Multisource Data Fusion for Landslide Classification Using Generalized Positive Boolean Functions

Chang, Yang-Lang; Liang, Long-Shin; Han, Chin‐Chuan; Fang, Jyh-Perng; Liang, Wen-Yew; Chen, Kun-Shan

doi:10.1109/tgrs.2007.895832

Cited by 51 publications

(34 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The major limitation of the post-classification approach is with the difficulty in excluding other types of spectrally similar land cover, such as human settlements, roads, riverbeds, or fallow lands. In addition, the post-classification approach requires the use of substantial ground truth data and other data, such as digital elevation model and stream networks, in order to achieve a satisfactory outcome (Barlow et al, 2006;Chang et al, 2007;Tarantino et al, 2007).…”

Section: Introductionmentioning

confidence: 99%

Using multi-temporal remote sensor imagery to detect earthquake-triggered landslides

Yang

Chen

2010

International Journal of Applied Earth Observation and Geoinfor

View full text Add to dashboard Cite

a b s t r a c tLandslides are a major type of geohazards claiming thousands of casualties and billions of dollars in property damages every year. Catastrophic landslide activities are often triggered by some extreme events such as earthquakes, excessive precipitations, or volcanic eruptions. Quickly identifying the spatial distribution of landslides induced by these extreme events is crucial for coordinating rescue efforts and planning in situ investigations. In this study, we propose an automated method for detecting the spatial distribution of earthquake-triggered landslides by examining after-event vegetation changes. Central to this method is the use of pre-and post-event remote sensor images covering the same area. Geometric correction and radiometric normalization are performed before deriving a vegetation index from each image. Then, an image differencing procedure is applied to the two derived indices. With the resultant difference image, an initial landslide distribution map is generated by highlighting the pixels with a threshold percentage decrease in the brightness values as a direct result of the image subtraction. The threshold percentage value is interactively determined by using a visual interpretation method. The final landslide distribution map is produced after using a modal filter to suppress boundary errors in the initial map. This method has been implemented in a test site, approximately 30 km from the epicenter of the Sichuan earthquake (7.9 Ms) that struck on 12 May 2008. A pre-event Thematic Mapper image and a post-event Advanced Spaceborne Thermal Emission and Reflection Radiometer scene are used. The thematic accuracy assessment indicates that 90% of the landslides have correctly been mapped. Given the relatively simple procedures and the good mapping accuracy, the image processing and change detection method identified in this study seems to be promising from an operational perspective.

show abstract

Section: Introductionmentioning

confidence: 99%

Using multi-temporal remote sensor imagery to detect earthquake-triggered landslides

Yang

Chen

2010

International Journal of Applied Earth Observation and Geoinfor

View full text Add to dashboard Cite

show abstract

“…Ideally, pre-event and post-event images should be acquired at the same time of the year and with similar view angle and solar illumination, but this is often not feasible (Guzzetti et al, 2012). Semiautomated approaches to landslide mapping can be classed, according to the type of image element used, as "pixel based" (e.g., Chang et al, 2007;Yang and Chen, 2010;Chini et al, 2011;Cheng et al, 2013;Mondini et al, 2013Mondini et al, , 2011a or "object based" (e.g., Aksoy et al, 2012;Holbling et al, 2012Holbling et al, , 2015Lacroix et al, 2013;Lahousse et al, 2011;Lu et al, 2011;Martha et al, 2010Martha et al, , 2011Martha et al, , 2013Stumpf et al, 2011Stumpf et al, , 2014Van Den Eeckhaut et al, 2012). When applied to very high spatial resolution images, pixel-based methods often exhibit a "salt and pepper" appearance (Van Westen et al, 2008;Guzzetti et al, 2012) which requires image post-processing.…”

Section: Automated Methods For Landslide Mappingmentioning

confidence: 99%

Semiautomated object-based classification of rain-induced landslides with VHR multispectral images on Madeira Island

Heleno

Matias

Pina

et al. 2016

Nat. Hazards Earth Syst. Sci.

View full text Add to dashboard Cite

Abstract.A method for semiautomated landslide detection and mapping, with the ability to separate source and run-out areas, is presented in this paper. It combines object-based image analysis and a support vector machine classifier and is tested using a GeoEye-1 multispectral image, sensed 3 days after a major damaging landslide event that occurred on Madeira Island (20 February 2010), and a pre-event lidar digital terrain model. The testing is developed in a 15 km 2 wide study area, where 95 % of the number of landslides scars are detected by this supervised approach. The classifier presents a good performance in the delineation of the overall landslide area, with commission errors below 26 % and omission errors below 24 %. In addition, fair results are achieved in the separation of the source from the run-out landslide areas, although in less illuminated slopes this discrimination is less effective than in sunnier, east-facing slopes.

show abstract

“…In landslide susceptibility assessment, NDVI can be utilized as one of the important indicator of possible vegetation stress such as landslide events. Chang et al (2007) point out that the loss of vegetative cover is one of the major causes of landslides. In the literature, it is reported that landslide-prone areas were usually located in grassland, afforested area and bare soils (Ercanoglu 2005;Yilmaz 2010).…”

Section: Landslide-related Parametersmentioning

confidence: 99%

An assessment of multivariate and bivariate approaches in landslide susceptibility mapping: a case study of Duzkoy district

2014

View full text Add to dashboard Cite

Landslide susceptibility maps are valuable sources for disaster mitigation works and future investments of local authorities in unstable hazard-prone areas. However, there are limitations and uncertainties inherent in landslide susceptibility assessment. For this purpose, many methods have been suggested and applied in the literature, which are generally categorized as bivariate and multivariate. Here, in this paper, the most popular and widely used multivariate [support vector regression (SVR), logistic regression (LR) and decision tree (DT)] and bivariate methods [frequency ratio (FR), weight of evidence (WOE) and statistical index (SI)] were compared with respect to their performances in landslide susceptibility modeling problem. Duzkoy district of Trabzon Province was selected due to its unique topographical and lithological characteristics, magnifying shallow landslide risk potential. Slope, lithology, land cover, aspect, normalized difference vegetation index, soil thickness, drainage density, topographical wetness index and elevation were employed as landslide occurrence factors. Accuracy measures based on confusion matrix (i.e., overall accuracy and Kappa coefficient) and receiver operating characteristic (ROC) curve were employed to compare the performances of the methods. Furthermore, McNemar's test was employed to analyze the statistical significance of differences in method performances. The results indicated that multivariate approaches (i.e., SVR, LR and DT) outperformed the bivariate methods (i.e., FR, SI and WOE) by about 13 %. Within the multivariate approaches, SVR method performed the best with the highest accuracy, while FR method was the most effective and accurate bivariate method. Interpretation of AUC values and the McNemar's statistical test results revealed that the SVR method was superior in modeling landslide susceptibility compared with the other multivariate and bivariate methods.

show abstract

Multisource Data Fusion for Landslide Classification Using Generalized Positive Boolean Functions

Cited by 51 publications

References 39 publications

Using multi-temporal remote sensor imagery to detect earthquake-triggered landslides

Using multi-temporal remote sensor imagery to detect earthquake-triggered landslides

Semiautomated object-based classification of rain-induced landslides with VHR multispectral images on Madeira Island

An assessment of multivariate and bivariate approaches in landslide susceptibility mapping: a case study of Duzkoy district

Contact Info

Product

Resources

About