A model is only as good as the goodness of fit of the variables used and the information fed into it. The lack of proper regression variables to get a good R^2 is the death of the model. Furthermore, the inputs into the model, such as accurate temperature data, is key. Otherwise it is a GIGO model.
Our inability to understand exactly what effects local climates is dependent on more variables than we can account for. Thus, local forecasting models are faulty.
As you can see from the climate models thus far, they have *all* been wrong. All of them. There isn't a single one that has even come close to predicting the temp, or the sea rise, or the glacial thickness. None.
Why? Because the science is "settled". Yet the regression sucks balls.
The science is *NOT* settled.