Spatial Constraint Modeling and Continuous Quantification of Loess Slope Topography Based on an Adaptive Dynamic-Window DEM

Abstract

Addressing the core need for precise quantification of the topographic characteristics of the Qin Straight Road, this study developed an improved dynamic-window continuous DEM model aimed at quantitatively analyzing the complex topographic features of loess slopes. Based on high-resolution topographic data and ground-penetrating radar monitoring data from the Yan'an region of Shaanxi, this study overcomes the limitations of traditional DEM models that rely on fixed-window calculations. By introducing a terrain complexity assessment mechanism, it dynamically adjusts the calculation windows for gentle and complex terrain areas, and utilizes field-validated calibration coefficients to correct elevation interpolation errors. The quantification process focused on extracting three core indicators: slope gradient, elevation difference along route segments, and distance to the watershed, and successfully exported the quantification results for ten designated points to a specified document. Experimental results show that the model's average relative error for quantification at designated points is only 2.1%, with a goodness-of-fit exceeding 0.94, representing a 73.75% improvement in accuracy compared to traditional models. This study not only achieves continuous mapping of topographic features but also accurately identifies multiple potential points of topographic abrupt change, providing a high-precision spatial constraint benchmark for subsequent route optimization and facility planning.