Yemen's PV Atlas

Best estimate: 586 MWp, with a 90% range of 480–774 MWp. The range covers three measured sources of uncertainty, not everything that could move this number. Below: what is inside the range, what is not, and what calibration areas it rests on.

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.
• Which farm areas the farm ground-PV estimate was fitted on. Its ratio of mapped panel area to detected area is refitted on random re-draws of the completely mapped farm areas (bootstrap over 6 mapped farm areas in Tihama; 1 in 'Amran (fewer than 3) a stated x/÷2 judgement).

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

• The calibration areas were chosen by hand, not drawn at random, so the range is not a statistical margin of error for the country. Re-drawing them shows how much the areas disagree with each other, not how far the places the estimate is applied to differ from all of them. Only a random sample of areas can measure that.
• "Every panel mapped" holds as of the mapping imagery, which is older than the satellite imagery. Panels installed in between are missing from the reference: the model's detections of them count as false alarms and its misses of them go uncounted. Each of these makes this figure more likely too low than too high.
• Small household systems on roofs are not on this page yet. One area of Yemen, Al Wehdah in Sana'a, has been mapped panel by panel, but a per-building rooftop classifier has not yet been fitted to it. Rooftop solar here 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. Small solar-irrigation arrays beside fields are estimated, but only in farm cells that resemble the mapped farm areas.
• Imagery covers where people live, plus every mapped solar site. Map cells were imaged if they hold at least 1,000 buildings or any hand-mapped solar installation. A utility-scale park built on empty land that nobody has mapped yet falls outside both rules and is not on this page.
• Solar parks the model outlines but nobody has mapped are not counted. The detector draws the Aden plant and smaller parks on the Red Sea coast, but they are not in OpenStreetMap, and a model outline above 100,000 m² only counts when it carries a mapped plant geometry. The same rule removes the dark rock outcrops of that size the model mistakes for panels. Mapping the parks would bring them into the verified layer.
• OpenStreetMap coverage is uneven. Hand-mapped solar is concentrated where volunteer mappers have worked, so the verified layer says as much about mapping effort as about where panels are.

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.

The ground-truth areas: seven farm areas, drawn and mapped by hand, not a random sample. The farm ground-PV figure is calibrated on the test ground around Huth in 'Amran, traced panel by panel, and on six areas of a Tihama region near Zabid where every installation was checked by hand, about 460 km² in all. Most of Tihama's installations were mapped in 2023 as rectangles around the whole site rather than the panels; those count at the open-ground rate, not as panel area. The areas lie in two regions, so the range measured over them says how much the result moves between farms in those two landscapes, not across Yemen, and the estimate is only applied to map cells whose share of cropland, density of buildings and elevation resemble them (lowland plain for Tihama, highland for 'Amran). A further area, Al Wehdah in Sana'a, is mapped panel by panel for a rooftop classifier that has not been fitted yet; no figure uses 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.