MS Plan B Defense: Estimating marine atmospheric boundary layer stratification with synthetic aperture radar data

Jonathan B. Chapman, PE
Department of Ocean & Resources Engineering
University of Hawai‘i at Mānoa

Location Information
**This defense will be held both in person (Kuykendall Hall 101) and over Zoom**
Meeting ID: 914 1679 5922
Passcode: 808795
https://hawaii.zoom.us/j/91416795922

Uncertainties in the lower atmosphere’s stratification, which is the balance between buoyancy and shear, lead to large uncertainties when determining air-sea fluxes. Previous works show that synthetic aperture radar (SAR) sea surface roughness images show atmospheric phenomena that are known to be related to stratification. In this project, we hypothesize that physics-guided neural networks (PGNNs) can be used to estimate Marine Atmospheric Boundary Layer (MABL) stratification parameters directly from the SAR sea surface roughness scenes. This could offer a new perspective of the global MABL because of the SAR’s ability to capture MABL properties in remote and sparsely sampled locations. The PGNN inputs are added separately to determine the most appropriate set of parameters to estimate the MABL stratification. Throughout most of the study, we use a reanalysis dataset as the MABL stratification reference because there is no other readily available global reference. The model is verified against measurements from the Station PAPA buoy in the North Pacific. Since most SAR images are visually dominated by atmospheric effects, we are able to map SAR sea surface roughness to MABL stratification. This suggests that SAR data contains important information about the vertical structure of atmospheric circulation.