Remote Sensing of Global Surface Shortwave Radiation and PAR l)ver the Ocean: a Sequoia Testbed by Catherine Gautier Michael Byers SEQUOIA 2000 Technical Report 94142 ** No page found ** Remote Sensing of Global Surface Shortwave Radiation and PAR Over the Ocean: a Sequoia Testbed by Catherine Gautier Michael Byers Report #94142 January l 994 SEQUOIA 2000 Technical Report 557 Evans Hall Berkeley, California 94720 (415) 642-4662 ** No page found ** SEQUIOA 2000 TECHNICAL REPORT Remote Sensing of Global Surface Shortwave Radiation and PAR Over the Ocean: a Sequoia Testbed Catherine Gautier Earth Space Research Croup Computer Systems Laboratory University of California SantaBarbara,CA 93106 and Michael Byers Santa Barbara Research Center 75 Coromar Drive Coleta, California 93117 ** No page found ** ABSTRACI During the past few years many methods have been proposed for estimating surface radiative fluxes (shortwave radiation, Photosynthetically Active Radiation - PAR) from satellite observations. We have developed algorithms for computing the shortwave radiative flux (shortwave irradiance) at the ocean surface from visible radiance observations and they have been found to be quite successful under most atmospheric and cloud conditions. For broken clouds, however, the simple plane parallel assumption for solving the radiative transfer equations may need to be corrected to account for cloud geometry. The estimation of PAR is simpler because the most commonly used satellite radiance measurements cover a similar region of the solar spectrum. We are in the process of producing global llSW and PAR as a contribution to the Sequoia 2000 project (to implement a distributed processing system designed for the needs of global change researchers). Results from our algorithms developed for Sequoia and preliminary global surface solar irradiance and PAR fields will be presented and discussed. 1. Introduction An accurate knowledge of the sun's influence upon the earth's biological, physical, and chemical processes and the radiative response of the surface- atmosphere system are vital for a complete understanding of the earth system's state and evolution. While it has long been recognized that the sun's energy drives the earth system, it is only recently that we have begun to produce accurate, large-scale surface radiation flux data sets and are just learning how to employ them in scientific studies. While climatologies of surface radiation fluxes have been available 1, their accuracies are limited and they do not offer the time and space resolution now required in numerical climate models (e.g., one month and 5ø x 5ø grids), or basin-wide ocean models. 1.1 Net fluxes The Earth's radiation budget is becoming better understood since some of its components are being measured accurately and globally from satellites (e. g., Earth Radiation Budget Experiment - ERBE). Quantifying the Earth's radiative processes at its surface is fundamental to understanding the origin and nature of physical and biogeochemical change. Today, using satellite information, combined with diagnostic and modeling studies, these radiative processes relatin~ to ~lobal climate change can be investigated2. 1.2 Incoming SW and PAR Several parameters relating to surface radiation processes can be derived from satellite observations. Greatest accuracies have been demonstrated for the parameters of the incoming shortwave radiation (I~SW) and Photosynthetically Active Radiation (PAR), the main modulating factor for the Earth's biological primary production3 4. These accuracies are particularly applicable to estimations of these parameters over land and ocean, although problems still remain for highly reflective surfaces, such as snow and ice. For the estimation of 11 SW, and PAR at the surface, many approaches5 require the determination of the presence (or absence) of clouds. In section 2 we discuss the sensitivity of clear sky 1~ SW and PAR to such factors as water vapor and ozone amount, aerosol types, surface effects, and solar zenith angle. Section 3 deals with the challenge of cloud detection and effects of broken clouds on incoming radiation. Section 4 outlines some methods for determining surface fluxes, their results and uncertainties encountered for all sky conditions. Section 5 describes the current effort at UCSB in the Sequoia 2000 project; the objective is a global data set for PAR and Net Primary Production (NPP). Section 6 summarizes. 