Multisatellite observed responses of precipitation and its extremes to interannual climate variability

The consistency of precipitation variability estimated from the multiple satellite-based observing systems is assessed. There is generally good agreement between TRMM TMI, SSM/I, GPCP and AMSRE data sets for the inter-annual variability of precipitation since 1997 but the HOAPS data set appears to overestimate the magnitude of variability. Over the tropical ocean the TRMM 3B42 data set produces unrealistic variability. Based upon deseasonalized GPCP data for the period 1998-2008, the sensitivity of global mean precipitation (P) to surface temperature (T) changes (dP/dT) is about 6%/K, although a smaller sensitivity of 3.6%/K is found using monthly GPCP data over the longer period 1989-2008. Over the tropical oceans dP/dT ranges from 10 to 30%/K depending upon time period and data set while over tropical land dP/dT is -8 to -11%/K for the 1998-2008 period. Analyzing the response of the tropical ocean precipitation intensity distribution to changes in T we find the wetter area P shows a strong positive response to T of around 20%/K. The response over the drier tropical regimes is less coherent and varies with data sets, but responses over the tropical land show significant negative relationships over an interannual time-scale. The spatial and temporal resolutions of the data sets strongly influence the precipitation responses over the tropical oceans and help explain some of the discrepancy between different data sets. Consistency between data sets is found to increase on averaging from daily to 5-day time-scales and considering a 1 degrees (or coarser) spatial resolution. Defining the wet and dry tropical ocean regime by the 60th percentile of P intensity, the 5-day average, 1 degrees TMI data exhibits a coherent drying of the dry regime at the rate of -20%/K and the wet regime becomes wetter at a similar rate with warming.