ssobie
commented
2 years ago This is a known artifact in these downscaled files and comes from both the target dataset and the downscaling method. In ANUSPLIN there can be sharp or discontinuous boundaries between different magnitudes of precipitation or temperature. It most frequent in precipitation where there can be adjacent values of 0 and larger non-zero precipitation. The reranking step of BCCAQv2 can also adds to this when days are reordered within each month. The daily spatial fields can end up with local boundary effects when the method accounts for the Canada-wide spatial analogues. It won't affect derived indices but it does make individual daily fields look unusual.
Looking at a few maps I saw some patterns that seemed unphysical. There are sometimes weird lines that show up creating discontinuity in tasmin and pr. I have put a few examples below.
Maybe this is a known artefact of the bias correction method? I wanted to flag it just in case, as it seemed a bit strange to me.
The code to reproduce the figure is in section Potential Issue 1 of the QAQC-CMIP6-BCCAQv2.ipynb.