Development of a GIS-based failure investigation system for highway soil slopes

Abstract

A framework for preparation of an early warning system was developed for Maryland, using a GIS database and a collective overlay of maps that highlight highway slopes susceptible to soil slides or slope failures in advance through spatial and statistical analysis. Data for existing soil slope failures was collected from geotechnical reports and field visits. A total of 48 slope failures were recorded and analyzed. Six factors, including event precipitation, geological formation, land cover, slope history, slope angle, and elevation were considered to affect highway soil slope stability. The observed trends indicate that precipitation and poor surface or subsurface drainage conditions are principal factors causing slope failures. 96% of the failed slopes have an open drainage section. A majority of the failed slopes lie in regions with relatively high event precipitation (P>200 mm). 90% of the existing failures are surficial erosion type failures, and only 1 out of the 42 slope failures is deep rotational type failure. More than half of the analyzed slope failures have occurred in regions having low density land cover. 46% of failures are on slopes with slope angles between 20° and 30°. Influx of more data relating to failed slopes should give rise to more trends, and thus the developed slope management system will aid the state highway engineers in prudential budget allocation and prioritizing different remediation projects based on the literature reviewed on the principles, concepts, techniques, and methodology for slope instability evaluation (Leshchinsky et al., 2015).

Graphical abstract

Keywords

soil slope / slope management system / geographic information system / hazard mapping

1 Introduction

All slope movements are a manifestation of slope instability. It is a well documented fact that slope failures can result in extensive property damage and loss of life (USGS, 2004; BBC, 2014; KING5, 2014). In 2004, the National Research Council estimated that landslides in the United States cause more than $2 billion in property damage and claim 20‒25 deaths annually (NRC, 2004). Given the increasing economic and societal cost of landslides, there has been an urgent need for improved protection against landslides (He and Beighley, 2008).

Investigation of slope instability and landslide hazard has sparked significant interest and is the primary focus of various research initiatives around the world. Numerous publications have directed efforts toward discussion of the different scales of landslide investigation and slope instability analysis (Selby, 1993; Cruden and Varnes, 1996; Dikau et al., 1996; Popescu, 2002; Glade et al., 2005). A considerable amount of publications, reports, and books discuss in detail the different aspects involved in developing a predictive model (Leroi, 1996; van Westen et al., 1997; Aleotti and Chowdhury, 1999; Chung and Fabbri, 1999; Dai and Lee, 2002). With the current trend being toward developing early-warning systems, Geographic Information Systems (GIS) have become an important and powerful tool in landslide hazard assessment.

Based on the literature reviewed on the principles, concepts, techniques, and methodology for slope instability evaluation (Carrara et al., 1999; Chung and Fabbri, 1999; Guzzetti et al., 1999; Cavallo and Norese, 2001; Cardinali et al., 2002; Clerici et al., 2002; Glade et al., 2005; Wang et al., 2005), all slope instability mapping techniques can be broadly classified into qualitative and quantitative analysis. Qualitative analysis involves techniques such as geomorphological mapping, landslide inventory mapping, hueristic analysis, and qualitative index overlay.

Geomorphological mapping relies on information about the surface topography and relief features of the site in question. It is the easiest method for mapping instability and was widely used in the past (Kienholz, 1978; Fenti et al., 1979; Rupke et al., 1988). Landslide inventory mapping systems use information available relating to slope failure events that have occurred on the slope in the past to develop an inventory. However, they only concentrate on slopes with failure histories (He and Beighley, 2008). Heuristic, or index-based analysis, uses a combination of expert opinion and past experience to analyze slopes (Anbalagan and Singh, 1996; Gupta and Anbalagan, 1997; Wachal and Hudak, 2000; Morton et al., 2003).

Qualitative index overlay, or factor mapping, is commonly used in the initial stage of regional assesment (Glade et al., 2005). It involves identifying the spatial distribution of one or more causative factors or a combination of causative factors, and investigating their influence on slope stability. Turner and Schuster (1996) and Guzzetti et al. (1999) studied the effect of a variety of parameters on slope instability, and provided a comprehensive list of causative factors influencing slope stability.

Quantitative analysis can further be classified into statistical techniques and physical geotechnical models. Statistical analysis makes use of regression functions and distribution curves to predict slope failure based on data collected on site or in the laboratory. Correlation between different physical factors and previous slope failures are mapped using discriminant analysis. Then, quantitative or semiquantitative estimates are made for those slopes without failure histories (Dai and Lee, 2002). Statistical methods are more appropriate for slope instability mapping as they eliminate the subjective bias present in qualitative analysis (Fall et al., 2006).

