Improving Weather Forecasts with AI
I did my physics master’s thesis at Uppsala University in cooperation with the Swedish Meteorological and Hydrological Institute (SMHI). In the thesis I show that neural network based post-processing of meteorological temperature forecasts is a usable method to improve forecast accuracy.
The neural network based method had better performance than the contemporary error-filtering methods used at SMHI. The neural network architecture used was a convolutional neural network followed by fully connected neural network.
The full thesis is available here.
