The use of interpolation methods for the modelling of environmental data

摘要

The purpose of the study was to provide some practical examples of using interpolation methods for modelling one-dimensional environmental data. Based on the measurement data acquired from the automatic air monitoring station in Warsaw (Poland), simulation of different variants of interpolation of the 1h average values of air temperature and SO2 concentration was performed. The interpolation was done by three methods, namely, by a linear method with a cubic polynomial and with a cubic spline. The simulation of supplementing the missing values was conducted for test vectors with a varying measurement gap length and the variable margin of extreme adjacent values. Comparison of the obtained modelling results with previously removed actual data was also made. For the assessment of the modelling error, the mean absolute error, the root mean squared error and Willmott's Index of Agreement were used. High accuracy of modelling was obtained, both for the short and the longer test vector variants, whereas the best environmental data modelling accuracy was obtained in short time intervals. It has been found that the use of interpolation for modelling a given type of environmental data should be preceded by the assessment of the accuracy of the employed methodology. In the framework of this study, a function for carrying out simulation of the accuracy of environmental data modelling was prepared within the MATLAB software program.