Modelling is the only possible way to study a system this complex.
The basis of our knowledge of climate is millions of bits of data. Only computer models are capable of making sense of these datasets. To discount computer modeling would be to discount climate science per se.