Intro:
I have evaluated extreme edge detector classifiers to program pcb fault detector machine learning algorithms in matlab. I have also worked with GIS data to determine crop data from spectral data in collaboration with the EPA in R. I am comfortable with the technology and this is a subject that has much research devoted to it.
Methds:
Often times complex system modeling can be reduced to 1 dimension to determine an optimal solution. One approach to modeling flood simulation using GIS is the one-dimensional hydrodynamic modelling system, HEC-RAS. The algorithm first implements svm to identify significant features to do the number crunching. The resulting batch is then passed through a mlp which compares the features to aggregated data to determine statistically significant line approximation.
Concl:
Matlab, mathematica, R, and python all have extensive machine learning libraries. This is not the part of the proposal where I try to sell the software but I have used all of the above in previous cases and each has it's own paradigm in computation. I invite you to contact me if you have any questions.
Peters, R., G. Schmitz, and J. Cullmann. "Flood routing modelling with Artificial Neural Networks." Advances in Geosciences 9 (2006): 131-36. doi:10.5194/adgeo-9-131-2006.