Currently, the State of Maryland does not have an assessment or predictive model to identify vulnerable highway soil slopes. Such a model would be able to highlight soil slopes that are more susceptible or vulnerable to movement or failure in comparison to the other slopes along highways. It was the intent of this study to lay the framework for setting up a robust model to facilitate the agencies in Maryland in prioritizing and optimizing their response to slope failures well in advance. The scale of assessment adopted for this study was in the regional scale. The method of assessment used was a semi-qualitative index overlay due to availability of limited data on historic slope failures.

2 Description of study area

The State of Maryland is located in the Mid-Atlantic region of the United States. It is the 9th smallest state by area, but the 19th most populous and the 5th most densely populated of the 50 states in the US (US Census Bureau, 2011). The total study region covers an area of 27,076 km². The mean elevation of the State of Maryland is 106 m above sea level, ranging from 1,000 m to mean sea level at the Atlantic ocean. Figure 1 shows the study area along with the recorded highway slope failures. The state is divided into 5 distinct physiographic provinces: Appalachian Plateaus Province, Ridge and Valley Province, Blue Ridge Province, Piedmont Plateau Province, and the Atlantic Coastal Plains Province.

The Coastal Plain Province is underlain by a wedge of unconsolidated sediments including gravel, sand, silt, and clay, which overlaps the rocks of the eastern Piedmont along an irregular line of contact known as the Fall Zone (McGee, 1888; Freitag et al., 2009). Eastward, this wedge of sediments thickens to more than 2,400 m at the Atlantic coastline. Beyond this line is the Atlantic Continental Shelf Province, the submerged continuation of the Coastal Plain, which extends eastward for at least another 120 km where the sediments attain a maximum thickness of about 12 km (Edwards, 1981).

The Piedmont Plateau Province is composed of hard, crystalline igneous and metamorphic rocks, and extends from the inner edge of the Coastal Plain westward to Catoctin Mountain, the eastern boundary of the Blue Ridge Province. Bedrock in the eastern part of the Piedmont consists of schist, gneiss, gabbro, and other highly metamorphosed sedimentary and igneous rocks of probable volcanic origin. Unlike the Coastal Plain and Piedmont Plateau Provinces, the Blue Ridge, Ridge and Valley, and Appalachian Plateaus Provinces are underlain mainly by folded and faulted sedimentary rocks. Figure 2 shows the generalized geological map for the state of Maryland (Edwards, 1981).

Despite its small size, Maryland exhibits considerable climatic diversity. Temperatures vary from an annual average of 9°C in the extreme western uplands to 15°C in the southeast, where the climate is moderated by the Chesapeake Bay and the Atlantic Ocean. Monthly average temperatures range from a high of 30.6°C to a low of −4.3°C. The average annual precipitation values for the eastern half of the state of Maryland range from 1,067 mm to 1,320 mm per year. Precipitation averages about 1,245 mm annually in the southeast, but only 914 mm in the west. Higher values of average annual precipitation are observed in the westernmost tip of the study region.

3 Physical parameters and data sources

The study area was examined in detail using the ArcMap GIS software. The input map layers were imported into ArcMap in their original format for verifying data compatibility and integrity. A wide array of physical parameters were considered as causative factors in this study based on literature (Turner and Schuster, 1996; Guzzetti et al., 1999). Due to non-uniformity in the quality of data and the level of resolution, only a handful of parameters were shortlisted.

Several physical parameters or factors that influence slope stability, directly or indirectly, have been used in different methods of analysis while mapping slope instability (Cavallo and Norese, 2001; Sakellariou and Ferentinou, 2001; Bhattarai et al., 2004; Chau et al., 2004; Saboya et al., 2006; He and Beighley, 2008; Singh and Huat, 2008), and the following ones were considered in the current study: (i) elevation, (ii) slope angle, (iii) land cover, (iv) storm event precipitation, (v) slope history or failure inventory, and (vi) surface geology, consistent with the approach undertaken in past studies (Turner and Schuster, 1996; Guzzetti et al., 1999; Bhattarai et al., 2004; Chau et al., 2004; He and Beighley, 2008; Singh and Huat, 2008).

