Evaluation of a High-Resolution Replay Dataset for Applications to Extreme Precipitation

Datasets that accurately represent extreme precipitation at fine spatial scales have numerous applications including, for example, ongoing efforts at NOAA to modernize estimates of Probable Maximum Precipitation (PMP). To this end, we introduce a high-resolution (3-km grid spacing) UFS-based “replay” dataset spanning the continental United States (CONUS). The replay methodology calculates increments between the background UFS forecast and an external pre-calculated analysis (in this case ERA5), then applies those increments to adjust the forecast. Once completed, the replay is expected to encompass at least 30 years (1994-2024).

Preliminary results will be shown evaluating the replay against the NCEP Stage IV Precipitation Analysis to identify trends and statistical patterns in extreme precipitation, in addition to case studies of individual extreme events. The high-resolution replay dataset will also be compared to a previously-developed coarse resolution replay with global coverage and 0.25° grid spacing. Preliminary analysis of this coarser dataset shows that it tends to underestimate both precipitation extremes and mean monthly trends in precipitation.

Future work will involve a more comprehensive and quantitative evaluation of the replay’s performance for precipitation extremes, with focus on verifying annual recurrence intervals (ARIs) and potentially, Positive Extreme Fields (PEFs). It is expected that the replay will serve a variety of future purposes such as the aforementioned PMP modernization effort while offering a suitable alternative to traditional reanalyses.