Hello James, this is a very interesting question. It is difficult to define this beforehand, without information of where landslide actually is ocurring at the local scale. However, if your inventory helps you determine where landslides are more common, perhaps you can use that as a reference. Taking myself a chance, I would think that slope, soil, geology  and perhaps landcover can can be quite probable to have the highest weights. Roads for example would depend more on the type of road and mitigation strategies that were used in their construction. I would recommend you to locally get expert advise from local key informants (experts in this issue), make a first proposal from this consensus and then validate and correct your weights according to observed versus predicted vulnerability.

Other factors that could be considered  stratigraphy, thick colluvial deposits, hydrostatic levels, the amount of rainfall.

Most important factors are slope (above a critical value), type of soil (absorption of water) and rainfall intensity.

In India we have prepared landslide zonation (risk) maps for Himalayan region and the factors taken were:

Type of Soil

Water holding capacity

Slope in degrees and topography

Rainfall patterns (intensity & duration) using 25, 50 years of data

Land use/land cover, mainly vegetation type and cover, built up areas etc.

These databases were created and using a GIS model, the landslide zonation in categories of high., medium, low and nil was generated. These results were calibrated with past landslides in the region and results fine tuned.

These maps are being used for development of areas, laying of roads, siting of important infrastructure like townships, hydroelectric plants, hospitals etc.

Best regards.

y

Also consider earthquake-shaking potential as a factor and possibly volcanic activity (increased instability of slopes with changes in activity).

See our agency's statewide map for California at: http://www.consrv.ca.gov/cgs/information/publications/ms/Documents/MS58.pdf

Slope, height, mechanical characteristics rocks, pore water pressure, but also triggers such as rainfall, seismic shocks (earthquakes or vibrations), overloads.

Dr. Karim, I think you are absolutely right. It is not so easy to determine already in advance which factors play a crucial role in landslides. I was not aware of this fact when I raised the question. I discovered that it unquestionably depends on the study area, available Landslide Inventories, factor maps and their quality as I learnt from my short project recently concluded for St. Lucia in the Caribbean. Although we received many Inventories, factor maps and various other information related to the Island country from different agencies, the thematic and geometric quality of data was dubious. Due to constraints in time, it was not possible to drastically improve their quality when we tried on our own.

Nevertheless, you, Prof. Balakrishnan and Prof. Maria have correctly pointed out that Slope is a major contributor in deciding landslides. There are other factors which you have mentioned, which even I thought would determine the course of landslides like geology, soil, LULC etc. Unfortunately, these factors were very vague and of poor quality, therefore they were not considered, even though they were tried in the model to figure out their correlation with landslides. Similarly, rainfall data was too coarse for the study.

We could have considered the Seismic Acceleration - Initiation model which Mr. Higgins has specified but it was beyond the scope of this project. However, it is a relevant factor in the case of the Caribbean Islands, given their proximity to tectonic plate boundary and their history of volcanic origins. Thank you for the Californian case study link. They show very detailed susceptibility information which would be quite useful for Land use planning activities for CA. A good example of a Susceptibility Map indeed.

Renie, thanks for the demographics info. addition. I haven't started with the inclusion of population figures as the data was unavailable. But I did consider the number of buildings in various susceptibility zones, indirectly taking the demographics into account. I also believe without actual demographics the study would not make a huge impact. Risk Assessment is something which I'll be dealing with in the coming months. Therefore there is still scope to encompass it in the Disaster Risk Analysis.

We also tried out a Dynamic Run-out Modeling using Flow-R, a model developed in the University of Lausanne, Switzerland which runs in a Matlab environment to simulate types of mass movement on a regional scale. This dynamic method was purely experimental and was not documented in the Report. 

Thank you Dr. Andreea and Dr. Hamid for your answers and for the article link. They were useful for my study as a background literature.

With all of your expert perspectives my horizon of looking at Landslide Susceptibility Assessment has truly widened. With my next project undertaking they will form a beneficial part of my thinking process. I've attached a PDF which summarizes the findings for your scrutiny.

Data Landslide Susceptibility Assessment - St. Lucia Island, Caribbean Sea

This is a very critical problem of generating the landslide hazard map. Better to know triggering mechanisms and collected more related factors. 

Hi,

It depends on: 

• How do you perform the SMCE?
• By means of the auto generation?
• By means of decision maker participation?  Combining both?
• Who are the decision makers?

When SMCE is applied to processes usually the results depend not only on the geographical distribution of attributes but also on the value judgement of decision makers. The main disadvantage of this method is that subjective preference in the ranking of factors may differ from one expert to another. So intensity of importance can be judged using landslide inventory.