GIS applications for modeling geomorphic processes

Annotated Bibliography
Jeremy Adams, GEO 565, Winter 2007



La Conchita, CA Landslide 1995
U.S. Geological Survey photo by R. L. Schuster


GIS provides powerful spatial analysis capabilities useful for for understanding geomorphic processes and assessing physical variables that influence these processes.  Numerous models employing GIS techniques have been developed to assess landslide susceptibility and soil erosion risk, among other geomorphic processes.  The ability to model and predict the behavior of landslides and soil erosion is useful for attempting to understand how external influences such as land use modify the rate and magnitude at which these processes occur.    

 


Literature

Ayalew, L., Yamagishi, H.  2005.  The application of GIS-based logistic regression for landslide susceptibility mapping in the Kakuda-Yahiko Mountains, Central Japan.  Geomorphology, 65, 15-31.

 

The authors develop a GIS based logistic regression model to assess landslide susceptibility in the Kakuda-Yahiko Mountains of Central Japan.  Using logistic regression, the presence of landslides is correlated with independent physical parameters including lithology, elevation, lineaments, bed rock-slope relationship, slope gradient, aspect and proximity to roads.  Remote sensing edge detection techniques were used to isolate lineaments, and a 100m buffer was established to assess the influence of lineaments on landslide activity.  A 10m DEM was used to derive slope and aspect and landslide density was calculated for classified relief zones and slope.  Road networks were buffered to assess the relationship with landslide distribution.  Logistic regression was used to test and weight the relationship between landslide distribution and each independent physical parameter.  Based on this analysis, lineaments were deemed insignificant and road networks were identified as the most influential in landslide activity.  Coefficients for each parameter were developed and a landslide probability value was assessed and mapped. 

Curtis, J.A., Flint, L.E., Alpers, C.N., Yarnell, S.M.  2005.  Conceptual model of sediment processes in the upper Yuba River watershed, Sierra Nevada, CA.  Geomorphology, 68, 149-166.

 

Curtis et al. develop a GIS based model to identify erosion potential based on field derived measurements of hillslope erosion and channel storage sediment yields.  Hillslope erosion and sediment source was calculated using various layers including: soil erodability, road networks, bedrock geology, vegetation type and percent cover, slope and elevation, potential evapotranspiration and locations of existing mass failures.  The model spatially distributes major processes controlling sediment dynamics and is useful for identifying the spatial distribution of sediment sources.  This article is useful for understanding the integration of field derived sediment yield information with a GIS to calculate potential hillslope erosion.      

Dai, F.C., Lee, C.F.  2002.  Landslide characteristics and slope instability modeling using GIS, Lantau Island, Hong Kong.  Geomorphology, 42, 213-228.

 

The authors develop a GIS based logistic multiple regression model for assessing slope instability on Lantau Island.  The authors develop an algorithm to model slope instability using a GIS.  Historic landslides were digitized from aerial photographs and incorporated into a GIS.  Physical parameters including slope and aspect derived from a 20m DEM, land-use, terrain morphology, and geology were assessed.  In addition, the distance between landslide distributions and drainage networks was established.  Vector datasets were rasterized to the DEM spatial resolution to calculate landslide susceptibility as a function of the described physical parameters.  Using a GIS overlay, landslide distributions and frequency were correlated with slope gradient, elevation, aspect, geology, and land use.  Logistic multiple regression modeling was used to assess the weight of each physical parameter in contributing to landslide susceptibility.  Based on these coefficients, the authors developed an algorithm for use in a GIS.          

Gorsevski, P.V., Gessler, P.E., Boll, J., Elliot, W.J., Foltz, R.B.  2006.  Spatially and temporally distributed modeling of landslide susceptibility.  Geomorphology 80, 178-198.

 
Gorsevski et al. develop a dynamic landslide susceptibility model capable of predicting landslide activity over space and time.  Spatial and temporal variation is integrated into the model using the Soil Moisture Routing (SMR) model.  The SMR model calculates water balance per raster cell as a function of slope, soil depth, root strength, and vegetation surcharge.  The model dynamically integrates weather data to model temporal change in hydrologic conditions.  The model incorporates documented landslides, Landsat TM vegetation cover, soils data and weather information and dynamically simulates soil moisture.  The authors compare their model outputs against established landslide susceptibility models including: FSmet and SHALSTAB.  This article is useful for providing a thorough framework for developing a landslide susceptibility model within a GIS for a forested watershed.

Gritzner, M.L., Marcus, W.A., Aspinall, R., Custer, S.G.  2001.  Assessing landslide potential using GIS, soil wetness modeling and topographic attributes, Payette River, Idaho.  Geomorphology, 37, 149-165.

