGWAVE UP, DOWN, NET L t o,L* 0 iii TOP LONGWAVE UP, NET LeA • ATMOSPHERIC LONGWAVE NET a·s • SURFACE TOTAL NET 0*0 • TOP TOTAL NET a*A • ATMOSPHERIC TOTAL NET Fig. 2.1c. Schematic of GATE tropospheric bulk radiative heat budget methodology and RADBUD notational key. 19 indicates total solar absorption of over 4~. Although there is an ongoing debate as to the magnitude and cause of anomalous in-cloud shortwave absorption [observations exceed theory by up to IS'; see Welch, et ale (1980)] a magnitude of greater than 4~ is somewhat beyond the acceptable maximum observed levels of 35%. What all the above examples have in common is quite straightforward. The radiation budget parameters derived from the weather satellites are known to be only ballpark estimates. In the case of the NOAA data, the planetary albedo is derived by assuming no spectral difference between the radiometer band pass and the total solar spectrum, as well as assuming isotropy in calculating directional reflectances. Broad band longwave flux is estimated by assuming a quasi-linear relationship between the equivalent black body window temperature (T NB ) and the equivalent black body terrestrial temperature (T BB ), i.e.: (2.2) where the Wi'S are empirical coefficients and w2 is very small. In the Smith (1980b) analysis shortwave anisotropy was considered however the spectral transformations were fairly elementary [see also Gube (1982a), and Saunders, et ale (1983)]. Thus there are known shortcomings in the radiation budget data derived from weather satellite information. Much of the physics of the atmosphere so strictly adhered to in theoretical radiative transfer applications have been virtually ignored. It is essential, therefore, to place qualifiers on many of the weather satellite RADBUD studies as the input data have been indirect estimates with uncertain error bars. 20 2.4 A Parameterization Approach for Transforming Narrow Band Radiance Measurements to Broad Band Flux Estimates Transforming radiance measures to flux estimates can be considered a two-fold problem. The first deals with geometric relationships; the second with spectral relationships. In addition, both of these topics separate into either shortwave or longwave considerations. The longwave process is primarily one of emission and thus requires a different modeling approach than the shortwave process, which is partly a problem in absorption and partly a problem in reflection and scattering. Furthermore, the parameterizations are complicated by the fundamentally different radiative processes that occur in clear and cloudy atmospheres. Therefore, to place this general problem of spectral radiance to broad band flux transformation within a conceptual framework, it is described in terms of a three order parameterization problem involving two radiative processes, two radiative spectrums, and two media: Radiative Spectrum Radiative Process Geometric Spectral radiative radiative medium medium Shortwave a) clear a) clear b) cloudy b) cloudy Longwave a) clear a) clear b) cloudy b) cloudy Figure 2.2 illustrates the proposed transformation procedure in a schematic fashion. In this figure the X and ~ functions indicate shortwave and longwave geometric transformations respectively. The nsw and ~lw functions indicate the shortwave and longwave spectral transformations. GEOMETRIC CONSIDERATIONS Correction for Anisotropy ~ Angular Configuration and Surface Dependence CLOUDY ATMOSPHERE SHORTWAVE CLEAR ATMOSPHERE PROPOSED SOLUTION SPECTRAL COiiS I DERAIl ONS CLEAR ATMOSPHERE LONGWAVE GEOMETRIC CONSIDERATIONS Correction for Limb Darkening and Hemispheric Path Contributions J.. View Angle and Height Dependence CLOUDY ATMOSPHERE nCLR (PW T T > II LW 'atm' sur CLD - -fl LW (PW, Tatm , Tsur> I1CLD (PW a ,surface) SW ' 0 "CLR (PW a , surface) II SW '0 IN GENERAL THESE TRANSFORMATIONS CAN BE APPLIED IN A COMMUTATIVE FASHION SINCE THEY ARE DESIGNED TO BE SEPARABLE FUNCTIONS Fig. 2.2. Conceptual diagram of the geometric-spectral parameterization scheme. t-) ..... 22 2.4.1 Geometric Considerations In the classic treatment of the shortwave geometrical transformation problem, bi-directional reflectance normalization models are used to correct natural occurring anisotropy implicit in the raw radiance measurements. In the infrared spectrum, limb darkening models are used to correct any path length induced biases in the raw radiance measurements. For this parameterization experiment models have been constructed for both the shortwave spectrum and the longwave spectrum based on current data sets and synthetic radiative transfer calculations. 2.4.2 Shortwave Spectrum and Bi-Directional Reflectance Normalization Historically, bi-directional reflectance normalization models, devised for satellite radiation budget applications, have been designed in two ways. The first approach has been to describe a particular terrestrial surface (e.g. water, desert, or cloud) as an anisotropic reflector with invariant properties. This 'surface dependent' approach requires the generation of 'statistically stable' radiance fields describing the reflection process over the relevant set of earth surfaces occurring in the analysis domain. The requirement of 'statistical stability' suggests that this method of solution is affected by 'statistical noise'. That is relevant if the radiance fields are generated from actual measurements (satellite or aircraft) over a set of solar and observer zenith angles, and the associated relative azimuth angles. On the other hand, the radiance fields may be described synthetically by the application of intenSity-form radiative transfer models, which can be used to produce smooth fields. However, in this 23 case, the 'statistical stability' requirement is implicity embodied in the model assumptions, which must be stipulated very realistically if the resultant bi-directional reflectance normalization models are to have any validity. In general, it is not possible to define realistic reflective surface boundary conditions for natural occurring earth surfaces such as forest canopies or mixed snow-ice fields. That is, most naturally occurring earth surfaces have such highly complex topographic structure and inhomogeneous compositions that theoretical treatment is ruled out apriori. There are exceptions, such as the reflection properties of a still water surface (Fresnel reflection), however, very little of the earth's surface can be characterized by ideal reflectors, such as still water ponds or perfectly flat and uniform sand deserts. Thus, 'empirical' or measurement based models are frequently used and thus the concern with 'statistical stability'. It 1s very important in the generation of empirically derived bi­ directional reflectance normalization models, that the data sample sizes associated with the mean fields are sufficient to characterize the 'average' reflection properties of the surfaces. There are a variety of factors that perturb the average radiative appearance of a surface. These include composition changes (e.g. vegetative cycles), topographic changes (such as induced by variable wind fields), and general inhomogeneities (e.g. the highly variable nature of cloud top surfaces). Therefore, surface invariant models are intrinsically related to time and space scales, insofar as their validity in representing average reflection characteristics. 