The Quantization Algorithm Impact in Hydrological Applications: Preliminary Results.

A computationally efficient surface to groundwater coupled hydrological model is being developed based on the Extended Cellular Automata (XCA) formalism. The three-dimensional unsaturated flow model was the first to be designed and implemented in OpenCAL. Here, the response of the model with respect to small variations of the quantization threshold has been assessed, which is the OpenCAL's quantization algorithm's parameter used for evaluating cell's steady state condition. An unsaturated flow test case was considered where the elapsed times of both the non quantized execution and the execution run by setting the quantization threshold to zero (with respect to the moist content variable) were already evaluated. The model response has been assessed in terms of both accuracy and computational performance in the case of an MPI/OpenMP hybrid execution. Results have pointed out that a very good tradeoff between accuracy and computational performance can be achieved, allowing for a considerable speed-up of the model against a very limited loss of precision.