Improving Image Fusion of Planetscope Superdove and Sentinel-2 for Land use and Land Cover Mapping in Sustainability Studies

The increasing need for precise and detailed geospatial products in sustainability and land-use change research, essential for effective territory management, drives the exploration of alternative methods to harness the potential of geospatial data, particularly satellite imagery. Image fusion techniques have emerged as a promising approach to integrating multispectral satellite images into single images that encapsulate the best available data. This strategy has demonstrated success within single-image spectral bands and, to some extent, in multisensor contexts, producing accurate results that merit collaborative investigation. This study examines the efficacy of an enhanced image fusion method applied to 10m Sentinel-2 and the latest 3m PlanetScope generation for land use and land cover (LULC) mapping in Northern Colombia. The evaluation involves traditional fusion metrics and thematic accuracy indicators, explicitly focusing on optimizing band pairs for SWIR bands. The findings reveal that these methods significantly improve LULC mapping, particularly for dominant classes, and demonstrate that (i) the proposed method enhances image fusion quality over conventional techniques and (ii) the assessment of image fusion quality is more effective when using thematic accuracy metrics in addition to conventional image fusion indicators.