Cook, Ashton Robinson — Department of Defense SBIR Phase I: AF182-001

Description

Exploitation of weather and climatological information has historically provided significant (and at times decisive) advantages for Air Force and Department of Defense missions (e.g., WWII, 2003 Iraq, etc.) Current and short-term (0-7 day) and climatological (long-term average) weather information near the geographic locations of missions have often been the basis for these operations. However, a critical gap exists regarding extended-range forecast information for high-impact weather (occurring in the monthly to seasonal timeframe). This potentially valuable information can assist in extended-range mission planning while also mitigating impacts from severe and extreme weather. To address this need, WeatherDeep proposes to develop “L-model� as a flexible software platform for providing these types of extended-range forecasts. The L-model builds upon machine/deep learning concepts developed by Dr. Ashton Cook to provide forecasts of high-impact weather with potential lead times of 10-12 months. Initial products forecast tornadoes, damaging thunderstorm winds, and hail, although these products will be expanded to forecast additional perils impacting Air Force and Department of Defense missions (e.g., wildfires, tropical cyclones, dust storms, etc.)