Solcast Solar Irradiance and PV Power Forecasting API

Solcast was formed in early 2016 by meteorologists James Luffman and Dr Nick Engerer and has been part of DNV since January 2023. Operated by Solar and Storage Modelling Pty Ltd in Sydney, Australia, it provides irradiance, weather and power forecasting data that underpins more than 650 GW of solar capacity for customers in over 100 countries.

The platform processes global weather data every five minutes at 90 m to 2 km resolution and delivers live data, historical time series, typical meteorological years and forecasts up to 14 days ahead at 5, 15 or 60 minute resolution. Three PV power models cover different needs: Rooftop PV runs on location and capacity alone for sub-MW fleets, Advanced PV is a physics-based pvlib model with up to 20 tunable parameters plus soiling and snow factors, and Premium PV is trained by DNV on a plant's own SCADA history with probabilistic outputs. Data is delivered through REST APIs, a toolkit and an SDK; logos include Enel, Engie, Canadian Solar, Taiwan Power Company and KEPCO.

• Live and forecast data: irradiance and weather every 5 minutes, forecasts up to 14 days ahead.
• Rooftop PV model: six-parameter model for low-spec, sub-MW fleets.
• Advanced PV model: pvlib-based physics model with soiling and snow-cover factors.
• Premium PV model: site-trained forecasts with probabilistic P10 to P90 outputs.
• Grid and portfolio models: aggregated forecasts for markets, portfolios and behind-the-meter solar.
• Developer tools: REST JSON or CSV API, toolkit for bulk downloads and an SDK for tuning.

Solar farm owners could submit day-ahead and intraday schedules using site-trained forecasts. Power distribution companies could estimate rooftop solar output across their networks with aggregated grid models. Energy management system vendors could embed the API to schedule battery charging and factory loads around expected solar output.

Gap-free estimated actuals and forecasts give operators a reference output even when on-site sensors fail. Site-trained Premium PV forecasts with P10 to P90 ranges support bid sizing and penalty management. An attention-based model improved day-ahead solar forecast accuracy by about 9% in North America, and Solcast recorded the lowest error rate in a 12-week EPRI trial. Free data requests let teams test the API before buying.

Grid operators increasingly require accurate solar schedules and penalise deviations, and the rapid growth of behind-the-meter rooftop solar makes it harder to forecast net load. Vietnam's large rooftop base, monsoon cloud variability and the move toward market-based power trading increase demand for API-based irradiance and power forecasts.

Solcast was formed in early 2016 by James Luffman and Dr Nick Engerer and was acquired by DNV in January 2023; in 2025 it became part of DNV's Green Data Products. The operating entity is Solar and Storage Modelling Pty Ltd, Sydney, Australia. It reports more than 80,000 users, 600+ companies using its data and 26 million API calls per day.