• Deep learning - is a data driven learning method successful at learning non-linear spatial and temporal dependencies from data (LeCun, et al. 2015). [1]
  • Automatically learn covariate relationships (feature selection) and exhibits ability to generalise to multiple sites (as opposed to Statistical and Dynamical Models).
  • CNN-LSTM applied to forecasting setting but not to the solar downscaling task.
  • If successful such a model, driven by a GCM ensemble, will be able to characterise the variability of solar radiation at the regional scale.
  • Estimate model uncertainty - important to assess future projections and prediction intervals.

    • [1]
    LeCun, Y, Bengio, Y and Hinton, G (2015 ). Deep learning. Nature. 521 436–44

    Creative Commons License
    Downscaling Global Climate Models with Convolutional and Long-Short-Term Memory Networks for Solar Energy Applications by C.P. Davey is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.