Atmospheric soundings from AIRS/AMSU/HSB Dr. Joel Susskind and Dr. Robert Atlas Laboratory for Atmospheres NASA Goddard Space Flight Center Popular summary This is an extended abstract for a paper to be presented at the SPIE Defense and Security Symposium entitled "Atmospheric soundings from AIRS/AMSU/HSB." AIRSIAMSUMSB is an advanced Wmicrowave atmospheric sounding system launched on the Eos Aqua satellite on May 4,2002. The goals of AIRS/AMSU/HSB are to produce atmospheric soundings of temperature profile with RMS errors of tropospheric 1 km layer mean temperatures of lK, and 1 km layer integrated precipitable water vapor of 20%, in cases of up to 80% cloudiness. Soundings with this accuracy are needed to be able to significantly improve forecast skill. The status of AIRS soundings, as of March 15,2004, is presented. Results show that accurate AIRS soundings can be obtained in up to 80% cloud cover, and CLLII-dCC a-s-s-ll:li-ii-au on of AIRS temperature soundings significantly improves forecast skill. Atmospheric soundings from AIRS/AMSU/HSB Joel Susskind and Robert Atlas NASA Goddard Space Flight Center, Greenbelt, MD, USA 2077 1 ABSTRACT AIRS was launched on EOS Aqua on May 4, 2002, together with AMSU A and HSB, to form a next generation polar orbiting infrared and microwave atmospheric sounding system. The primary products of AIRS/AMSU/HSB are twice daily global fields of atmospheric temperature-humidity profiles, ozone profiles, sea/land surface skin temperature, and cloud related parameters including OLR. The sounding goals of AIRS are to produce 1 km tropospheric layer mean temperatures with an rms error of lK, and 1 km tropospheric layer precipitable water with an rms error of 20%, in cases with up to 80% effective cloud cover. Pre-launch simulation studies indicated that these results should be achievable. Minor modifications have been made to the pre-launch retrieval algorithm as alluded to in this paper. Sample fields of parameters retrieved from AIRS/AMSU/HSB data are presented and temperature profiles are validated as a bction of retrieved effective fractional cloud cover. As in simulation, the degradation of retrieval accuracy with increasing cloud cover is small. Select fields are also compared to those contained in the ECMWF analysis, done without the benefit of AIRS data, to demonstrate information that AIRS can add to that already contained in the ECMWF analysis. Assimilation of AIRS temperature soundings in up to 80% cloud cover for the month of January 2003 into the GSFC FVSSI data assimilation system resulted in improved 5 day forecasts globally, both with regard to anomaly correlation coefficients and the prediction of location and intensity of cyclones. Keywords: infra-red, microwave, remote sensing, temperature, moisture, high spectral resolution, clouds, forecast, meteorology 1. INTRODUCTION AIRS/AMSU/HSB is a state of the art advanced infra-red microwave sounding system that was launched on the EOS Aqua platform in a 1:30 AMPM sun synchronous orbit on May 4,2002. An overview of the AIRS instrument is given in Pagano et al’. The sounding goals of AIRS are to produce 1 km tropospheric layer mean temperatures with an rms error of lK, and layer precipitable water with an rms error of 20%, in cases with up to 80% effective cloud cover. Aside from being part of a climate mission, one of the objectives of AIRS is to provide sounding information of sufficient accuracy such that when assimilated into a general circulation model, significant improvement in forecast skill would arise. The pre-launch algorithm to produce level 2 products (geophysical parameters) using AIRS/AMSU/HSB data, and expected results based on simulation studies, are given in Susskind et al? The results of that simulation indicate that the sounding goals of AIRS/AMSU/HSB should be achievable. In that simulation, perfect knowledge of the instrumental spectral response functions and the inherent physics of the radiative transfer equations were assumed. Therefore, if the true state of the atmosphere and underlying surface were known perfectly, one could compute the radiances AIRS, AMSU, and HSB would see exactly up to instrumental noise. Susskind et a].’ alluded to the fact that this is not the case in reality, and additional terns would have to be included in the retrieval algorithm to account for systematic differences (biases) between observed brightness temperatures and those computed knowing the “true” surface and atmospheric state, as well as for residual computational errors after that systematic bias is accounted for (computational noise). In this paper, we show results based on the algorithm we were using to analyze AIRS/AMSU/HSB data on March 10,2004, which we will refer to as Version 3.5. This algorithm is very similar to the pre-launch version, with the major differences attributable to the factors described above. JPL delivered an earlier version of the algorithm, Version 3.0, to the Goddard DAAC, for the earliest near real time processing of AIRS level 2 products starting in August 2003. We have used Version 3.5 to analyze data for the AIRS focus day September 6,2002, and all of January 2003 for use in a forecast impact experiment. Research to further improve the results of analysis of AIRS/AMSU/HSB data is continuing. JPL plans to deliver an improved version to the DAAC in late 2004 to be used to process near real time AIRS data from that point forward, as well as reprocess all AIRS data from September 2002, when the instrument became stable. 2. OVERVIEW OF THE AIRS TEAM RETRIEVAL, ALGORITHM The AIRS team retrieval algorithm is basically identical to that described in Susskind et a1.* The key steps are outlined below: 1) Start with an initial state consistent with the AMSU A and HSB radiances3; 2) Derive IR clear column radiances Rovalid for the 3x3 AIRS Fields of View (FOVs) within an AMSU A Field of Regard (FOR) consistent with the observed radiances and the initial state; 3) Obtain an AIRS regression guess4 consistent with Ro *1 using 1504 AIRS channels; 4) Derive Ri consistent with the AIRS radiances and the regression guess; 5) Derive Ri all surface and atmospheric parameters using for 415 AIRS channels and all AMSU and HSB radiances; 6) Derive cloud parameters and OLR consistent with the solution and observed Ri; 7) Apply quality control, which rejects a solution if the retrieved cloud fraction is greater than 80% or other tests fail. In the event that a retrieval is rejected, cloud parameters are determined consistent with the initial microwave state and observed AIRS radiances. Figure 1 shows a typical AIRS spectrum and indicates by different colors the AIRS channels used in different retrieval steps which are performed sequentially. 3. RESULTS USING VERSION 3.1 Figure 2 shows the number of cases for each retrieved effective fractional cloud cover, in 0.5% bins, for the whole day September 6, 2002. The effective fractional cloud cover is given by the product of the fraction of the field of view covered by clouds and the cloud emissivity at 11 pm. The average giobai effective cioudiness was dekrmirzd to be 41.52%. Also shown is the percent of accepted retrievals as a function of retrieved effective cloud cover. Roughly 95% of the cases with retrieved effective cloud cover 5% were accepted, falling to 35% at 40% effective cloud cover, and to 18% at 8OYO effective cloud cover. All cases with retrieved effective cloud cover greater than 80% are rejected.' The average effective fractional cloudiness for all accepted cases was 25.1 8%. 300 280 260 240 220 I Percent Accepted vs. Effective Cloud F r a c ~ i o n 6500. '100.0 6000. 7500, ' 90.0 7000. ' 80.0 m 6500. 0 al a 6000. 70.0 I 0 "0 5500. u 5000. ' 60.0 c 4500. 0 50.0 4000. I c 3500. 40.0 c 0 8 a 3000. al E 2500. 30.0 ," I z3 2000. al ' 20.0 1 1500. 1000. ' 10.0 500. 0.0 0.0 ( 10. 20. 30. 40. 50. 60. 70. ,*.\,0*-0 n. nm 90. lo[1 . Percent C l o u d Fraction C I o b d T r'b%-t an -Number of Cases 41.52% -Percent Cases Accepted 25.18% Figure 2 Soptambar 6. 