UK solar generation by location, pitch and orientation
625 modelled values — 25 locations, five pitches, five orientations — as one openly licensed file. This is the dataset both our calculators run on, so you can check them against it.
625 values: 25 locations × 5 pitches × 5 orientations.
Our calculators run on this file. Publishing it is the point — you can check what they tell you against the numbers underneath, which is not a thing most solar calculators let you do.
Download
UK solar generation by location, pitch and orientation
625 rows — zone, location, pitch in degrees and as a label, orientation in degrees from south and as a label, and annual output in kWh per kWp.
Download the CSV · 33 kB · CC BY 4.0
Credit as: solar.org.uk, modelled from PVGIS 5.3 (European Commission, Joint Research Centre). Contains modified Joint Research Centre data.
What it shows
The annual total is the number everyone quotes and the monthly shape is the one that decides things. Two thirds of the year’s generation arrives between April and September, which is why self-consumption is hard in winter and why battery sizing is set by November rather than July.
Columns
| Column | What it is |
|---|---|
zone | The MIS 3002 zone the location sits in |
location | The reference location modelled |
pitch_degrees | 0, 20, 35, 60 or 90 degrees from horizontal |
pitch_label | flat, vertical, or the pitch in degrees |
orientation_degrees_from_south | 0, 45, 90, 135 or 180 |
orientation_label | south, south-east or south-west, east or west, north-east or north-west, north |
kwh_per_kwp_per_year | Modelled annual AC output per kWp |
Orientation is symmetrical about south, so 90 degrees covers both east and west. That is a property of the model, not a simplification we added.
How it was made
scripts/data/fetch-irradiance.py runs PVGIS 5.3 for every combination — 1 kWp,
crystalline silicon, building-mounted, 20% system losses — and
scripts/data/build-dataset-csv.py emits the CSV. The script checks that every location
returns exactly five values per pitch and exits rather than write a short file.
Building-mounted matters: it runs hotter than a free-standing array and PVGIS models that, so the figures describe a roof rather than a field.
Sources
- PVGIS 5.3 Our own runs: 1 kWp, crystalline silicon, building-mounted, 20% system losses. Contains modified Joint Research Centre data.
- MIS 3002: The Solar PV Standard, issue 6.0 The method. MCS's own Kk lookup tables are published under terms that do not permit republication, so we model our own figures using the published method instead.
Sorry to hear that. What was the problem?