Adv. Space Res. Vol. 9, No. 7, pp. (7)311 (7)316, 1989 Printed in Great Britain.
0273 1177/89 $0.00 +.50 1989 COSPAR
THE G L O B A L PRECIPITATION C L I M A T O L O G Y PROJECT Phillip A. Arkin Climate Analysis Center, NMC/NWS/NOAA, Washington, DC 20233, U.S.A.
ABSTRACT The Global Precipitation Climatology Project of the World Climate Research Programme is an effort to use a variety of satellite estimates of rainfall together with station observations to derive a 10-year climatology of monthly rainfall for areas of 2.5 ° latitude/longitude. The Project will use rainfall estimates from both geostationary and polar orbiting satellites along with station observations to perform the global analysis. In the tropics, the primary data source used will be geostationary IR imagery, while passive microwave radiometric observations will be used over the extratropical oceans. A mix of data sources will be used over the continents. In this paper, we describe the various components of the Project, describe the process by which the operational geostationary satellite data are merged and the rainfall estimates derived, and show examples of tropical rainfall variability during 1987. BACKGROUND The need for analyses of large-scale areally averaged precipitation for periods of 5 days to monthly covering at least the tropics to satisfy the requir~ents of the World Climate Research Programme (WCRP) became clear in early 1985 as the Tropical OceanGlobal Ahaosphere (TOGA) Programme was beginning. A Workshop on Global Large-Scale Precipitation Data Sets for the WCRP was held in Camp Springs, MD, USA in July 1985, and recommended /i/ that a Global Precipitation Climatology Project (GPCP) be instituted to use available algorithms and data to produce these analyses. Its objectives include: a. the ~roduction of estimates of tropical precipitation for areas of 2.5 ° latitude x 2.5 longitude for periods of 5 days using geostationary satellite data; b. the derivation of rainfall estimates for similar spatial scales and for months from passive microwave radiometric data; c. the use of these data together with conventional rainfall data to produce analyses of monthly rainfall on a near-global scale; and d. the development of a capability to validate and calibrate the various satellite estimates. ORGANIZATION The organization of the GPCP is shown in Figure i. The production of estimates from geostationary data requires the routine production of statistics, including histograms, means and spatial variances, of IR brightness temperatures for 5-day periods from each of the available satellites. This activity is carried out at the several Geostationary Satellite Data Processing Centres (GSDPCs). The integration of these data into rainfall estimates for the global tropics is done by the Geostationary Satellite Precipitation Data Centre (GSPDC). Data from the SSM/I (Special Sensor ~icrowave/Imager) instr~ent on the U.S. Defense Meteorological Satellite Program polar orbiting satellite is processed at the Polar Satellite Data Processing Centre (PSDPC) add sent to the Polar Satellite Precipitation Data Centre (PSPDC), where it is used to obtain estimates of monthly oceanic precipitation. Both the Precipitation Data Centres use calibration and validation information provided by the Surface Reference Data Centre (SRDC) , where developmental work on new methods of measuring rainfall over the oceans is also coordinated. The Global Precipitation Climatology Centre (GPCC) receives the various satellite estimates and, by combining them with conventional rainfall measurements, produces monthly analyses of near-global rainfall.
GEOSTATIONARY SATELLITE DATA PROCESSING CENTRES
GEOSTATIONARY SATELLITE PRSCIPITATION DATA CENTRE
POLAR SATELLITE DATA PROCESSING CENTRE SURFACE REFERENCE DATA CENTRE
i POLAR SATELLITE pRECIPITATION DATA CENTRE
GLOBAL PRHCIPITATION CLIMATOLOGY CENTRE
WORLD DATA CENTRES FOR METEOROLOGY
Figure i. Organization and data flow of the Global Precipitation Climatology Project.
CURRENT STATUS Geostationary Component The GSDPCs for each of the geostationary satellites are currently producing the statistics required by the GPCP, and, with the exception of that for INSAT, have begun deliveries of data. The GSDPC for GOES is co-located with the GSPDC and has been producing data routinely for both satellites since July 1987. The GSDPCs for Meteosat and (~4S, operated by the European Space Agency under contract from Eumetsat and by the Japan Meteorological Agency (JMA) respectively, have been delivering data since late 1987. JMA has completed post-processing and delivery of older data as well. India has not yet agreed to participate in the GPCP. However, the functional duties of an INSAT GSDPC have been carried out since June i, 1986 by the India Meteorolog'ical Deparhnent as a result of a collaborative effort under a joint Indo-U.S. research program. The GSPDC has been staffed and has begun processing data. A microcomputer workstation has been obtained to make the development of merging and quality control algorithms feasible, and results have been obtained from two 30-day periods during 1987. Rainfall estimates are computed from each satellite for each 5-day period, and then merged into an analysis covering the global tropics. The initial merging algorithm uses a weighted average whenever estimates from two satellites overlap, with the weights inversely related to the distance from the sub-satellite point. Figure 2 shows the results for July 30-August 28, 1987. The gap in the Indian Ocean and over India results from the absence of data from INSAT. Figure 3 shows a similar result for the period December 2-31, 1987, with the exception that no data were available from Meteosat or INSAT. In an attempt to remove the gap due to the lack of geostationary data from one or more satellites, the use of estimates based on histograms of outgoing longwave radiation derived from IR window channel observations on the NOAA polar orbiting satellites is being tested. Figure 4 shows the result of using such estimates when geostationary estimates wore unavailable. While the polar orbiter estimates suffer from relatively poor sampling of the diurnal cycle, they do appear to provide usable auxiliary information. Histograms from two polar orbiting satellites are available beginning in June 1988, and should give improved results.
Global Precipitation Climatology Project
.~.,,". . . . . . . .
..,-~_-.-~,_..,. ,r~-<.~. ......