2002 1:30 Phi . . AIRS Cloud Paramoierm (mb) AIRS 477 to 5J5 nb Tlpmraturo (IC) m11 I I I I I l l 1 I I 2d.O 256.0 254.0 CLOlAL EAR= 25I.II STANDARD DEV= 0.49 ", AIRS Surfaoo Skin Tmporafurm (I() ~,A IRS Total Proofpltoblm Wator (om) 80 I n a 0 a *I a rrr EO ro 80 S *O 5 D **f. . .. i ilsal l 2l75l I tlolt l 1l171 183 iss $05 2I l l 1 CLOBAL YEAH= 2I1.18 STAIDARD DEV= 14.78 CLO8AL YEAH= 2.S7 STAIDARD OW= 1.S6 Figure 3 Figure 3a shows the retrieved effective cloud top pressure and effective cloud fraction for ascending orbits on September 6,2002. The global mean cloud effective cloud fraction and its spatial standard deviation are indicated in the figure. The results are presented in terms of cloud fraction in 5 groups, 0-20%, 20-40%, etc. with darker colors indicating greater cloud cover. These groups are shown in each of 7 colors, indicative of cloud top pressure. The reds and purples indicate the highest clouds, and the yellows and oranges the lowest clouds. Cloud fields are retrieved for all cases in which valid AIRS/AMSU observations exist. Gray means no data was observed. Figure 3b shows the retrieved 1 km thick 422 mb 535 mb layer mean temperature field. Gray indicates regions where either - no valid observations existed or the retrieval was rejected, generally in regions of cloud cover 80-1 00%. Figures 3c and 3d show retrieved values of surface skin temperature and total precipitable water vapor above the surface. Retrieved fields are quite coherent, and show no apparent artifacts due to clouds in the field of view. Water vapor has considerable more fine scale structure than temperature and contains some very large spatial gradients. Figure 4 shows the RMS difference between retrieved I km layer mean temperatures and the collocated ECMWF 3 hour forecast for all accepted cases as a hnction of retrieved effective cloud fraction. Results are shown for each of the 8 lowest 1 km layers of the atmosphere. Agreement degrades with increasing cloud cover, but only very slowly except in the lowest 1 km of the atmosphere. RMS temperature differences from ECMWF at all levels are somewhat larger than the 1 K goal for retrieval accuracy. Part of this difference can be attributed to the fact that the ECMWF forecast is not perfect. It is also possible that the accuracy of the ECMWF forecast may be somewhat poorer with increasing cloud cover. The increase in RMS temperature differences at 0% cloudiness is somewhat misleading because a large percentage of clear cases occurred over Antarctica on this day. AIRS RMS Temperature Difference from ECMWF vs. Effective Cloud F r a c t i o n - - 778 to 684 mb 684 to 618 mb -6 18 to 535 mb -5 35 to 477 mb 477 to 407 mb 407 to 344 mb Figure 4 2002 1:JO PM 477 to A5 k5S m rbn f1n m=iIr' (I0 GLOBAL IIEUL. -0.1) Slb 1-18 COIR= 0.8D ELODAL MEA* 0.S2 STD= 0.8s COIR= 1.60 863 to A7I7R7S drnf nTmr8r pEaCr@ '. ('1 Toto1 PArI ISo lpmlltnoubrl t?Wr (-1 i l l I I , I I I I I -!!?!!??.I -0'.5 d.J -6.1 6.0 6 . f - 0 GLOBAL YE* 0.18 STP 1.01 CORb 1.00 CLODAL YEAN= 0.02 STP 0.U CORR- 0.D8 Figure 6 Figure 7 Figure 7 shows the number of cases as a function of differences from ECMWF for ocean surface skin temperature 50iN-50iS, and for global values of 1 km layer mean temperatures. Also indicated on this figure for each is the percentage of cases with differences from the mean greater than 3K (outliers). The distribution of the differences is close to Gaussian, with only a very small percentage of outliers. This makes the data suitable for use in satellite data assimilation, even under cloudy conditions. 4. FORECAST IMPACT EXPERIMENTS The data assimilation system used in the experiments is FVSSI which represents a combination of the NASA Finite Volume General Circulation Model (FVGCM)’ with the NCEP operational Spectral Statistical Interpolation (SSI) global analysis scheme implemented at lower than the operational horizontal resolution - T62. The basics of the finite-volume dynamical core formulation are given in DAO s Algorithm Theoretical Basis Document (see httD://Doiar.gsf~nasagov/sci research/atbd.ehD), and the FVGCM has been shown to produce very accurate weather forecasts when run at high resolution.’ The AIRS temperature profiles produced by SRT were presented to the SSI analysis as rawinsonde profiles with observational error specified at 1 iK at all vertical levels. Results are presented for three sets of experiments in which data was assimilated for the period January 1 January - 31,2003. Five day forecasts were run every two days beginning January 6,2003 and forecasts every 12 hours were verified against the NCEP analysis, which was taken as truth . In the first experiment, called control , all the data used operationally by NCEP was assimilated, but no AIRS data was assimilated. The operational data included all ronventiom! hi., TGVS ZK! .