Description of Products
Experimental gridded high resolution Global Forecast System (GFS)-based model output statistics (HRMOS) Probabilistic Quantitative Precipitation Forecasts (PQPFs) estimate 6- and 12-h probabilities for multiple precipitation exceedance thresholds in 4-km boxes of the Hydrologic Rainfall Analysis Project (HRAP) grid. The PQPFs are interpolated to the 2.5 km continental United States (CONUS) National Digital Forecast Database (NDFD) grid for public dissemination. PQPFs are disseminated for ≥0.10, ≥0.25, ≥0.50, and ≥1.00 inches for the 6-h periods, and ≥2.00 inches is added for the 12-h periods. The probabilities are issued twice daily at about 0530 UTC (0000 UTC cycle) and 1730 UTC (1200 UTC cycle) for 6-h projections in the 12 - 156 h range.
Technical Basis of PQPFs
The HRMOS PQPFs were developed with MOS procedures used at the NWS Meteorological Development Laboratory (MDL). The PQPF predictand was specified from a CONUS Stage IV composite of a Stage III precipitation analysis produced at each of 12 NWS River Forecast Centers (RFCs) for a subset of the 4 km HRAP grid. In the eastern US, the Stage III analysis is produced from radar-estimated and gage-measured precipitation amount together with human quality control. In the mountainous western US, the Stage III analysis consists of an automated (.Mountain Mapper.) analysis of precipitation gage observations that incorporates gridded monthly Parameter-elevation Regressions on Independent Slopes Model (PRISM) precipitation climatolgy. The Stage IV composites are produced at the NWS National Centers for Environmental Prediction (NCEP), after which MDL applies supplemental quality control.
HRMOS PQPF predictors were developed from a combination of GFS model forecasts together with fine scale precipitation climatologies and topography. The precipitation climatologies consist of seasonal predictand relative frequencies based on the Stage IV precipitation grids, gridded monthly predictand relative frequencies based on the US cooperative-station hourly precipitation network, and PRISM monthly climatic precipitation. These precipitation climatology and topography data are dynamically interacted with GFS forecasts to effectively enhance the spatial resolution of the GFS-based predictors. Details of overall development of HRMOS PQPF model are provided in http://journals.ametsoc.org/doi/pdf/10.1175/2010MWR3224.1.
An important feature of the HRMOS PQPF model is its geographical regionalization, whereby separate sub-models are developed for each of thirteen geographical areas with similar precipitation climatology to minimize regional biases in the GFS-based PQPFs. To prevent spatial discontinuities in the PQPFs across regional boundaries, the regions are over lapped slightly, and multiple PQPFs in the overlap zones are objectively weighted to produce seamless PQPF patterns. Details of the regionalization technique are documented in http://journals.ametsoc.org/doi/pdf/10.1175/2010MWR2926.1.