2. Clear Skv Sensitivities Earth's atmosphere is comprised of major constituents like nitrogen and oxygen and trace gases such as carbon dioxide, ozone, methane, and others6. Even in the absence of clouds, these gases, along with aerosols and water vapor, modify incoming solar radiation by scattering and absorption. Surface effects in the form of surface albedo also play a role through multiple reflections between the surface and lower atmosphere, and is most pronounced where aerosol optical depths are large (l=0.6 at ~=0.51lm). Figure 1 indicates the sensitivity of 1~ SW to (a) water vapor amount, (b) ozone amount, (c) aerosol type and amount, and (d) surface albedo. Water vapor concentrations vary zonally from very dry (1.2 g cm-2) at the poles to moist (4.5 g cm-2) at the equator. The graph indicates that a span such as this yields surface irradiance differences of about 6%, but water vapor amounts within regimes, (e.g., tropical or polar) vary much less and hence, the effect on llSW are only on the order of 1%. Ozone concentrations have virtually no effect in this wavelength region, but aerosol types and amounts (expressed as visibility) can have rather substantial effects (15%-20%) when greater optical depths are present. Surface albedo can affect surface irradiance by as much 10%; e.g., the difference between ocean (oc=0.05) and extremely bright snow or ice (a=0.9). However, since broadband ocean albedo varies only slightly for low and middle latitudes (and outside regions of sun glint), this effect is inconsequential for most oceanographic applications. 1 ooo 980 --> 3 960 940 9ZO 900 ~5 880 R~0 \ \, US 62 ATMOSPHERE SOLAR ZENITH ANGLE=30' uARlTlUE AEROSOLS Vl!ilRlLlTY=23 KM OZONE AMOUNT=0.3~. CM.ATM _ 1 000 _ 980 ' S 960 _ 900 ~ US 62 ATMOSPHERE 880 _ 50LAR ZENITH ANGLE=30- I ~.20 0.25 0.30 0.35 0.40 01 2 3 4 5 v.~u U.~J V.JV (a) WATER VAPOR AMOUNT (G.CM-2) (b) OZONE AMOUNT (CM.ATM) ~OOOr ,~ , i 950, LIJ 900 ~ 850 .s L ~ 800, ------CONTINENTAL AEROSOLSl 1100 MARITIUE AEROSOLS ~VATER VAPOR iMOUNT-1.42 G.C4-2) OZONE AMOUNT=0.34 C?.l.AThl 700 1050 1 000 950 0 20 40 60 80 100 0 20 40 60 80 100 o 0 0.2 0.4 0.6 0.8 1.O (C) V15181LITY (KM) (d) SURFACE AL8EDO Figure 1. Effects on incoming solar irradiance by (a) water vapor amount, (b) ozone amount, (c) aerosol type and amount, and (d) surface albedo. Figure 2 shows the dependence of irradiance on solar zenith angle. The dashed line indicates the effect at the top of the atmosphere and the solid line shows surface flux which has been influenced by scattering and absorbing processes. TOA flux and surface flux are most modified by the sun's zenith angle. In fact the dependence is so strong that surface flux can be approximated by a function of the cos(sza). Our recent study using the Tanre' 5S model showed that PAR for clear sky conditions exhibits a linear relationship to PAR(sza=0) x cos(sza) up to sza = 70 degrees. This relation is plotted in ~lgure 3 using maritime aerosols and 23 km visibility. Similar relations hold for other aerosol types and visibilities. Such relationships are important in view of our global processing requiring a large number of computations. 1 400 1 200 ~~ 1000 Z 800 600 40 U~ qnn 200 TOP OF ATMOSPHERE ~ SURFACE _ '~ - ~\\\\ _ \~ _ \\\ \\\ - US 62 AThlOSPHERE \ \ MARITI~,IE AEROSOLS YISI~IUTY=23 KM _ ~ATER VAPOR AMOUNT= 1.42 G.CM-2 OZONE A~IOUNT=0.34 CM.ATM , , ~ , I I , , , 0 20 40 60 80 SOLAR ZENITH ANGLE (ø) Figure 2. Solar irradiance varies most as a function of solar zenith angle. For Maritime Aerosols, 23km visibility o~01 . 500 E 300 C 200 ~ 200 300 400 500 Estimated Irradiance in Watts/sqmeter Figure 3. Linear relationship between modeled surface irradiance (PAR) and irradiance estimated by PAR = PAR(sza=0) x cos(sza), for clear sky conditions. From what has been presented so far, .