Experimental LAMP 1-h Probability of Precipitation Guidance for the CONUS in Support of the National Weather Service’s National Blend of Models

The National Blend of Models (NBM) developed by the National Weather Service (NWS) provides a nationally-consistent and skillful starting point for making National Digital Forecast Database (NDFD) forecasts. Beginning in 2017, the NBM provided hourly guidance for Precipitation Potential Index (PPI) and Quantitative Precipitation Forecast (QPF) guidance over the CONUS through 36 hours by using an ad hoc expert-weighting scheme applied to each NBM grid point. While the hourly PPI and QPF were designed to assist forecasters with the timing of weather events in the Day 1–2 period, these products were uncalibrated and often exhibited biases inherent in the underlying model QPFs used as input. Additionally, sharpness of the NBM output in time and space did not reflect forecast uncertainty. Here we present 1-h Probability of Precipitation (hereafter “PoP01”) that is now serving as calibrated input to the NBM version 3.2 implemented in Fall 2019.

The Localized Aviation MOS Program (LAMP), a critical component of the NBM in the short-range period through 36 hours, implemented an upgrade of the LAMP suite of guidance in Summer 2019 in response to growing NBM requirements. This upgrade included 1) expansion of the Gridded LAMP (GLMP) spatial extent over the CONUS to cover the full NBM domain, 2) expansion of forecast projections out to 38 hours for several elements, and 3) addition of PoP01 guidance over the CONUS. LAMP PoP01 provides gridded probabilities of measurable precipitation over the full NBM CONUS domain for 1-h periods through 38 hours.

The LAMP PoP01 predictand is derived from Multi-Radar Multi-Sensor (MRMS) 1-h gauge-corrected Quantitative Precipitation Estimates (QPE). Candidate predictors for “Base LAMP” PoP01 include North American Mesoscale (NAM)-based MOS, European Centre (ECMWF)-based MOS, and initial and advected observations of MRMS 1-h QPE and MRMS 1-h maximum composite reflectivity. We used the GLMP “Meld” approach for the development of PoP01 regression equations by statistically combining Base LAMP PoP01 with High Resolution Rapid Refresh (HRRR)-based MOS to yield calibrated and skillful forecasts. Further details on the development procedure, verification results, application to case studies, and plans for future enhancements of the product will be presented.