P3.20 THE USAGE OF GOES IMAGER CLEAR-SKY RADIANCE IN … · Figure 2 The clear-sky frac tion vs....
3
* Corresponding author address: Xiujuan Su, EMC (201), 52 00 Auth Rd., Ca mp Sp rings, MD 20746; e-mail [email protected] P3.20 THE USAGE OF GOES IMAGER CLEAR-SKY RADIANCE IN THE NCEP GLOBAL DATA ASSIMILATION SYSTEM Xiujuan Su 1 *, John Derber 2 , Jim Jung 3 , Yoshihiko Tahara 4 , Dennis Keyser 2 , Russ Treadon 2 1: SAIC, Camp Springs, Maryland 2: EMC/NCEP/NWS/NOAA, Camp Springs,Maryland 3: CIMSS, Camp Springs, Maryland 4: UCAR, Camp Springs, Maryland 1. INTRODUCTION The National Environmental Satellite, Data, and Information Service (NESDIS) and the Coop erative Ins titute for Me teorologic al Satellite Studies (CIMSS) have been producing Geostationary Operational Environmental Satellite (GOES) imager clear sky brightness tempe ratures (C SBT) s ince No vemb er 2001 . The data includes one visible and four infrared channels. Th e dat a we re ex amin ed fo r pos sible use in the NCEP Global Data Assimilation System (GDAS). Severe cloud contamination was found in the surface channel data when the satellite observations were compared to Aviation six-hour forecast. Feedback from NCEP and ECMWF prompted NESDIS and CIMSS to reexamine their cloud clearing scheme. Several changes were made to the CSBT production algorithm and the new CSBT product has been generated since July 2002. The product was examined with the GDAS system, and it was found that the quality has greatly improved, espec ially for surfac e chan nels. 2. DATA The CSBT data are generated hourly, but only data closest to the analysis times: 00, 06, 12, and 18 UTC are used. For the surface channels 2, 4, and 5 (3 .9 :m,10.7 :m, and 12.0 :m wavelength respectively), the data over land are n ot ex amin ed be cau se of unce rtaintie s in land surfa ce cha racteristics . In addition, da ta from channel 2 are discarded when the solar zenith angles are less than 90 degrees because of sun glint e ffects. The data were examined by comparing brightness temperatures derived from the GDAS profiles with the CSBT. The background departures (differences between observations and first guess) of the surface channels show that large negative values usually occur in areas whe re the clear sky fractio n is sm all, as show n in Figure 1. Figure 2 is a scatter plot of the bac kgro und depa rture s for th e sur face chan nel 4 verse clear-sky fraction, which shows that there is a larger tail at lower clear-sky fractions. The scatter plot of background departures for the channel 3 (6.7 :m, water vapor channel) vs. clear-sky fraction also have a tail at lower clear- sky fractions and are shown in Figure 3, although some spread at larger clear-sky-fractions also exists. These background departure features make it likely that the observations with larger negative background departure values are cloud contaminated. Time series of all background departures were examined. There are diurnal variations of background departures for surface channels. Closer examination showed a portion of the variation may have been caused by the lack of a diurnal variation in the SST used in the GDAS. Also, the time series confirms the mid- night effect for GOES-8 (Koepken, 2002, personal comm unication). 3. THE DATA QUALITY CONTROL Since the data show possible cloud contam ination, the n ext step is to perform data quality control to remove the problematic data. For this p urpose , an eight-d ay period , Augus t 3 to 10 2002, was chosen. Only quality control procedures and res ults for GOES-10 are
Transcript of P3.20 THE USAGE OF GOES IMAGER CLEAR-SKY RADIANCE IN … · Figure 2 The clear-sky frac tion vs....
* Corresponding author address: Xiujuan Su,EMC (201), 52 00 Auth Rd., Ca mp Sp rings, MD 20746; e-mail [email protected]
P3.20 THE USAGE OF GOES IMAGER CLEAR-SKY RADIANCE IN THE NCEP GLOBAL DATAASSIMILATION SYSTEM
Xiujuan Su 1*, John Derber2, Jim Jung3, Yoshihiko Tahara 4, Dennis Keyser2, Russ Treadon2
1: SAIC, Camp Springs, Maryland 2: EMC/NCEP/NWS/NOAA, Camp Springs,Maryland 3: CIMSS, Camp Springs, Maryland 4: UCAR, Camp Springs, Maryland
1. INTRODUCTION
The National Environmental Satellite, Data,and Information Service (NESDIS) and theCoop erative Ins titute for Me teorologic al SatelliteStudies (CIMSS) have been producingGeostationary Operational EnvironmentalSatellite (GOES) imager clear sky brightnesstempe ratures (C SBT) s ince No vemb er 2001 . The data includes one visible and four infraredchanne ls. Th e data we re ex amin ed fo r pos sibleuse in the NCEP Global Data AssimilationSystem (GDAS). Severe cloud contaminationwas found in the surface channel data when thesatellite observations were compared to Aviationsix-hour forecast. Feedback from NCEP andECMWF promp ted NE SDIS a nd CIM SS toreexamine their cloud clearing scheme. Severalchanges were made to the CSBT productionalgorithm and the new CSBT product has beengenerated since July 2002. The product wasexamined with the GDAS system, and it wasfound tha t the quality has greatly improved,espec ially for surfac e chan nels.
