Over the last few months, we have re-programmed our forecast data calculation process chain by implementing the latest technologies, ensuring further improved accuracy and a fast loading speed on our website.
Besides the technology improvements, additional data sources have been added to the forecast data calculation, mostly aiming at improving our nowcasting. Nowcasting calculations are made for the next 1-6 hours using data from observed (satellite, radar) or measured (weather stations) datasets processed on the fly. These data are especially useful for improving the forecasting of current weather conditions and the upcoming hours.
The following list shows a few samples of how the upgrade improves the forecast data:
- Satellite radiation data for North and South America were added to the calculation, resulting in a better radiation forecast.
- IFS ensembles precipitation data have been added to our MultiModel precipitation calculation. This data is then further updated with additional radar datasets, which we have also extended (e.g. the Czech radar is now included).
- These additional datasets also allowed us to improve the calculation of precipitation probability.
- The temperature data calculation for high altitudes and for mountain peaks was improved.
- We introduced nowcasting for dewpoint temperature, which is then further used to calculate other weather variables (e.g. relative humidity).
Parts of the project are funded by the EU project “Global Earth Monitoring (GEM)”, which aims at developing and improving models to monitor Earth observation data by applying the latest machine learning techniques.
These data are now available on our website and will soon be added to our app. We will also publish a validation study by comparing them with measurement data to assess our forecast data quality publicly.
