Currently available long-term satellite rainfall data sets from infrared and passive microwave based retrieval techniques show an increase in tropical precipitation associated with the El Niņo/Southern Oscillation (ENSO). This suggests an enhanced hydrologic cycle associated with the warm phase of ENSO, which is not captured by current global climate models. There are, however, significant differences in ENSO related variability of tropical rainfall between current retrieval algorithms. Of particular concern are large differences in the response of tropical rainfall to the 1997/98 El Niņo between state of the art retrievals based on passive microwave and precipitation radar estimates from the TRMM microwave imager (TMI) and precipitation radar (PR). While both instruments indicate significant increases in east Pacific rainfall and a decrease in rainfall over the warm pool region during El Niņo, there is a significant increase in global tropical rainfall in the TMI estimates, but little or none in the PR estimates. Relatively good agreement elsewhere indicates that the issue of interannual variability of global tropical rainfall is primarily dependent on variability within the east and west Pacific regions.

To understand the reason for this discrepancy in the interannual rainfall variability of satellite retrieval algorithms and to determine the true nature of tropical rainfall variability it is important to determine the sensitivity of retrieval algorithms to variations in cloud properties. This includes such factors as freezing height, ice content, ice density, drop size distribution etc. One result of particular interest to this issue is the discovery of a much larger difference between the height of the freezing level, or zero degree isotherm, and the radar bright band, or melting layer, in the east Pacific than in the west. This discrepancy, however, disappears during El Niņo. Since passive microwave retrieval techniques from SSM/I and TMI measure the total column liquid water, an accurate determination of the height of the rain column is critical to the determination of the surface rain rate. Current retrieval algorithms for SSM/I and TMI use the freezing level height in the calculation of a surface rain rate. Because the melting layer is lower, however, this leads to an underestimate in the surface rain rate. In addition, the east/west variation in the height of the melting layer relative to the freezing level will result in a decrease in the interannual variability of global tropical rainfall from the passive microwave estimates. Other issues such as differences in the drop size distributions between the two regions are also being investigated to understand what the true response of the total global tropical rainfall is to changes associated with ENSO.