2. DATA
The CSBT data are generated hourly, butonly data closest to the analysis times: 00, 06,12, and 18 UTC are used. For the surfacechannels 2, 4, and 5 (3 .9 :m,10.7 :m, and 12.0:m wavelength respectively), the data over landare n ot examin ed be cause of unce rtaintie s inland surfa ce cha racteristics . In addition, da tafrom channel 2 are discarded when the solarzenith angles are less than 90 degrees becauseof sun glint e ffects.
The data were examined by comparingbrightness temperatures derived from the GDASprofiles with the CSBT. The backgrounddepartures (differences between observationsand first guess) of the surface channels showthat large negative values usually occur in areaswhe re the clear sky fractio n is sm all, as show n inFigure 1. Figure 2 is a scatter plot of thebackgro und depa rture s for th e sur face chan nel 4verse clear-sky fraction, which shows that thereis a larger tail at lower clear-sky fractions. Thescatter plot of background departures for thechannel 3 (6.7 :m, water vapor channel) vs.clear-sky fraction also have a tail at lower clear-sky fractions and are shown in Figure 3, althoughsome spread at larger clear-sky-fractions alsoexists. These background departure featuresmake it likely that the observations with largernegative background departure values are cloudcontaminated. Time series of all backgrounddepartures were examined. There are diurnalvariations of background departures for surfacechannels. Closer examination showed a portionof the variation may have been caused by thelack of a diurnal variation in the SST used in theGDAS. Also, the time series confirms the mid-night effect for GOES-8 (Koepken, 2002,personal comm unication).
3. THE DATA QUALITY CONTROL
Since the data show possible cloudcontam ination, the n ext step is to perform dataquality control to remove the problematic data.For this p urpose , an eight-d ay period , Augus t 3to 10 2002, was chosen. Only quality controlprocedures and res ults for GOES-10 are
Figure 1 The background departure forchannel 4 on 2002080306
Figure 2 The clear-sky fraction vs.the background departure forchannel 3 0n 2002080306
discussed here since the same principles areapplied to GOES-8. The general statistics for thedifferences between the observed and firstguess CSBT are listed in Table 1. There is anegative bias for the surface channels (2, 4, and5) and positive bias for the water vapor channel(3) between observed and the GDAS calculatedCSBT. Generally speaking, the large positivedepartures ( >=5K for the surface channels,>=10 .0 K fo r the W V channel) us ually o ccu r incoastal regions where the specification ofsurface properties (temperature, land/ocean,emissivity, etc.) is uncertain, and the largenegative departures usually occur where theclear sky fractions are small. To develop aquality pro cedure , we look ed at the s catter plotsof background departures verse clear skyfractions, and calculated the mean values ofbackground departure for every 10% clear skyfraction. Afte r exam ining these results, the datawith clear sky fractions less than 40% areconsidered to be cloud contaminated. Thecorrelation between background departure forthe surface channels (2, 4, and 5) and clear-skyfractions are much smaller ( 0.32, 0.46, 0.42 vs.0.24, 0.34, and 0.30) after the data with lessthan 40% clear-sk y fraction a re remo ved. Inaddition, the histograms of backgrounddepartures for all channels are more normaldistributed according to the P2 good ness of fittest. After removing the data which may becloud contaminated, the data with thebackground departures greater or less than twotimes sta ndard d eviations are also re move d. The ge neral statis tics for the d ata after qu alitycontrol are listed in Table 2. The backgrounddepartur e for c hannel 4 a fter qu ality contro l isshow n in Figure 4.
The quality control procedure describedabove is quite con servativ e, som e good datamay b e remo ved thro ugh the p rocedu re. Withthe incorporation of quality indicators in theCSBT, the usage of data and our quality controlproced ures m ay be m odified in the fu ture. Weare p rese ntly testing the assimilation of thewater vapor channel in our GDAS and theevaluating its impact on our forecast. We arealso planning to test, and eventually, assim ilatethe surface channels because of their significantnear su rface m oisture sig nal.
Table 1 The statistics of background departures (observed minus first guess)after our quality
Channel No. 2 3 4 5
Sample size 284266 747433 680571 680571
Mean (K) -0.65 2.96 -1.07 -0.90
Std . Dev . (K) 0.58 2.01 1.01 1.19
Max. Value(K) 24.01 17.32 18.04 15.56
Min . Va lue (K) -19.56 -16.56 -15.30 -26.96
Table 2 The statistics of background departures
(observed minus first guess)after our qualitycontrol procedures
Channel No. 2 3 4 5
Sam ple size 128050 376943 340411 339531
Mean (K) -0.50 2.98 -0.69 -0.49
Std Dev. (K) 0.37 1.56 0.58 0.68
Figure 3 The clear-sky fraction vs.the background departure for channel3 on 2002080306
Figure 4 The background departure forchannel 4 after quality control on
2002080306
![What causes the excessive response of clear-sky greenhouse ... · the response of clear-sky greenhouse effect to El Nin˜o warming. [5] The clear-sky greenhouse effect depends on](https://static.fdocuments.in/doc/165x107/5f1cfe30efefbd70c37e526c/what-causes-the-excessive-response-of-clear-sky-greenhouse-the-response-of-clear-sky.jpg)
![[Gokigenyou]_One Shot_Snowflakes Fluttering Down Through the Clear Sky](https://static.fdocuments.in/doc/165x107/577cc7791a28aba711a10a04/gokigenyouone-shotsnowflakes-fluttering-down-through-the-clear-sky.jpg)
