Unaddressed imbalance of multitemporal land cover (LC) data reduces deep learning (DL) model usefulness to forecast changes. To manage geospatial data imbalance. there is a lack of specialized cost-sensitive learning strategies available. Sample weights are typically derived from training instance frequencies which disregard spatial pattern complexities. https://storeanimalwiz.shop/product-category/water-treatments/