[RESOLVED: Flowx uses raw model values which can be expected to show large differences versus actuality]
1715 UTC Jan 1 2022
Android 12
Pixel 4a 5G
Flowx v3.352
Temperature disagreements: NOAA GFS vs CMC GDPS vs actual. I’m brand new to Flowx (and really like it!) so may not yet understand what to do in a case like this.
(Have screenshots but guess I’m at Trust level 0 and mustn’t post them. The colors on map correspond approximately to actual Wunderground temperature.)
I don’t really need help today, but thought the problem might signal something interesting or anomalous to experienced users. I rebooted the phone but that did not change the incorrect temperature values, nor anything else I could detect.
The temperature value that is incorrect in the two models below is the number shown in the summary status ribbon atop the map and below the temperature and cloud cover graphs at top.
(there is no doubt of Wunderground temperature being very close to actual: the forecast has been for such temps and I confirmed it on Accuweather and Weather.com)
also keep in mind temp, rain a oh etc. are going to very from between weather sites 1 might say 50deg another might say 49deg etc. and Data may very from location to location aka micro climates. like for my location I’m 14+ miles from local airport where they have a weather station they use to provide weather data for the area but at my house {micro climate} everything varies it can be raining/snowing north end of town and nothing at my house temp, humidity etc. can very +or - 5 to 10 degrees.
Thanks, that sounds like what I’ve noticed, too, over many years watching the weather: a few degrees plus/minus range.
In the present case, the NOAA temp anomaly was +23 degrees F. And the CMC temp was either missing, or was meant to be zero, which would be a +9 degree F anomaly. Without question a difference that is outside ordinary range.
One might say that +14 F was the forecast temp as of 0000 UTC today but that can hardly be right as all services during the past several days were forecasting temps substantially below zero F for this morning in Minneapolis, and those forecasts turned out to be correct.
You are comparing apples (GFS, GDPS) and oranges (wunderground, accuweather). GFS, GDPS are computer simulations of the weather. Wunderground, Accuweather are statistical corrections of predictions from the computer simulations. This is called MOS (Model output Statistics).
Most people think this means GFS and GDPS is inferior, whne in fact it forms the basis for most, if not all, weather predictions in the world. GFS, GDPS, etc… are the pinnacle of human achievement in weather predictions.
Thank you for that link, it explained things. I had not realized that the raw weather models needed that kind of tweaking (MOS). I do understand I was comparing a forecast with actual conditions, just didn’t know how much massaging the raw models need.