Practical considerations for using remotely sensed precipitation estimates—TMPA and 1DD
As such, it is important to summarize the means, extremes, and uncertainties of the two data sets and examine the underlying basis for the differences. In general the two data sets give a consistent picture. One result of the comparison is that the longest span of consecutive dry days in a year is sensitive to the details of the retrieval algorithms. Another is confirmation that several of the parameters, including frequency of precipitation and 95th percentile values are sensitive to the spatial scale. Because the datasets depend on different monthly gauge analyses, the bias characteristics of the two are somewhat different. Several dataset-specific issues exist. For example, the 1DD typically shows anomalously high fractional coverage in the latitude bands 40-50 deg N and 40-50 deg S due to a smoothing operator that is applied at these latitude bands to accommodate the transition from IR-based to sounding-based estimates. As well, the TMPA tends to have drier estimates than the 1DD at higher latitudes, ~40-50 deg, particularly in the winter hemisphere, where the microwave algorithms currently lack sensitivity to the reduced precipitation signals.
The status of pending upgrades to both datasets will be discussed in the context of expected impacts on the results presented for the current datasets.