Determination of radiative forcing of Saharan dust using combined TOMS and ERBE data

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 105, NO. D16, PAGES 20,649-20,661, AUGUST 27, 2000 Determinatin f radiative frcing f Saharan dust using cmbined TOMS and ERBE data N. Christina Hsu Raythen ITSS, Lanham, Maryland Jay R. Herman NASA Gddard Space Flight Center, Greenbelt, Maryland Clark Weaver SM & A Crpratin, Arlingtn, Virginia Abstract. We determine the direct radiative frcing f Saharan dust aersls by cmbining aersl infrmatin derived frm Nimbus-7 TOMS with radiatin measurements bserved at the tp f atmsphere (TOA) by NOAA-9 ERBE made during February-July Clud parameters and precipitable water derived frm NOAA-9 HIRS2 were used t aid in screening fr cluds and water vapr in the analyses. Our results indicate that under "clud-free" and "dry" cnditins there is a gd crrelatin between the ERBE TOA utging lngwave fluxes and the TOMS aersl index measurements ver bth land and cean in areas under the influence f airbrne Saharan dust. The ERBE TOA utging shrtwave fluxes were als fund t crrelate well with the dust lading derived frm TOMS ver cean. Hwever, the calculated shrtwave frcing f Saharan dust aersls is very weak and nisy ver land fr the range f slar zenith angle viewed by the NOAA-9 ERBE in Sensitivity factrs f the TOA utging fluxes t changes in aersl index were estimated using a linear regressin fit t the ERBE and TOMS measurements. The rati f the shrtwave-t-lngwave respnse t changes in dust lading ver the cean is fund t be rughly 2 t 3 but ppsite in sign. The mnthly averaged "clear-sky" TOA direct frcing f airbrne Saharan dust was als calculated by multiplying these sensitivity factrs by the TOMS mnthly averaged "clearsky" aersl index. Bth the bservatinal and theretical analyses indicate that the underlying surface prperties, dust layer height, ambient misture cntent, and the presence f clud all play imprtant rles in determining the TOA direct radiative frcing due t mineral aersls. 1. Intrductin Trpspheric aersls perturb the radiative energy balance by directly interacting with slar and terrestrial radiatin [Penner et al., 1994]. Hughtn et al. [1990] shwed that the aersl frcing f climate is cmparable in magnitude t the frcing caused by current anthrpgenic greenhuse gases but ppsite in sign. The Intergvernmental Panel n Climate Change (IPCC) estimated that the glbal-mean direct frcing due t anthrpgenic sulfate and bimass cmbustin aersl ranges frm -0.3 t -1.5 W m -2 [IPCC, 1994], while Hansen and Lacis [1990] reprted that the calculated frcing due t anthrpgenic greenhuse gases is abut 2.3 _ W m -2. These results suggesthat the aersl cling effect may partially cunteract the warming cntributed by greenhuse gas increases ver the past few decades. Amng all f the natural and anthrpgenicmpnents f trpspheric aersls, mineral aersls (dust) play an impr- Nw at Science Systems and Applicatins, Inc., Lanham, Maryland. Cpyright 2000 by the American Gephysical Unin. Paper number 2000JD /00/2000JD ,649 tant rle in climate frcing thrughut the entire year. Several recent papers have mdeled the direct frcing f such aersls using cmpiled ptical prperties f aersls frm varius measurements [Tegen et al., 1996; Sklik et al., 1998]. Hwever, these studies indicate that there are large uncertainties in the estimatin f bth shrtwave and lngwave climate frcing f mineral aersls. Even the sign f the net frcing is nt well determined. A thrugh reginally dependent understanding f the chemical, physical, and ptical prperties f aersls (e.g., chemical cmpsitin, size distributin, particle shape, and refractive index) and their spatial and tempral distributins (e.g., aersl ptical thickness r mass lading) is required befre we can accurately evaluate aersl effects in the climate mdels. Since trpspheric dust aersls have a relatively shrt lifetime (a few hurs t abut a week), there are large variatins in their tempral and spatial distributin. In rder t study their direct and indirect effects n the reginal and glbalscale energy balance, satellite data are needed t btain cntinuus tempral and spatial sampling f bth the radiative quantities (such as albed and fluxes) and the aersl abundance. Radiatin measurements frm the Earth Radiatin Budget Experiment (ERBE) have prvided well-calibrated measurements f the radiative quantities. Hwever, there are

2 . 20,650 HSU ET AL.: RADIATION FORCING FROM TOMS AND ERBE DATA n channels in these instruments fr the detectin f aersls, Essentially, the AI is a measure f the wavelength-dependent s the scenes with aersl cntaminatin can be identified. T change in Rayleigh-scattered radiance frm aersl absrptin calculate aersl radiative frcing, additinal surces fr their relative t a pure Rayleigh atmsphere. The aersl index is detectin must be used. defined s that psitive values generally crrespnd t UV- There have been limited attempts t study the reginal radiative frcing based upn a few days f cllcated AVHRR and ERBE measurements n NOAA 9 [Ackerman and Chung, 1992; Christpher et al., 1996]. These studies prvided valuable infrmatin n the satellite estimate f aersl frcing by absrbing aersls and negative values t nnabsrbing aersls [Hsu et al., 1996; Herman et al., 1997; Seftr et al., 1997; Trres et al., 1998]. Within the range f aersl ptical depths bserved frm grund-based Sun phtmeter instruments under bimass burning and regins cvered by African dust [Hsu airbrne dust frm sandstrm events in the Saudi Arabian et al., 1999], the TOMS AI measurements are linearly prprpeninsula and smke frm bimass burning in Suth America. Hwever, the technique f separating cluds frm aersls in the visible wavelengths is time cnsuming and cannt be easily dne n a rutine basis. Als, the separatin f the signal due t mineral aersls frm that due t the backgrund terrestrial envirnment can be difficult, particularly ver semiarid regins [Ackerman and Chung, 1992; Tanre et al., 1992]. In this paper we use TOMS ultravilet radiance measuretinal t the aersl ptical thickness (AOT). These results are verified by theretical calculatins [Trres et al., 1998]. Because the reflectance ver cluds has nly a small wavelength dependence, the TOMS AI measurements using tw ultravilet channels can distinguish between clud-filled and aersl-filled scenes. This technique can als be used t detect absrbing aersls ver arid and semiarid regins. Unlike the thermal-cntrast methd using infrared channels, the detecments t detect airbrne mineral dust ver bth land and tin f mineral aersls by TOMS is nt susceptible t water cean [Hsu et al., 1996, 1999; Herman et al., 1997; Trres et al., 1998]. We will study the TOA direct radiative frcing f airbrne