Gap filling in time series: A new methodology applying spectral analysis and system identification

The presence of gaps in time series makes the data analysis process difficult. Although there are several methods for filling such gaps, they do not present satisfactory results as the gap widens. The proposal of this paper is to present a new methodology that uses techniques of extraction of characteristics and identification of systems to fill the missing data. The proposed methodology was applied in time series of physical and chemical variables related to the water quality and behavior of the Paraguay River, and the effectiveness of the internal data forecast was proven.