What's the chances home weather stations have some bias in their signal (e.g. likely to be mounted on a hot roof) and that these data points get mixed into a larger corpus and someone ends up drawing incorrect conclusions from that corpus? Hopefully not high?