Saharan dust by cmbining TOMS, ERBE, and HIRS2 data t separate the aersl pixels frm the clear-sky pixels and t islate the aersl effect n the upward shrtwave and lngwave fluxes at the tp f the atmsphere (TOA). The paper is structured as fllws: vapr absrptin and surface temperature variatin. Even in the mist Sahel regin during the summer mnths, the crrelatin is high between TOMS AI and the grund Sunphtmeter-measured aersl ptical thickness [Hsu et al., 1999]. Using theretical mdel simulatins, hwever, Trres et al. [1998] have demnstrated that the TOMS AI depends n nt nly aersl ptical thickness (AOT) but als n single- 1. After brief descriptins f the satellite measurements scattering albed and aersl height. Althugh the height f used in the analyses in sectins , we will demnstrate the Saharan dust layer changes frm winter t summer, fr a the impact f airbrne Saharan dust n the daily bservatins given lcatin the dust-plume heights are usually repeatable n f the ERBE TOA utging lngwave and shrtwave fluxes in a seasnal basis [see Hsu et al., 1999, Figure 6]. sectin ERBE Radiatin Measurements 2. In sectin 3 we describe hw we use these daily satellite measurements t estimate the sensitivity f the TOA utging flux t changes in dust lading ver bth land and cean. 3. Calculatins f the mnthly averaged TOA direct frcing f airbrne Saharan dust are discussed in sectin In sectin 5 we estimate the changes f the TOA ut- ERBE include shrtwave (0.2-5/ m), lngwave (5-50/ m), and ttal radiatin (0.2-50/ m) wavelength channels. T btain radiatin measurements with the smallest ftprint, we use the ERBE scanner data. The ftprint size f ERBE scanner measurements is km by 45 km at nadir, which is clse ging lngwave and shrtwave fluxes per unit aersl ptical t that fr Nimbus-7 TOMS. thickness fr sites where Sun phtmeter aersl ptical thick- The details f the inversin algrithm t cnvert radiance ness measurements are available. int radiant exitence (flux) at the tp f atmsphere are de- 5. We will als cmpare these satellite-bserved values with the calculated values derived frm radiative transfer mdels. scribed by Barkstrm et al. [1989], Smith et al. [1986], and Suttles et al. [1991]. These inversin prcedures essentially 6. Cnclusins are summarized in sectin 6. cnstitute tw key elements: scene identificatin and the use f angular distributin mdels. Scene identificatin is an imprtant step in the determinatin f the radiant exitence at the tp 2. Satellite Observatins f atmsphere, since selectin f spectral crrectin, anistr TOMS Aersl Measurements pic factrs, and angular distributin mdels are all scene dependent. Fur different TOMS instruments have prvided nearly 20 years f daily glbal-aersl measurements since Nvember 1978, with a gap frm January 1995 t August In this In the ERBE inversin algrithm the surface scene types are classified int five different categries: cean, land, desert, snw, and castal. Each surface type is then further categstudy we use data frm Nimbus-7 TOMS (Nvember 1978 t rized int clear, partly cludy, mstly cludy, and vercast. In May 1993), whic has a ftprint size f 50 x 50 km: at nadir the case f heavy smke r dust, the ERBE algrithm culd and 100 x 200 km: farthest ff nadir (35 crss-rbit scans). misidentify these pixels as partly r mstly cludy, resulting in The TOMS aersl data are given in terms f an aersl index errrs caused by assuming the wrng limb darkening and bi- (AI), determined frm the Nimbus-7 TOMS 340 and 380 nm directinal reflectance mdels fr the cnversin. On the basis radiances (which have negligible dependence n zne absrp- f the results f Dieckmann and Smith [1989] the errrs due t tin) and is defined as scene misidentificatin are negligible in the ERBE TOA lng- AI = - 100{lgl [(I340/I380)... ] -- 1g10 [(1340/1380)calc]}, (1) wave flux estimate, while the errrs in the shrtwave fluxes are less than 14%. In this study we will use the archived ERBE where I... is the backscattered radiance measured by TOMS TOA albed and flux measurements and assume that the effect at a given wavelength, and Ica c is the radiance calculated using f the misclassificatin n the satellite-estimated aersl frca radiative transfer mdel fr a pure Rayleigh atmsphere. ing is small.

3 HSU ET AL.: RADIATION FORCING FROM TOMS AND ERBE DATA II ) /t I i! LI..,,! t 340 (],,).,_ 320 (.,,ci ' 2,,0, c:::: 12 6

4 20,652 HSU ET AL.' RADIATION FORCING FROM TOMS AND ERBE DATA! //. tt/ ' t / i1 tt '*- _6 180 (, u.. E ' u3 IO

5 HSU ET AL.: RADIATION FORCING FROM TOMS AND ERBE DATA 20, HIRS2 Clud and Water Vapr Measurements tivity. The areas included in these images are centered ver the T islate the aersl effect n the radiative budget, the screening f scenes cntaining cluds and high water vapr cntent is imprtant. In ur study, infrmatin n clud frac- Atlantic just west f Africa, since the respnse f the TOA shrtwave fluxes derived by ERBE t changes in dust lading is nt clearly seen ver land in ur analyses. This is cnsistent tin, clud-tp pressure, surface temperature, and precipitable with the finding f Ackerman and Chung [1992] using ERBE water was prvided by data frm the NOAA-9 HIRS2 instru- and AVHRR measurements ver the Saudi Arabian peninment. TOMS reflectivity data were als used as an additinal sula. On February 4 the effect f the dust plume was apparent surce fr clud screening. The measurements f clud pa- in the TOA shrtwave fluxes in the Atlantic just west f Senerameter and precipitable water used are part f the TOVS gal. There were cluds present in the nrthwest crner f the Pathfinder Path A data set develped by Susskind et al. [1997]. map, resulting in an increase in the shrtwave fluxes in that The fields are gridded n a 1 ø x 1 ø latitude-lngitude grid. regin. Cmparisns f these HIRS prducts against radisnde data On average, the TOA utging shrtwave fluxes increased and ECMWF reanalysis indicate that the quality f the HIRS due t the presence f dust aersls frm W/m 2 t as prduct is adequate fr use in climate studies [Susskind et al., much as 180 W/m 2. This dust plume cntinued t mve t the 1997]. west, and n February 6, aersls were nly detected in the tp and left-bttm part f the map. Cluds were als present in 2.4. Examples f Satellite Daily Maps the tp part f the map. The increase in the TOA shrtwave T demnstrate the effect f airbrne Saharan dust in the flux bserved in the dusty regin n this day was abut reginal radiatin budget n a daily basis, examples f the daily W/m 2. On February 13, mst f the regin was dust-free, and bservatins f ERBE radiant flux and TOMS aersl index the values f the TOA shrtwave fluxes drpped back t the are shwn fr the regin f interest (which includes the Saha- cean backgrund level f abut 100 