~I~BSUM OCEAN SZA=60 UVV(100%)=2.102 DISORT it can be concluded that l~SW and PAR can be accurately computed over the ocean using a radiative transfer model such as 5 S and climatological values for precipitable water, aerosols and ozone concentrations. Another ap- proach that has recently been pro- posed to compute SW in clear conditions is based on top of the atmosphere broadband irradiances derived from ERBE data7. This approach relies on the linear relationship that exists between top of the atmosphere absorbed SW and the surface absorbed SW (or net SW). In the case of oceanic conditions (plotted in figure 4), this relationship is very straight forward and can be used directly . roP NEr 620.00 600.00 580.00 560.00 540.00 520.00 500.00 480.00 460.00 440.00 420.00 400.00 380.00 360.00 340.00 320.00 300.00 280.00 260.00 240.00 ~f) nn UW= 4.2040 l I I U~ _ ~. ,. rl ~ UW_ 3.1530 _ /,. ',~ .' U-W-.-2'10-2-0- _ ./~ ' ~ . _ UW.- 1.0510- =~ ~ ~ ~_-UW=-0-52-55- ~7~.~= ~E ~t I Bar NET Figure 4. Linear relationship between TOA net SW and surface net SW. 3. Clouds In cloudy atmospheres, radiative transfer modeling is much more complex owing to absorption by water vapor, and scattering and absorption by liquid water droplets (and possibly aerosols) within the clouds. The simplest approach to solving the transfer of shortwave radiation through clouds assumes that clouds are homogeneous and plane parallel8~9-10 l1. It nevertheless requires rather complex computations and the knowledge of cloud microphysical properties (single scattering albedo, asymmetry parameter and particle size distribution). Stephens9 proposed simplifications by means of parameterizations of cloud albedo and absorption as a function of optical thickness, single scattering albedo, scattering phase function, and solar zenith angle. These allow us then to relate cloud radiances measured from space to cloud macrophysical properties (e.g., albedo). In spectral regions for which there is no absorption by cloud liquid water (e.g., PAR), the cloud transmission is simply related to cloud albedo. It is the existence of a simple relationship between cloud albedo and transmission that is a~ the heart of the success of satellite-based retrieval methods. Broadband transmission computations require, in addition, a knowledge of cloud absorption. This presents a more difficult problem but, fortunately, absorption by cloud droplets is usually small compared to scattering (single scattering albedo close to unity) and simple cloud absorption parameterization can provide reasonable results5. Also, because water vapor dominates above clouds and water droplet absorption prevails within clouds, it is possible to decouple liquid water and water vapor absorptionl2. Then, second order corrections to the transmission/albedo relationship may be needed for broken cloud conditions or surface effects, particularly in the case of highly reflecting surfaces. Finally, because total absorption in the atmosphere is small, theoretical results demonstrate that it is possible to decouple clear and cloudy conditions, further simplifying the calculation of 11 SW and PAR. While a second order issue in our problem, the impact of finite clouds and broken cloudiness on reflected radiation and surface solar radiative flux (and PAR) has been seriously examined by many reSearCherS13,14,15,16,17,18,19,20,2l,22 This has been achieved principally by means of Monte Carlo calculations, which are applied to photons in order to compute radiation scattering by finite size clouds and assess radiation diffusion through cloud sides and cloud-cloud interactions. Different (but simplistic) cloud shapes and liquid water distributions were tested, all demonstrating markedly different directional reflectance variations with sun zenith angle between parallel and finite clouds. From such results, for instance, Coakley and Kobayashi22 predicted a bias on the surface solar irradiance computations when incorrectly assuming overcast conditions; Schmetz23, on the other hand, found no alteration