Modelling the Impacts of Predicted Environmental Change on the Frequency and Magnitude of Rainfall Induced Landslides in Central Kenya
|Faculty/Professorship:||Fakultät Geistes- und Kulturwissenschaften: Abschlussarbeiten|
|Author(s):||Mwaniki, Mercy Wanjiru|
|Publisher Information:||Bamberg : opus|
|Year of publication:||2016|
|Pages:||xviii, 213 ; Illustrationen, Diagramme, Karten|
|Supervisor(s):||Möller, Matthias ; Schellmann, Gerhard|
Dissertation, Otto-Friedrich-Universität Bamberg, 2016
|Licence:||Creative Commons - CC BY-NC-ND - Attribution - NonCommercial - NoDerivatives 3.0 International|
The central highlands of Kenya frequently suffer the impacts of rainfall-induced landslides resulting from the interaction of slope stability and elements of environmental change (land-use and climatic variables). The impacts of rainfall-induced landslides affect the country’s fight against poverty, bearing in mind the limited budgets to cope with the socioeconomic losses incurred by landslide hazards. On the other hand, a fast population growth rate puts pressure on the country’s resources which is majorly agricultural based, thus contributing to more people settling on steep slopes and increasing their vulnerability to rainfall landslide hazards. Thus, this research sought to contribute to the mitigation measures by mapping the landslide areas, performing landslide susceptibility assessment, and investigating the impacts of predicted environmental change on the frequency and magnitude of rainfall-induced landslides. The role of environmental change was investigated using specific objectives which assessed the impacts of land-use on slope stability, and the impact of precipitation characteristics on landslide susceptibility. Several data types ranging from topographic, soil and geology, land-use land-cover (LULC), hydrology, and precipitation landslide controlling factors were mapped and used in the modelling process.
The methodology comprised of LULC change detection with Landsat multitemporal data for the years 1995, 2002, 2010 and 2014; structural geology and soil mapping; landslide inventory creation with Landsat multitemporal data for the years 1995, 2000, 2010 and 2014; landslide susceptibility mapping with Combined Hydrological and Slope stability Model (CHASM) and landslide modelling with Artificial Neural Network (ANN) model. The success of mapping and visualizing geology lineaments was owed to the digital image enhancement methods involving band ratioing, False Colour Composites (FCC), feature data transformation and data reduction methods of principal and independent component analysis. In addition to the feature data transformation and data reduction, the landslide inventory mapping was enhanced by utilizing a Normalized Difference Mid-Red (NDMIDR) spectral index involving Landsat geology and red bands.
The key results of this research indicated that human activities relating to land-use (mostly agricultural) did aggravate the landslide processes on the sloppy terrain. This was confirmed by the CHASM model results where forested slopes maintained low landslide susceptibility levels. In addition, the ANN model rated LULC, rainfall, and proximity to drainage network factors high in contributing to landslide occurrence in the study area. Thus, majorly shallow types of landslides dominated, although the ANN model mapped some areas with deep-seated landslide areas along lineament features. The impacts of heavy precipitation were observed to increase slope instability, especially in bare land covers and high density drainage network areas due to rapid soil saturation, while prolonged precipitation increased infiltration thus maintaining high landslide susceptibility levels. The effects of climatic variables were associated with increased rock weathering observed on bare volcanic rocks, hence high instability rates around such areas. Landslide hazard zonation with ANN model captured several landslide types and the stability classification. The results of this study can guide targeted policies on land-use management as it has been established that rainfall induced landslides are a result of the interactions of land-use, slope and rainfall landslide conditioning factors. Moreover, creating a landslide inventory which can be updated with landslide attributes was a success since this had not been done in this geographical location to indicate the potential of landslide reactivation.
|GND Keywords:||Kenia; Regen; Bevölkerungswachstum; Umweltveränderung|
|Keywords:||Landslides, susceptibility assessment, Environmental change, Landsats, Geo-hazard, Normalized Difference Mid Red (NDMIDR) spectral index, Artificial Neural Network (ANN) model, Digital Image Processing (DIP) and enhancements, Digital Image ProcCombined Hydrology and Slope Stability model (CHASM)|
|DDC Classification:||910 Geography & travel|
|RVK Classification:||RS 56841|
|Year of publication:||14. December 2016|
originated at the
University of Bamberg
University of Bamberg