Solar irradiance mapping for development site energy-yield potential

Satellite-derived solar irradiance climatologies, combined with topographic shadow modelling, let developers estimate annual PV yield potential before a single ground measurement is taken. The method turns energy physics into residual land-value arithmetic.

Sensors

Meteosat SEVIRI (Spinning Enhanced Visible and Infrared Imager): Provides cloud-cover fraction and surface solar irradiance (SSI) estimates over Europe, Africa and the Middle East at roughly 3 km spatial sampling and 15-minute temporal resolution. The EUMETSAT Climate Monitoring SAF uses SEVIRI to produce the CM SAF SARAH-3 irradiance climatology, which spans back to 1983 and is the primary input for European site assessments.
Himawari AHI (Advanced Himawari Imager): Japan Meteorological Agency's geostationary imager covering East Asia, Southeast Asia and Oceania at 2 km visible resolution and 10-minute full-disk cadence. Its cloud-fraction products feed the CAMS solar radiation service for sites east of roughly 60°E, filling the gap left by SEVIRI's footprint.
CAMS McClear / CAMS Solar Radiation Service: Copernicus Atmosphere Monitoring Service clear-sky and all-sky irradiance products, derived from SEVIRI and Himawari cloud observations combined with aerosol optical depth from CAMS atmospheric reanalysis. Delivers global horizontal irradiance (GHI), direct normal irradiance (DNI) and diffuse horizontal irradiance (DHI) at up to 1-minute time steps, with a typical bias below 5 W/m² for monthly means over well-observed regions.
Sentinel-3 SLSTR (Sea and Land Surface Temperature Radiometer): Provides top-of-atmosphere and surface reflectance data used to cross-validate aerosol and cloud retrievals that feed irradiance models. Its 500 m resolution thermal channels also support surface albedo corrections relevant to bifacial PV modelling, with a revisit of roughly one day at mid-latitudes.
SRTM / Copernicus DEM: Not a satellite sensor in the radiometric sense, but the 30 m Copernicus Digital Elevation Model (derived from TanDEM-X) is the standard topographic input for horizon-angle and cast-shadow calculations that determine effective irradiance on sloped terrain. Accuracy is typically better than 4 m vertical RMSE over open land.

What forty years of geostationary data actually tell you

A development site's energy yield potential is not a forecast. It is a climatological statement derived from decades of observed cloud cover and atmospheric aerosol loading. The CM SAF SARAH-3 dataset, produced from SEVIRI observations, covers Europe and Africa with a continuous record from 1983 onwards. That depth matters: a single year of ground measurements can be misleading if it coincides with an anomalously cloudy or sunny period, whereas a 40-year satellite climatology captures the interannual variability that a lender or grid operator needs to price risk correctly.

The CAMS McClear model separates the problem into two parts. First it computes clear-sky irradiance from first principles, accounting for Rayleigh scattering, ozone absorption and aerosol optical depth drawn from CAMS reanalysis. Then cloud-fraction retrievals from SEVIRI or Himawari modulate that clear-sky baseline into all-sky GHI, DNI and DHI. The split matters for developers because DNI and diffuse fractions determine whether a site suits fixed-tilt, single-axis tracking or concentrated solar configurations, each of which has a different capital cost profile.

Terrain shadows are where desktop estimates go wrong

Flat-site PV modelling is relatively forgiving. Sloped or valley sites are not. A south-facing hillside in a narrow Welsh valley may receive 30 to 40 percent less annual irradiation than the open-sky figure suggests, because surrounding ridgelines block low-angle winter sun for several hours each day. Conversely, a south-tilted slope at the right gradient can outperform the regional average by 10 to 15 percent.

Horizon-angle modelling from the 30 m Copernicus DEM calculates the elevation angle to the skyline in every azimuth direction for each point on the site. Combined with the hourly sun-position ephemeris and the CAMS irradiance time series, this produces a shading loss factor that is specific to the exact parcel geometry. The honest caveat: 30 m DEM cells average over sub-pixel relief, so narrow gullies or built structures taller than a few metres require a higher-resolution input, typically LiDAR or photogrammetric point clouds, to avoid overestimating yield.

From irradiance to residual land value: the arithmetic that planners miss

Grid connection costs for new residential or mixed-use developments are rising sharply in many markets as network operators enforce tighter capacity constraints. On-site generation offsets the required grid import capacity, which directly reduces the connection charge. A site that can demonstrate, through a credible irradiance assessment, that rooftop and ground-mounted PV will cover 40 percent of peak demand may qualify for a smaller, cheaper grid connection. That saving feeds directly into residual land value.

The satellite-derived annual yield estimate, expressed in kWh/kWp, is the starting number for that calculation. It feeds a standard PV simulation tool such as PVsyst or SAM (System Advisor Model), which converts irradiance into AC output accounting for inverter efficiency, wiring losses and temperature derating. The satellite data does not replace those tools; it provides the irradiance input that is often the largest single source of uncertainty in early-stage modelling. Reducing that uncertainty from the ±15 percent typical of interpolated ground-station data to the ±5 to 8 percent achievable with CAMS products over well-covered regions is what makes the difference between a bankable and a speculative yield estimate.