The elevation data set is obtained from the National Elevation Data set (NED) 1/3 Arc-Second coverage in raster format. The data set has a resolution of 10 m × 10 m and was downloaded from the United Staes Geological Survey (USGS) website. The slope angle data set was derived from this layer using spatial analysis tools available in the ArcMap software. The derived slope angle data layer was also resampled to a resolution of 10 m× 10 m.

The land cover data layer was obtained from the National Land Cover Database (NLCD) 2006 edition at 30 m resolution. The NLCD data set was reclassified into six different values: grass, shrubs, woodland, cultivated land, developed land, and other. The precipitation data was obtained from the National Oceanic and Atmospheric Administration (NOAA) Atlas 14, Volume 2 (Ohio River Basin and Surrounding States) data set. The precipitation listed in this document provides frequency estimates, with upper and lower bounds of the 90% confidence interval, in grid format, and are resampled at 30 m resolution at the time of data extent specification. Data are available for the 1, 2, 5, 10, 25, 50, 100, 200, and 500 year storm events and for 6, 12, 24, and 48 h durations. For this study, the estimates for a 2 year 24 h duration storm event and a 100 year 24 h duration storm event were chosen to understand the short term and long term, respectively, effects of precipitation on soil slope failures (He and Beighley, 2008; Singh and Huat, 2008).

The slope history or failure inventory data was derived from a GIS database developed to record slope failures (Ramanathan, 2012). The surface geology data set consists of 2 layers. The first layer depicts the extents of the different physiographic provinces in the state of Maryland. This shapefile was obtained from the Maryland Geological Survey. The second layer is the geological map of the state of Maryland which is obtained from the USGS mineral resources spatial database. This layer provides details regarding the superficial and bedrock geology of the state of Maryland. Both data sets are in vector format in 1:250,000 scale.

4 Data analyysis

For collecting baseline data related to highway slope failures, the data collection process was optimized and streamlined by setting up two components: the failure field sheet and the GIS database. The failure field sheet was used to collect relevant information relating to slope failures at the failure site. This data was then stored in a GIS database to track, evaluate, and monitor the highway slopes (Ramanathan, 2012).

Failure sites were digitized and stored as points because the area and perimeter of slope failures in comparison to the size of the study area is very small. A total of 48 slope failure cases occurring between 2008 and 2012 were recorded. Based on the comprehensive information for the 48 slope failures and using spatial analysis tools available in ArcMap software (ver. 10), trends in failure distribution in relation to the selected parameters were established, and are discussed below (Ramanathan, 2012).

4.1 Elevation and slope angle

Elevation and slope angle are the two most widely chosen parameters considered to influence slope stability while mapping regions vulnerable to failure (Sakellariou and Ferentinou, 2001; Chau, et al., 2004; Saboya et al., 2006; He and Beighley 2008; Singh and Huat, 2008). In this study, elevation as a separate parameter does not exhibit a strong correlation with slope instability. As shown in Fig. 3(a), 56% of the total number of slope failures has occurred on slopes between 30 m and 90 m in height and nearly a fourth of the total number of failures have occurred on slopes with heights between 10 m and 30 m. Skempton (1953) and Brundsen (1973) developed and modified, respectively, the relationship between slope angle and slope height in terms of potential failure mechanisms. However, no clear trend or correlation could be observed between slope height and soil slope failures in Maryland, most probably due to the limited amount of data collected so far.

Figure 3(b) shows the failure distribution for the slope angle subcategories. Nearly 46% of the failures have occurred on slopes with slope angles between 20° and 30°. For all engineering and analyses purposes, all man-made highway slopes had a 2H: 1V slope; however, few of the natural slopes deviated from this ratio, as seen in Fig. 3(b).

4.2 Storm event precipitation

It is evident from past events across the globe that failure is more likely to occur in areas with high estimation of precipitation values. Figure 4(a) shows the failure distribution pattern for the 2 year 24 h storm event. 87% of the total number of slope failures have occurred in regions estimated to have 80‒85 mm of precipitation. A similar trend can be observed for the 100 year 24 h storm events:>80% of the total slope failures have occurred in regions expected to have a heavy amount of rainfall during a storm event (Fig. 4(b)). He and Beighley (2008) also showed that, in general, areas receiving higher rainfall relative to the region have a higher probability of landslide occurrence. Similarly, slope movements triggered by heavy rains or storm events were observed in Japan, Malaysia, and Nepal (Bhattarai et al., 2004; Singh and Huat, 2008).

