Approaches and Data Quality for Global Precipitation Estimation

Abstract:

The space and time scales on which precipitation varies are small compared to the satellite coverage that we have, so it is necessary to merge “all” of the available satellite estimates. Differing retrieval capabilities from the various satellites require inter-calibration for the satellite estimates, while “morphing”, i.e., Lagrangian time interpolation, is used to lengthen the period over which time interpolation is valid. Additionally, estimates from geostationary-Earth-orbit infrared data are plentiful, but of sufficiently lower quality compared to low-Earth-orbit passive microwave estimates that they are only used when needed. Finally, monthly surface precipitation gauge data can be used to reduce bias and improve patterns of occurrence for monthly satellite data, and short-interval satellite estimates can be improved with a simple scaling such that they sum to the monthly satellite-gauge combination.

 The presentation will briefly consider some of the design decisions for practical computation of the Global Precipitation Measurement (GPM) mission product Integrated Multi-satellitE Retrievals for GPM (IMERG), then examine design choices that maximize value for end users. For example, data fields are provided in the output file that provide insight into the basis for the estimated precipitation, including error, sensor providing the estimate, precipitation phase (solid/liquid), and intermediate precipitation estimates. Another important initiative is successive computations for the same data date/time at longer latencies as additional data are received, which for IMERG is currently done at 6 hours, 16 hours, and 3 months after observation time. Importantly, users require long records for each latency, which runs counter to the data archiving practices at most archive sites. As well, the assignment of Digital Object Identifiers (DOI’s) for near-real-time data sets (at 6 and 16 hours for IMERG) is not a settled issue.