Lauren Sullivan
3/18/2010
Research Brief
Introduction
Landslides are one of the most destructive forms of geological hazard (Fernandez 2003). Several different types have the ability to ravage various terrain and each type has differing requirements in order to initiate the slide. Yet, despite the risk, the Index to Landslide Maps provided as a courtesy by the State of California on their website indicates that most of the maps referenced were made about 20 years ago. Because GIS allows us to map quite easily what might have been difficult when the State made the last set of Landslide Hazard maps, an updated analysis is now not only necessary, but much more feasible than when the last maps were commissioned.
The California Geological Survey defines several major types of Landslide Maps: Landslide-Inventory Maps, Landslide-Hazard Maps, Landslide-Risk Maps and Landslide-Zone Maps. Landslide-Inventory maps are essentially maps of past landslides, while Landslide-Hazard maps attempt to effectively predict where new landslides will occur. Landslide-Risk maps focus on the potential for landslides to destroy residents or their property, while Landslide-Zone maps merely define and depict areas in general that have a higher landslide potential. For the purposes of this project, I chose to create Landslide-Hazard Maps. This category has two further divisions, Landslide-Susceptibility maps and Landslide-Potential maps, which differ in defining areas of landslide potential based on native aspects of the area (slope, soils, etc.) versus the potential for a landslide due to an inducing event, such as an earthquake or heavy storm. Because the data that I could acquire for this project had a time limit and needed to be publicly (and somewhat easily) available, I decided to combine these map types to make an overall Landslide-Hazard Map for San Diego County.
There are many great articles currently available that discuss the benefits of GIS in landslide hazard analysis. Researchers use a variety of different methods, but many have found that, even after using a complex analysis with up to 17 different input layers (Remondo et al. 2003), only a few different datasets determine most of the probability of a landslide (Fabbri et al 2003), with layers such as geology, surface soil material and depth, and land use playing a secondary role in the determination of areas prone to sliding. Knowing this, I gathered my data to use for my analysis. I imagined that areas with the steepest slopes, heaviest rainfall and the areas affected most by recent wildfire would be most susceptible to landslides.
Methods
I used six different layers in order to analyze landslide risk in San Diego County accurately. I first located a DEM dataset for the County, and used it to make a new layer to depict slope. Next, found the locations of fault lines within San Diego, and buffered them to 1 kilometer, which I believe are areas most likely to experience the effects of such close proximity to faults. Though I would have loved to find a more accurate figure, I had a hard time looking it up so I estimated instead. These fault lines are relatively small and there are many of them in one central area, so my reasoning was that those close enough to be affected were probably within a kilometer of any fault. My third layer was a rainfall layer that I obtained from the SanGIS website for the County. Layer four was the areas of wildfires in the County. Because most of San Diego vegetation is chaparral, which is notorious for growing in soil that becomes highly impermeable following a wildfire, I thought this aspect was important enough to warrant a separate layer analyzing it, even though other researchers did not include it. My fifth layer was a dataset that classified all of the San Diego County soil types by erodibility: Slight, Moderate, or Severe. Surprisingly, much of the county soils were in the Severe category. My last layer was areas of geohazards in San Diego—locations of past landslides, areas prone to soil liquefaction, coastal bluffs (likely to collapse from erosion) and other slide-prone formations. This layer assisted me by providing the necessary data on previous landslides, which was a major portion of every analysis by other landslide researchers.
After converting necessary layers to a raster format, I also reclassified several layers as necessary. First, I created a slope raster that focused on slopes greater than 30 degrees. Though sliding does occur in slight or moderately steep slopes that are less than 30 degrees, I needed to create a threshold and used the definition from Endo et al., who defined generally hazardous slopes as those greater than 5 meters in height and a slope of 30 degrees. I also reclassified the soil erodibility layer and fire layer so that pixel values for areas with no data became zero. The layers analyzed in the Raster Calculator must be of the same extent (which I did not realize) so I learned how to assign values to areas that were not previously included in the analysis. This, for example, explains which much of the desert area of San Diego in the main Landslide Analysis map appears pixellated and much darker, indicating a low landslide probability. While the desert not only had fewer, lower slopes and less rainfall, making it a poor candidate for landslides, but it was also missing data—such as the soil erodibility layer. I did not worry about this lack of data, however, as it only occurred in the desert where it appeared that few landslides took place anyway.
