Seeking scientific knowledge from environmental time series data

paper faces the problem of inferring a differential equation representative of the dynamics of an environmental phenomenon. Starting from measured data, an evolutionary multiobjective modeling strategy, the Evolutionary Polynomial Regression, is used to construct the differential equation that rules the dynamics of a shallow groundwater system. A pre-processing of data is made in order to construct the differential terms which are used as input to the process. The Evolutionary Polynomial Regression returns a set of non-dominated models, as ordinary differential equations, which can be chosen on the basis of their performance against a test set and trying to assign a physical meaning to the terms.