East Saddlebrook Boulevard (AZ)
The forecast for this particular location is often several degrees lower than other predictions and actual. Not sure why. Was not the case in the winter.
@rlink57 When looking at weather models, it is important to compare models cause sometimes one or the other may give false predictions. Have you tried using GFS or NAM for temperature?
And you don’t say if the forecast is for the current day, the next day or for 10 days ahead. The differences between model forecasts decrease as the forecast date gets closer.
Then you must consider that ECMWF is updated twice a day only. So it is possible that sometime the forecast is 12h older than other models.
As you can see here, the forecasted temp. vs observed is not so different
now, lets zoom in
... nearby locations may have temperature differences up to 5 degrees.
So, one observer may say the model is right while, at the same time, another observer thinks (the model) is wrong.
But, there are also some inaccuracies in models, because of:
a) sparse grid
b) urban heat island (see: https://en.wikipedia.org/wiki/Urban_heat_island )
c) the "smoothing" of model's forecasted peak values
Especially in the summer, the diurnal temperature variation is bigger than in the winter.
Thus, the "smoothing" make us thinking that the model underestimates max temp
and overestimates min. temp.