Remote Sensing and Machine Learning-Based Analysis of Land Use and Vegetation Change in Dhaka District, Bangladesh

This study investigates the rapid urban transformation of Dhaka District, Bangladesh, between 2019 and 2024. The researchers utilized high-resolution satellite imagery from Sentinel-2 MSI and Landsat 8, processed within the Google Earth Engine (GEE) cloud platform. To quantify land-use changes, the team employed three supervised machine learning classifiers—Decision Tree, K-Nearest Neighbors (KNN), and Random Forest—and computed spectral indices including the Normalized Difference Vegetation Index (NDVI), Normalized Difference Built-up Index (NDBI), and Normalized Difference Water Index (NDWI). The performance of these models was validated using confusion matrices and kappa statistics.

The analysis reveals a dramatic shift in land cover, characterized by a 59.5% expansion of built-up urban infrastructure over the five-year period. This rapid urbanization has come at the expense of natural landscapes, resulting in an 8.46% decline in vegetation and a 7.77% reduction in water bodies. The study identifies the conversion of vegetated and aquatic areas into impervious urban surfaces as the dominant land-use trend. Among the evaluated machine learning algorithms, the Random Forest classifier demonstrated superior accuracy in distinguishing between these land cover types, confirming its effectiveness for monitoring complex, rapidly changing urban environments.

The findings highlight the severe environmental pressures facing Dhaka due to unregulated urban growth. By providing a high-resolution, spatiotemporal assessment of land loss, this research offers actionable data for policymakers and urban planners. The study demonstrates that integrating remote sensing with machine learning on scalable platforms like GEE is a vital, cost-effective strategy for monitoring ecological health, enforcing land-use regulations, and promoting sustainable development in rapidly growing megacities.