of the linear relationship between cloud field albedo and absorption (in overcast conditions) by the presence of broken clouds. More recently, Wielicki and Parker24, suggested that the broken cloud effects may be smaller than previously predicted because the shape of clouds in nature is quite different from that assumed in most of these Monte Carlo computations. The clouds for which the predicted effects are the largest are those which the horizontal size is similar to their vertical one (e.g., cumulus clouds). These clouds have usually been modeled as cubes or cylinders but observations suggest that their edges are thinner (have a smaller optical depth) than their core and therefore the side effects may be less pronounced than presently simulated. While this issue is not yet entirely resolved, it is nevertheless a second order effect in the estimation of the shortwave radiation at the surface (or PAR). 4. Model s All methods for deriving surface solar radiative flux from satellite measurements rely on the high correlation between planetary albedo and surface solar irradiance, which is relatively independent of cloud amount and typeS~23-25. Similarly, a correlation exists between TOA spectral radiance and downwelling surface solar irradiance, resulting from the fact that absorption of 1~ SW by the atmosphere is relatively small and negligible in the visible (or PAR) wavelengths . 4.1 Statistical methods Based upon the high correlation mentioned above, purely statistical methods for remotely sensing 1~ SW were the first to be developed. In these methods, pairs of co-located visible brightness and surface solar irradiance measurements are used to derive an empirical regression which is subsequently applied to independent satellite measurements. While proven useful in regional studies26-27~28, these approaches are limited29-30 because they can only be applied in regions and seasons for which regression coefficients have been derived . 4.2 Physicallv-based methods Physically-based methods also rely on the high correlation between planetary albedo and surface insolation but replace the statistical relationship by a quantitative expression which involves bulk atmospheric properties. Since cloud transmission can be reasonably expressed as a function of planetary albedo, cloud absorption can be parameterized as a function of albedo. Most physically- ha~ed methods make use of these relationships in one form or another. Physical methods vary in complexity and procedure with some handling the clear and cloudy atmospheric conditions in differing manners. Some of the simplest methodsS~3 I ,32 are based on energy conservation principles that account for the most significant contributions to an atmospheric column's radiative energy budget. The radiative transfer treatment is well developed for clear conditions. Cloudy downwelling irradiance is simply the product of computed clear irradiance and cloud transmission which is estimated from satellite observations following a parameterization similar to Stephens33. In the case of Gautier et al5, once overcast regions have been delineated based on a cloud threshold approach, clear and overcast conditions are handled separately. Despite model simplicity and limitation, the computed surface solar irradiance is surprisingly accurate with reported uncertainties on the order of 8%-10% on a daily basis and better on a monthly basis. Pinker and Ewing34 base their method on a three-layer Delta Eddington approximation. The model is run to construct look-up tables relating cloud optical thickness to TOA spectral reflectance for a variety of observational, sun zenith angle, and surface albedo conditions. Together with this look-up table, the satellite data are subsequently used to estimate cloud optical depth, which is then employed by the radiative transfer model to compute the surface solar irradiance. Again, accuracies are on the order of 10% daily. While the above methods were initially based on geostationary satellite observation, Darnell et al.35 proposed a simple method using sun-synchronous TOVS and AVHRR data. Broadband