Spatial Modelling of Long-Term Urban Vegetation Greening and Browning in Ibadan, Nigeria Using Sentinel-1 and Sentinel-2 Time-Series Data

Abstract Rapid urbanisation can produce contrasting patterns in vegetation greening, browning and stability in tropical regions, yet the environmental factors associated with these differences remain insufficiently understood. This study modelled the spatiotemporal trend in vegetation dynamics across urban and peri-urban local government areas of Ibadan metropolis, Nigeria, using Sentinel-1 and Sentinel-2 time series observations from 2015 to 2025. A random forest was utilised to predict spatial variation in long-term NDVI Theil–Sen slope, incorporating the mean conditions, temporal variability, spectral bands, moisture (NDWI), disturbance (NBR), and radar variables as predictors. The random forest performed strongly, explaining 84.3% of the variance in decadal NDVI Sen’s slope trend in the Ibadan metropolis (R² = 0.843, RMSE = 0.0044, MAE = 0.0034). NDWI Sen's slope was the strongest predictor (β = −0.841, 60.0% RF importance), followed by NBR Sen's slope predictor β = +0.379, 27.0% RF importance), indicating a strong association between moisture, disturbance and long-term vegetation trend. In contrast, the Sentinel predictors contributed negligible importance to vegetation dynamics in the metropolis. Greening was more extensive in peri-urban LGAs, ranging from 50.81% in Egbeda to 78.03% in Ido, while most urban-core LGAs recorded less than 50% greening. Greening, browning and stability coexist across the metropolis, revealing a clear rural-urban gradient in browning and greening rather than a uniform or linear pattern. The findings provide spatially explicit evidence for prioritising urban greening interventions, particularly in underserved, densely developed urban core areas, with implications for the 3-30-300 urban greening framework.