Dr. Cheng Da Granted Funding via NASA’s Early Career Investigator Program in Earth Science


headshot from Cheng Da
Dr. Cheng Da

In February 2024, Dr. Cheng Da from the ocean assimilation group at the GMAO, was selected as part of NASA’s Early Career Investigator Program in Earth Science (Element A.39 of ROSES-2023). His proposal aims to improve analyses of the integrated Earth system by using strongly coupled, non-Gaussian data assimilation of NASA IMERG precipitation, with particular attention to the variables near the air-sea interface.

“Direct assimilation of precipitation retrievals has generally been limited to atmosphere models, while not applied so commonly in ocean or land models. Unlike the atmospheric model, where surface precipitation is a diagnostic variable, precipitation over the ocean and land serves as a major forcing for those uncoupled models.” Cheng said. “One unique aspect of this proposal is that, by using a strongly coupled data assimilation approach with the Ensemble Kalman Filter and a coupled system we can directly assimilate precipitation into the near-surface ocean layer, to improve its analysis by taking advantage of ensemble error covariance for cross-domain variables.”

Previously, Cheng developed a data assimilation testbed for a coupled quasi-geostrophic atmosphere-ocean model. He systematically compared the performance of different coupled assimilation strategies for both ensemble and variational assimilation methods. Cheng stated that during the research leading to the development of this testbed, he “found that while weakly and strongly coupled approaches generate atmosphere analysis with similar accuracy, the strongly coupled approach can significantly improve the ocean analysis. We also found that strongly coupled ensemble Kalman Filter has a similar accuracy as strongly coupled 4D-Var, and that both are both more accurate than the ‘outer-loop coupling’ approach adopted by the most state-of-the-art coupled analysis CERA, generated by the European Centre for Medium-Range Weather Forecasts. That is why we want to adopt the strongly coupled approach in our proposal.”

slide graphic from Cheng Da
Figure 1: Analysis RMSE of atmosphere (left) and ocean (right) from different SCDA methods and CERA-like approach (4D-Var for the atmosphere and 3D-FGAT for the ocean). The SC ETKF shows the smallest RMSE in both the atmosphere and ocean, comparable to the SC 4D-Var and CERA-like approach.

Besides the findings of the research itself, this proposal includes an educational component: creating more open-science computational projects on numerical weather prediction and data assimilation. “I am very grateful to my Ph.D. advisor, Professor Eugenia Kalnay, who always has a tremendous passion for teaching, especially sharing the latest research results with us students,” Cheng said. “She is also very open that she posted all the learning materials for data assimilation on her website. I learned data assimilation even before I became her student. I will do the same thing to help others.”


NASA Early Career Investigator Program in Earth Science for ROSES-2023 (https://nspires.nasaprs.com/external/solicitations/summary!init.do?solId={33447CE4-8120-C775-0339-014A07C87B51}&path=open)