Machine Learning for Operational Weather

Raytheon has built a broad foundation of capabilities in Artificial Intelligence (AI) and Machine Learning (ML) for an array of Department of Defense (DoD) and Intelligence Community (IC) applications. We are now actively exploring how those capabilities can be ported and adapted into the operational weather value chain. Specific focus areas include: potential decision support enhancements to nowcasting in severe weather events; bringing non-traditional types of data (social media, traffic patterns, etc.) together with weather data to better understand public responses to weather, watches, and warnings; and improved efficiency in handling the vast amounts of weather observations that are collected on a continuous basis. In this presentation, we discuss: how AI and ML are defined; what role they should play in any decision support application; the major categories of challenges in successfully utilizing ML; examples of how Raytheon has addressed those challenges in the DoD and IC mission domains; and how similar solutions can translate into the weather mission, including better forecasts and nowcasts, watches and warnings, and interactions with industries affected by weather (aviation for example). The goal is to further the dialogue on how ML can revolutionize the weather enterprise while still reserving ultimate decision authority for human experts.