surface solar irradiance is estimated as the product of TOA irradiance, and clear sky and cloud transmittances. Reported accuracy is on the order of 19% daily and 2.7% monthly. This method suffers uncertainty when short term estimations are required because, at most, three observations per day are available. All the methods described above are based on plane parallel cloud assumptions and clear/overcast conditions. None of them model the complexities of three dimensional solar radiation fields in cloudy conditions, in part because observations do not provide the necessary information to support such modeling. Nonetheless, their accuracies are equivalent and quite remarkable and justify, in large part, their extreme simplicity. 5. Global mapping of PAR The Sequoia 2000 project is a cooperative venture among several campuses of the University of California. Combining the skills of computer scientists at Berkeley, and earth scientists at UCSB, UCLA, and elsewhere, the project goals are to support global change research by providing a distributed processing system sufficiently large and fast to afford users in diverse locations real time access to and processing of global scale data. As our contribution to this project, we have undertaken the task of mapping global PAR to be used as input to Net Primary Productivity algorithms for the world's oceans36 and land areas. Surface UV and incoming SW irradiances are also produced since the algorithms are sufficiently similar and data is common. The basic strategy for determining these parameters is outlined in figure 5. As a starting point, International Satellite Cloud Climatology Project (ISCCP) B3 (raw radiance at 32km resolution every 3 hrs) data will be input into a cloud detection algorithm from which values for cloud and clear sky surface radiances emerge. These must be transformed into spectral or broadband angular integrated reflectance values using Bidirectional Reflectance Distribution Funtions (BRDF). This output becomes the reflectance (or albedo) values used in the calculation of each desired parameter, 11 SW, PAR, UV (figure 5). In each equation SWClr (PARclr~ UVclr) is the theoretical value derived from the radiative transfer model for cloudless skies but accounting for known (or assumed, based on climatology) gaseous and aerosol conditions. Cloud BRDF Cloud Detection Algorithm Bc d y Cloud Brightness Cloud Reflectance dy = f(BCdy) Rcdy AcRpar Ruv BClr Clear Brightness Surface Reflectance Rs As RparRuV 1 - Ac- ab SW = SWclr 1 - Ac As PAR = PARClr lR Rparc 1 - parcdy parclr clr 1 R u v cdy R u v c I r S SM/I [H20] Aerosol Climatology TOMS [ø3] Surface BRDF Figure 5. Outline of the methodology being used to find incoming shortwave, PAR, and UV radiation at the earth's surface. For calculations involving clear sky PAR the Tanre' 5S model is being used with climatological conditions for gas contents (e.g., water vapor, and ozone) and an assortment of type and amounts for aerosols. The model is run for every (integer) solar zenith angle (sza) in each season, equinox and solstice. The theoretical data of solar irradiances then consists of twelve tables, each containing a certain season, aerosol type, and amount. The largest value obtained for surface irradiance (in the PAR wavelengths) is about 500 Wm-2 and occurs for thin aerosols and overhead sun. As shown in figure 2 in sensitivity study plotting the data shows that the most notable trend is that of PAR = f[cos(sza)]. Even for solstice conditions for which atmospheric gases are unequal in the zonal belts (north and south), the Tanre' model does not show a significant skewing of the cosine function. The sensitivity study in section 2 predicts this to be the case. This leads to the simple parameterization, where: PARClr = A x [500*cos(sza)] + B The parameterization is particularly attractive when the need for PARClr is global, necessitating only a simple algebraic computation in lieu of complex model for each pixel on the Earth. Still ahead lies the task of quantifying the coefficients with a global aerosol database. Fortunately these