A landslide hazard probability map can help planners (1) prepare for, and/or mitigate against,
the effects of landsliding on communities and infrastructure, and (2) avoid or minimise the
risks associated with new developments. The aims of the project were to establish, by means
of studies in a few test areas, a generic method by which remote sensing and data analysis
using a geographic information system (GIS) could provide a provisional landslide hazard
zonation map. The provision of basic hazard information is an underpinning theme of the
UN’s International Decade for Natural Disaster Reduction (IDNDR). It is an essential
requirement for disaster preparedness and mitigation planning. This report forms part of BGS
project 92/7 (R5554) ‘Rapid assessment of landslip hazards’ Carried out under the ODA/BGS
Technology Development and Research Programme as part of the British Government’s
provision of aid to developing countries. It provides a detailed technical account of work
undertaken in a test area in Viti Levu in collaboration with Fiji Mineral Resources
Department. The study represents a demonstration of a methodology that is applicable to
many developing countries.
The underlying principle is that relationships between past landsliding events, interpreted
from remote sensing, and factors such as the geology, relief, soils etc provide the basis for
modelling where future landslides are most likely to occur. This is achieved using a GIS by
‘weighting’ each class of each variable (e.g. each lithology ‘class’ of the variable ‘geology’)
according to the proportion of landslides occurring within it compared to the regional
average. Combinations of variables, produced by summing the weights in individual classes,
provide ‘models’ of landslide probability. The approach is empirical but has the advantage
of potentially being able to provide regional scale hazard maps over large areas quickly and
cheaply; this is unlikely to be achieved using conventional ground-based geotechnical
methods.
In Fiji, landslides are usually triggered by intense rain storms commonly associated with
tropical cyclones. However, the regional distribution of landslides has not been mapped nor
is it known how far geology and landscape influence the location and severity of landsliding
events. The report discusses the remote sensing and GIS methodology, and describes the
results of the pilot study over an area of 713 km2 in south east Viti Levu. The landslide
model uses geology, elevation, slope angle, slope aspect, soil type, and forest cover as
inputs. The resulting provisional landslide hazard zonation map, divided into high, medium
and low zones of landslide hazard probability, suggests that whilst rainfall is the immediate
cause, others controls do exert a significant influence. It is recommended that consideration
be given in Fiji to implementing the techniques as part of a national strategic plan for
landslide hazard zonation mapping.