Terrestrial LiDAR Scanning Research Articles

PypeTree: a tool for reconstructing tree perennial tissues from point clouds.

The reconstruction of trees from point clouds that were acquired with terrestrial LiDAR scanning (TLS) may become a significant breakthrough in the study and modelling of tree development. Here, we develop an efficient method and a tool based on extensive modifications to the skeletal extraction method that was first introduced by Verroust and Lazarus in 2000. PypeTree, a user-friendly and open-source visual modelling environment, incorporates a number of improvements into the original skeletal extraction technique, making it better adapted to tackle the challenge of tree perennial tissue reconstruction. Within PypeTree, we also introduce the idea of using semi-supervised adjustment tools to address methodological challenges that are associated with imperfect point cloud datasets and which further improve reconstruction accuracy. The performance of these automatic and semi-supervised approaches was tested with the help of synthetic models and subsequently validated on real trees. Accuracy of automatic reconstruction greatly varied in terms of axis detection because small (length < 3.5 cm) branches were difficult to detect. However, as small branches account for little in terms of total skeleton length, mean reconstruction error for cumulated skeleton length only reached 5.1% and 1.8% with automatic or semi-supervised reconstruction, respectively. In some cases, using the supervised tools, a perfect reconstruction of the perennial tissue could be achieved.

Assessment of tree structure using a 3D image analysis technique—A proof of concept

Efforts to improve the efficiency and efficacy of tree structure and crown architecture measurement are necessary to reduce error associated with indirect estimation of volume, which affects biophysical and ecosystem modelling, as well as resource assessment. In this short communication, we test the potential for a commercial SfM-MVS (Structure from Motion coupled with Multiple-View Stereophotogrammetry) software package, Photoscan-Professional, to accurately determine tree height, stem diameter, and eventually volume. SfM is a technique in computer vision, which calculates the 3D position of objects in a scene from a series of photographs. It uses a technique assuming that an object in a 3D scene is located on a vector between the image of the object in the camera and the object itself. The technique allows the construction of a 3D pointcloud, such as the one produced by laser technologies – terrestrial or airborne LiDAR. SfM requires no camera calibration or control points to construct an initial model. Moreover, it is a low-cost alternative to laser technologies. The second part of our method – MVS – is a visualization technique that reconstructs a 3D textured mesh of the scene from the SfM-derived pointcloud and RGB photographs. As a proof of concept, our methods were limited to two scenarios: (1) a single potted tree in a lab environment, where exact measurements could be made, and (2) two trees of different species and size in natural environments to test feasibility outside the laboratory. Precise measurements of tree height and stem diameter were compared with estimates obtained from the 3D model created using SfM-MVS. The results indicate that the SfM method is a promising – and inexpensive – alternative to terrestrial LiDAR and 3D scanners. Tree height estimates had error of 2.59%, while stem diameter estimates had error of 3.7%. The MVS algorithm used in this study was developed for plan surfaces such as topography or ‘compact’ objects and does not provide a representative 3D mesh for slender trees, although it works well for large stems. The authors link this disparity to the complex branching structure of trees.Future work requires (1) the development of effective automated volume reconstruction software specific to trees, and (2) the validation of the wide-scale applicability of the technology, which must include trees of various size, species, and growth forms. Moreover, future testing must include more complex environments, such as heterogeneous urban sites or closed-canopy forest sites where the proximity of other features may limit the utility of the new technology.

On seeing the wood from the leaves and the role of voxel size in determining leaf area distribution of forests with terrestrial LiDAR

Terrestrial LiDAR scanners have been shown to hold great potential for estimating and mapping three dimensional (3-D) leaf area distribution in forested environments. This is made possible by the capacity of LiDAR scanners to record the 3-D position of every laser pulse intercepted by plant material. The laser pulses emitted by a LiDAR scanner can be regarded as light probes whose transmission and interception may be used to derive leaf area density at different spatial scales using the Beer–Lambert law or Warren Wilson's contact frequency method among others. Segmenting the canopy into cubic volumes –or voxels- provides a convenient means to compute light transmission statistics and describe the spatial distribution of foliage area in tree crowns. In this paper, we investigate the optimal voxel dimensions for estimating the spatial distribution of within crown leaf area density. We analyzed LiDAR measurements from two field sites, located in Mali and in California, with trees having different leaf sizes during periods with and without leaves.We found that there is a range of voxel sizes, which satisfy three important conditions. The first condition is related to clumping and requires voxels small enough to exclude large gaps between crowns and branches. The second condition requires a voxel size large enough for the conditions postulated by the Poisson law to be valid, i.e., a turbid medium with randomly positioned leaves. And, the third condition relates to the appropriate voxel size to pinpoint the location of those volumes within the canopy which were insufficiently sampled by the LiDAR instrument to derive reliable statistics (occlusion effects). Here, we show that these requirements are a function of leaf size, branching structure, and the predominance of occlusion effects. The results presented provide guiding principles for using voxel volumes in the retrieval of leaf area distributions from terrestrial LiDAR measurements.