W/m 2. On February 27, ra/sahel areas and the eastern Atlantic). Plate 1 (tp) depicts anther intense dust strm ccurred near western Sahara and maps f the daily TOA utging lngwave fluxes bserved by the NOAA-9 ERBE scanner in February 1985, while the crrespnding daily cmpsites f aersl index and reflectivity Mauritania, resulting in increases f mre than 80 W/m 2 in TOA utging shrtwave fluxes ver a vast area f water suth f Tenerife and Canary Islands. derived frm TOMS are shwn in Plate 1 (bttm). Cluds are usually assciated with higher reflectivity values and are represented by the gray clr shwn in the bttm left clr bar. 3. Methd and Results Accrding t the Natinal Centers fr Envirnmental Prtectin (NCEP) reanalysis data fr early February 1985, an anticyclne centered near Spain was bserved with ridge axes extending t western Africa. On the 500-hPa surface, this an- In this sectin we discuss hw the radiative frcing f mineral aersls is calculated using the daily satellite measurements described abve. The resulting respnse f TOA radiatin due t the presence f mineral aersls will be shwn ver ticyclne was sandwiched by a strng deep trugh ver the bth land and cean. Since bth TOMS and ERBE data are Nrth Atlantic and anther deep trugh centered near Cyprus. This cmbinatin created a large pressure gradient and strng surface winds, which lifted dust frm surces in West Africa and near Lake Chad. Cnsequently, intense dust plumes were seen ver Mauritania, western Sahara, Senegal, Mali, and snapshtypes f measurements taken at near nn and 1430, respectively, n time averaging is invlved in ur estimate f dust radiative frcing. Therefre the radiative frcing in this study is referred as "direct" r "instantaneus" frcing by aersls. Chad. The cntrast f TOA lngwave fluxes between the dustladen and clear-sky cnditins ver land is shwn by the first tw maps (i.e., February 3 and 12) n the tp panel. Cmpared t February 12, the TOA utging lngwave flux n February 3 decreased by W/m 2 in the territries ver Mauritania, Mali, and Algeria. These lcatins are cnsistent with areas where TOMS, as shwn n the bttm panel, detected dust aersls. Mre dust strm events were bserved ver the Sahel The Level 2 TOMS aersl and ERBE flux measurements fr the perid frm February 1985 t July 1985 were binned int 1 ø x 1 ø gridded data sets. Since the ftprint sizes f bth TOMS and ERBE are x 45 km, the sensitivity t the presence f inhmgeneus subpixel cluds and the underlying grund is rughly the same fr bth instruments. Fr this study, the ERBE data were screened fr bth clud and water vapr cntaminatin using the 1 ø x 1 ø gridded HIRS2 data set. The threshld used fr filtering data due t clud cntaminatin is a clud fractin f 0.1, while the threshld fr high regin in mid-february On February 14, TOMS detected high dust lading in the regin near Niamey, Niger, as shwn in Plate 1. A reductin in TOA utging lngwave flux was as water vapr cntaminatin is a ttal precipitable water cntent f ver 2 cm fr February and 3 cm fr July. We chse the much as W/m 2 in this area. As dust plumes swept water vapr threshld by balancing the requirement f having thrugh the cntinent int the Atlantic, the aersl effect n the TOA lngwave flux was als bserved ver cean. One example, using satellite measurements taken n February 27, a relatively dry scene with the requirement f having an adequate sampling f data pints. The TOMS data were screened fr clud cntaminatin by requiring the TOMS reflectivity t shws gd crrelatin between the TOMS-bserved aersl be less than 12%. The TOMS and ERBE data were then distributin and the changes in TOA lngwave fluxes ver the Atlantic just west f Senegal. The TOA utging lngwave flux merged int ne cmbined data set. Figure 1 (tp) shws the plt f the gridded daily ERBE in the eastern Atlantic was reduced frm t TOA upward lngwave fluxes versus TOMS AI measurements W/m 2. A band f cluds suth f the maximum dust regin using the methd mentined abve fr a 2 ø x 2 ø bx ver ver the Atlantic, as seen in the TOMS data, is reflected by lw values f the lngwave flux bserved in the ERBE data. Plate 2 shws the images f ERBE TOA utging shrtwave fluxes and the crrespnding TOMS aersl index and reflec- Africa (29ø-31øN; 5ø-7øE) fr the mnth f July. It is apparent that there is a gd crrelatin between the TOA utging lngwave flux and TOMS AI at this lcatin. The linearregressin fit t these data is represented by the slid line with

6 20,654 HSU ET AL.: RADIATION FORCING FROM TOMS AND ERBE DATA 38O " , øo ø q % - F = * AI Aersl Index F = * AI Aersl Index Figure 1. (tp) Relatinship f the daily gridded ERBE TOA lngwave fluxes and TOMS AI measurements fr a 2 ø x 2 ø bx in Africa (29ø-31øN; 5ø-7øE) fr the mnth f July. The linear-regressin fit t these data is represented by the slid line. (bttm) The relatinship f the daily gridded ERBE TOA shrtwave fluxes and the TOMS AI measurements fr a 2 ø x 2 ø bx in the Atlantic (25ø-27øN; 19ø-21øW) fr the mnth f July. slpe and intercept f -9.9 (_+0.9) and (_+ 1.8), respectively. The 1-rr (ne standard deviatin) errrs assciated with the regressin line cefficients are given in parentheses. The pixels with cluds and high water vapr cntent were filtered ut f the analyses. It is apparent that the ERBE TOA upward lngwave fluxes decrease with increasing dust lading. A similar plt fr the ERBE TOA shrtwave fluxes versus TOMS AI data in the Atlantic (25ø-27øN; 19ø-21øW) is shwn in Figure 1 (bttm). The pixels with cluds and high water vapr cntent were als filtered ut f the analyses. The slpe and the intercept f the linear-regressin line fr the shrtwave are 21.4 (_+2.7) and 91.4 (_+4.4), respectively. As seen frm this figure, the TOA utging shrtwave fluxes crrelate well with the TOMS AI ver cean. The ERBE TOA upward shrtwave fluxes appear t increase with increasing dust lading. If we define the dust radiative frcing as the difference between the ERBE flux measured under the dust-laden cn- ditins t that under clear-sky, dust-free cnditins and assume such frcing can be apprximated as a simple linear functin f the amunt f airbrne dust, the relatinship between the TOA upward flux and the dust lading can be expressed as fllws: F t TOA = K' --clear 1' TOA sky + sensitivity*ai, where F 1' TeA dentes the TOA ERBE measuredaily flux Jr7? TeA during daylight, and-c... ky represents the TOA ERBE flux btained under clear-sky cnditins. If statistically significant, the regressin line slpes in Figure 1 represent the sensitivity factr f the TOA ERBE flux t changes in aersl lading and the intercept value is the TOA flux in the dust-free cnditin. On the basis f ur analysis f the satellite measure- ments the linear assumptin hlds fr mst f the AOT range, usually