Variogram Analysis for Assessing Landscape Spatial Heterogeneity in NDVI: an Example Applied to Agriculture in the Jiansanjiang Reclamation area, Northeast China

Abstract

It is necessary to characterize and monitor the spatial heterogeneity of agricultural landscapes, to assist the spatial understanding of some related agricultural processes and to predict food production patterns. In the study, variogram models fitted to empirical variograms are proposed to retrieve spatial heterogeneity characteristics, and they are applied to NDVI data generated from Landsat TM imagery at a 30 m pixel scale. Four representative study sites with distinct landscape patterns, located in the Honghe farm of Jiansanjiang Reclamation Area, Northeast China, were selected for spatial heterogeneity testing and analyzing. The results provided quantitative agricultural landscape knowledge: (1) the agricultural landscape heterogeneity at different directions within the study area could be effectively quantified with the variogram model and then interpreted; (2) the spatial heterogeneity of the dry land matrix was normally larger than that of the paddy field matrix; and (3) the agricultural spatial heterogeneity was affected by two factors: the land use type and the distribution pattern of land use types. As remote sensing techniques can now provide various types of surface monitoring data, we argue that quantitative variogram analysis of the data for spatial heterogeneity can help identify explain related ecological phenomena. It could also be used to improve quantitative agricultural remote sensing monitoring in a spatial heterogeneity area as well.