Wavelet analysis of a Sentinel-2 time series to detect land use changes in agriculture in the Vega Alta of the Guadalquivir River: Cantillana case study (Seville)

Abstract

Historically, the Vega Alta of the Guadalquivir River (southern Spain) has been an anthropized space. Over time, the dominance of latifundia agriculture has evolved towards more intensive citrus-based agriculture. In this study, wavelet algorithms applied to Sentinel-2 time series were used to determine both the expansion of citrus plantations and the level of intensification of these plantations within the municipality of Cantillana. Sentinel-2 provides comprehensive global coverage from March 2017 to the present. Our study applied a 90% power wavelet transformation for the creation of a wavelet-smoothed time series for four years of Sentinel-2 NDVI data. Based on the data, it can be stated that within our research region covering 5000 hectares of agricultural land, over a span of four years (2017 to 2020), more than 980 hectares of native vegetation and pasture were transformed into citrus orchards, giving rise, at the end of 2020, to a total area of 3250 ha. Analyzing unique spatial patterns within a wavelet-smoothed time series data is very useful for land management, as it allows land use changes to be controlled. For this reason, it becomes feasible to assess the reliability of the wavelet method using both remote sensing and GIS tools.