Sea Ice Modeling and Data Assimilation (SIMDA)

Integrated Foundations of Sensing, Modeling, and Data Assimilation for Sea Ice Prediction
Department of Defense Multidisciplinary University Research Initiatives program through the Office of Naval Research

Parameter selection of regularized inverse problems

20 Oct 2022 - Bosu Choi

OT

B. Choi, Y. Lee and J. Han are investigating when an improper combination of data from multiple sources can degrade the reconstruction quality. In such case, they suggest an alternative way of using each source for disparate purposes: using the partial data promoting the well-posedness of a problem for the reconstruction, and using the remaining data for the cross validation to identify proper parameters which affect the reconstruction quality when solving the regularized inverse problems.

The plots show the comparison of reconstructions using the noisy single-type data and combined data. After fine-tuning the regularization parameter, λ, it is observed that the single-type data has the smaller reconstruction error. Further, the validation measure in their approach successfully identifies λproducing the reconstruction of quality better than the one from combined data.