between 0 and 2, fr dust in this area [Hsu et al., 1999]. Nte that aersl frcing defined in this fashin is ppsite in sign t the quantities usually used by climate mdels (i.e., Ft cl... TeA ky -- F t TEA). We will apply this regressin analysis fr each 1 ø x 1 ø grid bx t calculate the lngwave and shrtwave sensitivity factrs thrughut the entire regin f interest TOA Flux Sensitivity t Dust Lading Over Land Figure 2 shws the lngwave sensitivity factrs, averaged ver a 5 ø latitude band, derived ver regins where significant amunts f airbrne dust are ften bserved fr February (tp) and July (bttm) Since the maximum dust belt mves nrthward frm February t July, the February case was represented by the Sahara/Sahel regin (15 ø t 20øN; 0 ø t 15øW) and the July case was represented by the Sahara regin (29 ø t 33øN; 5øW t 15øE). Fr February 1985 the values f the TOA lngwave sensitivity t airbrne dust ver land are arund -16 _+ 4 W/m 2 per unit AI. These values were fund t be fairly cnsistenthrughut the Sahara/Sahel regin during February except fr the regin near equatrial Africa where lw values f lngwave sensitivity were bserved. These lw values ver the equatrial frests are mst likely due t the effect f smke generated frm bimass burning that ften ccurs in this regin during February. The values f lngwave sensitivity factrs fr July 1985 are apprximately -9 _+ 1.5 W/m 2 per unit AI ver land as shwn in Figure 2 (bttm). These values are cmmnly fund in the entire regin nrth f 20øN except in the muntain areas where higher values are btained. The crrelatin f a high-sensitivity factr and terrain pressure culd be due t artifacts assciated with algrithms ver high-altitude terrain. Fr the regin suth f 20øN in July, there are nt enugh data pints that pass the screening f lw-precipitable water (<3 cm) t cnduct a regressin analysis. If nly a clud screening is applied in ur analysis, regardless f the water vapr cntent, the change in lngwave radiatin due t dust strms is fund t be very small in the Sahel (suth f 17øN) in July. This culd be due t -5 ili N Latitude ver Land February... i{ i Lngitude (degree) -.!iii'... i " ' 'I'"... ß.'.. ' "'' " ' ' " ' ' '' ' ' "" Lngitude (degree) Figure 2. Values f the lngwave sensitivity factr derived ver land as a functin f lngitude. The sensitivity factrs derived fr each 1 ø x 1 ø grid bx were averaged ver a 5 ø latitude band fr 15ø-20øN during February (tp), and 29 ø- 33øN during July (bttm). Circles representhe mean values, and bars dente 1-rr f the sensitivity factrs derived within the latitude band.

7 HSU ET AL.' RADIATION FORCING FROM TOMS AND ERBE DATA 20,655 cancellatin between the effect f airbrne dust and the water vapr during dust strm events. Figure 3 depicts the seasnal variatins (February t July 1985) f ttal precipitable water frm HIRS2 measurements fr the suthern zne (15ø-20øN; 5ø-10øW) and the nrthern zne (27ø-32øN; 5ø-10øW). Because the Intertrpical Cnver gence Zne (ITCZ) is near its suthernmst psitin in February, the water vapr cntent is generally lw nrth f 10øN in *' øø the regin f interest. Thus the analyses f the sensitivity factr %.,, øo are less susceptible t prblems arising frm water vapr cntaminatin. Hwever, as the psitin f the ITCZ mves nrthward frm winter t summer, the crrespnding water O O vapr cntent becmes significantly higher in the suthern ø zne (15ø-20øN; 5ø-10øW) in July cmpared t that in Febru- ary. The ttal precipitable water increaseslightly in the nrth ern zne frm February t July, prducing a strng nrthsuth gradient in the ttal precipitable water frm 15øN t 30øN during July. Figure 4 depicts the relatinship between the 440-nm aer- 3 i i sl ptical thickness and the precipitable water measured by the grund Sun phtmeter at Bidi Bahn, Burkina Fas (14øN, 2.4øW), during June-July There is clearly an anticrre- 440 nm AOT 1.5 latin between the aersl ptical thickness and the precipita- Figure 4. Relatinship between the 440-nm aersl ptical ble water during this perid. The ambient atmsphere is gen- thickness and precipitable water measured by the grund Sun erally very mist (ttal precipitable water ften >6 cm) in the phtmeter at Bidi Bahn, Burkina Fas (14øN, 2.4øW), during Sahel during the mnth f July. As the dust plume passes by June-July The aersl ptical thickness appears t be the Sun phtmeter site, it brings in a large amunt f airbrne anticrrelated with the precipitable water during this time perid. dust as well as drier air. Accrding t Tanre and Legrand [1991] the radiative effect f a variatin f 1 g/cm 2 in precipitable water crrespnds t a variatin f 1 in aersl ptical thickness. Therefre accurate water vapr measurements are crucial in the estimatin f aersl frcing, particularly in the regin suth f 20øN during the summer mnths. respnse f the TOA shrtwave flux t dust lading is very weak ver the desert surface fr the range f slar zenith angle viewed by the NOAA-9 ERBE in 1985 (i.e., slar zenith angle (SZA) ø fr trpics). In cntrast t the lngwave respnse, the shrtwave re TOA Flux Sensitivity t Dust Lading Over Ocean spnse is difficult t determine ver land fr bth February and July Figure 5 shws the relatinship between the Cmparisns f the mnthly averaged lngwave and shrt- ERBE TOA upward shrtwave flux and the TOMS aersl wave sensitivity factrs f the ERBE TOA upward fluxes ver index fr the same regin ver Africa (29ø-31øN; 5ø-7øE) as in Figure 1 (tp) fr the mnth f July. It is apparent that the cean in February are shwn in Figure 6. The values presented in the figure are the averaged sensitivity factrs ver the latitude band 15ø-20øN as a functin f lngitude. The lngwave sensitivity factrs ver the cean are arund -10 +_ 2.5 W/m 2 Jan Feb Mar Apr May Jun Jul Aug per unit AI. Cmpared t the February value ver land shwn in Figure 2, there is clearly a difference in the respnse f the 27-32N, 5-10W TOA upward lngwave flux t changes in dust lading between N, 5-10W Bidi Bahn (June-July 1996) Over Land... ;:i days in = 400 > 350 c _- %m ø %O ø OO 0% ø F = 402.7,5 r * AI Aersl Index Figure 3. Seasnal variatins (February t July 1985) f ttal precipitable water frm HIRS2 measurements fr the suth Figure 5. Relatinship f the daily gridded ERBE TOA zne (15ø-20øN, 5ø-10øW) (pluses) and the nrth zne (27 ø- shrtwave fluxes and TOMS AI measurements fr the same 32øN, 5ø-10øW) (asterisks). Because the psitin f the ITCZ regin in Africa (29ø-31øN; 5ø-7øE) as in Figure 1 (tp) fr the mves nrthward frm winter t summer, the crrespnding mnth f July. The respnse f the TOA shrtwave flux t dust water vapr amunt is significantly higher in the suth zne lading is bserved t be weak ver land cmpared t that ver cean. (15ø-20øN, 5ø-10øW) during July cmpared t that in February.

