The lower boundary conditions of the computer models are also sources of error, again because the model has to represent a continuously-varying Earth by discrete grid points.
In the case of topography, the model cannot include every peak and valley; only those features that are several tens of kilometers across can be represented by the model. Below is a figure showing the natural variation in height on the Earth, and what that same terrain looks like in a computer model.
Any weather caused by the small-scale peaks and valleys in the model are simply not simulated by the computer model.
Until about 10 years ago, California's Central Valley did not exist in any forecast models. The model topography simply rose from the Pacific Ocean to the Sierra Nevada. So the dense fog that formed and collected in the Central Valley for days at a time was completely unforecasted by the models. People in Salt Lake City had similar problems. Now, at least those geographical features "exist", but smaller-scale features are still missing.
A related problem occurs on coastlines. If a model's grid point spacing is 40 km, then the transition from a grid point over land to a grid point over water takes place over that distance: 40 km. In the real world, though, the coastal transition zone can be as small as 100 m. So while real sea breezes can form right on the coastline and can feature narrow, sharp temperature contrasts, the forecasted sea breezes have to form with horizontal scale closer to 100 km, making them unrealistic.
Here's a similar case: suppose that after a particularly strong cold front there's a strong wind from the north across Texas with cloudy skies and very cold temperatures, say 30F. As the cold air gets blown across the Gulf it gets heated by the warm Gulf waters. So a grid point 25 km onshore would have a temperature of 30F and a grid point 25 km offshore might have a temperature of 46F. Interpolating the model output to the coastline, halfway between the two grid points, gives a temperature of 38F.
So, what is the most likely temperature at the coast?