Filling gaps in ecological time series by means of twin surrogates

This paper addresses the problem of reconstructing missing data in time series by the twin surrogates method. We consider a set of time series, collected by a sensor network, representing the same state variable measured at different locations of a spatially distributed system. Assuming the presence of coherent dynamics among the available data, the method of twins is applied as an input–output model using the information provided by the complete time series for filling the incomplete recordings. The method is applied to the measurements of Dissolved Oxygen concentrations collected in the lagoon of Orbetello (Italy) and to time series obtained from the logistic chaotic map. Both cases show satisfactory performances in the reconstruction of missing data.