8 ... 20,656 HSU ET AL.: RADIATION FORCING FROM TOMS AND ERBE DATA 50? : N Latitude Over Ocean!i February land and cean. This difference is mst likely due t the difference in the surface temperature between land and cean [Ackerman and Chung, 1992]. The values fr the shrtwave respnse ver cean are ---25_+ 6 W/m 2 per unit AI in February. The rati f the abslute value f TeA shrtwave t lngwave frcing is abut 2.5 ver cean in February. This indicates that there is a net cling effect due t the presence f mineral aersls in the atmsphere ver cean in February. Figure 7 shws the lngwave and shrtwave respnse t dust lading ver the latitude band 25ø-30øN fr July The general pattern f the land/cean cntrast in the lngwave respnse t dust lading is again bserved during the mnth f July. The values fr the lngwave respnses in July are arund W/m 2 per unit AI, which are smaller by as much as 30% than the values ver land, as shwn in Figure 3. The values fr the shrtwave respnse ver cean are _+ 3 W/m 2 per unit AI in July. The rati f the abslute value f the TeA shrtwave t lngwave frcing is less than 3 ver cean in July. Because there are ften cluds present ver cean in the regin suth f 20øN in the summer mnth, the number f the measurements made under clud-free cnditins in this regin was nt enugh fr a definite determinatin f the radiative frcing f mineral aersls. The difference in sensitivity in lngwave radiatin between February and July may be due t a number f factrs, including differences in dust height r differences in aersl characteristics. Our preliminary analysis indicates that the lngwave sensitivity t the TOMS aersl index is weakly dependent n the height f the dust layer. The February/July sensitivity difference is therefre due t ther factrs, which may include the fllwing: 1. the rigin f the dust is different fr the tw mnths (the Sahel in February versus the Sahara in July), which may lead t differences in aersl characteristics. 2. Althugh we attempted t cnfine ur analysis t dry regins, the water vapr cntent fr July (mstly between 2 and 3 cm) is still slightly higher than in February (mstly between 1 and 2 cm), which culd lead t lwer values fr lngwave radiatin. 4. Mnthly Averaged TeA Radiative Frcing f Saharan Dust T btain the mnthly averaged TeA "clear-sky" radiative frcing f dust (using the methd described abve), we multiply the mnthly mean sensitivity factr f TeA utging lngii Shrtwave wave and shrtwave flux with the mnthly mean TOMS aersl :: index taken n clud-free days as defined in sectin 3. Plates 3a and 3b shw the TeA instantaneus lngwave and shrtwave :..,:'.':: frcing f the Saharan dust bserved by ERBE at 1430 LT fr February 1985, respectively. The mnthly TeA "clear-sky" lngwave frcing can be as much as -15 t -30 W/m 2 ver land and -10 t -15 W/m 2 ver cean. Fr the shrtwave, the - 0 :... Ln wave February TeA frcing is as much as 10 t 30 W/m 2 ver cean.... Thus the Saharan dust results in a cling effect n the atm t6 -t4 spheric radiatin ver cean in February by as much as 15 Lngitude (degree) W/m 2 n a mnthly basis. Figure 6. Cmparisns f the mnthly averaged Insave The regin suth f 10øN is usually under the influence f and shr ave sensitivity factrs f the ERBE TeA upward smke generated frm bimass burning frm December t fluxes ver cean in Februa. Circles represent the mean March and is nt shwn in Plate 3. On the basis f the TOMS sensitivity values ver 15ø-20øN latitude, and bars dente 1- satellite data and grund Sun phtmeter aersl ptical f the sensitivi factrs derived within the latitude band. thickness data, the ccurrence f the sandstrm events was quite frequent in February 1985 [Hsu et al., 1999]. There is a latitude shift frm land t cean in the maximum dust frcing that crrespnds t a similar shift in the distributin f dust. This is due t the directin f the prevailing winds at this time f the year. The July mnthly averaged TeA "clear-sky" lngwave frcing is generally frm -10 t -25 W/m 2 ver land and -3 t -12 W/m 2 ver cean (see Plate 4). Cmpared t the February case, the maximum dust frcing ccurs at higher latitudes near 25øN. The increased misture in the Sahel regin during July causes there t be nly a small amunt f lngwave frcing even in the presence f significant amunts f dust plumes in the regin suth f 20øN shwn in Plate 4. Fr the shrtwave the July TeA dust frcing is as much as W/m 2 ver cean. There is a net cling effect f as much as 25 W/m 2 due t the presence f the Saharan dust ver cean in July. It is imprtant t nte again that the TeA dust frcing shwn in Plates 3 and 4 is fr clud-free cnditins. Only satellite measurements btained n cludless days were included in the calculatin f the mnthly-mean dust frcing N Latitude Over Ocean July ' ';... '; ' ; i: ::::i Shrtwave Lngwave ::: ::t -20 ::... ;... ; ;,... L...,., :... J...,,..., Lngitude (degree) Figure 7. Similar t Figure 6 except fr the latitude band 25ø-30øN ver cean fr July 1985.

9 ß ß HSU ET AL.: RADIATION FORCING FROM TOMS AND ERBE DATA 20,657 (a) Direct TOA LW Frcing (February 1985) I i Watts/m (b) Direct TOA SW Frcing (February 1985).~ I I *. ' ' Watts/m Plate 3. Mnthly averaged TOA instantaneus lngwave (a) and shrtwave (b) frcing f the Saharan dust fr February The Saharan dust results in a mnthly-average cling effect n the atmspheric radiatin ver cean during February by as much as 15 W/m 2. Nte that the quantitieshwn here are in abslute units. Therefre a negative sign must be applied t lngwave frcing befre cmbining with the shrtwave frcing t derive the net effect.

