Code for my MSc thesis "Developing a deep learning forecasting system for short-term and high-resolution prediction of sea ice concentration" written in collaboration with MET Norway for the program Computational Science: Geoscience offered the University of Oslo
In the Regrid to Patch script, a ValueError is encountered when trying to generate patches for data in the dates specified in the title ONLY for the above threshold. This results in 12 days of missing data for the full Dataset and should be inspected.
As the regrid should be correct, the error is likely to arise when determining the patches (though this is working everywhere else??).
The impact this has on the final model is unknown, as at the time of writing the Dataset is randomly selecting patches with replacement during iteration. Also, only some dates are missing. However, the time period that is missing is during the September minimum. This is either valuable information, or an edgecase that the model can work adequately without.
In the Regrid to Patch script, a ValueError is encountered when trying to generate patches for data in the dates specified in the title ONLY for the above threshold. This results in 12 days of missing data for the full Dataset and should be inspected. As the regrid should be correct, the error is likely to arise when determining the patches (though this is working everywhere else??). The impact this has on the final model is unknown, as at the time of writing the Dataset is randomly selecting patches with replacement during iteration. Also, only some dates are missing. However, the time period that is missing is during the September minimum. This is either valuable information, or an edgecase that the model can work adequately without.