Analyzing Spatial Autocorrelation for the Hypsometric Integral to Discriminate Neotectonics and Lithologies Using DEMs and GIS

Abstract

This investigation documents the sensitivity of the hypsometric integral (HI) and its relationship to neotectonics and lithology. We used digital elevation models (DEMs) of 30 and 90 m spatial resolution from the Shuttle Radar Topographic Mission (SRTM) and the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) to calculate HI values in the Hindu Kush and its vicinity (Northwest Pakistan and Northeast Afghanistan). We used an analysis grid of regular squares of different sizes to calculate maximum, minimum, and mean elevations. The spatial distributions of HI do not show clear spatial patterns and correlation with mean elevation or relief amplitude. We applied spatial pattern analysis using Local Indices of Spatial Autocorrelation (LISA) to measure the degree to which our HI distribution was clustered, dispersed, or randomized. LISA analysis shows that the data are autocorrelated because of high positive z-scores. Hot spots (clusters with high HI values) are consistent with tectonic uplift and show a strong correlation with the different structural domains in the region. Cold spots represent recent sedimentation close to faults and coincide with shallow earthquake clusters in the region. The HI values do not show any correlation with relative topographic position or lithology. Analysis of HI distribution shows that they are robust and independent of digital elevation model (DEM) resolution but are strongly scale dependent. The LISA technique allows the extraction of clusters of the HI that reveal recent tectonic processes; otherwise it is difficult to interpret the high variability of HI values. The scale dependency of the HI may reflect the varying importance of drainage network and hillslope processes.