Skip to main content
. 2021 Feb 15;379(2194):20200099. doi: 10.1098/rsta.2020.0099

Figure 9.

Figure 9.

Evaluation metrics of our post-processing framework across all lead times on 200 random forecast scenarios. We compare our QRF post-processed output to the raw NWP ensemble in terms of continuous ranked probability score (CRPS, a) and logarithmic score (b). (Online version in colour.)