In this document
We will explain how we approach the forecast modelling at Solargis.
Solargis Forecast approach
PV power forecast is calculated from the relevant solar and meteorological primary data parameters. The forecast data streams are systematically quality-controlled and their accuracy is enhanced through internal validation procedures. These procedures rely on solar resource data from a satellite-based model, as well as meteorological data from advanced weather models. This multi-layered approach ensures that the forecasts are both reliable and robust.
Solar data, which serves as the main input for PV forecast calculations, is sourced from two key models: the Solargis Cloud Motion Vector (CMV) model and Numerical Weather Prediction (NWP) models. The choice between these models depends on the forecast time horizon. For immediate and short-term forecasts, the CMV model is used, while NWP models are applied for longer-term forecasts.
Forecasted meteorological data is obtained from NWP models. Specifically, Solargis uses the IFS model for forecast horizons up to DAY+3, and the GFS model for forecasts from DAY+4 to DAY+14. More detailed information about these models and their applications can be found here.
Solargis Cloud Motion Vector model
Solargis Nowcast is based on the Solargis Cloud Motion Vector (CMV) model. This technique is designed to track the movement of clouds across the sky by analyzing satellite imagery. The CMV model identifies cloud features and calculates both the direction and speed at which these clouds are moving. Using this information, Solargis can estimate solar irradiance levels with high temporal resolution. The data is updated every 5, 10, or 15 minutes, depending on the specific location, ensuring that forecasts remain current and responsive to rapidly changing weather conditions. Further details about Nowcasting data availability and update delays are available here.
Numerical Weather Prediction (NWP) models
Numerical Weather Prediction (NWP) models are sophisticated mathematical representations of the atmosphere. They use equations based on fundamental physical principles to simulate and forecast weather conditions. These models incorporate a wide range of variables, including temperature, humidity, atmospheric pressure, wind speed, and more.
NWP models begin by assimilating observational data from multiple sources, such as weather stations, satellites, aircraft, balloons, and radar. This data is used to establish the initial atmospheric conditions. The models then solve atmospheric equations on a grid that covers the Earth's surface and the atmosphere above, advancing in time steps that range from minutes to weeks.
At each step, the NWP model calculates how atmospheric variables evolve, taking into account physical processes such as radiation, convection, and turbulence. The output includes forecasts for solar irradiation, temperature, wind speed, precipitation, and pressure at different locations and altitudes. Model results are regularly validated against real-world observations to identify any biases and to continually improve forecast accuracy.
Solargis Forecast is based on the following NWP models:
Note: You can read more information about meteorological NWP models here.