The IMERG Experience in Building Precipitation Products that Users Want (Invited Presentation)

Satellite-based precipitation products hold great promise for facilitating a diverse set of societal benefit areas, including agriculture, public health, disasters, and water management. Many of these applications hinge on getting information on excessive precipitation events. The relatively small scale of such storms sets a severe limit on the permissible time/space scales of observational sampling, and the search for “extremes” requires decades of these fine-scale observations in order to adequately characterize the tail of the distribution. The present constellation of precipitation-relevant satellites and its historical record approximately fulfills the sampling requirements, retrieval schemes for the individual sensors continue to improve, and several multi-satellite combination algorithms are in development and use for quasi-global precipitation estimation. In parallel with this scientific work, developers need to work with the user communities to implement features in the data system that make the datasets as useful as possible to the non-expert user.

Starting with the Global Precipitation Climatology Project (GPCP), then with the Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) and Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (GPM) mission (IMERG), the multi-satellite precipitation group at NASA/GSFC has sought to advance the science and the utility of global precipitation datasets. On the science side, it is key to intercalibrate the various sensor estimates to ensure that variations in precipitation are not dominated by changes in satellite data source. However, this intercalibration can impact the interannual global variations. On the user side, differing user latency tolerances have been addressed by providing multiple runs, with TMPA providing both a near-real-time and a post-real-time product, and IMERG providing two near-real-time and one post-real-time products. We have learned that users need all product runs to be provided for the entire data record, which in the case of IMERG now stretches for almost 20 years, so that they can characterize the statistics of the particular run that they are using. As well, new data fields have been added to the products in response to user requests. Since the start of IMERG, we have added fields for a “quality index” and diagnostic “probability of liquid phase precipitation”. User interactions have led us to move into the difficult polar regions more quickly than planned to support Alaska wildfire managers, and to provide a variety of “value-added” products.

The talk will end with remarks on the need for a more-systematic summary of user activities than now exists.