Search

Gridded fields (analyses) of global monthly precipitation have been constructed on a 2.5 [degrees] latitude-longitude grid for the 17-yr period from 1979 to 1995 by merging several kinds of information sources with different characteristics, including gauge observations, estimates inferred from a variety of satellite observations, and the NCEP-NCAR reanalysis. This new dataset, which the authors have named the CPC Merged Analysis of Precipitation (CMAP), contains precipitation distributions with full global coverage and improved quality compared to the individual data sources. Examinations showed no discontinuity during the 17-yr period, despite the different data sources used for the different subperiods. Comparisons of the CMAP with the merged analysis of Huffman et al. revealed remarkable agreements over the global land areas and over tropical and subtropical oceanic areas, with differences observed over extratropical oceanic areas. The 17-yr CMAP dataset is used to investigate the annual and interannual variability in large-scale precipitation. The mean distribution and the annual cycle in the 17-yr dataset exhibit reasonable agreement with existing long-term means except over the eastern tropical Pacific. The interannual variability associated with the El Nino-Southern Oscillation phenomenon resembles that found in previous studies, but with substantial additional details, particularly over the oceans. With complete global coverage, extended period and improved quality, the 17-yr dataset of the CMAP provides very useful information for climate analysis, numerical model validation, hydrological research, and many other applications. Further work is under way to improve the quality, extend the temporal coverage, and to refine the resolution of the merged analysis.

Comparisons of rain gauge and microwave satellite data for monthly rainfall indicate that these data sources cannot be used individually to produce global precipitation figures. At least five gauges are required to produce average monthly rainfall data for a grid of 2.5 degrees latitude by 2.5 degrees longitude. Microwave based satellites provide spatial coverage from 60 degrees north to 60 degrees south while infrared based data is available from 40 degrees N to 40 degrees S. Microwave data is suitable for tropical and subtropical ocean regions while gauge observations are most useful for tropical and subtropical land regions.

The Global Precipitation Climatology Project (GPCP) was established by the World Climate Research Programme to produce global analyses of area- and time-averaged precipitation for use in climate research. To achieve the required spatial coverage, the GPCP uses simple rainfall estimates derived from IR and microwave satellite observations. In this paper, we describe the GPCP and its first Algorithm Intercomparison Project (AIP/1), which compared a variety of rainfall estimates derived from Geostationary Meteorological Satellite visible and IR observations and Special Sensor Microwave/Imager microwave observations with rainfall derived from a combination of radar and raingage data over the Japanese islands and the adjacent ocean regions during the June and mid-July through mid-August periods of 1989. To investigate potential improvements in the use of satellite IR data for the estimation of large-scale rainfall for the GPCP, the relationship between rainfall and the fractional coverage of cold clouds in the AIP/1 dataset is examined. Linear regressions between fractional coverage and rainfall are analyzed for a number of latitude-longitude areas and for a range of averaging times. The results show distinct differences in the character of the relationship for different portions of the area. In general, to the south and east of the mountainous axis of Japan, rainfall and fractional coverage are highly correlated for thresholds colder than 245 K, and correlations can be increased by averaging in space and in time up to the dominant period of the precipitation events. To the north and west of the axis, the correlations between rainfall and fractional coverage, while generally smaller for all scales, are highest for thresholds warmer than 245 K. The proportional coefficients relating rainfall to fractional coverage at cold thresholds, however, differ greatly between the two periods and both differ significantly from those found for the GARP (Global Atmospheric Research Program) Atlantic Tropical Experiment. These results suggest that the simple IR-based estimation technique currently used in the GPCP can be used to estimate rainfall for global tropical and subtropical areas, provided that a method for adjusting the proportional coefficient for varying areas and seasons can be determined.