24 The second approach for generating bi-directional reflection models is best referred to as a 'climatological dependent' method. Here, the radiance fields over discrete regions of the globe are characterized for given periods of time by angular models describing the average spatial and temporal reflectance conditions. This method attempts to average the effects of the underlying surfaces and the overlying atmosphere, for all meteorological conditions (clear and/or cloudy), during the specified time period. This methodology has been proposed for application with the upcoming Earth Radiation Budget Experiment (ERBE) based on composited radiance data from the Nimbus-7 ERE scanning instrument. The main advantage of this approach is that it circumvents the problem of having to discriminate what type of surface is being viewed (e.g. a warm land surface versus a low warm stratus cloud deck). The main disadvantages of this approach are 1) that the time/space scales of model utilization must conform to the time/space scales over which the models were generated originally; and 2) cloud situations are not discriminated from clear conditions. In addition there is a tacit assumption that angular biases cannot be overcome in the day-to-day treatment of radiance measurements. It is important to note at this point that bi-directional reflectance normalization models are dependent upon the amount of atmosphere overlying the surface in question. There are two extreme cases; the top-of-atmosphere case and the directly-above-surface case. Although bi-directional normalization reflectance models can be defined at any level in-between these two extremes, they are of limited interest. In general, top-of-atmosphere models are useful in transforming remotely sensed radiance measurements to flux quantities; 2S thus correcting for naturally occurring anisotropy. Directly-above- surface models are useful when specifYing the lower surface boundary condition for multi-stream radiative transfer models. An interesting theoretical problem concerns the retrieval of a surface bi-directional reflectance normalization model from a model specified at some arbitrary level above the surface (including above the top of atmosphere). This is an inversion problem in which the effect of the atmosphere must be removed so as to arrive at the actual boundary condition. This problem can usually be solved iteratively, using an isotropic surface condition as an initial guess. For this investigation, since the concern is with high time and space resolution computations (e.g. diurnal variations at a 1/2 degree grid scale), the requirement is for 'surface dependent' models. These have been constructed from the composited data of Minnis and Harrison (1984), and Davis and Cox (1981, 1982). Bi-directional reflectance normalization models are applied for four individual earth surfaces and for clouds. The four earth surface types and their associated geographic domains are as follows: 1. Water (Arabian Sea, Bay of Bengal, Indian Ocean) 2. Desert (Arabian Peninsula, Southern Asia East Africa) 3. Semi-Arid Continent (East African Highlands, Southern ASia, Indian Sub-continent) 4. Partially Snow Covered Mountain Range (Himalayas) The decision to select only 4 surface models and a single cloud model was dictated by the availability of data sets over the Southwest Summer Monsoon region characterizing the angular variability of the 4 types of surfaces. The Minnis and Harrison (1984) composites of GOES Satellite VIS channel data have been used to specifY angular models for. 26 water and cloud. Davis in his Ph.D. dissertation [Davis and Cox (1981)], composited CV-990 bugeye data, for the Arabian desert, the Himalayas, and parts of the Indian Sub-continent. The Indian model is referred to as the semi-arid case for purposes of this investigation. These data were taken prior to the monsoon and thus prior to the burst of vegetation that evolves in July. The models are given as a function of solar zenith angle (9 ), o satellite zenith angle (9s )' and sun-satellite relative azimuth angle (fr ). A bi-directional reflectance normalization coefficient is expressed as: (2.3) where i indicates the surface type category. The X's represent ratios of integrated directional reflectance factors [(R i (9 0 )] to bi-directional reflectance values [Pi(9 0 ,Ss,tr )] multiplied by the isotropic scale factor (n), i.e.: Thus any particular spectral radiance [N4A (9 0 ' Ss' t r )] can be transformed to a spectral flux estimate [F4A (9 0 )] by utilizing the appropriate bi-directional reflectance normalization coefficient in conjunction with the isotropic scale factor, i.e.: (2.4) (2.S) Now it is easily seen that X is a measure of how much to increase or reduce a particular isotropic flux estimate (n·N) based on the bias due to angular orientation. That is, if the radiance is larger than the isotropic magnitude (e.g. in a forward scatter peak), the X factor 27 would be less than 1.0 so as to decrease the isotropic estimate. If the radiance is smaller than the isotropic magnitude (e.g. in a side lobe), the factor would be greater than 1.0 so as to increase the isotropic estimate. It should be noted that a graph of l/X would then represent the invariant angular radiative appearance of a model surface. In Fig. 2.3 contoured diagrams of l/X fields have been provided for various of the model surfaces. at discrete solar zenith angles. The key feature to note in these diagrams is that the ocean model emb