India’s solar ambition is written into its landscape. Since the launch of the Jawaharlal Nehru National Solar Mission in 2010, the country has pursued one of the most aggressive renewable energy expansions in history — targeting 500GW of solar capacity by 2030. But the geography of that expansion follows a troubling logic. The regions chosen for development are not simply sunny — they are also among the most water-stressed, the most ecologically fragile, and the most socially vulnerable in the country.
India’s clean energy transition is being built on landscapes that were never empty, and on communities that were never consulted.
3/ 4 layers : thats why the reason to study them balh balhh below is the metho
Phase 1
Research by Yenneti (University of Birmingham) and cases documented by Land Conflict Watch reveal a consistent pattern: solar development has concentrated in regions already marked by agrarian distress, pastoral displacement, and contested land tenure. Adivasi and pastoral communities report loss of common grazing land, disrupted water access, and broken promises of employment and electricity. The social cost of clean energy is not evenly distributed — it is absorbed by those with the least power to refuse it.
Spatially, this literature finds its confirmation in the data. ERA5 solar irradiance maps show the highest GHI values concentrated across Rajasthan, Gujarat, and the Deccan plateau — precisely the zones where WRI Aqueduct records Extremely High baseline water stress. JRC Global Surface Water occurrence data, drawn from four decades of Landsat imagery, reveals that land classified as wasteland for solar development carried seasonal and ephemeral water presence long before development began. OSM solar park locations, overlaid on these layers, make the argument unavoidable: the sun is here, the water is already scarce here, there was water here to begin with — and this is exactly where solar is being built. (shown in the map above)
Together they build a sequential spatial argument: the sun is here → the water is already scarce here → there was historical water here → and this is exactly where solar is being built.
Phase 2
To move beyond points on a map, solar parks need to be seen as the physical infrastructures they are. Using instance segmentation — a YOLOv11 model trained on ~1000 annotated satellite images sourced from Roboflow Universe and GitHub — individual solar park polygons are extracted across India. These boundaries become the analytical units for everything that follows: buffers, zonal statistics, hydrological overlap. The model achieves 89.7% mAP@50, sufficient precision to distinguish park footprints from surrounding terrain at Landsat-compatible scales.
Phase 3
