**3. Materials**

#### *3.1. Historical Landslides in Kenya*

A landslide inventory usually portrays the date, location, cause, type, and geometry of landslides. A detailed landslide inventory is a mandatory input for LSM using quantitative methods. Landslide inventories can be produced through field surveys, review of relevant documents, and remote sensing (satellite images, aerial photos, etc.). Since compiling a detailed landslide inventory remains a time- and labor-consuming task there has been no reliable landslide inventory with national coverage available in Kenya until now.

Considering the conditions mentioned above, the landslide inventory utilized in the present study was composed of three subsets from di fferent sources, as illustrated in Figure 1. The first subinventory (LS1) was obtained from the Kenya Open Data of the Ministry of Information, Communications, and Technology of Kenya. LS1 contained 39 historical landslides that occurred in Kenya during the 1999–2013 period. Landslide information regarding the specific longitude/latitude, dates of occurrence, number of people a ffected, and estimated economic losses were provided in the LS1. The second subinventory (LS2) was derived from the global fatal landslide database (GFLD) (version 2) extracted from Froude and Petley [28]. Similar to LS1, LS2 also provided the location, date of landslide occurrence, induced fatalities, general description, and related reports of landslides. In total, 63 landslides were recorded in LS2. The third subinventory (LS3) was extracted from a landslide inventory of Africa, which was compiled by Broeckx et al. [29]. Within Kenya's territory, 323 landslides were detected in LS3. The majority of landslides in LS3 were mapped through visual interpretation of Google Earth imagery. This was time-consuming work and involved subjective interpretation. In contrast to LS1 and LS2, only the locations of landslides were provided in LS3. For the present study, a total of 425 landslides were stored in the inventory.

As indicated in Figure 2, the landslide occurrences were mostly concentrated in the periods from April to May and from November to December. It should also be noted that the monthly distribution of landslides was consistent with the monthly distribution of precipitation.

#### *3.2. Landslide Contributing Factors (LCFs)*

From the perspective of geotechnical engineering, landslides are a comprehensive consequence of several contributing factors. Nevertheless, there are no global rules for selecting these factors. In a typical LSM, such LCFs are chosen on the basis of data availability, characteristics and scale of the study area, as well as the expert knowledge or experience. In this study, ten LCFs were utilized in LSM over a nationwide area of Kenya, including four topographic factors (namely, the altitude, aspect, slope, and curvature), two hydrological factors (namely, the topographic wetness index (TWI), stream power index (SPI)), soil texture, precipitation, land use, and landform (as shown in Figure 3). To perform the LSM, all factors were rasterized into 1 × 1 km grids and classified into several classes using ArcGIS software (Version 10.2). In what follows, a brief description of each landslide contributing factor is given.

#### 3.2.1. Mean Annual Precipitation (MAP)

Rainfall is among the dominant inducing factors for landslides not only in Kenya but also in many countries because it increases the soil mass and decreases the soil shear strength. As shown in Figure 2, there are strong correlations between landslide occurrences and precipitation in Kenya. For this study, the MAP data were adopted and categorized into nine levels using a 400 mm interval, as shown in Figure 3a. To initially obtain the MAP factor, monthly total rainfall data from 87 meteorological stations in Kenya were collected and filtered. Only data from stations operated in all types of weather were kept. Then, the filtered monthly rainfall data were summed and averaged annually for each station. Finally, through the inverse distance weighting (IDW) of data for each station, the MAP map of Kenya was derived.
