Verification of Spatial Forecasts of Continuous Meteorological Variables Using Categorical and Object-Based Methods
Technical Report,01 Oct 2015,31 Jul 2016
US Army Research Laboratory Adelphi United States
Pagination or Media Count:
Spatial forecasts from Numerical Weather Prediction NWP models of meteorological variables to support Army operations on the battlefield have become an integral part of the products available for the Staff Weather Officer to use in providing mission planning and execution forecasts. These forecasts are ingested by Army tactical decision aids TDAs and are fused with information on the operational weather thresholds, which impact the performance of Army systems and missions. The TDA generates spatial forecasts of these impacts for user-specified systems andor missions. This report presents methods to verify forecast fields of meteorological variables that have been filtered by the application of a threshold, the same way as that used by the TDA. A threshold applied to a continuous variable field becomes a categorical forecast for which there are traditional and nontraditional methods for verification. This study evaluates the ability of the NWP model to predict a category of the spatial variable. Preliminary results suggest a combination of a traditional technique for assessing categorical forecasts with a nontraditional object-based, forecast-evaluation technique has great potential in assessing forecasts of continuous variables that have been filtered by the application of a threshold.