Spatial-temporal modeling of land-use dynamics at the agricultural-forest interface: insights from Ntchisi District, Malawi

Understanding land-use dynamics in heterogenous agricultural-forest landscapes is crucial for a targeted approach to mitigating persistent environmental degradation and restoring degraded forests. This study uses Landsat 8 OLI/TIRS and Sentinel-2 imagery from 2015, 2019, and 2022 to investigate spatial-temporal patterns of land and forest degradation, underlying drivers, and future land cover projections in Ntchisi District, Malawi's Central Region. A machine learning Random Forest supervised classification was utilized to quantify the extents of six land-cover classes identified as: tree cover, shrubland, cropland, built up, water and wetlands. A Multilayer Perceptron (MLP) model was deployed to analyze land-cover transition sub models based on key drivers such as elevation, slope, soil texture, population density, distance from major roads and proximity to villages from Ntchisi forest reserve, and from rivers. The root mean squared error (RMSE) for sub models in MLP ranged from 0.40 to 0.48 while the Markov Chain model predicted a future land-cover for 2030 with a Kappa index of 0.96. Our findings indicate substantial increases in cropland and built up between 2019 and 2022, primarily driven by high probability of land cover conversions from tree cover (at 60 %), shrubland (40 %), and wetlands (over 50 %) to cropland. Moreover, the main drivers of these land cover transitions included proximity to villages, population density, distance from major roads, slope, and elevation. Projections for 2030 land use and cover change suggest an overall decline in tree cover, even within forest reserves and targeted restoration zones, due to the expansion of cropland across Ntchisi District. These findings provide critical insights into addressing land degradation and restoration hotspots. Taking the underlying drivers of local land use change into consideration, our study offers valuable guidance for assessing strategies for scaling on-going restoration efforts across Malawi.