parameters are not significantly different from solstice to equinox, with the greatest discrepancy occurring in optically thick Continental aerosols where the ~ B is 8 Wm-2. The geometry of the earth's motion relative to the sun gives the solar zenith angle for any given day, hour, and location on the surface. Figure 6 below shows a preliminary mapping of net surface solar irradiance. NET SURFACE SOLAR IRRADIANCE 80 160 240 320 wm -2 Figure 6. Global net surface solar irradiance derived using the Gautier et al method and ISCCP-B3 data. 6. SummarY Working from well known principles of radiative transfer and available regional climatologies for atmospheric gases and aerosols, researchers have devised algorithms for computing the surface solar irradiance (1~ SW) and PAR from satellite derived data. Since these fluxes are little affected by atmospheric gases and most radically modified by changes in the solar zenith angle, simple parameterizations have been devised for clear sky surface fluxes. In cloudy conditions, cloud effects can be accounted for through a parameterization of the transmission as a function of cloud albedo, which can be estimated from space observations. This requires an estimation of cloud absorption by liquid water and its parameterization still remains problematic. But absorption is small compared to reflection and thus its uncertainties have only a small effect on the estimation of SW. Broken cloud effects have been observed and numerical simulations have suggested that they may be important. However, recent results suggest that they might have been overestimated by using clouds with unrealistic shapes. High correlation exits between planetary albedo (determined from space) and surface fluxes; and a similarly high correlation exits between TOA radiance (measurable from space) and downwelling fluxes. Statistical and physically- based methods for determining surface fluxes make use of these two relationships, and even in their simplicities, yield results which agree among one another within acceptable error bounds, on the order of 10%. Many surface radiation studies have been performed on a regional basis, and these have evolved into the models referred to above. The Sequoia 2000 project at UCSB will attempt to apply the Gautier et al., method on a global scale, specifically, to compute a global field for PAR. However, this is not the end product; we will use this field in calculations of global NPP, using NDVI over land ~nd Morel's model for oceanic production. Acknowledgments: Part of this research has been supported by the NOAA TOGA project office under grant # A-447633-22584. We wish to thank Digital Equipment Corporation which has provided us with a hardware grant in support of the Sequoia 2000 project. References 1. Budyko, M.I., 1963. Atlas of the heat budget of the earth (Russian). Kartfabrika Gosgeoltehiydata, 41 pp. 2. Baker, D.J., 1990. Planet Earth, The View from Space. Harvard University Press, Carnbridge. 189pp. 3. Whitlock, C.H., et al., 1990. AVHRR and VISSR satellite instrument calibration results for both CIRRUS and marine stratus IFO periods, FIRE Science Report 1988, NASA Conference Proceedings, 3083, Washington, D.C. 4. Frouin, R., D.W. Linger, C. Gautier, K.S. Baker, and R.C. Smith, 1989. "A simple analytical formula to compute clear sky total and photosynthetically available solar irradiance at the ocean surface", J. Geophysical Research - Oceans, V94:9731-9742. 5. Gautier, C., G. Diak and S. Masse, 1980. "A simple model to estimate incident solar radiation at the surface from GOES satellite data", J. Applied Meteorology, V19:1005-1012. 6. Liou, K. 1980. An Introduction to Atmospheric Radiation, Academic Press, Inc. 392 pp. 7. Cess, R.D., I.L Vulis, 1989. "Inferring surface solar absorption from broadband satellite measurements" 7. Climate, V2:974-985. 8. Stephens, G.L., 1978. "Radiation profiles in extended water clouds, I: Theory" J. Atmos Sci, V35(11):2111-2122. 9. Stephens, G. L., 1978. "Radiation profiles in extended water clouds, II: Parameterization schemes" J. Atmos Sci, V35(11):2123-2132. 