you can pretty well reproduce the same trends with ~300 stations in random positions around the globe too, as long as you can find a set of 300 that are all unc

you can pretty well reproduce the same trends with ~300 stations in random positions around the globe too, as long as you can find a set of 300 that are all unchanged over the duration you look at. Using thousands and doing a proper grid is good, but it is illustrative of how robust the trend is that you can pick it up with even an amateur effort.