Airborne lidar point cloud-based below-canopy line-of-sight visibility estimator

Point cloud data collected by small-footprint lidar scanning systems have proven effective in modeling the forest canopy for extraction of tree parameters. Although line-of-sight visibility (LOSV) in complex forests may be important for military planning and search-and-rescue operations, the ability to estimate LOSV from lidar scanners is not well developed. A new estimator of below-canopy LOSV (BC-LOSV) by addressing the problem of estimation of lidar under-sampling of the forest understory is created. Airborne and terrestrial lidar scanning data were acquired for two forested sites in order to test a probabilistic model for BC-LOSV estimation solely from airborne lidar data. Individual crowns were segmented, and allometric projections of the probability model into the lower canopy and stem regions allowed the estimation of the likelihood of the presence of vision-blocking elements for any given LOSV vector. Using terrestrial lidar scans as ground truth, we found an approximate average absolute difference of 20% between BC-LOSV estimates from the airborne and terrestrial point clouds, with minimal bias for either over- or underestimates. The model shows the usefulness of a data-driven approach to BC-LOSV estimation that depends only on small-footprint airborne lidar point cloud and physical knowledge of tree phenology.

Accuracy assessment of a mobile terrestrial lidar survey at Padre Island National Seashore

The higher point density and mobility of terrestrial laser scanning (light detection and ranging (lidar)) is desired when extremely detailed elevation data are needed for mapping vertically orientated complex features such as levees, dunes, and cliffs, or when highly accurate data are needed for monitoring geomorphic changes. Mobile terrestrial lidar scanners have the capability for rapid data collection on a larger spatial scale compared with tripod-based terrestrial lidar, but few studies have examined the accuracy of this relatively new mapping technology. For this reason, we conducted a field test at Padre Island National Seashore of a mobile lidar scanner mounted on a sport utility vehicle and integrated with a position and orientation system. The purpose of the study was to assess the vertical and horizontal accuracy of data collected by the mobile terrestrial lidar system, which is georeferenced to the Universal Transverse Mercator coordinate system and the North American Vertical Datum of 1988. To accomplish the study objectives, independent elevation data were collected by conducting a high-accuracy global positioning system survey to establish the coordinates and elevations of 12 targets spaced throughout the 12 km transect. These independent ground control data were compared to the lidar scanner-derived elevations to quantify the accuracy of the mobile lidar system. The performance of the mobile lidar system was also tested at various vehicle speeds and scan density settings (e.g. field of view and linear point spacing) to estimate the optimal parameters for desired point density. After adjustment of the lever arm parameters, the final point cloud accuracy was 0.060 m (east), 0.095 m (north), and 0.053 m (height). The very high density of the resulting point cloud was sufficient to map fine-scale topographic features, such as the complex shape of the sand dunes.

Coseismic slip variation assessed from terrestrial lidar scans of the El Mayor–Cucapah surface rupture