10 20,658 HSU ET AL.' RADIATION FORCING FROM TOMS AND ERBE DATA (a) Direct TOA LW Frcing (July 1985) b.r,? 9 --, -,4 b b b ;::> Watts/m (b) Direct TOA SW Frcing (July 1985), I ' ' ' ' II ' I :! i ' : U U cn cn m n Watts/m * Plate 4. Similar t Plate 3 except fr July There is a net cling effect f as much as 25 W/m 2 due t the presence f the Saharan dust ver cean in July. As in Plate 3, a negative sign must be applied t lngwave frcing befre cmbining with the shrtwave frcing t derive the net effect.

11 HSU ET AL.: RADIATION FORCING FROM TOMS AND ERBE DATA 20, Dust direct frcing at tp f atmsphere 80 Ocean/Feb Land/Feb Ocean/Jul 60 SW sw 40 I 20 sw 0-4O P -60 f Land/Jul I! I LW Figure 8. Summary f the estimated TOA lngwave and shrtwave direct frcing f dust per unit AOT ver land and cean surfaces fr February and July. Therefre any cmparisns between climate mdel results with the results shwn here must take this factr int accunt. 5. Radiative Frcing f Dust in Terms f Aersl Optical Thickness While there are nearly 20 years f aersl index measurements taken frm TOMS, extensive grund-based Sun phtmeter measurements f AOT have nly becme available recently. The aersl index has been shwn t be prprtinal t aersl ptical thickness (AOT) using grund-based Sun phtmeter data [Hsu e! al., 1999]. We will use the results f Hsu et al. t derive the TOA radiative frcing as a functin f AOT. These results, which are based upn the Earth-Prbe TOMS data, and the Sun phtmeter data frm AERONET [Hlben e! al., 1998] can be summarized as fllws: Cape Verde (cean) (Feb.-March 1997): AOT (_+0.10) ß AI (_+0.14) Cape Verde (cean) (May-Aug. 1997): AOT (_+0.04) * AI (+_0.07) Bidi Bahn (land) (Feb.-March 1997): AOT = 0.85(_+0.05) ß AI (_+0.09) Bidi Bahn (land) (May-Aug. 1997): AOT = 0.30(+_0.05) ß AI (+_0.11) On the basis f these relatinships and the aersl wavelength adjustment cnstant between Nimbus-7 and Earth- Prbe TOMS [Hsu e! al., 1999], we cnverted the dust frcing per unit AI shwn in the previus sectin int per unit AOT fr the Sun phtmeter sites. Figure 8 shws the TOA lngwave and shrtwave frcing fr bth land and water fr tw different mnths. The latitude fr the land and cean case is 10ø-20øN fr February and 25ø-30øN fr July. The TOA lngwave frcing is apprximately -32 W/m:/AOT ver land in February and -36 W/m:/AOT ver land in July. The TOA lngwave frcing ver cean is abut -20 W/m:/AOT in February and -24 W/m:/AOT in July. The shrtwave frcing ver cean is rughly 52 W/m:/AOT in February and 60 W/m:/AOT SW in July. The slar zenith angles crrespnding t these ERBE measurements are -- 20ø-30 ø. It shuld be nted that a bias may be intrduced int these results since the relatinship between the aersl index and the ptical thickness depends n the aersl vertical prfile. 6. Theretical Simulatins f TOA Direct Frcing f Airbrne Dust These bserved results can be cmpared with radiative calculatins using a ne-dimensinal versin f the NASA Gddard Earth Observing System Data Assimilatin System (GEOS 2) GCM radiative transfer mdel [Chu, 1992; Chu and Suarez, 1994]. In this mdel the slar spectrum was divided int eight bands in the UV and visible regin and three bands in the near-infrared regin. Fr the terrestrial radiatin there are nine bands in the infrared regin. T calculate the TOA lngwave and shrtwave fluxes, we chse vertical prfiles fr temperature, misture, and clud fractin n August 8, 1988, at 24øN, 25øW (cean) and 24øN, 12øW (land) frm the GEOS-2 assimilatin. The infrmatin regarding aersl prfiles used in this study was based upn the utput f the GSFC dust transprt mdel (P. Ginux, persnal cmmunicatin, 2000). These prfiles were calculated frm the cnditins ccurring n August 8, There are seven dust particle size ranges in the GSFC mdel. The effective radii fr the bins are centered at 0.15, 0.25, 0.4, 0.8, 1.5, 2.5, and 4/ m. Fr this study, the dust mass distributin was centered at the 1.5/ m bin and Mie scattering thery was assumed. The dust ptical parameters such as single-scattering albed and asymmetry factr, which are dependent n wavelength and particle size, are the same values used in recent radiative calculatins by Tegen and Lacis [1996][see Tegen and Lacis, 1996, Figure 1]. The differences between the clean and the dusty atmsphere TOA upward flux (defined in the same fashin as in sectin 3) are shwn in Figure 9 fr the lng and shrt wave bands. The surface reflectivity is assumed t be 3% ver cean and 24% ver desert. The ERBE bservatins were made at LT lcal equatr crssing time, which is clse t 1500 GMT fr the prfile lcatins. As shwn in the lng-dashed curve f the left panel, there is a 75 W/m: increase TOA shrtwave flux ver cean fr the first AOT. This is cnsistent with the bserved 60 W/m: increase in SW ver cean in Figure 8. The TOA upward shrtwave flux als increases ver cean with increasing AOT at all zenith angles. Hwever, ver land (as shwn in the right-hand panel) the situatin is mre cmplex. At lw zenith angles, absrptin by dust reduces the TOA utging shrtwave radiatin (dtted line). At high zenith angles enugh f the radiatin that wuld have been absrbed at nn is nw scattered back t space, and the TOA upward shrtwave flux actually increases (slid and dtted-dashed lines). The TOA shrtwave flux ver land is nt sensitive t atmspheric dust lading at 1500 GMT (lngdashed line), which explains the pr relatinship between the ERBE-bserved TOA shrtwave flux and the AI ver land. It is nted that in this study we assume a surface reflectivity f 24% fr desert. In the case f the desert surface reflectivity higher than 24%, it wuld require a less absrbing type f dust aersls t achieve the weak shrtwave respnse as bserved fr the same slar zenith angle. The middle panel shws the sensitivity f the TOA lngwave radiatin ver land t elevatin f the dust layer, the misture

