Detecting Spatial Impacts of Violence through Remote Sensing: A Case Study of Conflict-Affected Regions in Central Mali

Sammanfattning

This study investigates if socio-economic indicators can reliably be derived from remotely sensed data and can serve as proxies for detecting and monitoring the spatial impacts of the violent conflict in the Mopti region of Mali. Traditional conflict event databases - like the Armed Conflict Location & Event Data Project (ACLED) – provide point-based event data but often fail to capture gradual spatial shifts that occur when insurgent groups like Jama’at Nusrat al-Islam wal-Muslimin (JNIM) gain territorial control. This study addresses this gap by integrating geospatial intelligence (GEOINT) using remote sensing to track changes in cropland activity, night-time light emissions and changes in built-up area in the Mopti region between 2016 and 2023. The methodology in this study consists of a multi-indicator approach in which natural environmental variability like seasonal rainfall is filtered out. This was done to ensure that the observed changes can be attributed to the conflict dynamics. Agricultural activity was analysed using the 3-Period TimeScan (3PTS) method, which uses Sentinel-2 NDVI data to visualize the agricultural cycle and makes it possible to identify land abandonment. Population displacements and economic stability were assessed through built-up area growth by using Impact Observatory’s Land Classification Land Use (LCLU) deep learning AI algorithms and NASA’s Black Marble night-time light radiance values. Different statistical methods were used to evaluate the relationship between these indicators and recorded conflict events in the ACLED event database. The results of this study suggest that cropland activity serves as a useful indicator of violence-related spatial change, as demonstrated by the identification of villages like Bobosso and Doukoro. In these villages significant agricultural abandonment was directly linked to violent actions by insurgency actors. While a two-way fixed effects model indicated a correlation between night-time light fluctuations and conflict frequencies, this is caused by structural overfitting due to the fixed effect dummies. The variations in night-time light also suggest a broader institutional grid instability due to maintenance issues and fuel shortages. While horizontal urban built-up area growth is evident within the study areas cities and timeframe, a predictive relationship between IDP numbers and built-up area growth could not be established in statistically significant terms. Qualitative research confirmed that an influx of displaced persons does not immediately contribute to an expanded built-up area footprint. IDPs initially find shelter with friends or family or in other communal existing structures.