We analyze high-resolution (>103points/m2) terrestrial lidar surveys of the 4 April 2010 El Mayor–Cucapah earthquake rupture (Baja California, Mexico), collected at three sites 12–18 days after the event. Using point cloud-based tools in an immersive visualization environment, we quantify coseismic fault slip for hundreds of meters along strike and construct densely constrained along-strike slip distributions from measurements of offset landforms. Uncertainty bounds for each offset, determined empirically by repeatedly measuring offsets at each site sequentially, illuminate measurement uncertainties that are difficult to quantify in the field. These uncertainties are used to define length scales over which variability in slip distributions may be assumed to reflect either recognizable earthquake mechanisms or measurement noise. At two sites characterized by 2–3m of concentrated right-oblique slip, repeat measurements yield 2σ uncertainties of ±11–12%. Each site encompasses ∼200m along strike, and a smoothed linear slip gradient satisfies all measurement distributions, implying along-fault strains of ∼10−3. Conversely, the common practice of defining the slip curve by the local slip maxima distorts the curve, overestimates along-fault strain, and may overestimate actual fault slip by favoring measurements with large, positive, uncertainties. At a third site characterized by 1–2.5m of diffuse normal slip, repeat measurements of fault throw summed along fault-perpendicular profiles yield 2σ uncertainties of ±17%. Here, a low order polynomial fit through the measurement averages best approximates surface slip. However independent measurements of off-fault strain accommodated by hanging wall flexure suggest that over the ∼200m length of this site, a linear interpolation through the average values for the slip maxima at either end of this site most accurately represents subsurface displacement. In aggregate, these datasets show that given uncertainties of greater than ±11% (2σ), slip distributions over shorter scales are likely to be less uneven than those derived from a single set of field- or lidar-based measurements. This suggests that the relatively smooth slip curves we obtain over ∼102m distances reflect real physical phenomena, whereas short wavelength variability over ∼100–101m distances can be attributed to measurement uncertainty.

A fine-scale architectural model of trees to enhance LiDAR-derived measurements of forest canopy structure

Forest canopy structure consists of the complex spatial arrangement of the foliage, branches, and boles of trees. Measuring forest canopy structure is an active research question because of its influence on a wide range of biophysical and ecological processes. Light detection and ranging technology (LiDAR), both with airborne and terrestrial instruments, has been used to assess canopy structure and estimate vegetation attributes. Although LiDAR has proven to be an effective tool for mapping the forest canopy, its ability to probe all canopy elements is limited by the signal occlusion that occurs when multiple objects lie between the sensor and target area. This limitation can be alleviated by comparing realistic fine-scale 3D canopy architectural models with terrestrial and airborne LiDAR point-cloud data sets. We propose a method that mitigates the limitations of object occlusion and allows detailed description of 3D canopy structure from small-footprint airborne LiDAR (ALiDAR). Our approach involved reconstructing stem-mapped forest stands with catalogs of terrestrial LiDAR (TLiDAR) scans of individual trees, using an architectural model called L-Architect (LiDAR data to vegetation Architecture). L-Architect has been developed to link TLiDAR scans to tree-level structural attributes that reproduce the branching structure and foliage spatial distribution of a tree according to information gathered from detailed mensurational data. This study has further developed the potential of L-Architect to reconstruct individual trees to model the structure of a stand. The method was evaluated by comparing vertical profiles of material density for a forest stand calculated from TLiDAR scans and L-Architect's canopy model, resulting in a Pearson's correlation coefficient of 0.95. Most individual tree structures were reproduced with high accuracy, with a relative deviation of <10% on tree diameter at breast height (DBH) and total leaf area. The use of the ALiDAR data together with L-Architect allowed quantifying how significant object occlusion affects untreated data. We obtained mean correlations between ALiDAR data sets and L-Architect derived profiles of 0.84 and 0.43 for the vertical and horizontal material distribution, respectively. We demonstrate how L-Architect can be used to make accurate 3D maps of stand components from LiDAR data.

CLOSE RANGE HYPERSPECTRAL IMAGING INTEGRATED WITH TERRESTRIAL LIDAR SCANNING APPLIED TO ROCK CHARACTERISATION AT CENTIMETRE SCALE

Abstract. Compact and lightweight hyperspectral imagers allow the application of close range hyperspectral imaging with a ground based scanning setup for geological fieldwork. Using such a scanning setup, steep cliff sections and quarry walls can be scanned with a more appropriate viewing direction and a higher image resolution than from airborne and spaceborne platforms. Integration of the hyperspectral imagery with terrestrial lidar scanning provides the hyperspectral information in a georeferenced framework and enables measurement at centimetre scale. In this paper, three geological case studies are used to demonstrate the potential of this method for rock characterisation. Two case studies are applied to carbonate quarries where mapping of different limestone and dolomite types was required, as well as measurements of faults and layer thicknesses from inaccessible parts of the quarries. The third case study demonstrates the method using artificial lighting, applied in a subsurface scanning scenario where solar radiation cannot be utilised.

