A rice field by the sea may look green, but the stalks are stunted and the grains are shriveled. Salt has silently poisoned the soil. In Satkhira district in southern Bangladesh, every farmer knows this scene. Salty water comes in with the tides, seeps into the ground, and the land stops yielding. To fight the problem, you first need to know where the salt is hiding beneath the roots.

Researchers collected 205 soil samples in 2024–2025 and measured their salt content. They then matched these numbers with Landsat satellite images. The satellite doesn't see the salt itself but indirect clues: how the vegetation color changes and how the soil reflects light in different parts of the spectrum. The XGBoost model (a very diligent agronomist that learns from hundreds of examples) learned to predict salinity from these clues. Another model — generalized additive model — helped refine the prediction.

Maps built over ten years (2014–2023) painted a worrying picture: patches of high salinity are not just staying put but spreading across the central part of the district. Southern and coastal areas suffer more, while northern inland areas are less affected. The researchers used spatial cross-validation (a clever way to keep the model from being fooled by neighboring points) and bootstrap (repeatedly shuffling the data) to assess how much they could trust the predictions. It turned out that uncertainty is higher in coastal zones, but when data from several years are combined, the predictions become more stable. This approach is like a regular health checkup for the land: you can keep an eye on salt constantly, without expensive field trips, just by updating satellite images.