I naively thought that most 'simply' lifted it from whichever model they use so, to take the case of GFS-based forecasts which @Man With Beard was helping me with earlier, it's less than suitable for local forecasting. It would certainly *appear* (I may very well be wrong) that sites like XCWeather and, dare I say it, this one do exactly that for their 'local' forecasts because, for the 5 or 6 locations I was following today (including Birmingham, Bristol, Colchester and Hull - so a fair spread), they seemed to bear very little resemblance to reality.