Remote Sensing: Earthquake Induced Potential Landslide Site Detection Through NDVI

Landslide Study Area
[Term Paper on the remote sensing class I took this semester. I did receive a lot of feedback regarding the contents but haven't made the changes yet. Those will be made when I do submit it to the Journal. The paper is aimed at people who have basic knowledge on terminologies associated with remote sensing and are aware of satellites that are being used by the United States Geographical Survey, USGS]

Full Paper [HERE]

Abstract and Introduction portion provided down below:


A Simple Method for Rapid Assessment of Landslide Prone Areas in an Event of an Earthquake

Abstract: This study presents a simple method for rapid assessment of landslide prone areas in an event of an earthquake. The data acquired is based on the April 2015 earthquake that occurred in Nepal and is analyzed through vegetation phenology using Normalized Difference Vegetation Index (NDVI). Although landslide-vegetation index relationship is clear, rapid assessment procedure using NDVI to locate potential earthquake induced landslide prone areas is not available. The implications could include assisting rescue agencies to relocate limited resources to affected areas or to re-vegetate potential landslide site.
                Keywords:  Landslide, Earthquake, NDVI

I. Introduction
            The April 2015 earthquake of magnitude 7.8M and depth of 15km was Nepal’s worst earthquake since 1934. The earthquake took 9,000 lives while injuring more than 23,000 individuals. The casualties were exacerbated because rescue efforts were highly centralized inside Kathmandu Valley. The regions outside the valley received aid after roads leading outside the capital were cleared of debris. The need for quick response to other regions was clear, however, rescue teams needed access to reliable information on areas that required the most attention.

            Nepal’s geographical structure poses serious challenges during times of crisis. Earthquakes are known to trigger catastrophic landslides (Harp and Jibson, 1996; Kieffer et al., 2006; Lin, 2008). In addition, reconnaissance study has estimated that at least 75 percent of all landslides in Nepal are natural (Laban, 1979) and landslides in Nepal should be considered normal rather than exception (Brundsen et al., 1981). Naturally, threats from landslide during an earthquake need to be analyzed as rapidly as possible.

            One way to estimate casualty areas after an earthquake is to overlap landslide prone areas to population dense areas. The information provides rescue teams to make rational decisions to allocate limited resources in identified critical areas. However, this analysis can be completed before a disaster by identifying potential landslide areas beforehand. Mitigation strategies to reduce landslide damages by localized vegetation and population relocation can be done to ensure casualties after an earthquake is minimized.

            To detect possible landslide areas, a large area has to be studied. Field surveys and remote sensing can be used to map spatial distribution of landslides. Nepal’s topography poses major logistical challenges for field surveys, however, remote sensing through the use of space borne cameras provide a good alternative to field surveys for understanding topography. Satellite imagery has been used for landslide mapping since late 1970s when Landsat became available (Sauchyn and Trench, 1978; Cochrane and Browne, 1981). Data from satellites have application in documenting landscape changes through surface erosion (Metternicht and Zinck, 1998; Pickup and Marks, 2000; Meyer et. al., 2001) and landslide mapping (Lillesand and Kiefer, 1999; Singhroy and Mattar 2000). Factors that cause landslides can be identified as well (McKean et.al., 1991).

            Satellite images from multispectral sensors provide the best option to locate landslide. Vegetation indices based on red (R) and near-infrared (NIR) band of the electromagnetic spectrum are commonly used to study vegetation phenology (Jensen, 2000). Healthy vegetation is highly sensitive to NIR region while burnt, dying, or diseased vegetation has decreased reflectance on the same region (Vohora and Donoghue, 2004). Normalized Difference Vegetation Index (NDVI) is a common index that can be used because NDVI data has important relationship with landslide (Zhang et.al, 2010) and has shown high correlation with occurrence of landslide (Kim et. al., 2014).

            Although landslide- vegetation index relationship is clear, rapid assessment procedure using NDVI to locate potential earthquake induced landslide prone areas is not available. This paper studies a landslide caused by the April 2015 earthquake in Nepal and characterizes why that particular area was affected by the disaster. That characteristic can then be applied on a lower spatial resolution satellite imagery to locate and identify future potential landslide hazard sites. The overall process is simple, clear and quick and can be utilized not only before an earthquake but after the disaster in a time-constraint period.

Rest of portion:
Full Paper [HERE]

Comments