IMPLICIT SHAPE MODELS FOR OBJECT DETECTION IN 3D POINT CLOUDS

Abstract. We present a method for automatic object localization and recognition in 3D point clouds representing outdoor urban scenes. The method is based on the implicit shape models (ISM) framework, which recognizes objects by voting for their center locations. It requires only few training examples per class, which is an important property for practical use. We also introduce and evaluate an improved version of the spin image descriptor, more robust to point density variation and uncertainty in normal direction estimation. Our experiments reveal a significant impact of these modifications on the recognition performance. We compare our results against the state-of-the-art method and get significant improvement in both precision and recall on the Ohio dataset, consisting of combined aerial and terrestrial LiDAR scans of 150,000 m2 of urban area in total.

Canal Reach Deformation Monitoring Technology Based on Terrestrial LIDAR Scanning Technology

Canal reach soil of the south to north water transfer project has a certain expansibility. Concrete lining plates is puring on the soil. After about half a year of water immersion test, a certain plastic deformation has happened on the lining plates because of the expansion in the soil. In order to monitoring the whole lining plates deformation pre and post water immersion test, point clouds is collected by Terrestrial LIDAR and total station. After constructing of triangular mesh and endowing the specific color, the two group of mesh can be overlapping displayed. Deformation area can be found easily by color distortion on superposition chart. And then deformation tendency is concluded.

Hyperspectral image analysis of different carbonate lithologies (limestone, karst and hydrothermal dolomites): the Pozalagua Quarry case study (Cantabria, North‐west Spain)

AbstractGround‐based hyperspectral imaging combined with terrestrial lidar scanning is a novel technique for outcrop analysis, which has been applied to Early and Late Albian carbonates of the Pozalagua Quarry (Cantabrian Mountains, Spain). An image processing workflow has been developed for differentiating limestone from dolomite, providing additional sedimentary and diagenetic information, and the possibility to quantitatively delineate diagenetic phases in an accurate way. Spectral absorption signatures can be linked to specific sedimentary or diagenetic products, such as recent and palaeokarst, hydrothermal karst, (solution enlarged) fractures and different dolomite types. Some of the spectral signatures are related to iron, manganese, organic matter, clay and/or water content. Ground‐truthing accessible parts of the quarry showed that the classification based on hyperspectral image interpretation was very accurate. This technique opens the possibility for quantitative data evaluation on sedimentary and diagenetic features in inaccessible outcrops. This study demonstrates the potential of ground‐based imaging spectroscopy to provide information about the chemical–mineralogical distribution in outcrops, which could otherwise not be established using conventional field methods.

Comprehensive Utilization of Temporal and Spatial Domain Outlier Detection Methods for Mobile Terrestrial LiDAR Data

Terrestrial LiDAR provides many disciplines with an effective and efficient means of producing realistic three-dimensional models of real world objects. With the advent of mobile terrestrial LiDAR, this ability has been expanded to include the rapid collection of three-dimensional models of large urban scenes. For all its usefulness, it does have drawbacks. One of the major problems faced by the LiDAR industry today is the automatic removal of outlying data points from LiDAR point clouds. This paper discusses the development and combined implementation of two methods of performing outlier detection in georeferenced point clouds. These methods made use of the raw data available from most time-of-flight mobile terrestrial LiDAR scanners in both the temporal and spatial domains. The first method involved a moving fixed interval smoother derived from the well-known position velocity acceleration Kalman Filter. The second method fitted a quadratic curved surface to sections of LiDAR data. The combined use of these routines is discussed through examples with real LiDAR data.

Estimating leaf area distribution in savanna trees from terrestrial LiDAR measurements

Vegetation structure parameters are key elements in the study of ecosystem functioning and global scale ecosystemic interactions. The detailed retrieval of many of these parameters by direct measurements is impractical due to the quantity of plant material in trees. Terrestrial LiDAR Scanners (TLSs) have been shown to hold great potential as an indirect means of estimating plant structure parameters with a high level of detail, while some studies identified a number of challenges inherent to this approach. In this study we investigate the use of a voxel-based approach to retrieve leaf area distribution of individual trees. The approach is based on the contact frequency method applied to co-registered TLS returns from two or more scanning positions. The contact frequency was computed for voxels being 10, 30, and 50 cm in size and subsequently corrected for the influence of occlusion effects, leaf inclination, the presence of non-photosynthetic material, and the laser beam size. The leaf area of voxels for which occlusion effects were too pronounced was estimated using modeled values based on the availability of light. We compared the TLS derived leaf area estimates against direct measurements, obtained by the harvesting of leaves, in a broad-leaved savanna of central Mali. The measured leaf area values of the sampled trees ranged from 30 to 530 m 2, and crown LAI values between 0.8 and 7.2. The leaf area estimates lay on average 14% from the reference measurements (general bias). Our method provides vertical as well as radial distributions of leaf area in individual trees, and lends itself to the estimation of savanna vegetation structural parameters with a high level of detail.

