India's PV Atlas

Best estimate: 103,037 MWp, with a 90% range of 83,676–132,969 MWp. The range covers two measured sources of uncertainty, not everything that could move this number. Below: what is inside the range, what is not, and how it compares with other sources.

Inside the range:

• The two area-to-power constants. Panel area on a roof converts at about 0.18 kWp per m², the rating of a typical module; the outline of a ground-mounted park converts at about 0.05 kWp per m², measured on parks whose capacity is known. Each is varied over its own range (90%: 0.15–0.21 and 0.035–0.075 kWp per m²).
• How reliable the segmentation model's detections are. How often its detections turn out to be real, and how many mapped arrays it misses, were measured size class by size class against mapped installations. Those measurements have their own sampling error, which is drawn into the range.

Outside the range, and not something more arithmetic can add back:

• The hand-mapped areas here do not check these numbers yet. Two areas of this country have been mapped panel by panel and declared complete, but no figure on this page uses them yet. The detectors' measured accuracy is still carried over from the countries where it was measured, and how well it transfers here has not itself been measured.
• The small-installation half of this page does not exist yet. One area of India, in Gurgaon, has been mapped panel by panel, and one area is not enough to calibrate the per-building classifier that supplies most of Pakistan's estimate. Everything on the map is either hand-mapped in OpenStreetMap or an array the segmentation model outlined directly, which in practice means installations of roughly 400 m² and up. The household rooftop systems that dominate India's recent rooftop programme are mostly below that size.
• Imagery covers where people live, every mapped solar site, and the land around them. Map cells were imaged if they hold at least 1,000 buildings or any hand-mapped solar installation, or lie within three cells (about 33 km) of one. A park built on empty land farther out than that is not on this page; hatched cells were never imaged.
• Solar parks the model found that nobody has mapped count once a person has judged them. A model outline larger than 10 hectares is as often bright bare ground or rock merged into one shape as a solar park OpenStreetMap does not hold yet, so it counts only once it is mapped or someone has judged it solar on the imagery the model saw. In India 1,640 of 1,789 such outlines have been judged, and the 1,427 found to be parks count in full, about 28,900 MWp of the best figure. The other 149 were left out of the review by the filters on the lead files (steep slopes, open ground at low confidence, outlines beyond the border) and count only if someone maps them. None of these parks is in the hand-mapped figure.
• A mapped outline drawn around a whole planned zone counts only where panels are seen. Seven ground outlines larger than 5 km², among them two at Khavda in Gujarat, are mostly still empty land or were not fully imaged. Each counts at the share of it the model detects as panels, plus any plant mapped inside it, in both the hand-mapped and the best figure. Rewa Ultra Mega Solar is missing: its cells were never imaged properly.

Treat this as an early estimate from an active research pipeline, not a census. What it offers is a reproducible way to estimate distributed solar from free satellite imagery and open-source models where official statistics are thin or missing. Expect the figures to move as the evidence behind them grows.

Nothing on this page rests on hand-mapped ground truth from this country yet. Two areas have been mapped panel by panel (Gurgaon, Haryana, Purba Champaran, Bihar (farm, no PV)): every panel in it drawn by hand, the ground truth that turns a detection into a capacity estimate by measuring how much of a flagged roof is actually covered in panels and how many real installations the detectors miss. Once a classifier is fitted to them, these figures will rest on them. More such areas are the most useful thing anyone can contribute.

Against India's register, ground-mounted solar sits in the right states and falls about a fifth short. MNRE's state-wise register, republished by MoSPI's eSankhyiki, records 110.6 GW of utility-scale solar on the ground (ground-mounted plants plus the solar part of hybrid plants) at the centre of the imagery window, January 2026. This page's best estimate of ground-mounted solar is about nine tenths of that, and across the 19 states with at least 100 MW registered it ranks them almost exactly as the register does (rank correlation 0.97). The register mostly quotes AC capacity and this page DC, so the real shortfall is nearer a fifth, mostly parks this page does not see: cells never imaged, such as Rewa's, and parks built during the window. Rooftop is the large gap: the register holds 24.3 GW, most of it household systems of a few kW, and this page sees less than a tenth of it.

Layers. "Show" picks the quantity the cells are coloured by: the Best estimate, the OSM-mapped capacity alone, the satellite detections above 400 m², rooftop PV below it, farm ground PV below it (arrays beside fields, where the farm detector has run), or the gap between the Best estimate and what is already mapped, which is where mapping effort pays off most. Each cell shows its total in MWp. The colour scale is logarithmic and is refitted to whichever quantity is shown; the legend always states its range.

Your own data. "Layers" → "Add a layer" overlays a GeoJSON file, a CSV with latitude/longitude columns, a GeoJSON URL, an XYZ raster tile URL or a WMS (use {bbox-epsg-3857} in its GetMap URL). Files are read in your browser and are not uploaded anywhere. The live OpenStreetMap overlay queries the public Overpass API for the area in view when you zoom in, so it shows mapping done after this page was built.

Exports. Every figure downloads as PNG (for slides), SVG (vector, for papers) and CSV (the numbers behind it); the map as PNG with its legend and attribution, and the grid as GeoJSON or CSV. Basemap imagery and tiles keep their own providers' terms: OpenStreetMap data is © OpenStreetMap contributors (ODbL), the grey canvas and the satellite imagery © Esri and its partners.