Blog

Solcast has improved day-ahead solar forecast accuracy by ~9% in North America using an attention-based ensemble correction layer that targets systematic errors in NWP models. The update improves the reliability of short-term forecasts used for dispatch planning and operational decision-making.

Coastal East Asia experienced one of its sunniest Mays in years, with irradiance gains across Japan, Korea, eastern China, Taiwan and the northern Philippines, while inland China saw heavier late-month rainfall.

The developing El Niño is expected to strengthen through 2026, and analysis of past strong El Niño events in 2009, 2015 and 2023 shows how sharply solar resource shifts between regions during the July–September window.

As solar projects grow more complex, SolarFarmer’s 3D API brings bankable yield modelling into Python workflows, making sensitivity analysis, verification and portfolio-scale automation easier to run and repeat.

‍

A comparison of Solcast’s Advanced and Premium PV Power models—how they work and when to use them. Designed to help you choose the right approach for your data, resources, and operational needs.

‍

Southeast Asia recorded above-average solar irradiance in April 2026, with Bangkok experiencing its sunniest April on record, while Japan and South Korea saw weaker solar performance amid storm activity.

Introducing Premium PV & Wind — our most comprehensive forecasting and first dedicated wind product, now on the Solcast API.

The June, July and August 2026 seasonal outlook points to a broadly neutral to positive irradiance conditions across many key markets, with clearer regional patterns emerging as the season progresses.

March saw mixed irradiance as cyclones and polar winds shifted cloud patterns—southern Australia trended below average, while coastal regions recovered between systems.

‍

Solcast is updating its Typical Meteorological Year (TMY) dataset to include data from 2025, this change will take effect on the 15th April, 2026. As a result, TMYs generated after this change will not be identical to TMYs generated before the change, because there is new data to select from and long term averages have changed.

‍