*.TC)VS :dirt-.ces fGi XCM-:4, 15, aid 16, cloud iracked winds, SSMii rotai precipitable water and surface wind speed over ocean, QuikScat surface wind speed and direction, and SBUV ozone profiles. In the second and third experiments, called clear AIRS and all AIRS , temperature profiles retrieved from AIRS soundings were assimilated in addition to the data included in the control experiment. Clear ocean included all accepted temperature retrievals derived from AIRS over ocean and sea ice in cases where the retrieved effective cloud fraction derived from AIRS was less than or equal to 2%, while the all ocean experiment assimilated accepted AIRS temperature soundings over ocean and sea ice for all retrieved cloud fractions. Figure 8 shows anomaly correction coefficients of forecast sea level pressure verified against the NCEP analysis for both Northern Hemisphere extra-tropics and Southern Hemisphere extra-tropics for both the control and all AIRS experiments. In the Northern Hemisphere, addition of all AIRS soundings resulted in an improvement in average forecast skill of the order of 1 hour or less, but an improvement in average forecast skill in the Southern Hemisphere on the order of 6 hours results from assimilation of AIRS soundings. Assimilation of AIRS soundings under essentially clear conditions (not shown), resulted in somewhat poorer forecasts than using all AIRS soundings. It should be noted that the Aqua orbit (1 :30 ascending) is almost identical to that of NOM 16 carrying HIRS3, AMSU A and AMSU B, so AIRS/AMSU/HSB soundings are providing additional information to that contained in the AMSU NAMSU B radiances on NOAA 16 in the same orbit. Figure 9 shows the RMS position error (km) and magnitude error (ma) for 5 day forecasts of extra tropical cyclones in the three experiments. It is apparent that addition of AIRS soundings improved RMS forecast skill for both the position and magnitude of extra-tropical cyclones globally, and addition of AIRS soundings in partially cloudy areas further improved forecast skill as compared to use of soundings only in essentially clear conditions. Several thousand cyclones verifications are included in these statistics. Addition of AIRS data did not improve forecasted cyclone position and intensity for each cyclone. Some were improved substantially however. Figure 10 shows the impact of AIRS data on the 24 hour forecast of position and intensity of tropical storm Beni, which was centered roughly 4i east of New Caledonia on January 3 1, 2003 with a central pressure of 990 mb (see Figure 1Od). The control forecast (Figure loa) produced a relatively weak cyclone (1007 mb) displaced considerably to the northwest, while the 24 hour forecast using AIRS data (Figure 1O b) was much more accurate in both position and intensity (995 mb). It is significant to note that our forecast using AIRS data was more accurate in both position and intensity than the NCEP operational forecast (Figure 1Oc) in this case, which, even though it used a higher resolution model and analysis system, did not have the benefit of AIRS data. The results shown indicate the potential of AIRS soundings to improve operational forecast skill. We are working with NCEP to arrange an experiment to incorporate the addition of AIRS temperature soundings to the NCEP operational analysis and compare the results to the otherwise equivalent run on the NCEP computing system to see the extent, if any, that operational forecast skill can be improved upon. Figures 5a and 5b show RMS differences of temperature and moisture profiles from the truth with both simulated and real data. The gray and black curves reflect all accepted cases, and the pink and red curves are cases identified as clear, for simulated and observed radiances respectively. For temperature, I km layer mean differences from the truth are shown, and