Enhancing weather & climate information

Post-processing, calibration, and downscaling improve the accuracy and reliability of weather and climate predictions. Although numerical weather prediction models and observational data have advanced, errors persist due to atmospheric uncertainties, model limitations, and initial condition errors. Statistical techniques like machine-learning, bias correction and ensemble-based methods can reduce errors and provide better characterisations of uncertainty and the extremes of the distribution. Post-processing can also customise forecasts for specific locations, weather variables, and lead times, to meet the specific needs of users.

Related Publications