Abstract: Investigation of loess microstructure can reveal information on the seepage of loess masses and the formation and evolution of sliding surfaces in loess landslides at both macroscopic and mesoscopic levels. A loess microstructure connectivity model, both with and without fractures, is established to investigate the relationships among pore connectivity p, pore variation coefficient λ , maximum fracture span ω , and fracture network penetration q. The connectivity thresholds of these two models are verified through scanning electron microscopy analysis and field seepage tests of loess microstructure. The results show that the existence of fractures, especially their time spans, substantially affects the morphology, size, and distribution of pores in loess. Fractures are important channels for preferential flow and significant contributors to loess preferential seepage under low porosities. The research conclusions from the loess microstructure preferential connectivity model studies agree with the loess seepage test results. These results explain the negative effects of the “flooding-type” irrigation in Nanyuan, Jingyang County, on the stability of high-angle slopes containing fractures and the occurrence of major landslides in Dongfeng and Jiangliu villages.

Key words: loess, preferential flow, connectivity, microstructure, geographic information system(GIS)