Datasets used in the AI-powered spatiotemporal imputation and prediction of chlorophyll-a concentration in coastal oceans

We conducted a benchmarking study of STIMP over four representative global coastal oceans, including northern Gulf of Mexico, Chesapeake Bay, the Yangtze River Estuary, and the Pearl River Estuary.

Each of the datasets contains the following files:

1. Chla.npy: the original observation of  chlorophyll from the satellite remote sensing.
2. is_sea.npy: indicates where the ocean is and where the land is.
3. adj.npy: the spatial graph constructed from the spatial coordinates of positions.
4. mean.npy, std.npy, min.npy, max.npy: contains the mean Chl_a, standard variation of Chl_a and max and min value of Chl_a for each position.