Validation of Near-Term Rainfall Forecasts in Data Scarce Coastal North Carolina
Abstract
Near-term (i.e., hourly to daily) weather forecasts are a valuable decision-making resource for agricultural and aquacultural producers and managers. However, in situ data used to validate and refine weather models are often collected in data-abundant locations, and therefore, may not be as accurate in data-scarce regions where agricultural and aquacultural production occurs. Here, we validate the National Digital Forecast Database (NDFD) 1-, 2-, and 3- day probability of precipitation (POP) and quantitative precipitation frequency (QPF) forecasts against a collection of in situ precipitation observations measured by automated rain gauges and community-supported rain gauges (e.g., the Community Collaborative Rain, Hail and Snow Network; CoCoRAHS) in North Carolina from January 1 to December 31, 2016. We calculated standard validation metrics including Briar score, root mean squared error, and mean bias error. Furthermore, we explored relationships between NDFD POP and QPF performance and data scarcity by comparing validation metrics in data abundant (i.e., major North Carolina cities) and data scarce locations (i.e., parts of the North Carolina coastal plain). We interpreted results in the context of aquacultural management decision-making. Specifically, the results of this study will inform near-term forecasting of temporary shellfish harvest closures, which impact shellfish operations in coastal North Carolina. While centered in North Carolina, this study serves as a representative case for the assessment of near-term NDFD weather forecasts in other data-scarce regions.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2020
- Bibcode:
- 2020AGUFMH170.0014S
- Keywords:
-
- 1833 Hydroclimatology;
- HYDROLOGY;
- 1880 Water management;
- HYDROLOGY;
- 1922 Forecasting;
- INFORMATICS;
- 6309 Decision making under uncertainty;
- POLICY SCIENCES & PUBLIC ISSUES