Una metodología para la caracterización del uso del suelo mediante imágenes de media resolución espacial

Abstract

Introduction: The characterization of land uses represents one of the essential inputs for the management of natural resources at different scales. Objective: To develop a methodology to characterize land use in the upper creek basin from the Azul stream (Buenos Aires, Argentina), through the fusion of satellite images with a medium spatial resolution. Materials and methods: A time-series of 23 images was used from the Normalized Difference Vegetation Index (NDVI) of the MODIS-Terra satellite (product MOD13Q1) for the period May 2015 - May 2016. Landsat 8 images were used to discriminate some categories difficult to classify with NDVI-MODIS. The final cover map was validated regarding verification points independent to the classification process; its accuracy was evaluated by means of the Kappa statistic. Results and discussion: The NDVI time series allowed to recognize the phenological patterns of the covers and land use of greater representativeness in the region. Seven land cover were discriminated; the agricultural uses represented 81.5 % of the surface, double-crop wheat-soya (soybean in Argentina) system predominated (39.4 %). The overall accuracy of the final map was high (88.9 %, Kappa coefficient = 0.86). Conclusion: The methodology used has the advantage of being quick and replicable, to characterize the land uses of a given region and to evaluate its potential changes over time.