关键词: 逆断层, 智能化提取, 地貌参数, 活动构造, 高精度数字地形数据

Abstract:

Digital topographic analysis, an important means in the research of active tectonics and tectonic geomorphology, has increasingly become one of the principal tools in the identification of active tectonic features and understanding of the development of the earth’s surface process. Indoor interpretation of surface fault trace plays a key role in the digital topographic analysis as it can provide the foundation for setting priorities and defining strategies in the subsequent field investigation. At present, the extraction of fault traces is often realized by assisting the traditional visual interpretation through the image enhancement method. The relevant subjective assessments lead to the amount of work and usually cause different results due to the differences in the interpretation experience of actual operators. At the same time, the field of quantitative research on geomorphic parameters is evolving very rapidly with the advances in the popularity of high-resolution digital topographic data. Therefore, intelligent and automatic extraction of surface fault traces has gradually become a promising research direction. The methods based on machine learning often rely heavily on the good programming foundation of the operator, which is a visible technical barrier. We present a semi-automated method using an ArcGIS toolbox with a set of tools to extract surface fault traces based on geomorphic constraints. The Hutubi and Dushanzi faults are two typical thrust faults located on the northern piedmont of the Tianshan Mountains and are chosen as examples. Excellent exposure of the surface fault traces in these two regions permits detailed mapping of fault traces and deriving shape factors of faults with high-resolution DEMs(digital elevation models). Additionally, they are two of the most-studied thrust faults in this area. Large-scale geological and geomorphological mappings of them and numerous achievements have been published. This creates possibilities for us to conduct comparison analysis on different major methods. Based on typical morphology characteristics of fault scarps, appropriate geomorphic parameters are selected. In practice, reverse fault scarps are distinctly defined into forward and backward ones according to whether their dip is the same as that of the neighboring geomorphic surfaces. Based on two sets of geomorphic constraints,two approaches are then illustrated, including slope calculation, gully extraction, data density analysis and process modeling. Through a detailed comparison of the final extraction results and previous visual interpretations of remote sensing data and field geomorphic investigations, the validity of the method proposed in this study is proven. This method provides a set of tools with user-friendly interfaces to realize step-by-step interpretation and emphasizes the importance of field-based geomorphic constraints at the same time. Moreover, many subtle fault traces which have not been recognized before are simultaneously revealed in the Dushanzi research area. The high-resolution DEMs guarantee the realization of picking out finer bits of fault information. Compared to traditional ways of working, the method has the advantage of automatically delineating reverse fault traces on the earth’s surface. This advantage can significantly reduce the efforts to manually digitize geomorphic features and improve efficiency. But many basic manual adjustment options for recognizing target characteristics also need to be set in extraction, because the distinguishing criterion of fault scarp and surrounding geomorphic landforms vary among different areas. In different specific circumstances, users can manually adjust relevant parameters for the extraction during the modeling process. Generally speaking, the more detailed constraints, the more confidence in the final delineation of fault traces. Subjective judgments are therefore particularly critical for conducting extraction under complex backgrounds. But improving the degree of automation of the whole process is still an important study direction. Future work is thus recommended to employ machine learning and explore appropriate evaluation methods to search for the optimal solution of intermediate parameters.

Key words: thrust fault, automatic extracting, geomorphic parameter, active tectonics, high-resolution digital topographic data