Keywords

gully erosion, nonparametric modeling, cart, mars, roc curve

Start Date

1-7-2008 12:00 AM

Abstract

Gully erosion represents an important soil degradation process in rangelands. Inorder to take preventive or control measures and to reduce its environmental damages andeconomical costs it is useful to localize the points in the landscape where gullying takesplace and to determine the importance of the different factors involved. The study is carriedout in Extremadura, southwest Spain. The main objectives of this work are: 1) to model thespatial distribution of gullies, 2) comparison of two nonparametric schemes to constructpredictive models 3) to analyze the role of prevalence and of scale for susceptibility modelsin geomorphology, 4) evaluating the importance of the different factors involved and 5)implementing and mapping the results with the help of a Geographical Information System(GIS). Two methods were used to model the response of a dependent variable (gullying)from a set of independent variables: Classification And Regression Trees (CART) andMultivariate Adaptive Regression Splines (MARS). Three different datasets were used; thefirst one for constructing the model (training dataset) and the others for validating the model(external datasets). These datasets are formed by a target variable (presence or absence ofgullies) and a set of independent variables. The dependent variable was obtained mappingthe locations of gullies with the help of a GPS and high resolution aerial ortophotographs.We used 32 independent variables reflecting topography, lithology, soil type, climate andland use and vegetation cover of each area. To evaluate the performance of the models weused a non-dependent threshold method: the Receiver Operating Characteristic (ROC)curve. The results show a better performance of MARS for predicting gullying with areasunder the ROC curve of 0.98 and 0.97 for the validation datasets, while CART presentsvalues of 0.96 and 0.66.

COinS
 
Jul 1st, 12:00 AM

Use of Two Nonparametric Methods (CART and MARS) to Model the Potential Distribution of Gullies in Spanish Rangelands

Gully erosion represents an important soil degradation process in rangelands. Inorder to take preventive or control measures and to reduce its environmental damages andeconomical costs it is useful to localize the points in the landscape where gullying takesplace and to determine the importance of the different factors involved. The study is carriedout in Extremadura, southwest Spain. The main objectives of this work are: 1) to model thespatial distribution of gullies, 2) comparison of two nonparametric schemes to constructpredictive models 3) to analyze the role of prevalence and of scale for susceptibility modelsin geomorphology, 4) evaluating the importance of the different factors involved and 5)implementing and mapping the results with the help of a Geographical Information System(GIS). Two methods were used to model the response of a dependent variable (gullying)from a set of independent variables: Classification And Regression Trees (CART) andMultivariate Adaptive Regression Splines (MARS). Three different datasets were used; thefirst one for constructing the model (training dataset) and the others for validating the model(external datasets). These datasets are formed by a target variable (presence or absence ofgullies) and a set of independent variables. The dependent variable was obtained mappingthe locations of gullies with the help of a GPS and high resolution aerial ortophotographs.We used 32 independent variables reflecting topography, lithology, soil type, climate andland use and vegetation cover of each area. To evaluate the performance of the models weused a non-dependent threshold method: the Receiver Operating Characteristic (ROC)curve. The results show a better performance of MARS for predicting gullying with areasunder the ROC curve of 0.98 and 0.97 for the validation datasets, while CART presentsvalues of 0.96 and 0.66.