**7. Conclusions**

In this research, an integrated method of fuzzy theory and conventional AHP analysis was employed for the LSM of Kenya. A two-level hierarchical index system was established to predict landslide susceptibility with a GIS platform. Ten factors contributing to landslide occurrence were included in the first level of the evaluation system. These contributing factors included slope, altitude, aspect, SPI, TWI, curvature, land use, MAP, landform, and soil texture. For the second level, each of these factors was divided into several subclasses. The weights of these factors and their subclasses were determined using the adopted TFN-AHP theory. A nationwide landslide susceptibility map for the entire Kenyan territory was produced with five different levels ranging from extremely high susceptibility to very low susceptibility. Extremely high and high landslide susceptibility zones primarily covered the rift valley and its nearby regions. Validation results using ROC curves indicated that the TFN-AHP method performed well for developing LS maps of the study area. This method resulted in a higher AUC accuracy than the conventional AHP using the same datasets.

This study was the first attempt to identify landslide susceptibility zones in Kenya on a national scale. The produced map can be used as a general indicator of the relative landslide susceptibility for larger areas rather than an accurate susceptibility measure for each specific site. The results would be helpful in various land resources-related fields to inform decision making, such as regional landslide hazard mitigation, land use management, and infrastructure planning.

**Author Contributions:** X.T. contributed to the conception of the study, S.Z. (Suhua Zhou) performed the data analyses. S.Z. (Shuaikang Zhou) wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the National Natural Science Foundation of China (grant number 51708199); the Science and Technology Infrastructure Program of Guizhou Province (grant number 2020-4Y047: 2018-133-042); the Fundamental Research Funds for the Central Universities (grant number 531107050969); the Science and Technology Program of Beijing (grant number Z181100003918005). All these financial supports were acknowledged.

**Acknowledgments:** The authors would like to express their gratitude to Fang and Li Hongquan in Central South University for their kind help in evaluating the LCFs.

**Conflicts of Interest:** The authors declare no conflict of interest.
