Comparison of Global Precipitation Estimates across a Range of Temporal and Spatial Scales

Estimates of global precipitation differ widely from one product to the next. Some of the differences are likely due to differing goals in producing the estimates. High-Resolution Precipitation Products (HRPPs) aim to produce the best instantaneous precipitation estimate whereas Climate Data Records of precipitation emphasize homogeneity over instantaneous accuracy. Precipitation estimates from global reanalyses are dynamically consistent with the large scale circulation, but tend to compare poorly to rain gauge estimates as they are forecast by the reanalysis system and precipitation is not assimilated. Gridded estimates of global precipitation are becoming more and more necessary for applications such as model validation, input for land surface models or extreme event characterization. Without detailed knowledge about current precipitation distributions it is impossible to quantify changes in precipitation estimated by global warming scenarios. Extreme events are generally defined as percentiles of a distribution which assumes accurate or at least adequate estimates of that distribution exist. Here we consider three global HRPPs, three global Climate Data Records of precipitation, and five reanalysis precipitation estimates. Patterns of means are consistent among data sets. Variance is highest where means are large, and the same is true for the average spread between data sets. Differences in the means and variances between data sets in some regions are as large as the means and variances respectively. Biases in time evolution, temporal correlation, rain rate and amount distributions are explored using temporal and spatial averaging. Scales range from daily to annual time scales and from single grid point to continental and global averages. These estimates vary widely, even for global averages, highlighting the need for better constrained precipitation products in the future.