12 ._ 20,660 HSU ET AL.: RADIATION FORCING FROM TOMS AND ERBE DATA Changes in TOA flux frm African mineral aersl, August O sw TOA vercean... i... i... t... i... LW TOA ver land SW TOA verland... i... i... i... i x= ' 50 '., '., Optical Thickness.4-.7m Optical Thickness.4-.7m ptical Thickness.4-.7m cszen= 0.4 GMT 9, alb=0.03 cszen= 0.9 GMT 12, alb= cszen= 0.9 GMT 15, alb=0.03 cszen= 0.5 GMT 18, alb= Dusth= 0.3 km precw= 3.3g/cm2 n clud Dusth- 2.3 km precw= 3.3g/cm2 n clud Dusth= 0.3 km precw= 1.6g/cm2 n clud Dusth= 0.3 km precw= 6.Sg/cm2 n clud Dusth= 0.3 km precw= 3.3g/cm2 cldh-12.0km cszen= 0 6 GMT 9, alb cszen= 1.0 GMT 12, alb=0.24 cszen- 0.9 GMT 15, alb=0 24 cszen- 0.3 GMT 18, alb= Figure 9. Mdel-calculate differences between the clean and dusty atmsphere TOA upward lngwave and shrtwave flux (defined in the same fashin as in sectin 3). The ERBE bservatins were made at lcal equatr crssing time, which is clse t 1500 GMT fr the prfile lcatins. (left) The respnse f the TOA upward shrtwave flux t changes in AOT ver cean as a functin f slar zenith angle; (right) similar t (the left) except fr the land cnditin; (middle) the sensitivity f the mdel-calculated TOA instantaneus lngwave dust frcing t dust layer height, misture cntent, and the presence f cluds. cntent, and the presence f cluds. The slid line is the standard prfile (assuming dust height = 0.3 km, precipitable water = 3.3 g/cm 2, and n clud), and it shws a decrease f abut 17 W/m 2 in TOA lngwave flux per unit AOT. This is significantly less than the bserved 32 W/m 2 per unit AOT shwn in Figure 8. Hwever, the mdel-calculated respnse f the lngwave radiatin changes significantly with changes t the input parameters. Fr example, the decrease in the lngwave radiatin becmes 22 W/m 2 per unit AOT when the ambient precipitable water is decreased by 50% (lng-dashed line). Als, as the dust height was elevated frm 0.3 t 2.3 km, the TOA lngwave flux decreases by 27 W/m 2 per unit AOT under the same misture cnditin f 3.3 g/cm 2. Finally, placing a clud at 12 km (slid curve) significantly reduces the sensitivity f LW TOA t dust lading, which is cnsistent with the results f Lia and Seinfeld [1998]. 7. Cnclusins The effect f trpspheric aersls in climate frcing simulated by GCM mdels depends greatly n the input aersl Table 1. Summary f Daily and Mnthly TOA Direct Radiative Frcing f Airbrne Saharan Dust fr Lngwave and Shrtwave Fluxes Land Ocean 20---, 100 Daily, W/m 2 Mnthly (W/m 2) SW LW SW LW (Feb) (Jul) (Feb) (Jul) (Feb) (Jul) prperties and will nt be well determined in the absence f aersl data. In this study we derive the TOA direct frcing f airbrne Saharan dust using satellite data frm TOMS and ERBE. Our results clearly indicate that the signature f the mineral aersls can be traced in the lngwave and shrtwave fluxes at the tp f the atmsphere. Fr "clud-free and dry" cnditins the rati f the shrtwave t lngwave frcing due t Saharan mineral dust is fund t be ---2 t 3, but ppsite in sign, ver cean fr bth February and July. It indicates a net cling effect in the atmsphere ver cean. Hwever, ver land and when the slar zenith angle is abut ø, the respnse f the ERBE shrtwave flux t changes in dust lading seems t be weak and nisy, while the bservatinshw a strng lngwave frcing in the same regin. The results f the TOA direct frcing due t airbrne Saharan dust n the daily and mnthly basis are summarized in Table 1. Fr "mist" cnditins, such as in the Sahel during summertime, we fund that there is a very weak respnse f the lngwave radiatin t the passage f the Saharan dust layer. This may be due t the fact that the effect f mineral aersl nearly cancels ut with the water vapr effect n the lngwave radiatin, as a result f the anticrrelatin fund between the precipitable water and the aersl ptical thickness in the Sahel regin. Since the Nimbus-7 TOMS data were taken near nn and the NOAA-9 ERBE measurements were made at arund 1430 LT this time difference culd ptentially intrduce uncertainties in ur results. T investigate uncertainties due t this time difference, we cnducted similar analyses using the Nimbus-7 TOMS measurements and ERBE data frm the ERBS satellite taken during July 30 t August 6, The time difference between measurements frm these tw satellites during this time was less than 30 min. We fund that the respnses f the

13 HSU ET AL.: RADIATION FORCING FROM TOMS AND ERBE DATA 20,661 lngwave and shrtwave fluxes t dust lading derived frm this experiment are similar t thse determined abve. This is prbably due t the fact that althugh the tempral variatin in cluds is large, the amunts f airbrne dust assciated with large-scale dust strms d nt change significantly between nn and early afternn (n a 1 ø x 1 ø grid average scale). Other factrs, which culd ptentially impact the results f ur analyses, include the uncertainty assciated with the effects f randm nise and natural atmspheric variability. Hwever, they are expected t be small since we use a large amunt f satellite measurements made ver a vast area during several mnths (February t July). On the basis f in situ measurements taken in different regins f the wrld, there is a very wide range in the ptical prperties f mineral aersls [Sklik and Tn, 1999]. Our results prvide useful cnstraints n the mdel assumptins f these dust aersl prperties and further reduce the uncertainty in determining the aersl effect n radiatin budget fr reginal and glbal climate mdels. The effects f trpspheric aersls and water vapr in climate frcing are cmplex in the real atmsphere. The techniques we describe in this paper can be applied fr ther parts f the wrld, as lng as the aersl layer is nt t clse t the surface (abve 0.5 km). Strnger cntributins f ther aersls, hwever, may mask such results (such as the cmbinatin f sulfates with Asian dust). With