Assessment of environmental impacts of armed conflict in Mozambique using remotely sensed data

Armed conflicts have been among the most common human activities across the world, and their effects on the environment have increased with the scale and technologies of warfare. However, little is known in depth about these effects, especially in land use and land cover (LULC) fragmentation and change intensity. Here we assess the impacts of an armed conflict on fragmentation, change intensity, and the pattern and process of changes in LULC in three districts in Mozambique. To evaluate these effects, we used Multi-temporal satellite images (Landsat 5 TM and Landsat 8 OLI-TIRS) in combination with fieldwork, geographic information systems, landscape ecology metrics, and the Random Forest machine learning algorithm. We also evaluated the spatiotemporal pattern of density of conflict events and landscape changes using the Kernel density estimator model. Statistical models were used to investigate the relationship between the density of conflict events and landscape changes during conflict. We found that the fragmentation of forest, grassland, salt marsh, and cropland increased before conflict (1993–2005, 2005–2017), but it drastically declined during conflict (2017–2023). During conflict, forest increased at the expense of grassland and cropland, while before the conflict period, grassland and cropland were the main causes of forest loss. We recorded a significant concentration of built-up land and cropland in refugee areas. There was a significant positive relationship between the density of conflict events and landscape changes, with R2 = 0.57 and r = 0.75. Our research indicated that remote sensing technology can effectively facilitate understanding war-related effects in conflicted areas where access is limited. This research advances meaningful insights into the patterns and processes of LULC dynamics before and during conflict, which is crucial to devising adequate strategies for environmental management in conflicted areas.