This is exactly the limitation of weather models with large grid spacing - e.g., 25km for GFS. Meteorologists use statistics to compare models to real measurements. If the model consistently predicts 4C lower than reality, they will add 4C to the model predictions.
This is called Model Output Statistics (MOS).
The limitation of MOS is that it is localised to where you measure temperature which is usually at the airport for cities. So you get point values for cities instead of color maps.
Weather models are just that, a best-guess prediction based on accumulations of historical weather data and available computing power and computing time. Msm does not purposely lie, it does not tell the truth either, but rather attempts to forecast future weather from data that is thinly spaced in both space and time. Weather is a chaotic system. Local weather is affected by small topographic perturbations which are too small to resolve on the scale of most weather models. The cool thing with Flowx is that it shows graphically how a host of future weather parameters are likely to behave and allows you to compare several models at once. It shouldn’t take you long to figure out which of the many models available on Flowx is the most reliably predictor of your local weather. Where I live, the most accurate model is NOAA GFS ~25km and not Expedition Marine with a 9km range. The latter gives very high-resolution output but due to local topography, rain data is frequently displaced to the north by many 10s of kilometers.
Is it really that important if the forecast is out by 4 degrees? Lockscombi, if you have Flowx, what on Earth are you doing consulting Msm Weatherman/women?