| dc.description |
Kursah, M.B., School of Resources and Environment, University of Electronic Science and Technology of China, 2006 Xiyuan Avenue, West Hi-tech Zone, Chengdu, Sichuan 611731, China, Department of Geography Education, University of Education Winneba (UEW), Box 25, Winneba, Ghana; Wang, Y., School of Resources and Environment, University of Electronic Science and Technology of China, 2006 Xiyuan Avenue, West Hi-tech Zone, Chengdu, Sichuan 611731, China, Department of Geography, Planning, and Environment, East Carolina University, Greenville, NC 27858, United States; Bayoh, H.D., Pan African Climate Justice Alliance (PACJA), Freetown, Sierra Leone; Tarawally, M., Department of Science Education, Ernest Bai Koroma University of Science and Technology, Makeni Campus, Makeni, Sierra Leone |
en_US |
| dc.description.abstract |
Landslides have caused substantial economic loss, infrastructural damage, and threatened lives and human shelter in Western Area, Sierra Leone (Africa), requiring effective mitigations. Landslide susceptibility (LS) mapping is one effective and foremost step in reducing landslide disaster as it provides vital information to the local authority for land use planning and management in the future. A set of controlling factors and landslide inventory (LI) are typically used to generate LS. Thus, the LS accuracy can be affected by the LI availability and its source and timespan. Here, the impact of LI from two sources and timespans on the accuracy of LS is analyzed. The LIs are the archived inventory (archived-LI) and small baseline subset (SBAS) interferometry inventory (SBAS-LI). The former LI consisted of landslide events from 1945 to 2014. The latter was acquired by analyzing the ground deformation with multi-temporal Sentinel-1 SAR datasets and covered between 2015 and 2017. Eleven landslide controlling factors after multicollinearity diagnostics were selected. With the controlling factors in a GIS-based environment, the frequency ratio (FR) model is applied to both LIs and the factors, generating two LS maps. Each map accuracy was validated using recent landslides from 2017 to 2020. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve was 0.739 and 0.898 for the LS from the archived-LI and SBAS-LI, respectively. The degree-of-fit index of 61.7% for the LS from the archived-LI and 80.8% from the SBAS-LI was obtained. The SBAS-LI produced more reflective present LS and possibly for future susceptibility even though the number of events in the modeling was about one-third of that in the archived-LI. Thus, the current LI is preferred for LS analysis. � 2021, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature. |
en_US |