After correcting the layers, I used the Raster Calculator to evaluate the following expression:
2 * ([Fault Zones] + (2 * [Fire]) + (3 * [Prone to Sliding]) + (0.5 * [rain]) + (3 * [Slope]) + [Soils])
While other researchers clearly believed in the use of GIS to evaluate landslide probability, none of the research I dug up appeared to use this method of evaluation, so when it came to writing the Raster Calculator equation I was left to estimate based on common sense. Ultimately, the range of values for each of my layers was in landslide index units. For example, for soil erodibility, Slight = 1, Moderate = 2, and Severe = 3. The calculator essentially added all these indices (weighted) together. I determined that the highest-weighted layers (areas prone to sliding, slope and rainfall (because it had relatively high values for more inundated areas)) were most important in determining slope instability from reading the relevant research or from the California Geological Survey’s website.
Results
My final output map from this analysis is the map of Slope Instability in San Diego County. Areas of highest slope instability appeared to be along the beaches near La Jolla and Point Loma as well as select slopes inland, particularly around Palomar Mountain. It is also clear that several gorges in the Ramona/Julian area and between Ramona and Palomar Mountain have a high probability of collapsing or sliding. These areas have steep slopes, get lots of heavy rainfall, have a high soil erodibility and have recently burned (last 5-10 years). However, coastal areas prone to landslides are unstable for other reasons. Areas near Torrey Pines and La Jolla are centrally located over San Diego fault lines, have steep slopes, erodible soils, and are prone to liquefaction or sliding from a sandier soil. These areas have had landslides before. Though the areas of the map are red for different reasons, they are both prone to landslides.
I would not say my map is completely free from error, but I do think that it does a good job of representing areas where landslides might occur. Regarding its limitations, I would say that I would attempt to classify the slopes differently next time. Debris flows, which are a more common problem in mountainous areas like the San Gabriel Mountains, require steeper slopes than the kinds of landslides that take place along the coast. I think my map does underestimate to a small degree the hazard of landslides near the coast because the slopes are not as steep there and in comparison they appear less hazardous. In reality, these slopes do not need to be as steep as the mountainous areas to provoke a landslide, because the soils near the beach have very little clay helping them to bind together, which makes it difficult for the soil to break away in a slide. I would also investigate more thoroughly the use of aspect in landslide analysis. Several other analyses used slope aspect (which direction the slope faces) in their calculations but never disclosed exactly how aspect affects the slope instability. I could have integrated aspect into my own analysis but did not know exactly how the direction of some slope faces contributes to landslide probability more than others.
Conclusion
Overall, I believe my maps do an adequate job of predicting areas of potential slope instability, both on the coast and inland in San Diego County. Mountainous areas with steep slopes, heavy rainfall and a recent history of fire are prone to fast runoff and, when devoid of vegetation, the soil can easily slip from the mountainside. Though I believe that this analysis would be useful in any area with as diverse of topography as San Diego County, I would not hesitate to change the analysis based on either new data or a better understanding of the current information.
References
Barlow, J. and S. E. Franklin. 2007. Mapping Hazardous Slope Processes Using Digita Data. In Li, J., Zlatonova, S., and Fabbri, A. (eds.) Geomatics Solutions for Disaster Management. Lecture Notes in Geoinformation and Cartography Series. Berlin: Springer.
California Geological Survey. 2007. "California Geological Survey -- Landslides." http://www.conservation.ca.gov/cgs/geologic_hazards/landslides/Pages/Index.aspx
Endo, K., et al. 2001. Slope failure disaster prevention activities in local government using GIS. Proceedings, ESRI International User Conference. San Diego: July 9-13. http://proceedings.esri.com/library/userconf/proc01/professional/papers/pap996/p996.htm (Accessed 14 March 2010)
Fabbri, A. G., et al. 2003. Is prediction of future landslides possible with a GIS? Natural Hazards. 30: 487-499
Fernandez, T., et al. 2003. Methodology for landslide suceptibility mapping by means of a GIS. Application to the Contraviesa Area (Granada, Spain). Natural Hazards. 30: 297-308
Remondo, J., et al. 2003. Landslide susceptibility models utilising spatial data analysis techniques. A case study from the Lower Deba Valley, Guipuzcoa (Spain). Natural Hazards. 30: 267-279
0 comments:
Post a Comment