POSTER Session 1

High-resolution sea surface salinity from the fusion of GOCI and HYCOM data based on machine learning over East Asia

Sea surface salinity (SSS) with a high spatiotemporal resolution is crucial for understanding oceanic phenomena in both open ocean and coastal areas. Despite efforts to enhance satellite-based SSS resolution using ocean color data, the weak correlation between SSS and optical parameters in the open ocean has limited progress to coastal areas. To broaden the research scope, we propose a machine learning-based approach for SSS estimation by combining the Geostationary Ocean Color Imager (GOCI) and the Hybrid Coordinate Ocean Model (HYCOM). We used the GOCI remote sensing reflectance (Rrs), HYCOM SSS, and multi-scale ultra-high resolution sea surface temperature as input variables for a multi-layer perceptron (MLP) model to simulate the SSS from the Soil Moisture Active Passive (SMAP) sensor over East Asia. This approach produced high-quality SSS with high spatial (500 m) and temporal (daily) resolutions for both coastal and offshore areas across East Asia, exhibiting detailed seasonal and spatial variation in SSS. When compared to in-situ measurements, the proposed model exhibited better performance (R2 = 0.78 and root mean squared error (RMSE) = 0.78 psu) than SMAP (R2 = 0.71 and RMSE = 0.92 psu). Based on the Shapley Additive Explanations method, we found that there were positive correlations between SSS and shortwave bands and negative correlations between SSS and longwave bands in coastal areas. The findings contribute to not only generating high resolution of SSS in the East Asian region, but also enhancing our understanding of the correlation between ocean optical properties and SSS.

*So-Hyun Kim, Ulsan National Institute of Science and Technology, Republic of Korea

Taejun Sung, Ulsan National Institute of Science and Technology, Republic of Korea

Seongmun Sim, Ulsan National Institute Science and Technology, Republic of Korea

Daehyeon Han, Ulsan National Institute of Science and Technology, Republic of Korea

Eunna Jang, Korea Institute of Ocean Science and Technology, Republic of Korea

Jungho Im, Ulsan National Institute of Science and Technology, Republic of Korea

Code of Conduct
Privacy Policy