TOWARDS A SCORECARD CONFIDENCE METRIC USING A TIME-LAGGED-MULTI-MODEL (TLMM) ENSEMBLE

Abstract

Consistent and quantitative methods to characterize the run-to-run performance of the Coupled Ocean Atmospheric Mesoscale Prediction System Numerical Weather Prediction (COAMPS NWP) model have not been readily available to naval forecasters. Moreover, an ensemble version of COAMPS is not typically run for uncertainty calculations. Operational forecasters rely on COAMPS output to use tactical decision aids (TDA) and ultimately provide mission-critical forecasts. COAMPS NWP forecasts are deterministic and only provide end-users with a specific value for a given area and time. An ensemble model, such as a Time-Lagged-Multi-Model (TLMM), may provide an acceptable method to combine successive deterministic forecasts into probabilistic values with associated uncertainty. This research is focused on statistical methods to investigate the uncertainty and statistical properties of COAMPS NWP deterministic model data, and ultimately develop a new statistical and analytical approach to capture model uncertainty in parameters used for electromagnetic propagation predictions. This uncertainty could then be shared with the end-user for improved mission planning and decisions. The overall goal of this research is not to improve the performance of COAMPS NWP output, but to assess the performance and uncertainty of a deterministic forecast.