The field of earth observation and urban planning is moving towards more accurate and efficient mapping techniques, leveraging advances in deep learning and satellite imagery. Recent developments have focused on improving flood mapping, wildfire risk prediction, and urban heat stress mitigation.

Notable papers in this area include: FM-LC, which introduces a hierarchical framework for flood mapping by land cover identification, achieving average F1-score improvements of up to 29% across all land-cover classes. AlphaEarth Foundations, an embedding field model that enables accurate and efficient production of maps and monitoring systems from local to global scales, outperforming all previous featurization approaches tested on a diverse set of mapping evaluations. DeepC4, a novel deep learning-based spatial disaggregation approach that incorporates local census statistics as cluster-level constraints, enhancing the quality of urban morphology maps. MergeSAM, an innovative unsupervised change detection method for high-resolution remote sensing imagery, based on the Segment Anything Model, addressing real-world complexities such as object splitting and merging. GSM-UTCI, a multimodal deep learning framework designed to predict daytime average Universal Thermal Climate Index at 1-meter hyperlocal resolution, achieving near-physical accuracy and reducing inference time from hours to under five minutes for an entire city. FuseTen, a novel generative framework that produces daily Land Surface Temperature observations at a fine 10 m spatial resolution by fusing spatio-temporal observations derived from Sentinel-2, Landsat 8, and Terra MODIS, outperforming linear baselines with an average 32.06% improvement in quantitative metrics.