4.3 Land cover

In the current study, 56% of the total number of failures have occurred on slopes predominantly covered with grass (Fig. 5). Cross referencing this information with the vegetation density information recorded on site, it is evident that many failures have occurred on slopes with medium to low density grass. This trend highlights the importance of type of vegetation cover on highway slopes as an important factor of influence in slope vulnerability studies in Maryland. Lee and Choi (2004) also showed the probability of landslide occurrence in southern California to be the highest for grass lands and certain forest types. It must be noted that their findings may be a result of co-existing landscape characteristics. For example, they show a high probability of landslide occurrence for vegetation types found in steep and mountainous areas.

Figure 5 depicts that 23% of the slope failures have occurred on developed land or urbanized regions. This trend presents an interesting insight into the effect of urbanization and land use pattern on slope instability. This relatively large percentage occurrence of failures on developed land can be attributed to the increased amount of human activity such as blasting, drilling, traffic volume, and other construction activities.

4.4 Slope failure inventory

Slope instability classification systems are usually based on a combination of material and movement mechanism (Dai and Lee, 2002). For this study, the classification system proposed by Cruden and Varnes (1996) was slightly modified to reflect the failure conditions prevalent locally in the State of Maryland (Table 1).

Figure 5(b) shows the distribution pattern for the different types of failure as per the classification shown in Table 1. 90% of the slope failures are due to surficial erosion. Cross referencing with the GIS database, 80% of the slope failures have occurred during or after rainfall. Figure 6(a) shows the distribution pattern for the different types of slopes in the State of Maryland. This trend, when compared with the failure distribution pattern for the type of drainage section at failure sites, Figure 6(b) shows the influence of precipitation and drainage conditions on slope instability. A comparison of Figs. 6(a) and 6(b) clearly indicates that open section drainage can promote slope failures due to the presence of large amounts of water in the environment. The primary reason for the majority of slope failures to occur on highways with open section drainage in comparison to closed section drainage is that open sections introduce surface runoff directly onto the slope. The direct introduction of surface runoff onto the slope causes erosion of slope material.

4.5 Physiographic provinces and lithology

It may be reasonably expected that the properties of the slope-forming materials, such as strength and permeability, are involved in the failure, are related to the lithology, which therefore should affect the likelihood of failure (Dai and Lee, 2002). The statistic that 83% of the total number of slope failures recorded has occurred in the Atlantic coastal plains province highlights the effect of rock formation on highway slopes (Fig. 7(a)).

The Atlantic Coastal Plains Province predominantly consists of slopes with silty or clayey sand, gravelly sand, coarse sand, and gravel type soils. Nearly 55% of the total number of slope failures lay on slopes with gravel-sand formations (Fig. 7(b)). As mentioned above, the majority of failures observed in the current study were erosion type. It is widely known that sandy soils and sand-gravel formations can erode with the presence of water or open drainage conditions (US Army, 2003; Chase et al., 2005).

5 Slope instability mapping

Logistic multiple regression is a multivariate technique which considers several physical parameters that may affect probability (Gorsevski et al., 2006). The advantage of logistic multiple regression modeling over other multivariate statistical techniques is that the dependent variable can have only two values—an event occurring or not occurring, and that predicted values can be interpreted as probability since they are constrained to fall in the interval between 0 and 1 (Dai and Lee, 2002).

While this method of analysis is highly recommended for this scale of study and is most compatible with the format in which data are recorded, due to the inadequate sample size of slope failures, the application of this method to the current study was not feasible. The sample size of the number of failure cases needs to be exponentially larger (e.g., over 1,000 slopes according to Chau et al. 2004) for this model to be used. The time required for the GIS database to acquire the appropriate volume of data would render the application of such a statistical model out of the scope of the present study.

Instead, the qualitative index overlay was considered to be the most suitable method of analysis for the volume of data collected for this study (Glade et al., 2005). The general concept behind the qualitative index overlay analysis is to characterize both spatial and temporal conditions that have determined the occurrence of past instability events, and to use these characteristics to highlight the slopes with similar conditions that are vulnerable to failure. Chau et al. (2004) discussed the principle behind a weighted overlay of index or thematic maps using ArcGIS software. In the qualitative index overlay model, the whole study area of the instability map is denoted by A, and there are m layers of thematic spatial data (elevation, slope angle, lithology, and precipitation etc.) containing causal factors, c _i. A pixel p in A would have m pixel values, c₁, c₂,…,c _m. The model can be programmed to calculate the occurrence of failure in p in terms of conditional probabilities based on pixel values of the causal factors (Clerici et al., 2002). Figure 8 shows a schematic of the model principles.

