Summary
The component of the Earth System addressed in this project is snow cover in mountain regions. The main scientific and technical objectives of AI4Snow are the development and training of AI methods to greatly improve remote sensing-based snow cover products for mountain regions as well as the implementation of data cubes containing all necessary datasets to apply these AI methods to the desired study regions. The aim is to produce a consistent, gapless, high resolution set of snow products suitable for highly detailed analyses even within complex terrain. The data cubes constituting the basis for these products shall be designed in a way that makes the application of the AI methods easily scalable and transferrable to any desired region of interest. The training of the AI models will be performed relying on an innovative approach which combines a physical-based snow process model with the remote sensing-based and meteorological datasets. This approach provides a very high density of available training data covering three large test domains within Switzerland, ensuring that a huge variety of topographic, climatic, and land cover characteristics will be represented. The project will include a scientific application/a scientific case, where the results produced by the developed AI-models will be used in a hydrological model. The results from this model will be compared with the outputs based on traditional input data, which shall demonstrate the value of the AI4snow-developments.
