Mapping and modelling of landslide and flood hazards on St. Eustatius with openLISEM
Between 1900-2016 there have globally been more than 14,000 major events of natural hazards, like landslides and floods. The Caribbean is, partly due to the high abundance of islands, one of the most vulnerable areas of the world to natural hazards. The area experiences many different types of hazards with a high frequency of occurrence. When dealing with multiple hazards, it is important to notice that they are connected to each other. The research in the multi-hazard subfield is, however, still scarce and there are only a few real multi-hazard models.
St. Eustatius is an island in the Caribbean that experiences many different hazards. The information on the occurrences of these hazards is, however, very little, especially regarding landslides and floods. This study, therefore, investigated the historical occurrences of the landslide and flood hazards on St. Eustatius and their magnitudes for different return periods. This was achieved by constructing a hazard inventory and modelling the hazards with the deterministic model openLISEM. A new version of openLISEM has added the capability of modelling slope stability and debris flow simultaneously with the already available hydrological model.
The hazard inventory was constructed using field observations of historic landslide deposits. The flood hazard was not included in the inventory since it was too difficult to recognize historical flood events in the field. The data needed for the model input was constructed from available datasets and with the analysis of both field measurements and taken soil samples. After all the model input was constructed a sensitivity analysis was performed. This sensitivity analysis showed that the new openLISEM model is most sensitive to changes in the cohesion, internal friction angle, soil depth and initial soil moisture content.
After a qualitative calibration of the model with the constructed landslide inventory, the model was used with four model scenarios of design storms with a return period of 1-year, 5-years, 10-years and 50-years. The results of these model runs show that the slopes of the Quill volcano and the coasts of the island have the highest landslide hazard. The model results also indicate that there is no flood hazard on the island.
The model results were validated qualitatively with the constructed hazard inventory. A quantitative assessment could not be made since the landslide database was not detailed enough to perform this assessment. The assessment showed that it is highly likely that the slope failure and debris flow components of the model overestimate the intensity of the landslide hazard. However, the spatial occurrence of the slope failure events does coincide with the location of previous landslide events. ii
The predicted flood hazard could not be validated since there was no flood hazard inventory. The discrepancy between reality and the model results is partly due to problems within the model itself and partly due to the input data (meteorological data, DEM, geotechnical soil data and hydrological soil data) that was given to the model.
The results of this study should thus be interpreted with care. The new component of the openLISEM model still has some issues and is not suited, at the moment, to provide a detailed or accurate prediction of the hazards that occur on the island. For future research, it is important that a better hazard inventory of all the occurring hazards is constructed first. After this, a better DEM, more detailed data on soil characteristics of the island and a more suited model will also contribute to a better prediction of the hazards occurring on the island.