The AI model CHIMP – developed at Chalmers – is now being used at SMHI to improve the real-time analysis of precipitation. The model was trained on four years of satellite observations, with radar data used to confirm where precipitation fell. With higher quality of the analyses, the model also opens the door to better forecasts.
“It’s inspiring to see an AI model developed by us now taken into operational use at SMHI. That it improves SMHI’s real-time analysis and precipitation forecasts is important, as predicting precisely where precipitation will fall remains one of the biggest challenges in traditional weather forecasting,” says Patrick Eriksson, professor of global environmental measurements, at the division of Geoscience and Remote Sensing.
CHIMP was developed by Simon Pfreundschuh, former colleague at the same division, with this application for SMHI in mind. The integration at SMHI has been led by Bengt Rydberg, researcher in meteorological satellite applications, also with a background in the same research division at Chalmers.
“This is another step in our long-standing collaboration with SMHI, where machine-learning techniques from our group have been used for five years. We are currently selecting students for two master’s theses aimed at further improving precipitation forecasts, in which even more of the system will become AI-based,” Patrick adds.
More info:
Read more about the improved precipitation analyses in the SMHI website.
Contact:
- Full Professor, Geoscience and Remote Sensing, Space, Earth and Environment