None of the models ingests that data, mostly because there is no way to ensure that personal weather stations are sited correctly. So, for instance, an anemometer has to be around 30’ above ground with no obstructions to generate usable readings. Most people putting a personal weather station in the backyard aren’t going to the trouble to locate the station where it can provide data that are usable.
Plus, surface readings are only a small part of the atmospheric picture; you need weather balloons to generate a more holistic reflection of the atmosphere at a site. There are also data generated from aircraft, but those are more one-dimensional (in the sense that they are only sampling conditions at a single altitude) than twice-daily balloon launches from selected sites. Until someone decides to implement a private, nation-wide radiosonde network, there probably won’t be a numerical prediction model that operates independently of NOAA data.