The values of all the physical parameters were classified into subclasses or categories as shown in Table 2. Continous variables were classified into equal bins within a certain range to examine the failure distribution pattern at certain particular range of values for a given parameter. Classification using natural breaks in data was not preferred as it dilutes the range of interest.

A failure density index was assigned for each subclass. The purpose of assigning such an index to each subclass was to identify the unstable slopes along regions with no previous slope failure occurrence. The class intervals were decided using statistical tools available in the ArcMap software.

Eq. (1) was used to calculate the failure density index for each subclass given in Table 2. As evident from the equation, the failure density index provides the probability of failure occurring along highway slopes associated with each particular subclass. A normalized density index was calculated for a more conservative approach. For a particular factor, the density index for each subcategory was normalized by the maximum density index value for that factor. Figure 9 shows the variation of both the failure density index and the normalized failure density index for the different subclasses of parameters. This figure shows how the conservative index provides for more striking variation in failure density values for the same sample set of data when compared to the failure density index.

(1) \mathrm{F}\mathrm{a}\mathrm{i}\mathrm{l}\mathrm{u}\mathrm{r}\mathrm{e}\hspace{0.17em}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}\hspace{0.17em}\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}\left(v\right)={\displaystyle \frac{\mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\hspace{0.17em}\mathrm{o}\mathrm{f}\hspace{0.17em}\mathrm{s}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{e}\hspace{0.17em}\mathrm{f}\mathrm{a}\mathrm{i}\mathrm{l}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{s}\hspace{0.17em}\mathrm{i}\mathrm{n}\hspace{0.17em}\mathrm{s}\mathrm{u}\mathrm{b}\mathrm{c}\mathrm{l}\mathrm{a}\mathrm{s}\mathrm{s}}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\hspace{0.17em}\mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\hspace{0.17em}\mathrm{o}\mathrm{f}\hspace{0.17em}\mathrm{s}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{e}\hspace{0.17em}\mathrm{f}\mathrm{a}\mathrm{i}\mathrm{l}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{s}}}.

Figure 10 illustrates the variation of failure density indices of parameter subclasses over the area of the study region. The low sample size of slope failures for this study gives rise to insignificant failure density values for some parameter subclasses, as shown in Fig. 10. A conservative index allows for a well distributed model by giving a higher rating to slopes that have low failure density values due to lack of field data, but might have potential to fail based on spatial and temporal conditions.

A weighted mean of the normalized failure density index of the various factors gives the failure density value at any particular pixel. The weights were assigned based on expert opinion and the trends observed between the failure density index and the causative parameter. A weightage of 3 was applied to parameters exhibiting a clear trend between parameter data and the failure density index, while the weightage of 2 or 1 was provided to other parameters based on expert opinion. In essence, the failure density index provides the rating for subclasses, and the weights provide rating for the parameters as a whole. Four trials were conducted, and thus four failure density maps were generated. Table 3 gives the different weights assigned, w _i, to the different factors, v _i, used to calculate the failure density as defined in Eq. (2).

(2) \mathrm{F}\mathrm{a}\mathrm{i}\mathrm{l}\mathrm{u}\mathrm{r}\mathrm{e}\hspace{0.17em}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}={\displaystyle \frac{{\displaystyle {\sum }_{i=\mathrm{1}}^m{w}_i{v}_i}}{{\displaystyle {\sum }_{i=\mathrm{1}}^m{w}_i}}}.

Figure 11 shows the results of the four weighted overlay maps using the raster calculator function in ArcMap software. The selection of the weights based on expert opinion seems to affect the failure densities; however, in all four cases evaluated in this study, the slope failures were concentrated in the greater District of Columbia area. This area is a suburban area of Washington, DC, and includes a large number of roads with heavy traffic.

6 Practical limitations

The slope management system (SMS) defined in this study is still in a nascent stage. As frequently mentioned above, the full potential of the system can be realized with the inclusion of more slope failure cases. The SMS has recorded only information relating slope failures that have occurred between 2008 and 2012. With the passage of time and continual process of further population of the database, many more improvements can be made to the system to support the influx of new information, and analyze and establish more conclusive trends with regard to highway slope failures. The skeletal structure or framework of the system has been established and further improvements can be implemented to realize the full potential of such a tool.

