Session 7.2 Quantitative Precipitation Forecasts (QPF) Verification of DWFE

Tuesday, 2 August 2005: 10:45 AM
Empire Ballroom (Omni Shoreham Hotel Washington D.C.)
Meral Demirtas, NCAR, Boulder, CO; and L. Nance, L. R. Bernardet, Y. Lin, A. Loughe, R. L. Gall, S. E. Koch, and J. L. Mahoney

Presentation PDF (271.6 kB)

The Developmental Testbed Center (DTC) Winter Forecast Experiment (DWFE) was performed from 15th January to 31st March 2005. For this experiment, Continental US (CONUS)-wide, forecasts were generated daily (00 UTC cycle) using two dynamic cores of the Weather Research Forecast (WRF) framework: the Non-hydrostatic Mesoscale Model (NMM) developed by NCEP was run at NOAA/FSL and the Advanced Research WRF (ARW) developed by NCAR was run at NCAR. Forecasts for both WRF cores extended to 48h with output available at 3h intervals.

Two types of QPF verification techniques are used: Grid-to-grid and grid-to-point (station). For the grid-to-grid QPF Verification-1 (QPFV1) of NCEP, all the model precipitation forecasts and available corresponding observations were re-mapped to the same verification grid. Two verification grids with horizontal resolutions of 5km and 12km were used in this study. The verification domains consist of: CONUS, and additional Eastern_US, Central_US and Western_US sub-domains. Verification statistics were computed for 24 hour and 3hour precipitation accumulations. For the 24-hour accumulation, the National Weather Service's River Forecast Center (RFCs) dataset was used. The data consist of reports of 24h-accumulated precipitation ending at 12UTC each day. For the 3-hourly accumulation, Stage II data was used. Stage II is a national multi-sensor hourly precipitation analysis based on hourly radar precipitation estimates from 140 WSR-880D radars over the CONUS, and about 3,000 automated gauge reports.

The grid-to-point (station) QPF Verification-2 (QPFV2) of NOAA/FSL provides verification statistics for model forecasts interpolated to more than 4,000 Hydro-meteorological Automated Data System (HADS) gauge observation sites. Accumulation periods are 3, 6, 12, 24, 36 and 48 hours. Both verification techniques compute equitable threat scores (ETS) and bias frequency scores for a range of threshold values.

This presentation will outline some basics of these verification tools, and discuss various out comes of the statistical analysis of DWFE precipitation forecasts.

Supplementary URL:

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner