Landslide inventory for hazard assessment in a data-poor context : a regional-scale approach in a tropical African environment

Landslide hazard remains poorly characterized on regional and global scales. In the tropics in particular, the lack of knowledge on landslide hazard is in sharp contrast with the high landslide susceptibility of the region. Moreover, landslide hazard in the tropics is expected to increase in the future in response to growing demographic pressure and climate and land use changes. With precipitation as the primary trigger for landslides in the tropics, there is a need for an accurate determination of rainfall thresholds for landslide triggering based on regional rainfall information as well as reliable data on landslide occurrences. Here, we present the landslide inventory for the central section of the western branch of the East African Rift (LIWEAR). Specific attention is given to the spatial and temporal accuracy, reliability, and geomorphological meaning of the data. The LIWEAR comprises 143 landslide events with known location and date over a span of 48years from 1968 to 2016. Reported landslides are found to be dominantly related to the annual precipitation patterns and increasing demographic pressure. Field observations in combination with local collaborations revealed substantial biases in the LIWEAR related to landslide processes, landslide impact, and the remote context of the study area. In order to optimize data collection and minimize biases and uncertainties, we propose a three-phase, Search-Store-Validate, workflow as a framework for data collection in a data-poor context. The validated results indicate that the proposed methodology can lead to a reliable landslide inventory in a data-poor context, valuable for regional landslide hazard assessment at the considered temporal and spatial resolutions