A majority of the SMS reported in past studies have an inherent factor of subjectivity linked with their study. Although the issue of subjectivity while assessing failed slopes was not completely eliminated, the current study presented and reviewed procedures that can quantitatively analyze and rate slopes, thus reducing subjective or qualitative analysis to a minimum. Due to the small amount of data, a qualitative index overlay analysis was chosen, and a logistic multiple regression model or an improved qualitative index overlay analysis can be developed for a more robust quantitative mapping system when more data are collected. Such a model requires large sample data sets to accurately predict the probability of failure of any given highway slope. While this implementation cannot be adopted immediately, it is definitely imperative, because at the regional scale, only a quantitative mapping system would be ideal and accurate for use.

Slope data are collected on a systematic basis in the State of Maryland, and when sufficient data are available regarding dimensions of initiation sites and volume of debris in the future, it is recommended that the scale of study is leveled down to medium to small scale (analysis is performed at the district level or county level). This increases the accuracy of prediction, and presents a multitude of mapping techniques to be chosen from. Also the ratio between the total area of failure sites and the total study area becomes much more significant at this scale of study, thereby presenting conditions for susceptibility analysis or conditional probability analysis.

When the GIS database is populated with remediation details and maintenance information, the data can be analyzed to ascertain the most cost effective and efficient remediation methodology for a particular type of failure. The results from the data analysis could be used as input for developing an automated remediation response model which provides the most viable remediation option based on the set of parameters previously discussed. This model may also be able to perform benefit-to-cost ratio analysis, thereby providing district offices with significant results which may be used to allocate budgets and resources accordingly.

7 Conclusions

A framework for analyzing slope instability is proposed and developed for the State of Maryland. A total of 48 slope failures recorded using the GIS database were analyzed for emerging trends and patterns correlating physical parameters with slope instability.

In this study, six factors were considered to affect highway soil slope stability: event precipitation, geological formation, land cover, slope history, ground slope, and elevation. Overlaying statewide GIS data for these factors brings some interesting trends to light, and the significant trends are listed as follows.

1) Precipitation and poor surface drainage conditions are the principal factors causing slope failures. More than 80% of the total slope failures have occurred in regions expected to have a heavy amount of rainfall during a storm event. 96% of the failed slopes lie along roads with an open drainage section.

2) 90% of the existing failures are surficial erosion type failures, and only 4% of the slope failures are deep rotational type failures. Cross referencing these observations with the GIS database, it is found that 80% of the total number of slope failures have occurred during or after rainfall.

3) 58% of the existing slope failures have occurred in regions having low density land cover (grass plus cultivated land). This relatively large percentage occurrence of failures on developed land can be attributed to the increased amount of human activity, such as blasting, drilling, traffic volume, and other construction activities.

4) About 23% of the remaining slope failures have occurred on developed land or urbanized regions. This trend presents an interesting insight into the effect of urbanization and land use patterns on slope instability.

5) 56% of the total number of slope failures have occurred on slopes between 30 m and 90 m in height, and nearly a fourth of the total number of failures have occurred on slopes with heights between 10 m and 30 m. No clear trend or correlation could be observed at present between slope height and soil slope failures in Maryland.

6) Nearly 46% of failures have occurred on slopes with slope angles between 20° and 30° as these failures have occurred along highway slopes. For all engineering and analyses purposes, all highway slopes have a 2H: 1V slope unless explicitly mentioned. Thus, the analysis is congruent with field conditions.

7) The failure distribution pattern for elevation and slope angle correlates with the field conditions observed by engineers. Since a distinct pattern or correlation with slope instability is yet to be drawn with respect to these parameters, it can be concluded that these numbers, when combined with failure distribution patterns for other parameters, will yield a more conclusive result.

8) The statistic that 83% of the total number of slope failures recorded have occurred in the Atlantic Coastal Plains Province highlights the effect of lithology or soil type of the highway slopes. The Atlantic Coastal Plains Province predominantly consists of slopes with silty or clayey sand, gravelly sand, coarse sand, and gravel type soils.

9) The physical parameters listed above presently influence highway slope stability to a greater extent in relation to physical parameters such as elevation and slope angle. The influx of more data relating to failed slopes should give rise to more trends, and thus this system will aid the highway engineers in prudential budget allocation and prioritizing different remediation projects.