 
Gritzner et al. evaluate the effectiveness of the DYNWET model for use in GIS modeling of landslide susceptibility.  The DYNWET model is used to predict regions of surface saturation.  The authors assess the relationship between the DYNWET indicator and documented landslides and evaluate the influence of topographic variables on landslide events.  Using GIS, documented landslide locations were digitized as point locations and rasterized to compare with the DYNWET output and topographic variables.  Basic topographic variables including slope, aspect, elevation, and slope curvature were calculated.  Soil saturation was calculated using the DYNWET model.  A Bayesian model was integrated with a GIS and input variables were tested to assess landslide susceptibility.  Chi-square analysis was performed to identify the most influential variables.  The authors determine that elevation is most strongly linked with landslides.  In addition, it was determined that the soil saturation index does not improve landslide predictions.

Guzzetti, F., Reichenbach, P., Cardinali, M., Galli, M., Ardizzone, F.  2005.  Probabilistic landslide hazard assessment at the basin scale.  Geomorphology, 72, 272-299.


Guzzetti et al. present a model to represent the spatial and temporal dynamics of landslides.  The model predicts where landslides will occur, the physical extent of landslides and the frequency of landslides.  A multi-temporal assessment of landslide activity was created using aerial photographs.  Using GIS, the authors partition landslide susceptibility regions based on environmental conditions including morphology, lithology, and land use.  The model assesses frequency of occurrence and calculates the probability of a particular landslide size. 

Kheir, R.B., Cerdan, O., Abdallah, C.  2006.  Regional soil erosion risk mapping in Lebanon.  Geomophology, 82, 347-359.


Kheir et al. assess soil erosion risk through use of basic GIS overlay using multiple thematic variables characterizing slope, land cover, drainage density, geology, annual rainfall.  The authors develop three base maps to classify landscape sensitivity, runoff potential and soil erodibility.  Landscape sensitivity is assessed with an overlay of slope classes derived from a 50m DEM, drainage density and landcover density.  Runoff potential assesses hydrologic conditions as a function of annual precipitation and soil water retention and infiltration.  Erodibility is classified based on the quantity of material available.  The authors derive potential soil erosion by overlaying the landscape sensitivity map and the runoff potential map.  To determine erosion risk, regions of potential erosion are linked to regions of erodibility.  The model was derived for use in
Lebanon and may not be transferable to environments with dissimilar environmental characteristics; however the methodology provides a relatively simple approach for assessing soil erosion risk.      

Lee, S.  2004.  Application of likelihood ratio and logistic regression models to landslide susceptibility mapping using GIS.  Environmental Management, 34, 223-232.

 
Lee uses the likelihood ratio and logistic regression models coupled with a GIS to identifying regions susceptible to mass failure.  Using satellite imagery, landslides were identified and mapped.  Landslides were converted from vector to raster for incorporation into the GIS model.  Vector based thematic maps consisting of topography, soil, forest cover, and geology were incorporated for analysis.  Land use data was derived from classified Landsat TM imagery.  Lastly, a DEM was used to calculate slope, aspect and slope curvature.  An overlay between documented landslides and the thematic layers was conducted to analyze the spatial relationship between the features.  Using the likelihood ratio and logistic regression models, that spatial relationship between documented landslide and the thematic layers was quantified.  Based on these relationships, a formula was generated to define the probability for landslide occurrence.

Sun, G., McNulty, S.G.  1998.  Modeling soil erosion and transport on forest landscape.  In: Winning solutions for risky problems: Proceedings of conference 29; 1998 February 16-20; Reno, NV. Steamboat Springs, CO: International Erosion Control Association: 189-198.


The authors present a model to predict sediment erosion and transportation in response to various forest management activities based on a coupling of a GIS and the Universal Soil Loss Equation (USLE).  USLE is used to predict soil erosion per grid cell within a 30m DEM using a rainfall runoff factor, soil erosivity, slope length and steepness, forest cover and soil conservation practices.  Arc/Info GRID utilities and empirical models developed from field observation are used to model sediment routing from hillslope to channel.  The use of a GIS allows the USLE model to be applied per pixel area to allow managers to assess areas with the greatest potential for soil erosion. 

Wilson, C.J., Carey, J.W., Beeson, P.C., Gard, M.O., Lane, L.J.  2001.  A GIS-based hillslope erosion and sediment delivery model and its application in the Cerro Grande burn area.  Hydrological Processes 15, 2995-3010.


Wilson et al. present a GIS-based model to predict hillslope erosion and sediment yield from a burned catchment following simulated rain events.  The model generates flow pathways to model runoff depth and volume and incorporates the HEM model to simulate hillslope erosion and sediment yield as a function of gradient, soil type and erodibility, and vegetation cover.  The model is useful for simulating sediment routing within channel networks and models hillslope erosion as a function of hillslope profile and erosion resistant features.  Sediment routing within channel networks and diverse topographic terrain are represented by the model. 








Mail to: adamsjer@geo.oregonstate.edu