Although a landslide is an important factor in the development of landforms, it is also one of
the most re-occurring natural hazards affecting steep slopes around the world. The main
objective of the study was to assess the spatial and temporal probability to landslide
occurrence in the Bamenda mountain region which has witness many unexpected deathly
events in the last two decades. The study was carried out on large parts of the mountain area
(600 km2
) and some parts of the lowlands, giving a total area of approximately 913.7 km2.
A landslide inventory database was build from archives, aerial photographs and satellite
image interpretation and field work. A total of 355 landslides including 181 transaltional and
154 rotational landslides were registered and were validated through field suveys caried out
in 2009, 2010, 2013 and 2014. A landslide distribution map and a landslide density map for
each of the slope movements identified was constructed from the inventory.
A total of eleven geo-environmental factors believed to have an influence on slope instability
were examined and nine of them were retained after scrutiny using the relaibility and
accountability rates. The factors included: slope, curvature, aspect, lithology,
geomorphology, landuse, stream density, proximity to streams and superficial formation.
These factors served as conditioning factors used in analysing landslides activity in the
region. Two bivariate models (Informative Value and Weighted Index model) and one
multivaraite model (Logistic Regression model) were used to assess the spatial probability to
slope movements. The results from all three models show that the slope gradient is the most
important factor of slope instability. The impact of the other factors varied from one slide
type to the other with some factors being very influential in one slope movement and less
important in the other.
Daily rainfall from 1970 – 2013 was statistically analysed using the empiric Antecedant
Daily Rainfall Model and the Gumbel law. This data was used to assess the temporal
probability of rainfall on landslide activities. Each rainfall episode that triggered landslides in
the past had a return period in which it is assumed that landslides of the same type and
magnitude will again be triggered. The annual occurrence rate, the reoccurence interval and
exceedance probability of having one or more landslides occurring over a period of time was
assessed using the Poisson Distribution model. This was used to compute the landslide hazard
probability for different return periods. It was observed that most landslides were activated
when rainfall was above the threshold for the area. The ultimate goal was to produce a
xxiv
landslide hazard assessment that can be used to reduce the negative impacts of this
phenomenon in the area. This was done by making a probabilistic combination of the spatial
and temporal probabilities of landslides and this led to the construction of a landslide hazard
map for the area. This results can be used as a working tool for civil protection policy makers
when deciding which level of risk to accept and where and when action should be taken.
Finally, it was shown that quantitative methods can be used in predicting future landslide
events in the region.