PV yield forecasting and solar resource assessment

Solar yield forecasting covers two different purchases that are often confused. Developers and financiers need a long-term energy estimate to size a plant and structure debt, usually expressed as P50 and P90 production over the project life. Asset managers and grid operators need short-term forecasts of generation for the next hours or days, to schedule maintenance, manage storage and meet dispatch or imbalance obligations.

The input data decides the quality of both. Long-term assessment depends on satellite or reanalysis irradiance datasets, their period of record, bias correction against ground measurements and assumptions about interannual variability. Operational forecasting depends on numerical weather predictions, their update cadence and spatial resolution, plus a model of the plant's configuration and losses. Snow, soiling, curtailment, availability and degradation all have to be represented explicitly rather than hidden in a single uncertainty figure.

Failures are easy to spot in hindsight. A yield estimate built on one favourable year, a P90 that is simply P50 minus a fixed percentage, a forecast vendor that cannot report historical error against actual production, or a model that cannot be recalibrated once the plant is energised. Financiers have learned to ask for the assumptions behind the number, and operators increasingly ask the same of short-term forecasts.

Evaluate tools on source data and period, uncertainty methodology, editable loss assumptions, forecast horizon and cadence, measured accuracy and recalibration with operating data. The directory is still expanding its coverage of PV asset tooling; the entries listed here are the closest energy and grid-data tools available today, not PV forecasting products.