Evaluation of Fire-Weather Predictions by the Colorado Fire Prediction System (CO-FPS): Results and Lessons Learned

The Colorado Fire Prediction System (CO-FPS) is a coupled atmospheric and wildfire model designed to provide decision support for the Colorado Division of Fire Prevention and Control (DFPC). Working together, the National Center for Atmospheric Research (NCAR) and the DFPC’s Center of Excellence for Advanced Technology Aerial Firefighting (CoE) are iteratively developing this decision support system to meet end-user needs. The CAWFE® (Coupled Atmosphere-Wildland Fire Environment) model has been used as a basis to extend the capabilities of the Weather Research and Forecasting (WRF) model to develop WRF-Fire, which drives the CO-FPS predictions.

Critical to successful development and implementation of WRF-Fire is evaluation of the fire behavior it simulates. To test performance of WRF-Fire, 11 fires in Colorado in 2016 were simulated. The fires were chosen based on the availability of observations, specifically having at least two fire perimeter observations no more than 36 hours apart. The HRRR background weather analysis, the WRF weather analysis, and the WRF-Fire outputs were all analyzed for the cases. The Model Evaluation Tools (MET) software package was used to analyze the HRRR performance. To assess the performance of the WRF-Fire forecasts, model output for burned area was compared against observed boundary shapefiles and both contingency statistics and object-oriented calculations were used. During the evaluation process, a number of observation issues presented including a lack of observation consistency, no central location for accessing observations, long periods between observation updates, and inaccuracies in the observed data. To overcome some of these issues, new techniques such as using film capture in a Geographic Information System (GIS) platform are being explored.