10. Stephens, G. L., G.W. Partidge and C.M.R. Platt, 1978. "Radiation profiles in extended water clouds, III: Observations' J. Atmos Sci, V35(11):2133-2141. 11. Stephens, G.L. and P.J. Webster, 1984. "Cloud decoupling of the surface and planetary radiative bud~ets" J. Atmos Sci, V41(4):681-686. 12. Davies, R, W.L. Ridgeway and K.-E. Kim, 1984. "Spectral absorption of solar radiation in cloudy atmospheres: A 20 cm-l model" J. Atmos Sci, V41:2126-2137. 13. McKee, T.B. and S.K. Cox, 1974. "Scattering of visible radiation by finite clouds" J. Atmos Sci, V31:1885-1892. 14. Aida, M., 1976. "Scattering of solar radiation as a function of cloud dimension and orientation" J. Quant. Spectrosc. Radiat. Transfer, V17:303-311. 15. Davies, R., 1978. "The effect of finite geometry on the three dimensional cloud fields" J. Atmos Sci, V35:1712-1725. 16. Welch, R.M. and B.A. Wielicki, 1984. "Stratocumulus cloud field reflected fluxes: The effect of cloud shape" J. Atmos Sci, V41(21):3085-3103. 17. Harshvardhan and R.W.L. Thomas, 1984. "Solar reflection from interacting and shadowing cloud elements" J. Geophys. Res. V89:7179-7185. 18. Welch, R.M. and B.A. Wielicki, 1985. "A radiative parameterization of stratocumulus cloud fields" r. Atmos Sci, V42:2888-2897. 19. Joseph, J.H. and V. Kagan, 1988. "The reflection of solar radiation from bar cloud arrays" J. Geophys. Res. V93(D3):2405-2416. 20. Kobayashi, T., 1988. "A parameterization of reflectivity for broken cloud fields" J. Atmos Sci, V45:3034-3045. 21. Kobayashi, T., 1989. "Radiative properties of finite cloud fields over a reflecting surface" J. Atmos Sci, V46(14):2208-2214. 22. Coakley, J. A. and T. Kobayashi, 1989. "Broken cloud biases in albedo and surface insolation derived from satellite imagery data" J. Climate, V2:721-730. 23. Schmetz, J. 1984. "On the parameterization of the radiative properties of broken clouds" Tellus V36A:417-432. 24. Wielicki, B.A. and L. Parker. "On the estimation of cloud cover from satellite sensors: the effect of sensor spatial resolution" Submitted to J. Geophys. Res. 1991. 25. Ramanathan, V. 1986. Scientific use of surface radiation budget data for dimate studies. In "Surface radiation budget for climate applicafions" Suttles and Ohring, eds, NASA Pub No.1169, Washington, D.C. 26. Tarpley, J.D., 1979. ~Estimating incident solar radiation at the surface from ~eostationary satellite data" J. Applied Meteorology, V41(4):1172-1181. 27. Pastre, C., 1981. "Developpement d'une methode de determination du rayonnement solaire global a partir des donnees METEOSAT" Meteorologie, VI serie, (24):5-15. 28. Hay, J.E. and K.J. Hanson, 1978. "A satellite-based methodology for determining solar irradiance at the ocean surface during GATE" Bull. Amer Meteor. Soc, V59:1549. 29. Raphael, C. and J.E. Hay, 1984. "An assessment of models which use satellite data to estimate solar irradiance at the earth's surface" J. Climate and Applied Meteorology, V23:832-844. 30. Pesin, E. 1989. "Comparison of different models to estimate surface insolation" Proc. 7th METEOSAT Scientific Users Meeting, Madrid, 1988, Eumetsat P 04:107-111. 31. Dedieu, G., P.Y. Deschamps and Y.H. Kerr, 1983. "Solar irradiance at the surface Meteosat visible data" Proc. of Machine Processing of Remotely Sensed Data Symposium. 32. Le Borgne, P. and A. Marsouin, 1988. "Determination du flux rayonnement ondes courtes incident a la surface: mise au point d'une methode operationelle a partir des ~1~nnees du canal visible de METEOSAT" Meteorologie, VII serie, (20):9-19. 33. Stephens, G. L., 1978. "Radiative profiles in extended water clouds, II: Parameterization schemes" J. Atmos. Sci., V35(11):2123-2132. 34. Pinker, R.T. and J.A. Ewing, 1985. "Modeling surface solar radiation: Model formulation and validation" J. Climate and Appl. Meteor., V24(5):389-401. 35. Darnell, W.L., W.F. Staylor and S.K. Gupta, 1988. "Estimation of surface insolation using sun synchronous satellite data" J. Climate, V1:820-835. 36. Morel, A. 1991. "Light and marine photosynthesis: a spectral model with geochemical and climatological implications" Prog. Oceanog. V26:263-306.