Integration of AI/ML with Data Assimilation for Earth System Prediction

Abstract

Focal Area(s): Development of dynamically consistent AI/ML methods that can integrate with the data assimilation cycle to improve the efficiency and effectiveness of state estimation and prediction at short to medium range forecast horizons. Science Challenge: This white paper addresses a first step toward complete integration of AI/ML and DA. Here, we address model emulation and its application to simple coupled atmosphere-ocean dynamics for use in data assimilation applications.