There’s a lot more on this in the manual (especially the first part which recounts the history of how I developed WXSIM), but it’s a good question so I’ll attempt at least a brief answer here.
First, technically, most users (at least if they are using WXSIMATE) are not actually using READY data these days, but instead the convenient equivalent (or better) GFS model data culled from huge GRIB files. U.S. users can also use FOUS data from the NGM and/or NAM models to supplement this. Model data like this is vital for WXSIM to forecast cloud cover, rainfall, upper level temperatures, and wind direction. Without it, you’re right - these things pretty much just stay the same.
WXSIM started as a single site near-surface temperature model. It really calculates sun angle and all kinds of thermal processes on its own. A major enhancement of this is its advection routine (especially if you’re using regional data). This is still all WXSIM’s territory. Other things that WXSIM does very much on its own are forecasts of dew, fog, frost, diurnal variations in wind speed (though this can be overwhelmed by imported data), and precipitation type (though whether or not precipitation occurs is determined almost entirely by imported model data).
Actually forecasting large scale weather changes (passage of fronts, rain, cloud changes, etc.) is utterly beyond the ability of any single station model. Various governments and some universities run supercomputer models that take in huge data streams in order to come up with what have become pretty good forecasts. The U.S. government (and no others that I know of) has made its model data publicly, freely available, for the entire world. WXSIM quite “willingly” imports this data.
You have probably seen algorithms for forecasting weather based on conditions and trends from a single station. In fact some weather stations sport such forecasts. You must understand that these are only a little better than a guess. I’ve thought about trying to use such rules in WXSIM, but performance would be poor and it makes a lot more sense to use the excellent, freely available stuff. I recall back in the early-mid 90’s as I was starting to make WXSIM public (after over 10 years of developing it on my own and just showing friends, family, and students), there were a couple of “competitors” out there. These were forecasting programs that took things like barometric pressure and tendency, and wind direction, and tried to make a forecast. They didn’t use any physical calculations that I know of - just sort of an if-then decision tree. They basically did the same thing a little forecast icon on a weather station console, or a one-page list in a book, would do - again not very much and greatly lacking in specificity of timing and other details.
In cases (such as in the U.S.) where multiple models are available, WXSIM can import up to three at a time for a sort of average or consensus. Also, WXSIM has adjustments that can be made to this data to correct for biases in the models. These should be used with care, as I find the models are pretty good in the first place.
One way to look at this is that WXSIM is handling very local things, mainly temperature and humidity, on its own and borrowing from the big models what it needs and couldn’t do itself. The result is a sort of “symbiosis” with hopefully the best of both worlds (the local world of WXSIM and the global view of the models).
Bottom line: yes, WXSIM does heavily rely on outside model data for many important parts of its forecast, but it also does a great deal on its own (mainly regarding temperature and humidity), and tailors this specifically for the customized site.
One more thing: I didn’t intend WXSIM solely as a forecasting tool. It’s a modeling tool in which you can experiment with its virtual world of weather. This is why it retains (perhaps seldom used) tools for putting your own weather into it manually. YOU can make it rain, make the wind direction change, etc. and see how these affect things like temperature. It has tremendous educational potential and is completely unique, as far as I can tell, in this regard. Playing with it in this hypothetical mode can give you years (perhaps a lifetime) of experience in weather forecasting (again, mainly temperature) in a few weeks.
Hope that helps!