Variations of Daily Precipitation Intensity and Extremes in the Tropics from TRMM/TMPA and Other Satellite-Based Precipitation Datasets

Temporal variations of daily precipitation intensity and extremes in the tropics (30^oN-30^oS) are examined by means of the high-spatiotemporal resolution TRMM multi-satellite precipitation analysis (TMPA) and two other satellite-based products: GPCP-1DD and PERSIANN-CDR. Daily percentiles for each month are estimated from daily rain-rates for tropical land, ocean, and land+ocean, respectively, which are applied to represent a variety of precipitation intensity (and extremes). Consistency among these three products is assessed by comparing temporal variations in their respective percentiles. General consistencies can be obtained especially for the percentiles from the 70^th (Pct70) to 99.9^th over both tropical land and ocean. However, differences occur at both the very high and intermediate (low) ends, for instance, the 99.99^th and 60^th percentiles. Differences also appear in percentile magnitudes, suggesting possible discrepancies in these products' algorithms, input data resources, and native spatial and temporal resolutions, etc. Temporal inhomogeneity also exists specifically for percentiles corresponding to light rain events.

Primarily by means of TMPA, precipitation intensity variations are examined by focusing on the variations of precipitation percentiles on the interannual time scale. It is noted that the high percentiles such as the 99^th percentile (Pct99) tend to have a stronger positive correlation than tropical mean precipitation with tropical mean surface temperature that is dominated by the ENSO signal. Their corresponding interannual change-rates with surface temperature are about +14.0 and +6.0 %/K, respectively. There is also a strong negative correlation between intermediate percentiles, for instance Pct70, and surface temperature, resulting in a change-rate of about -22.3 %/K.

Comparisons with the results on the decadal/trend time scales from the monthly precipitation product (GPCP) are further made. It is found that the interannual relations between precipitation intensity and surface temperature derived from daily products are roughly in agreement with what are derived from monthly precipitation on interdecadal/trend time scale. Thus, despite the different processes and different temporal scales, precipitation intensity changes seem to have similar statistics in relation to surface temperature changes on two different time scales, though more detailed physical explorations are warranted likely through comparing with other precipitation products and climate model simulations.

The GPM IMERG product will be added to the daily statistics when available back to 1998.