Physical-Based Spatial-Spectral Deep Fusion Network for Chlorophyll-a Estimation Using MODIS and Sentinel-2 MSI Data

Satellite-derived Chlorophyll-a (Chl-a) is an important environmental evaluation indicator for monitoring water environments. However, the available satellite images either have a coarse spatial or low spectral resolution, which restricts applicability of Chl-a retrieval in coastal (e.g., less than 1 km from shoreline) large- and medium-sized lakes/oceans. Considering Lake Chaohu as study area, this paper proposes physical-based spatial-spectral deep fusion network (PSSDFN) using Moderate Resolution Imaging Spectroradiometer (MODIS) Sentinel-2 Multispectral Instrument (MSI) reflectance data. The PSSDFN combines residual connectivity attention mechanisms to extract effective features, introduces physical constraints, including response functions degradation model, reconcile information. fused MSI data were used input variables collaborative retrieval, while only retrieval. Combined with field data, comparison between was conducted four machine learning models. results showed that can greatly improve accuracy compared This research illustrates estimated (less