884 An Ensemble Reforecast-Based Precipitation Bias Analysis and Application for Sub-Seasonal Forecast

Wednesday, 9 January 2019
Hall 4 (Phoenix Convention Center - West and North Buildings)
Eric Sinsky, NOAA/NCEP/EMC and IMSG, College Park, MD; and Y. Zhu, Y. Luo, H. Guan, and W. Li

Precipitation is a challenging field in numerical weather prediction which is usually biased and therefore requires statistical post-processing. Ensemble-based reforecasts can be a valuable resource for precipitation to provide more reliable forecasts, and increase forecast skill for sub-seasonal prediction. In this study, the model bias and uncertainty of precipitation with extended model integration will be analyzed, the spatial distribution of bias will be investigated to feedback to our model developers, and the frequency-matching method (FMM) will be applied to improve week-2 and weeks 3-4 forecast.

The National Oceanic and Atmospheric Administration (NOAA) Environmental Modeling Center (EMC) has generated an 18-year (1999-2016) ensemble reforecast based on the Global Ensemble Forecast System Version 11 (GEFSv11) to support the SubX project. Furthermore, a 30-year (1989-2018) ensemble reforecast to support FV3 (GEFSv12) will be generated. The new reforecast is comprised of 11 members and is extended to 35 lead days. The Climatology-Calibration Precipitation Analysis (CCPA), which is available from 2002 to present, is used as a proxy “truth” for bias calculation and validation. The domain of this investigation is the contiguous United States (CONUS). Various methods for sampling of reforecast precipitation frequency bias will be compared and presented to determine the optimum configuration for future sub-seasonal applications. A comparison between the calibrated precipitation forecast from GEFSv11 and GEFSv12-based reforecasts will also be presented. In the verification, various measures are used to evaluate the bias and skill of the calibrated precipitation forecast, which include frequency bias index (FBI), equitable threat score (ETS) and true skill statistic (TSS) for the ensemble mean forecast. Probabilistic measures such as ranked probability skill score (RPSS) and reliability diagrams are used in the precipitation evaluation as well.

- Indicates paper has been withdrawn from meeting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