References

[1] Aleotti P, Chowdhury R (1999). Landslide hazard assessment: summary review and new perspectives. Bull Eng Geol Environ, 58(1): 21–44

[2] Anbalagan R, Singh B (1996). Landslide hazard and risk assessment mapping of mountainous terrains—A case study from Kumaun Hymalaya, India. Eng Geol, 43(4): 237–246

[3] Army U S (2003). Living on the Coast: Protecting Investments in Shore Property on the Great Lakes. Great Lakes Hydraulics and Hydrology Office, Detroit District, U.S. Army Corps of Engineers, Detrioit, MI.

[4] BBC (2014). Japan landslides kill 32 in Hiroshima prefecture.

[5] Bhattarai P, Tiwari B, Marui H, Aoyama K (2004). Quantitative Slope Stability Mapping With ArcGIS: Prioritize Highway Maintenance. Proceedings of ESRI’s 24th Annual International User’s Conference, San Diego. ESRI.

[6] Brundsen D (1973). Application of systems theory to the study of mass movement. Geologica e Idrogeologia, Univ. of Baria.

[7] Cardinali M, Reichenbach P, Guzzetti F, Ardizzone F, Antonini G, Galli M, Cacciano M, Castellani M, Salvati P (2002). A geomorphological approach to the estimation of landlside hazards and risks in Umbria, Central Italy. Nat Hazards Earth Syst Sci, 2(1/2): 57–72

[8] Carrara A, Guzzetti F, Cardinali M, Reichenbach P (1999). Use of GIS technology in the prediction and monitoring of landslide hazard. Nat Hazards, 20(2/3): 117–135

[9] Cavallo A, Norese M F (2001). GIS and Multicriteria Analysis to Evaluate Map Erosion and Landslide Hazards. Informatica., 12: 25–44 (Lithunian Academy of Sciences.)

[10] Chase R, Edil T, Kehew A, Keillor P, Mickelson D (2005). Stabilizing Coastal Slopes on the Great Lakes, University of Wisonsin Se Grant Institute, Madison, WI.

[11] Chau K T, Sze Y L, Fung M K, Wong W Y, Fong E L, Chan L C P (2004). Landslide hazardanalysis for Hong Kong using landslide inventory and GIS. Computers and Geosciences, 30(4): 429–443

[12] Chung C J F, Fabbri A G (1999). Probabilistic Prediction Models for Landslide Hazard Mapping. Photogramm Eng Remote Sensing, 65(12): 1389–1399

[13] Clerici A, Perego S, Tellini C, Vescovi P (2002). A procedure for landslide susceptibility zonation by the conditional analysis method. Geomorphology, 48(4): 349–364

[14] Cruden D M, Varnes D J (1996). Landslides: types and processes. In: Turner A K, Schuster L R, eds. Landslides: Investigation and Mitigation. Transportation Research Board Special Report, No. 247, 36–75

[15] Dai F C, Lee C F (2002). Landslide characterisitcs and slope instability modelling using GIS, Lantau Island, Hong Kong. Geomorphology, 42(3–4): 213–228

[16] Dikau R, Brundsen D, Schrott L, Ibsen M L (1996). Landslide Recognition: Identification, Movement and Causes. Chichester: John Wiley & Sons Ltd., 274 p

[17] Edwards J Jr (1981). A brief description of the geology of Maryland. Pamphlet Series. Maryland Geological Survey

[18] Fall M, Azzam R, Noubactep C (2006). A multi-method approach to study the stability of natural slopes and landslide susceptibility mapping. Eng Geol, 82(4): 241–263

[19] Fenti V, Silvano S, Spagna V (1979). Methodological proposal for an engineering geomorphological map, forecasting rockfalls in the Alps. Bulletin of the International Association of Engineering Geology, 19(1): 134–138

[20] Freitag B, Bolton S, Westerlund F, Clark J (2009). Floodplain Management: A New Approach for a New Era. Island Press, p 77

[21] Glade T, Anderson M, Crozier M J (2005). Landslide Hazard and Risk. West Sussex: John Wiley & Sons, Inc., 824 p

[22] Gorsevski P V, Gessler P E, Foltz R B, Elliot W J (2006). Spatial Prediction of Landslide Hazard Using Logistic Regression and ROC Analysis. Transactions in GIS, 10(3): 395–415