Where the method has limits and what to do about them

SEVIRI's 3 km pixel size means that irradiance values are spatially smooth. A site smaller than roughly one hectare sits inside a single retrieval cell, so any local shading or albedo anomaly is invisible to the satellite. This is acceptable for regional screening but not for detailed design. The workflow therefore has two stages: satellite climatology for site selection and early feasibility, then higher-resolution terrain modelling and, where budget allows, a short-duration ground pyranometer campaign to anchor the satellite estimates.

Cloud retrievals degrade over bright surfaces, particularly snow-covered ground and desert sand, where the contrast between cloud and surface is low. CAMS documentation acknowledges higher uncertainty in arid and high-altitude regions. Tropical sites with persistent convective cloud present a different problem: cloud fraction can vary at scales finer than SEVIRI's 15-minute cadence, introducing sampling errors in the daily irradiance integral. For sites in these environments, Himawari's 10-minute cadence reduces but does not eliminate that error. Sentinel-3 SLSTR aerosol retrievals can help cross-check the CAMS aerosol inputs, but the correction is modest outside heavily polluted or dust-prone regions.

Fitting this into a planning submission

Many local planning authorities in the UK now require a renewable energy statement alongside applications for major residential schemes. The statement typically needs to demonstrate that the site has been assessed for on-site generation potential and that the proposed scheme achieves a specified percentage of energy demand from renewables. A satellite-derived irradiance assessment, clearly sourced to CAMS and the CM SAF archive, satisfies the evidentiary standard for that statement at pre-application stage without the cost or delay of a bespoke ground campaign.

The output format matters. Planners and energy consultants generally want annual GHI in kWh/m², monthly profiles to check seasonal adequacy, and a shading-loss map overlaid on the site boundary in a GIS-compatible format. Satellize's analytics pipeline delivers exactly those layers, calibrated to the CAMS solar radiation service, as part of a broader site intelligence package. The Tonga crop-estimation programme demonstrated that the same climatological pipeline that handles agricultural irradiance inputs can be adapted to energy applications with modest reconfiguration. For developers who want to move from satellite data to a planning-ready energy statement, the concrete next step is a site-boundary submission for a scoped irradiance report.

Typical figures

Spatial resolution of irradiance retrieval	~3 km (SEVIRI over Europe/Africa); ~2 km (Himawari AHI over Asia-Pacific)
Temporal resolution of source observations	15 minutes (SEVIRI full-disk); 10 minutes (Himawari AHI full-disk)
Archive depth	1983 to present for CM SAF SARAH-3 (SEVIRI heritage); 2015 to present for Himawari AHI
Typical bias on monthly mean GHI	Below 5 W/m² over well-observed mid-latitude regions (CAMS published validation)
Topographic DEM resolution	30 m (Copernicus GLO-30); upgradeable to 1 m with client-supplied LiDAR
Irradiance components delivered	GHI, DNI, DHI (global horizontal, direct normal, diffuse horizontal)
Coverage	Global land surface; SEVIRI footprint covers Europe, Africa, Middle East; Himawari covers Asia-Pacific
Minimum site size for reliable retrieval	Approximately 1 ha for irradiance; smaller sites require terrain model upscaling
Delivery formats	GeoTIFF irradiance grids, CSV monthly/annual profiles, GIS polygon layers (GeoPackage or Shapefile), PDF feasibility summary
Latency for historical climatology report	5 to 10 working days from site-boundary submission

Analytics Satellize can run

Annual GHI and DNI climatology map	CAMS McClear clear-sky model modulated by SEVIRI/Himawari cloud-fraction retrievals; multi-decadal averaging with interannual variability quantification	GeoTIFF raster at 100 m resampled resolution, plus CSV of monthly mean and P50/P90 annual totals per site polygon
Topographic shading loss layer	Horizon-angle calculation from Copernicus DEM 30 m; cast-shadow integration over hourly sun-position ephemeris for full annual cycle	GIS polygon layer showing shading loss fraction (%) per parcel sub-zone, overlaid on site boundary
Effective irradiation surface for PV simulation input	Combination of CAMS all-sky irradiance time series with topographic shading loss; output formatted to PVsyst and SAM import specifications	Hourly TMY (Typical Meteorological Year) file in EPW or CSV format, ready for direct import into PV simulation software
Site yield-potential ranking for portfolio screening	Batch processing of multiple site polygons against the same CAMS climatology; ranked by annual GHI, shading-adjusted GHI and estimated kWh/kWp at standard system parameters	Tabular ranking report with map, suitable for investment committee or planning pre-application pack
Aerosol and turbidity correction assessment	Cross-referencing CAMS aerosol optical depth reanalysis with Sentinel-3 SLSTR surface reflectance retrievals to flag sites where dust or pollution loading may suppress DNI materially	Flagged site report with estimated DNI reduction range and recommended ground-validation trigger threshold
Planning-ready renewable energy statement data pack	Aggregation of irradiance climatology, shading model and indicative PV output into a structured evidence pack referencing CAMS and CM SAF source data	PDF technical annex and supporting GIS layers formatted to UK NPPF renewable energy statement conventions

Sources

Who does the work

We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.