Adding to this, the #downscaling method they use (I think cubic convolution) assumes a continuous underlying distribution of the data. This assumption is rarely satisfied over #urban areas, especially when detecting #hotspots.

This #paper has some good illustrations of this: https://doi.org/10.5721/EuJRS20164929

Almost the last trip of the year: To Liege to talk #PROTECT +#POLARRES with the #MAR group and to examine a PhD thesis. Looking forward to some good discussions with Xavier Fettweis and the rest of the group on coupled #ClimateModel development with #ocean #SeaIce + #IceSheet, next gen #RegionalClimateModels and how to choose #CMIP6 #ESMs for #downscaling.