[23] Gupta P, Anbalagan R (1997). Slope stability of Theri Dam Reservoir Area, India, using landslide hazard zonation (LHZ) mapping. Q J Eng Geol, 30(1): 27–36

[24] Guzzetti F, Carrara A, Cardinali M, Reichenbach P (1999). Landslide hazard evaluation: a review of current techniques and their application in a multi-scale study, Central Italy. Geomorphology, 31(1–4): 181–216

[25] He Y, Beighley E R (2008). GIS-based regional landslide susceptibility mapping: a case study in southern California. Earth Surf Process Landf, 33(3): 380–393

[26] Kienholz H (1978). Maps of geomorphology and natural hazard of Griendelwald, Switzerland, scale 1:10 000. Artic and Alpine Research 10: 169–184

[27] KING5 (2014). Oso landslide body officially identified, KING5.com

[28] Lee S, Choi J (2004). Landslide susceptibility mapping using GIS and the weight-of-evidence model. Int J Geogr Inf Sci, 18(8): 789–814

[29] Leroi E (1996). Landslide hazard-risk maps at different scales: objectives, tools and developements. In: Senneset K, ed. Landslides. Rotterdam: Balkema, 35–51

[30] Leshchinsky B, Olsen M, Tanyu B F (2015). Contour connection method for automated identification and classification of landslide deposits. Journal of Computers and Geosciences, 74: 27–38

[31] McGee W J (1888). The Geology of the Head of Chesapeake Bay. Geological Survey 7th Annual Report, 537–646

[32] Morton D M, Alvares R M, Campbell R H (2003). Preliminary Soil-Slip Susceptibility Maps, Southwestern California, US Geological Survey Open-File Report 03–17

[33] NRC (2004). Partnerships for Reducing Landslide Risk: Assessment of the National Landslide Hazards Mitigation Strategy. National Research Council, National Academies Press: Washington, DC; 131

[34] Popescu M E (2002). Landslide Causal Factors and Landslide Remedial Options. Proceedings 3rd International Conference on Landslides, Slope Stability and Safety of Infra-Structures, Singapore. 61–81

[35] Ramanathan R S (2012). Soil Slope Failure Investigation Management System. M.S. Thesis, University of Maryland, College Park, MD, 121 p

[36] Rupke J, Cammeraat E, Seijmonsbergen A C, van Westen C J V (1988). Engineering geomorphology of Widentobel Catchment, Appenzell and Sankt Gallen, Switzerland: a geomorphological inventory system applied to geotechnical appraisal of slope stability. Eng Geol, 26(1): 33–68

[37] Saboya F, da Glória Alves M, Dias Pinto W (2006). Assessment of failure susceptibility of soil slopes using fuzzy logic. Eng Geol, 86(4): 211–224

[38] Sakellariou M G, Ferentinou M D (2001). GIS-based estimation of Slope Stability. Nat Hazards Rev, 2(1): 12–21

[39] Selby M J (1993). Hillslope Materials and Processes. Oxford University Press, 451 p

[40] Singh H, Huat B B (2008). Slope Assessment Systems: a review and evaluation of current techniques used for cut slopes in mountainous terrain of West Malaysia. Electronic Journal of Geotechnical Engineering, 13: 1–24

[41] Skempton A W (1953). Soil Mechanics in relation to geology. Proceedings of the Yorkshire Geology Society, 29(1): 33–62

[42] Turner A K, Schuster R L (1996). Lanslides: Investigation and Mitigation. Washington D.C.: National Academy Press, 673 p

[43] U. S. Census Bureau (2011). “2010 Resident Population Data”. U. S. Census Bureau. Retrieved 26 December 2010.

[44] USGS (2004). Lanslides Types and Processes, Fact Sheet 2004–3072, U.S. Department of the Interior, U.S. Geological Survey

[45] van Westen C J, Rengers N, Terlien M T J, Soeters R. (1997). Prediction of the occurrence of slope instability phenomena through GIS-based hazard zonation. Geol Rundsch, 86: 404–414

[46] Wachal D J, Hudak P F (2000). Mapping landslide susceptibility in Travis County, Texas, USA. GeoJournal, 51(3): 245–253

[47] Wang H, Liu G, Xu W, Wang G (2005). GIS-based landslide hazard assessment: an overview. Prog Phys Geogr, 29(4): 548–567

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg