The Proa Forecasting System (PFS) uses multiple methods to provide an optimal combined forecast for all time scales. Each component forecasts is calculated independently, using independent input data, which creates redundancy and increases overall reliability.
- Live Data forecasts (LDF): real time monitoring of the generation of a single system, or a group of PV systems, can be used to estimate the current cloud cover for a location or region.
- Cloud Motion Vectoring forecasts (CMV): this method uses data from the Himawari-8 meteorological satellite, which provides high resolution images every 10 minutes. Our CMV algorithms detect the clouds, track their motion and forecast their position into the future.
- Numerical Weather Prediction models (NWP): this methods uses dynamic equations and physical models of the atmosphere to estimate cloud formation and dissipation, and hence solar irradiance. The PFS uses several global and mesoscale NWP models.
The PFS forecast combines each of these methods, with optimal weights as a function of the forecast horizon, so that the combined forecast has the best performance over all time scales.
Custom input data
We can also incorporate your custom data feeds to improve the forecast accuracy. For example, if you have sky camera feeds for your site, we can include these to improve the forecasts. Our forecast engine will automatically combine all forecasts with the optimal weighting for each forecast horizon. Contact us to discuss custom input data further.