Menu

Insights

March 1, 2016

Landslide susceptibility for Jamaica

landslide1

Landslide susceptibility for Jamaica using Conditional Probabilities in a GIS

Christopher Burgess, PE, MSc, Kristifer Freeman, BSc and Carlnenus Johnson, BEng

A Susceptibility spatial analysis was done to further the understanding of the landslide phenomenon in Jamaica. Such an understanding will aid in the reduction of loss of life, property damage, and social and economic disruption from hillslope movements, through both planning and mitigations measures.

Geospatial analysis of a landslide inventory of four parishes was used to calibrate the susceptibility model. The variables used in the model include: slopes, distance from faults, and roads and soil types. Slopes were derived from a radar dataset, and soils, roads and faults were derived from relevant national database. Once the model was satisfactorily tuned it was applied on an all island basis. A general trend of high susceptibility in St. Thomas, St. Mary and north St. Andrew was consistently revealed and coincided with previous works. When applied to the North-south highway of Highway 2000 project through Jamaica’s the predictions coincided with observations of geologists on brecciate landslide prone and landslides that were evident. The landslide susceptibility map identifies geographic areas where past landslides have occurred and future land slippages are possible at varying degrees of likelihood. Approximately 3% of the road network is susceptible to landslides. Having access to such key information, planners and engineers can be forewarned as to the consequences for roads and other development in such areas. Suitable planning and investigatory measures can be undertaken to address these concerns.

Introduction 

The geological processes and active tectonics which have produced the existing landscape on the island of Jamaica contributes to making it most vulnerable to natural hazards. Nonetheless, when natural disasters are discussed, one would most likely mention hurricanes, flooding and earthquakes. However, of these natural hazards, landslides are the most common in Jamaica but are often unobserved due to them commonly coinciding with other natural disasters by which they are triggered.

Landslide is a general term used to describe a wide variety of processes that result in visible downward and outward movement of sloping materials under the influence of gravity. This natural hazard occurs when the gravitational pull increases in the instance where weight is added to the rock mass by additional load (structures) or when the rock becomes saturated (rainfall). Landslides may also occur in submarine environments, along coastlines and on shelf regions, where they are generally triggered by earthquakes (e.g.) Port Royal, 1692. They can be classified by rate of movement, type of material and nature of movement. A major contributor of landslides is land use; the change of land use may induce movement of soil downslope creating land slippage (e.g.) excavation for a new road.

For purposes of this report, the term ‘landslide’ is defined as any gravity induced movement of natural rock, earth, debris or any combination of these materials down a sloped section of land, usually triggered by either geological characteristics (weak or fractured), morphological attributes (structure of land) or human activity. The moving soil mass may proceed in any one of three principal types of movement: falling, toppling, sliding, spreading, flowing or by their combinations (United States Geological Survey, 2013). Most landslides, however, are initiated as as thin earth (soil) slides or debris slides (soil with pieces of rock) (Ahmad, Miller, & Rowe, Landslides Related to Precipitation in Eastern Jamaica).

Geological information on locations within Jamaica which are vulnerable to landslide hazards, the methodology employed in the preparation of the hazard map and its advantages and limitations are presented in this study.

Presently, landslides are not agreeable to risk assessments due to there being no basis in determining the probability of landslides occurring within a given time period. However, hazard assessments are indeed feasible and may be substituted for risk assessments. Hazard assessments are estimations of an area’s vulnerability to landslides based on but not limited to three (3) intrinsic physical factors – distribution of historical landslides, steepness of slopes and geological structure of the bedrock. The zonation of Jamaica, according to these differing degrees of hazard, has led to the production of an island-wide landslide hazard map.

Landslide hazard maps provide information that can be used to identify various levels of risks generated by landslides. This facilitates citizens, planners, engineers, and developers with the implementation of appropriate structural and non-structural loss reduction strategies for both existing and future developments. The map generated in this paper was intended primarily for planning purposes on a regional scale; however, a comprehensive island-wide study was conducted. The map indicates the indirect extent and relative severity of landslide hazards and can be used in the preliminary selection of housing and infrastructure development areas. In addition, the map can identify regions where detailed geologic-geotechnical investigations are advantageous prior to development.

