Solar output calculator

The calculation runs in one direction, and every step is visible in the result: your location is resolved to a coordinate, that coordinate is matched to a long-term solar-resource dataset, and the resource is converted to electricity using your system size, panel tilt, panel orientation and a system-loss factor. Monthly figures and the annual total come from the same calculation, so the twelve months always add up to the year.

Long-term is the important word. The dataset describes what the sky above your location typically delivers across many years, not what it is delivering today. That is the right basis for the question “what will this system produce?” A cloudy week does not change how much a roof produces in a typical year, and a forecast tool answering “how much will I generate today?” is a different product with a different data requirement.

Location. This is the dominant factor and the reason a regional average is not good enough. The solar energy reaching a square metre varies by a factor of roughly two between northern Scandinavia and the desert south-west of the United States, and it varies materially inside a single country — a coastal site and an inland site at the same latitude can differ by ten per cent or more.

Tilt. A near-flat array loses direct sunlight in winter when the sun is low; a steep array gives up some summer output. Somewhere around the site latitude, less roughly ten degrees, maximises annual production for most locations, but any real roof pitch between about 15° and 45° performs within a few per cent of the optimum.

Orientation. The best direction is the one facing the equator: south in the northern hemisphere, north in the southern hemisphere. East or west-facing arrays typically give up 10–20% of annual production compared with an equator-facing array, and shift production towards morning or evening — which can suit self-consumption even when the annual total is lower.

System losses. Inverter conversion, wiring, module mismatch, soiling, snow and availability are bundled into a single loss factor. A documented 14% default is used here, which is the same figure NREL uses in PVWatts; you can change it under advanced settings if you have a better number for your installation.

What the model cannot see. Shading is not modelled. Neither is your exact roof geometry, your specific modules and inverter, or degradation over the years. These are real and sometimes large effects, which is why the result is presented as a typical estimate with sensible rounding rather than a precise promise.

If the location-specific source is unavailable, the calculator still returns a complete answer, but from a deliberately coarse latitude band rather than from your coordinate. Absolute latitude sets the broad resource level and the sign of the latitude sets the season, so the southern hemisphere gets a December peak rather than a June one. Nothing else is inferred: a coastal, a desert and a monsoon site at the same latitude all receive the same band figure.

That is the honest limit of a bundled dataset, and the reason fallback results are labelled as approximate and never described as location-specific. The ±25% attached to them is SnapEnergyLab’s own judgement of how wrong a latitude-only figure can be — it is not an empirical confidence interval, and it is not the interannual standard deviation PVGIS reports on the location-specific path. An earlier version of this fallback picked the geographically nearest named region instead, which mapped Delhi to the Nordics and northern Canada to New York — the latitude band is coarse, but it cannot fail that way.

Two credible estimates for the same roof can differ by twenty per cent without either being wrong. The usual causes are different irradiance datasets and reference periods, different assumed loss factors, DC versus AC ratings, and whether shading was modelled at all. A proposal that quotes one confident number without stating its loss assumption or its data source is not more accurate than a range — it is simply less transparent.

That is why every result on this page names its data source, its dataset and its reference period, and states plainly whether it came from location-specific data or from the latitude-band fallback.