An architectural model of trees to estimate forest structural attributes using terrestrial LiDAR

Terrestrial lidar (TLiDAR) has been used increasingly over recent years to assess tree architecture and to extract metrics of forest canopies. Analysis of TLiDAR data remains a difficult task mainly due to the effects of object occlusion and wind on the quality of the retrieved results. We propose to link TLiDAR and tree structure attributes by means of an architectural model. The proposed methodology uses TLiDAR scans combined with allometric relationships to define the total amount of foliage in the crown and to build the tree branching structure. It uses the range (distance) and intensity information of the TLiDAR scans (i) to extract the stem and main branches of the tree, (ii) to reconstruct the fine branching structure at locations where the presence of foliage is very likely, and (iii) to use the availability of light as a criterion to add foliage in the center of the crown where TLiDAR information is sparse or absent due to occlusion effects. An optimization algorithm guides the model towards a realistic tree structure that fits the information gathered from TLiDAR scans and field inventory. The robustness and validity of the proposed model is assessed on five trees belonging to four different conifer species from natural forest environments. This approach addresses the data limitation of TLiDAR scans and aims to extract forest architectural metrics at different structural levels.

Sensitivity of tree volume measurement to trajectory errors from a terrestrial LIDAR scanner

The use of terrestrial LIDARs in agriculture enables the measurement of structural parameters of the orchards such as the volume of the trees. The sequence of two-dimensional scans performed with a LIDAR attached to a tractor can be interpreted as the three-dimensional silhouette of the trees of the grove and used to estimate their volume. In this work, the sensitivity of the tree volume estimates relative to different error sources in the estimated spatial trajectory of the LIDAR is analyzed. Tests with pear trees have demonstrated that the estimation of the volume is very sensitive to errors in the determination of the distance from the LIDAR to the center of the trees (with errors up to 30% for an error of 50 mm) and in the determination of the angle of orientation of the LIDAR (with errors up to 30% for misalignments of 2°). Therefore, any experimental procedure for tree volume estimate based on a motorized terrestrial LIDAR scanner must include additional devices or procedures to control or estimate and correct these error sources.

Management tools for optimal allocation of wood fibre to conventional log and bio-energy markets in Ireland: a case study

A set of tools are described for optimal allocation of wood fibre at an operational planning level. These were applied to a case study in Ireland. Allocation was based on optimising net value recovery (delivered price minus harvesting and transportation costs) while meeting market demands and operational constraints (mainly crew capability and productivity limits). Two new models were developed to predict harvesting costs and transportation costs for Irish forest conditions. A new model was developed to link Sitka spruce biomass expansion factors to optimal log-making algorithms so that log and bio-energy product yields could be estimated for individual harvest areas. An existing operational allocation model based on a tabu search heuristic procedure was used. The case study included 16 forest harvest areas and 12 processing plants (saw logs, pallet logs, stakes, pulp, bio-energy slash bundles, etc.). New terrestrial lidar scanning procedures were used to obtain representative stem profiles from over 4,000 trees for the 16 forests. We demonstrated that optimal allocation of bio-energy and log products, while complex, can be achieved through the use of appropriate management tools.

