Trends of Heavy Precipitation Events in Global Observation and Reanalysis Datasets

Abstract

Trends of heavy-precipitation events represented by available global observations such as the Global Precipitation Climatology Project (GPCP) and four global reanalysis datasets are examined using two indices, OPC (occurrence of rainy days within precipitation class) and NPC (normalized precipitation within precipitation class), mainly over the tropics. These indices are defined by normalization with the total integrated precipitation amount for the analysis period, 1979-2001.
Over land, the observational datasets exhibit larger decreasing trends in the heavier classes. All but one of the reanalysis datasets reproduce these tendencies. Significant correlations for the heaviest class between the GPCP and the other datasets can be found.
Over sea, there was a large discontinuity around 1987 in the two indices for the GPCP pentad data, which was probably caused by the introduction of Special Sensor Microwave Imager (SSM/I) data. Restricted to data after 1988, all of the reanalysis datasets are seen to have positive and more increasing tendencies in the heavier classes, while the observational results show the opposite tendencies, although they are statistically insignificant.
To resolve this discrepancy in order to enable the use of reanalysis data in extreme-event studies, additional improvements will be needed, not only in reanalysis but also in observational datasets.