INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 30: 110–119 (2010) Published online 23 February 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/joc.1875 Validation of the abrupt change in GPCP precipitation in the Congo River Basin Xungang Yin* and Arnold Gruber Cooperative Institute for Climate Studies, Earth System Science Interdisciplinary Center, University of Maryland, College Park, MD 20740, USA ABSTRACT: The Global Precipitation Climatology Project (GPCP) monthly precipitation exhibits a significant negative trend during 1979–2004 over southern tropical Africa from the Congo River Basin to the east coast. This trend appears as a more than 20% drop beginning in 1992 in the 6-year and 9-year averages of the areal mean GPCP satellite-gauge precipitation, whose magnitude is largely determined by the gauge analyses. This paper’s analysis of satellite precipitation estimates, gauge precipitation analyses, and gauge coverage information suggests that the negative precipitation trend is only true in part of southern tropical Africa but the magnitude is much smaller than that calculated from the GPCP. In the eastern portion of the region, the precipitation drop in the GPCP is confirmed by the satellite-only estimates but the decrease of more than 16% is amplified by a change in gauge coverage. In the western portion of the region, basically the southern Congo River Basin, all gauge dependent products show a negative precipitation trend, which is much larger in the GPCP merged satellite-gauged data set, but not supported by the satellite-only precipitation estimates. In this study we conclude that for the Congo River Basin, where both the mean precipitation and its spatial gradient are high, the spurious negative trend detected in the GPCP precipitation is caused by a significant change in local gauge coverage and the methodology used by the GPCP to merge satellite and gauge data during the analysis period. Copyright  2009 Royal Meteorological Society KEY WORDS GPCP; the Congo River Basin; precipitation trend; satellite; gauge Received 3 September 2007; Revised 15 December 2008; Accepted 20 January 2009 1. Introduction Trend analysis, particularly for temperature and precipi- tation, is an important component in the study of global change (IPCC, 2007). Although it is widely accepted that the global mean surface temperature has increased by 0.6 ° C in the twentieth century, our knowledge of pre- cipitation trends during the same time is still limited. Since precipitation is highly variable and discontinuous in space and time, and gauge sampling is frequently inad- equate, changes in precipitation are difficult to detect. Nevertheless, efforts have been made on precipitation data mining to increase our understanding of future pre- cipitation trends under the scenario of climate change (e.g. Karl and Knight, 1998; Morrissey et al., 1996; Dai et al., 1997; New et al., 2001). The Global Precipitation Climatology Project (GPCP) data set (Huffman et al., 1997; Adler et al., 2003) is one of the few precipitation products that take advan- tage of both satellite estimates and gauge analyses to provide global coverage of monthly mean precipitation on 2.5 ° latitude/longitude grids. The GPCP combines the precipitation information available from each source (satellite infrared and microwave estimates of rainfall and gauge observations) into a final merged product, taking * Correspondence to: Xungang Yin, STG Inc., 151 Patton Ave, Asheville, NC 28801, USA. E-mail: xungang.yin@noaa.gov advantage of the strengths of each data type and remov- ing biases based on hierarchical relations in a stepwise approach (Adler et al., 2003). The last step in the bias removal is to adjust satellite estimates to the average of gauge measurements over a 5 × 5 grid box or a 7 × 7 grid box depending on the availability of gauges within the array. In the final step, the gauge adjusted satellite estimates and the gauge analyses at each grid box are combined in a weighted average (Huffman et al., 1995). This data set has been widely used in global change stud- ies (e.g. New et al., 2001; Hicke et al., 2002; Curtis and Adler, 2003; Seager et al., 2005; Lau and Wu, 2006; Smith et al., 2006; Gu et al., 2007) and also in social sci- ence studies (e.g. Miguel et al., 2004; Funk et al., 2005). Even with both conventional (gauge) and modern (remote sensing) approaches available for precipitation measurements, caution should be taken when using the GPCP data set, especially since it is a combination of var- ious data inputs. Similar to other analyses and reanalyses data sets, the GPCP result is only an approximation of the truth under the current knowledge and technology. At present, comparison and intercomparison with other pre- cipitation products are the most effective way to validate an analysis and reanalysis precipitation data set. In the past decade analysis and reanalysis, products have been greatly improved with the help of satellite observations, enhanced computer power, and improved analysis tech- niques. In a series of studies, the GPCP monthly mean Copyright  2009 Royal Meteorological Society