Results

Single Point Comparison

• Best performing models CNN-LSTM
• There were differences between sites, CNN-LSTM model performed best at the Springs site, and worst at the Carpentaria Downs site (the latter was best for GLM).
• CNN-LSTM exhibited larger mean error in October and December at both sites. GLM mean errors more distributed across summer and winter months.

Timeseries of the test set observed versus simulated radiation for each model at site Springs.

Back

Next

Deep Dive: Chapter 4.

---

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.