This data set serves as training data for modelling the temperature field emanating from a groundwater heat pump. It is simulated with Pflotran, preprocessed with python and saved in pt-format. It contains 100 data points, each consisting of one simulation run after 5 years - so not necessarily in steady state. Each datapoint measures 100 m x 1280 m with 20 x 256 cells. The varying parameters of the data set are pressure and permeability. Both are constant within a data point, but vary across the data set. Other parameters that define the data set, such as porosity, are chosen to be as close as possible to reality. Generated with scripts from Dataset generation with Pflotran (commit 94daf52) with arguments given in inputs/args.yaml.

The prepared dataset contains two input options: full fields in Inputs folder or reduced input of 4 parameters per datapoint in InputParams; and two label options: full fields in Labels and 4 parameters in LearnableParams.

More information about the specific parameters are in the readme file.