the advent f imprved data sets frm instruments such as the Cluds and the Earth Radiant Energy System (CERES) n TRMM and Terra, the Visible and Infrared Scanner (VIRS) n TRMM, ur ability t screen ut cludy and mist scenes will be significantly imprved, and we will be able t definitively answer questins n the effect f cluds and water vapr n radiative frcing. Acknwledgments. The authrs wuld like t thank J. Susskind and P. Pirain fr prviding the HIRS2 clud and water vapr data, D. Tanre and B. Hlben fr prviding the AERONET Sun phtmeter measurements, A. A. Lacis fr prviding the ptical prperties f dust aersls, and the NASA Langley Distributed Active Archive Center (DAAC) fr prviding the ERBE data. The radiative transfer cde was kindly prvided by Ming-Dah Chu. We wuld als like t thank the txv annymus reviewers fr their valuable cmments and suggestins, which helped t strengthen ur discussins and analysis. References Ackerman, S. A., and H. Chung, Radiative effects f airbrne dust n reginal energy budgets at the tp f the atmsphere, J. Appl. Meterl., 31, , Barkstrm, B. R., E. F. Harrisn, G. Smith, R. Green, J. Kebler, R. Cess, and the ERBE science team. Earth Radiatin Budget Experiment (ERBE) archival and April 1985 results, Bull. Am. Meterl. Sc., 70, , Chu, M., A slar radiatin mdel fr use in climate studies, J. Atms. Sci., 49, , Chu, M., and M. J. Suarez, An efficient thermal infrared radiatin parameterizatin fr use in general circulatin mdels, in Technical Reprt Series n Glbal Mdeling and Data Assimilatin, vl. 3, edited by M. J. Suarez, NASA Mem , Christpher, S. A., D. A. Kliche, J. Chu, and R. M. Welch, First estimates f the radiative frcing f aersls generated frm bimass burning using satellite data, J. Gephys. Res., 101, 21,265-21,273, Dieckmann, F. J., and G. L. Smith, Investigatin f scene identificatin algrithms fr radiatin measurements, J. Gephys. Res., 94, , Hansen, J. 15;., and A. A. Lacis, Sun and dust versus greenhuse gases: An assessment f their relative rles in glbal climate change, Nature, 346, , Herman, J. R., P. K. Bhartia, O. Trres, N. C. Hsu, C. J. Seftr, and E. Celarier, Glbal distributin f UV-absrbing aersls frm Nimbus-7/TOMS data, J. Gephys. Res., 102, 16,911-16,922, Hlben, B. N., et al., AERONET--A federated instrument netwrk and data archive fr aersl characterizatin, Remte Sens. Envirn., 66, 1-16, Hughtn, J. T., G. J. Jenkins, and J. J. Ephraums (Eds.), Climate Change: The IPCC Scientific Assessment, 362 pp., Cambridge Univ. Press, New Yrk, Hsu, N. C., J. R. Herman, P. K. Bhartia, C. J. Seftr, O. Trres, A.M. Thmpsn, J. F. Gleasn, T. F. Eck, and B. N. Hlben, Detectin f bimass burning smke frm TOMS measurements, Gephys. Res. Lett., 23, , ttsu, N. C., J. R. Herman, O. Trres, B. N. Hlben, D. Tanre, T. F. Eck, A. Smirnv, B. Chatenet, and F. Lavenu, Cmparisns f the TOMS aersl index with Sun phtmeter aersl ptical thickness: Results and applicatins, J. Gephys. Res., 104, , Intergvernmental Panel n Climate Change (IPCC), Radiative frcing, Radiative frcing f climate change and an evaluatin f the IPCC 1992 emissin scenaris, in Climate Change 1994, edited by J. T. Hughtn, L. G. M. Filh, J.P. Bruce, H. Lee, B. T. Callander, E. F. Hailes, N. Harris, and K. Maskell, 352 pp., Cambridge Univ. Press, New Yrk, Lia, H., and J. H. Seinfeld, Radiative frcing by mineral dust aersls: Sensitity t key variables, J. Gephys. Res., 103, 31,637-31,645, Penner, J. E., R. J. Charlsn, J. M. Hales, N. Laulainen, R. Leifer, T. Nvakv, J. Ogren, L. F. Radke, S. E. Schwartz, and L. Travis, Quantifying and minimizing uncertainty f climate frcing by anthrpgenic aersls, Bull. Am. Meterl. Sc., 75, , Seftr, C. J., N. C. Hsu, J. R. Herman, P. K. Bhartia, O. Trres, W. Rse, D. Schneider, and N. Krtkv, Detectin f vlcanic ash cluds frm Nimbus 7/TOMS, J. Gephys. Res., 102, 16,749-16,759, Smith, G. L., R. N. Green, E. Raschke, L. B. Avis, J. T. Suttles, B. A. Wielicki, and R. Davies, Inversin methds fr satellite studies f the earth's radiatin budget: Develpment f algrithms fr the ERBE missin, Rev. Gephys., 24, , Sklik, I. N., and O. B. Tn, Incrpratin f mineralgical cmpsitin int mdels f the radiative prperties f mineral aersl frm UV t IR wavelengths, J. Gephys. Res., 104, , Sklik, I. N., O. B. Tn, and R. W. Bergstrm, Mdeling the radiative characteristics f airbrne mineral aersls at infrared wavelengths, J. Gephys. Res., 103, , Susskind, J., P. Pirain, L. Rkke, L. Iredell, and A. Mehta, Characteristics f the TOVS Pathfinder Path A data set, Bull. Am. Meterl. Sc., 78(7), , Suttles, J. T., B. A. Wielicki, and S. Vemury, Tp-f-atmsphere radiative fluxes: Validatin f ERBE scanner inversin algrithm using Nimbus-7 ERB data, J. Appl. Meterl., 31, , Tanre, D., and M. Legrand, On the satellite retrieval f Saharan dust ptical thickness ver land: Tw different appraches, J. Gephys. Res., 96, , Tanre, D., E. Vermte, B. N. Hlben, and Y. J. Kaufman, Satellite aersl retrieval ver land surfaces using the structure functins, in Prceedings f the 12th Annual Internatinal Gescience and Remte Sensing Sympsium, IGARSS '92, 2, , Tegen, I., and A. A. Lacis, Mdeling f particle size distributin and its influence n the radiative prperties f mineral dust aersl, J. Gephys. Res., 101, 19,237-19,244, Tegen, I., A. A. Lacis, and I. Fung, The influence n climate frcing f mineral aersls frm disturbed sils, Nature, 380, , Trres, O., P. K. Bhartia, J. R. Herman, Z. Ahmad, and J. Gleasn, Derivatin f aersl prperties frm satellite measurements f backscattered ultravilet radiatin, Theretical basis, J. Gephys. Res., 103, 17,099-17,110, J. R. Herman, Cde 916, NASA/Gddard Space Flight Center, Greenbelt, MD N. C. Hsu, SSAI, Cde 916, NASA GSFC, Greenbelt, MD C. Weaver, SM & A Crpratin, Arlingtn, VA (Received July 19, 1999; revised February 3, 2000; accepted February 9, 2000.)