Methodology 

The approach taken to this assessment involved both qualitative and quantitative methods of data collection and analysis. The methodology employed involved: data collection and analysis, calibration and preparation of the susceptibility maps.

The specific set of input parameters included:

Within the landslide vulnerability study, a susceptibility raster matrix approach was implemented; this method relies on the inventory of historical landslides. Initially, factor maps were generated for slope angles, lithologic groups (soil types), distance to faults and distance to roads. The slope analysis was based on a digital elevation model (DEM) created from the Digital Globe satellite radar data, in order to highlight the slopes across the island. The slope angles were grouped into four (4) classes based on natural breaks in the frequency histogram.

The inventory of historical landslides were transposed into a raster format, reclassified and weighted into two (2) categories. These landslides varied in magnitude but were concentrated in the eastern part of the island.

The various soil types and textures were determined from the soils inventory obtained from Ministry of Agriculture. The soils data received had over 203 soil types within the zones where landslides were known to have occurred across the island. The probabilities of these soils causing a landslide were calculated based on data of existing landslides. This was then reclassified and weighted into three (3) classes.

All input parameters were then converted to a common format – raster maps. Each parameter was assessed using the assigned susceptibility weighting based on a conditional probability that the parameter has occurred given a landslide has occurred, using the following probability formula.

forumla
 

Each parameter was then equally weighted, from low to high susceptibility, and the product of the susceptibility of all the parameters was weighted in four classes varying from low to high susceptibility. By overlaying the maps of these five (5) factors, a landslide susceptibility map of composite categories were created according to the following raster equation:

Model Predictions

The model was calibrated using the existing landslide areas on the eastern end of the island, where landslide occurrences are high. The model was then applied to the entire island and the results plotted. The landslide susceptibility map for the island of Jamaica is shown in Figure 1. The vulnerability of landslides is slight throughout the island with scattered areas of moderate and high susceptibility. Most of the high susceptible areas are within close proximity to fault lines which can be due to the presence of brecciaed stones caused by the faults themselves.

 
landslide-susceptibility-map
Figure 1 Landslide susceptibility map generated from parameters

By observation, the eastern end of the island is the most vulnerable to landslides due to the steep terrain and soil types which exist. The central section of Jamaica is of relatively low susceptibility but as the mountainous areas are traversed, the probability of landslides occurring increases and the susceptibility enters the moderate to high risk classes accordingly.

Verification Process

With the final susceptibility map generated, the results may now be verified. The landslide susceptibility map generated demonstrates that there is a group of high vulnerable areas on the eastern end of the island which are prone to landslides. The regions predicted by the landslide model to be of moderate susceptibility has the highest occurrence of existing landslides, followed by the high susceptibility class.

A graphical representation of the model-predicted susceptible areas and their correlation to actual existing landslides is illustrated in Figure 2.

 
landslide-susceptibility-map-2
Figure 2 Landslide susceptibility shown correlating to existing landslides

The verification process was taken a step further and into the field. Using the landslide susceptibility map as a guide, highly vulnerable areas were explored and investigated in search of landslides. Having already observed the Mt. Rosser landslide, its location could be quickly identified and verified as being in a moderate susceptible area shown in Plate 1 and Plate 2. In addition, two landslides were found at the northern entrance to the Bog Walk Gorge one of which, according to the susceptibility map, occurred in a high susceptible area while the other in a slight vulnerable zone. These landslides are also illustrated in Plate 3 and Plate 4. Verification illustrates that most historical landslides have occurred in moderate to high susceptible areas.

 
landslides

Road Vulnerability to Landslides

The road network analysis and associated vulnerabilities was analyzed further. The road lengths related with the model predicted susceptibilities were identified and evaluated. The results showed that approximately 3% of the road network has high susceptibility to landslides.

Leave a Reply

Your email address will not be published. Required fields are marked *