I am working on a project focused on the semantic segmentation of water bodies and aim to leverage multi-modal data, specifically combining time-series meteorological data with static remote sensing imagery. The core challenge lies in effectively fusing these heterogeneous data types to enhance segmentation accuracy and robustness.

Could you provide insights or examples of existing studies that have successfully integrated similar multi-modal datasets for water body segmentation or related geospatial analysis tasks? I am particularly interested in the data fusion techniques employed to handle the temporal aspects of meteorological data alongside the spatial characteristics of remote sensing images.