A novel approach for extraction of Gaoshanhe-Group outcrops using Landsat Operational Land Imager (OLI) data in the heavily loess-covered Baoji District, Western China.

The final interpretation result of Gaoshanhe Gp. outcrops (anomalies), being overlaid on the natural color image background. • Our study objective was to conduct a case study of Landsat-based remote-sensing interpretation for the Gaoshanhe Group, which is an essential source bed for the mineralization of noble metals in central China. • Terrain analysis seems to be a significant factor in the high-precision interpretation of rock-minerals information. • The spatial aggregation of anomalous polygons is an important indication of a true anomaly. We studied the applicability of Landsat Operational Land Imager (OLI) data for extraction of Gaoshanhe-Group outcrops in a heavily loess-covered area in the northern Baoji district of China, which is always a crucial but challenging issue in geological remote-sensing mapping. An integrated approach was employed to process the remote-sensing data: first, the interpretation target area of interest was clipped using certain slope and elevation difference thresholds (topographical analysis); second, band ratio (B4/B2) and fractal summation algorithms were conducted in succession to extract the bedrock-induced anomalies; third, spatial clumping and hot-spot analysis were performed in order to eliminate residual false anomalies; and finally, we introduced high-resolution Worldview 2 data for visual interpretation and further spectral analysis. The findings show that the rocky outcrops were somewhat spectrally indistinguishable, and they were identified mainly due to their generally moderate degree of spatial clustering. Thirteen anomalous pixels were extracted from nearly 200,000 pixels covering the study area, and our field reconnaissance verified the results. This research constitutes the first useful case study on the applicability of Landsat data-derived lithology interpretation in China's loess plateau region. [ABSTRACT FROM AUTHOR]