for water vapor, % differences in total integrated water vapor in 1 km layers are shown. In simulation, the truth is known perfectly, while with real data, the 3 hour ECMWF forecast is taken as a proxy for truth . For real data, as in simulation, temperature retrievals under cloudy conditions (roughly 48% of all cases are accepted) degrade by only a few tenths of a degree compared to cases identified as clear (3.6% of the cases are identified as clear), while water vapor retrievals do not degrade at all. Differences from truth are poorer with real data than in simulation. Two major causes of degradation are: 1) perfect physics was assumed in simulation; and 2) the truth has errors in real data. The degradation of soundings in the presence of real clouds, as compared to soundings in clear cases, appears to be similar to that implied by simulation. LAYER MEAN RMS TEMPERATURE ('C?, GLOBAL DIFFERENCES FROU TRUTH 1 Km LAYER PRECIPITABLE_WATER PERCENT DIFERENCES FROM TRUTH" 2.15 20 1 I 3.34 235 5.87 Jon 11.0 A f 20.9 hn 359 E v 32.2 U 407 Ya w 44 1 p: v3In ) 60.9 3v) 515 w wv) pn: 103. pn: 596 160. 706 314. 802 535. 904 777. 1100. I I 1100 0.5 1 1.5 2 2.5 3 3:s 4 4.5 RMS - - - - - All Accepted Cases Simulated 63.9% - All Aooaptmd Cases (5lmulat.d) 63.9% - Clear Cases Simulated] 6.0% - Clear Coaes (Slmulotad) 8.0% All Accepted Cases Observed 48.1% AI1 Aooaptwd Cases (Obsarvrd) 48.1% Clear Cases Observed] S.6% Clear Casea (Observed) 3.6% Figure 5a Figure 5b Figures 6 a-d show ascending orbit differences from the 3 hour ECMWF forecast for ocean surface skin temperature, 477 mb - 535 mb temperature, 683 mb - 772 mb temperature, and totalprecipitable water. In temperature fields, white represents agreement with ECMWF forecast within OSK, shades of red indicate AIRS is warmer than ECMWF, and shades of blue indicate AIRS is cooler, with each color showing intervals of 1K (e.g. 0.5K-1.5K for the first shade of red). In the case of water vapor, each interval is 0.2 cm. The global mean difference from the ECMWF forecast, as well as the spatial standard deviation and the correlation coefficient is shown in each figure. Temperature differences from ECMWF are generally small and spatially coherent. Part of this is due to retrieval error, but part is also due to errors in the ECMWF forecast field, which is not perfect. The small biases in retrieval atmospheric temperatures are most likely due to real errors, and results from imperfect parameterization of the physics giving rise to the observations. We are continuing to work on improving this. Differences in total precipitable water vapor from ECMWF have much smaller scale and are due in part to the ECMWF forecast smoothing out the high spatial variability of water vapor. SEA LEWL PRESSURE Northern Hemisphere Global Extratropical Cyclone Forecast Error from 11 Five-day FvsSl Forecasts UR = Contrdtcleor AM( ~ow;I) ALL = Controltoll ARS (global) 325. h E 1 0.6 1 2 3 4 v DAY SEA LEVEL PRESSURE Southsrn Aemiqhere 1 2 3 4 Position Error Mognitude Error DAY Figure 9 Impact of AIRS on 24hr Forecast of Sea Level Pressure a Control 24hr Forecast b AIRS 24hr Forecast C d NCEP AVN 24hr Forecast Verifvina Analvsis January 31,2003 002 Figure 10 REFERENCES 1. Pagano, T. S., H. H. Aumann, D. E. Hagan, and K. Overoye, Prelaunch and in-flight radiometric calibration of the Atmospheric Infrared Sounder (AIRS). IEEE Tmns. Geosci. Remole Sensing, 41, 265-273, February 2003. 2. Susskind, J., C. D. Barnet, and J. M. Blaisdell, Retrieval of Atmospheric and Surface Parameters from AIRS/AMSU/HSB Data in the Presence of Clouds. IEEE Trans. Geosci. Remote Sensing, 41, 390-409, February 2003. 3. Rosenkranz, P. W., Retrieval of temperature and moisture profiles from AMSU-A and AMSU-B measurements. In Proc. IGARSS, 2000. 4. Goldberg, M. D., Y. Qu, L. M. McMillin, W. Wolff, L. Zhou, and M. Divakana, AIRS near-real-time products and algorithms in support of operational numerical weather prediction. IEEE Trans. Geosci. Remote Sensing, 41, 379-389, February 2003. 5. Lin, S.J., R. Atlas, and K. S. Yeh, January-February, Global weather prediction and high end computing at NASA. Computing in Science und Engineering, 29-35,2004.