| Tropical Rainfall Measuring Mission (TRMM) satellite measurements from the precipitation radar and TRMM microwave imager have been combined to yield a comprehensive 3-year database of precipitation features (PFs) throughout the global tropics.{09}This study presents an analysis of the diurnal cycle of the observed PFs' rainfall amount, frequency, intensity, convective-stratiform portioning, and convective intensity. Over the oceans, the diurnal cycle of rainfall has a small amplitude, with the maximum contribution to rainfall coming from Mesoscale Convective Systems (MCSs) in the early morning. Land areas have a much larger rainfall cycle than over the ocean, with a marked minimum in the mid-morning hours and a maximum in the afternoon, decreasing through midnight. MCSs over land have a convective intensity peak in the late afternoon, however their rainfall peaks later (near midnight) due to their life cycle.; An evaluation of the version 5 Precipitation Radar (PR) and TRMM Microwave Imager (TMI rainfall products is performed as a function of system type using the PF algorithm. The evaluation is performed by comparing long term TRMM rainfall products with rain gauge analyses, as well as other rainfall estimates, and by directly comparing rainfall estimates from the PR and TMI within the PR swath. The TMI overestimates rainfall in most of the tropics and subtropics with respect to both the gauges and the PR. The PR estimates are generally higher than the TMI's in midlatitude cold seasons, as well as in MCS dominated areas. The analysis by feature type revealed that the TMI overestimates relative to the PR are due to overestimates in MCSs and features with appreciable 85 GHz ice scattering, with negative biases in precipitation features without appreciable ice scattering. The TMI's distribution of rain rates in MCSs and features with ice scattering is biased high with respect to the PR mainly in PR-defined stratiform regions, while estimates from features without ice scattering are biased negatively with respect to the PR due to the TMI's sensitivity and beam filling effects. This preliminary study shows that the development of a storm-identifying algorithm may reduce regime dependent biases in a microwave precipitation algorithm. |