Jo. I; ~ 199E
,oto ,oo.o ,o, o
o _ . / " " - s.
- - - I'~'~ .
. . . .
Figure 2. Merged geostationary estimates for July 30-August 28, 1987 using data from GOES-E, GOES-W, QMS and Meteosat. Solid contours at 200 mm intervals beginning from 100 mm; dashed contours at 25 and 50 mm.
Polar Orbiting Component Estimates of rainfall based on passive microwave observations from the SSM/I (Special Sensor Microwave/Imager) instrument on the latest polar orbiting satellite of the Defense Meteorological Satellite Program offer considerable potential for supplementing and improving the geostationary estimates. The satellite was launched on June 19, 1987, and data began to be produced near the end of that month. No formal commitments for either the PSDPC or the PSPDC have been received, although initial data processing as well as development and testing of estimation algorit~s is under way. SSM/I data will be archived at NOAA/NESDIS, and the possibility that the Satellite Data Services Division of NESDIS could provide the PSDPC is being explored. In the interim, data for a period of up to one year from launch are being supplied by a NASA/NESDIS contractor to a group headed by T. Wilheit at NASA/Geddard Space Flight Center. This group has agreed to provide the PSPDC provided a satisfactory arrange~nent for supply of the data can be made. They have presently received data for approximately 60 days, and have processed roughly half into a form suitable for testing of the preliminary estimation algorithm. This algorithm, which uses 18 GHz data and a physically-based model of the absorption due to raindrops, is now being tested. Another algorithm, based on scattering due to ice at 86 GHz, is being developed by R. Spencer of NASA/Marshall Space Flight Center.
P . A . Arkin
EQ J00. O,
~ 3 o o . o--..~\\ ~;.~.~..p.~. o
;~,~--'-~--~:~.~ ~, ~=25.o
•tOS _ _
I~Q,. ' ~ u ,
400 0 '
,J .M ...........~
_3 .... 20N
Figure 3. As in Figure 2 except for the period December 2-31, 1987, and without Meteosat data.
Calibration and Validation The use of indirect estimates of rainfall (such as those based on observations of clouds or ice particles) requires calibration and validation in order to be usable with confidence. The Calibration/Validation (Cal/Val) Program of the GPCP consists of two principle efforts: the acquisition and organization of surface observations of rainfall from areas representing different climatic regimes; and the sponsorship and coordination of efforts to develop new methods of observing rainfall over the oceans. A Workshop of the Validation of Satellite-Based Precipitation Estimates for the GPCP /2/ was held in Nov~aber 1986. One of its principal recorrmendations was that a Surface Reference Data Center (SRDC) be established to coordinate this effort. The National Climatic Data Center of NOAA/NESDIS has expressed interest in providing the SRDC, and initial development is udder way. Efforts to acquire and format data from primary and secondary calibration/validation sites have begun. The Cal/Val Project has arranged for an experiment in the application of new rainfall measurement techniques to oceanic rainfall to be conducted during the June-September 1989 period. A subsurface acoustic sensor and an optical rain gauge are to be mounted on a buoy in the Atlantic Ocean near Kennedy Space Center (KSC) add within range of the KSC rain-measuring radar. If usable rainfall measurements can be obtained, the possibility of mounting such instruments on buoys in the tropical Pacific will be explored.
Global Precipitation Climatology Project
-- 1 00
~ 44 14
1I 0 OW W
W 88 ~4
~ 4 W
~oj~.y. ,oro~ ,.ol.o
300.0 400.0 EO
600.0 600 .~ l h 3O0.0
Figure 4. As in Figure 3 except using NOAA polar orbiting AVHRR data to fill in where geostationary data were unavailable. FUTURE DIRECTIONS A critical component of the GPCP not mentioned above is the Global Precipitation Climatology Center (GPCC). While the Federal Republic of Germany has offered to provide the GPCC, it has not yet begun operating. Its successful implementation is one of the major challenges of the coming year. The transition to routine operations will be a major goal for the GSPDC, with the emphasis on implementing a merging algorithm, completing the post-processing of data from 1986 and 1987 and meeting the nominal data processing schedules. The polar orbiting/microwave component of the GPCP will focus on getting c~naitments for the PSDPC and the PSPDC, and on developing and implementing preliminary algorithms based on absorption and scattering mechanisms. The Cal/Val Project will need a commitment for the SRDC and some preparatory work to begin the process of producing validation data sets, as well as a successful conclusion to the KSC buoy experiment and the development of a follow-up. While these operational objectives are critical to the success of the GPCP, it is important to focus on the research requir~nents as well. It is clear that the current state of the art in all of the activities of the Project is far from perfect. Research is needed to improve the currently used geostationary algorithm and to investigate new vis/IR-based techniques, to implement one or more microwave algorithms and produce monthly estimates from SSM/I data and to determine the characteristics required for usable validation data sets. The most critical research task facing the GPCP, and one which will be critical to the future of such WCRP programs as GEWEX, is the development of methods of analyzing rainfall which can make optimal use of multiple sources of information, such as station observations, radars, various satellite estimates, circulation data and model forecasts of precipitation. All of these data types contain (or at least will in the future contain) information on the distribution and intensity of large-scale rainfall. The challenge is to begin to learn how to use that information in some optimal fashion, possibly through some process analogous to that used by operational numerical weather prediction centers in producing circulation analyses to be used as initial conditions.
World Climate Research Programme, Global Large-Scale Precipitation Data Sets for the WCRP, WCP-III, ~ O / T D - No. 94, Geneva, Switzerland (1986)
World Climate Research Programme, Validation of Satellite-Derived Precipitation Estimates for the Global Precipitation Climatology Project of the WCRP (in press) , Geneva, Switzerland (1988)