Direct Georeferencing of Stationary LiDAR

Unlike mobile survey systems, stationary survey systems are given very little direct georeferencing attention. Direct Georeferencing is currently being used in several mobile applications, especially in terrestrial and airborne LiDAR systems. Georeferencing of stationary terrestrial LiDAR scanning data, however, is currently performed indirectly through using control points in the scanning site. The indirect georeferencing procedure is often troublesome; the availability of control stations within the scanning range is not always possible. Also, field procedure can be laborious and involve extra equipment and target setups. In addition, the conventional method allows for possible human error due to target information bookkeeping. Additionally, the accuracy of this procedure varies according to the quality of the control used. By adding a dual GPS antenna apparatus to the scanner setup, thereby supplanting the use of multiple ground control points scattered throughout the scanning site, we mitigate not only the problems associated with indirect georeferencing but also induce a more efficient set up procedure while maintaining sufficient precision. In this paper, we describe a new method for determining the 3D absolute orientation of LiDAR point cloud using GPS measurements from two antennae firmly mounted on the optical head of a stationary LiDAR system. In this paper, the general case is derived where the orientation angles are not small; this case completes the theory of stationary LiDAR direct georeferencing. Simulation and real world field experimentation of the prototype implementation suggest a precision of about 0.05 degrees (~1 milli-radian) for the three orientation angles.

The structural and radiative consistency of three-dimensional tree reconstructions from terrestrial lidar

A novel methodology is proposed to reconstruct 3D tree architectures from terrestrial LiDAR (TLiDAR) scans. The methodology is robust and relatively insensitive to wind- and occlusion-induced artefacts in the 3D TLiDAR point clouds. A quantitative evaluation of structural attributes, like the vertical foliage and wood area profiles, as well as the shoot orientation distribution, was performed. Due to the difficulties of acquiring reliable and accurate estimates of these parameters in the field, an original evaluation approach was chosen that reproduces the TLiDAR scanning and subsequent tree reconstruction process in a virtual environment. In a second step the reconstructed tree models were ingested in a validated 3D radiative transfer model to simulate both their reflectance signatures (observable by space borne instruments) and directional transmission properties (measurable during field campaigns) under various spectral, illumination and tree density scenarios. The results of these evaluations confirm the appropriateness of the proposed tree reconstruction model for the generation of structurally and radiatively faithful copies of existing plant and canopy architectures.

Instability investigation of cantilevered seacliffs

AbstractWave action is a fundamental mechanism in seacliff erosion, whereby wave undercutting creates an unstable cantilevered seacliff profile and can lead to large catastrophic cliff failures, thus threatening coastal infrastructure. This study investigated the instability of two such failures that occurred in Solana Beach, California, by combining terrestrial LIDAR scanning, cantilever beam theory and finite element analysis. Each landslide was detected by evaluating the surface change between subsequent high resolution digital terrain models derived from terrestrial LIDAR data. The dimensions of failed cantilever masses were determined using the surface change measurements and then incorporated into failure stress analysis. Superimposing stress distributions computed from elastic cantilever beam theory and finite element modeling provided a method to back‐calculate the maximum developed tensile and shear stresses along each failure plane. The results of the stress superposition revealed that the bending stresses caused by the cantilevered load contributed the majority of stress leading to collapse. Both shear and tensile failure modes were investigated as potential cliff failure mechanisms by using a comparison of the back‐calculated failure stresses to material strengths found in laboratory testing. Based on the results of this research, the tensile strength of the cliff material was exceeded at both locations, thus causing the cliffs to collapse in tension. Copyright © 2008 John Wiley &amp; Sons, Ltd.

Processes of coastal bluff erosion in weakly lithified sands, Pacifica, California, USA

Coastal bluff erosion and landsliding are currently the major geomorphic processes sculpting much of the marine terrace dominated coastline of northern California. In this study, we identify the spatial and temporal processes responsible for erosion and landsliding in an area of weakly lithified sand coastal bluffs located south of San Francisco, California. Using the results of a five year observational study consisting of site visits, terrestrial lidar scanning, and development of empirical failure indices, we identify the lithologic and process controls that determine the failure mechanism and mode for coastal bluff retreat in this region and present concise descriptions of each process. Bluffs composed of weakly cemented sands (unconfined compressive strength — UCS between 5 and 30 kPa) fail principally due to oversteepening by wave action with maximum slope inclinations on the order of 65 at incipient failure. Periods of significant wave action were identified on the basis of an empirical wave run-up equation, predicting failure when wave run-up exceeds the seasonal average value and the bluff toe elevation. The empirical relationship was verified through recorded observations of failures. Bluffs composed of moderately cemented sands (UCS up to 400 kPa) fail due to precipitation-induced groundwater seepage, which leads to tensile strength reduction and fracture. An empirical rainfall threshold was also developed to predict failure on the basis of a 48-hour cumulative precipitation index but was found to be dependent on a time delay in groundwater seepage in some cases.