Thing is, their adjustments are always up, not down, which isn't what you'd do in the case of a heat island.
This is a big problem I have with what I've been reading on adjustments. Geostatistically, there should be some downtrends as well, but they never seem to make it above statistical noise in the raw data to model conversions. Conversely, the heat island affect is much more pronounced in recent decades versus in older data, often due to very rapid development in station locations. If anything, I would expect to see some downtrends in aggregate data over the last few decades specifically correcting for heat islands springing up near WX stations.
It seems like they're spending more time correcting "older and inaccurate data" than they are ground-truthing current WX stations (of which there are many, many more) that may have had an asphalt pad laid next to them five years ago or whatever.