GIS-Based Rockfall Hazard Assessment in Support of Decision Making

When designing infrastructure, settlements or facilities in mountainous areas, rockfall hazard assessment is considered essential, as it is a major hazard worldwide. Rockfall hazard estimation can help greatly in the design of countermeasures, such as barriers and net fences, in order to protect the built environment, as well as for landuse planning. Rockfall modelling is considered an effective way of estimating rockfall hazard despite the complicated processes involved. In this paper, a new three-dimensional rockfall simulation model, developed in the GIS environment, is proposed as a tool for assessing rockfall hazard for a localor regional-scale area. The model, adopting a kinematic approach, can simulate rockfall trajectories based on the topography, the frictional characteristics of the ground, the magnitude and direction of the initial velocity and the restitution coefficients on the block velocity. The model is implemented in an application, called ROCKFALL ANALYSIS, running in the ArcGIS environment allowing stochastic analysis and, even more, threedimensional visualization and animation of rockfalls. By means of case studies we evaluate both the simulation model and the application as a tool assisting spatial analysis and planning, which can be used in decision-making and design concerning transportation infrastructure or large technical works, such as dams. the three-dimensional (3D) effect of the topography, an effect playing a major role in controlling the dynamics of falling blocks. This is a severe limitation especially in areas where minor changes in topography influence a rockfall, e.g. along steep channels, where the topography is concave or convex or at the apex of a fan (Crosta and Locatelli, 1999). 3D rockfall simulation models are better than 2D models as they take into account the 3D effect of topography on rockfalls. The most important effect is the “lateral dispersion” of rockfall trajectories (Crosta and Agliardi, 2004), i.e. the ratio of the lateral distance separating the extreme fall paths to the slope length (Azzoni et al. 1995). The occurrence of lateral dispersion makes it difficult to choose a priori the right rockfall path when a 2D approach is adopted (Agliardi and Crosta, 2003). 2 GIS-BASED ROCKFALL HAZARD ASSESSMENT Over the last decade, the widespread availability of Geographical Information Systems (GIS) and the improvement of GIS technology have given engineers the possibility to conduct different kinds of landslide hazard assessments for a localor even a regional-scale area by the use of simplified geotechnical models. GIS can now be the platform where hazard occurrence models can be developed, permitting evaluation of results and adjustment of the input variables (Sakellariou & Ferentinou 2001, Ferentinou 2004, Sakellariou et al. 2006). In addition, GIS allow management and analysis of large amounts of spatial data and support the zonation of large areas. In this paper an application called ROCKFALL ANALYSIS is presented. ROCKFALL ANALYSIS is developed in a GIS environment and is capable of performing rockfall hazard assessment for an area, localor regional-scale, in a 3D space. This assessment is based on a new 3D rockfall simulation model, first created in 2006 (Charalambous), which, adopting a kinematic approach, simulates 3D rockfall trajectories based on a number of spatial or non-spatial input data 2.1 The Application ROCKFALL ANALYSIS The application is developed within the software ArcGIS. Is written in VBA and is comprised by two basic sections. The first section, “ROCKFALL_ANALYSIS.mxd,” runs within the program ArcMap (Fig. 1a). The calculation of the 3D rockfall trajectories is performed in this section and simulated rockfalls are represented in 2D (plan view). Furthermore, extra output data, spatial or not, can be produced within this section. “ROCKFALL_ANALYSIS_3D.sxd” is the second section. It runs within the program ArcScene (Fig. 1b) and is responsible for the 3D visualization and animation of the simulated rockfalls and the 3D representation of any other spatial data. Non-spatial data can also be presented here. ROCKFALL ANALYSIS runs in each section using the toolbars (Fig. 1). Apart from the functions the application offers, ArcMap and ArcScene offer a vast range of other useful functions to the users (e.g. mapping, analysis, exporting, printing, etc.), helping them in achieving a more integrated and credible rockfall hazard assessment. Figure 1. a. The first section of ROCKFALL ANALYSIS running in ArcMap b. The second section of ROCKFALL ANALYSIS running in ArcScene. 2.2 The Rockfall Simulation Model 2.2.1 Assumptions, Input and Output Data Rockfall modeling is accomplished by the use of the proposed 3D rockfall simulation model. As it is impossible to model accurately any physical phenomenon a number of assumptions have been made regarding the model: ● The model deals only with single block falls and mass falls, also called “fragmental rock falls” (Evans and Hungr, 1993), where the interaction among the falling blocks is considered null or negligible. ● A kinematic approach, treating the falling block as a lumped mass, allows free fall, bouncing/impact, rolling and sliding motions modeling in a 3D framework. ● The influence of the shape, the size and the angular momentum of the rock boulders on rockfall trajectories is only taken into consideration when rolling occurs. While in air, a boulder’s velocity is considered only translational, i.e. rotational velocity is ignored. ● Topography is represented by a Digital Terrain Model (DTM) in a Triangulated Irregular Network (TIN) format, without resolution restrictions. In this way, the 3D effect of the topography is taken into account. ● A number of input data (e.g. geology, restitution coefficients) are spatially distributed within the study area, using the ESRI shapefile format. ● The rockfall source can be defined by a point, polyline, polygon, rectangle, circle or an ellipse. When the source area is described by a polyline or a surface a set of rock boulders (the number is specified by the user) can be thrown from points along the polyline or within the surface. In that way, a form of stochastic modeling can be performed. For the moment, the source area (initial position of rocks) is the only parameter of the model defined as a random variable. ● Air drag and block fracturing are not taken into account. The input data required by the model are divided into spatial and non-spatial data. The spatial data are: ● The TIN representing the 3D topography of the area; ● A polygon shapefile containing the friction characteristics of the ground (friction angle of the geological formations); ● A polygon shapefile containing the normal and tangential restitution coefficients of the ground (in terms of velocity) on the block velocity; ● The rockfalls’ source area. ● The model allows additionally the design and the implementation of barriers (3D polygon shapefile), vertical or non-vertical, as a countermeasure. The model takes into account the characteristics of the barrier (location, geometry, frictional characteristics and restitution coefficients) when calculating the rockfalls. The necessary non-spatial input data are the magnitude and the direction of the initial velocity and the magnitude of the velocity which defines the transition point between the projectile state and the state where the rock is moving too slowly and, thus, should be considered rolling, sliding or stopped; This limit velocity is called minimum velocity. While the application produces 3D output data as results, in order to enhance the information generated, qualitative data are also generated referenced to the output spatial data. Specifically, the output data are: ● A 3D point shapefile containing all the points created along the rockfall trajectory. For each point non-spatial data are provided into the shapefile’s Attribute Table, e.g. the Id of the rockfall, the location (X,Y,Z) of the rock boulder, the height above the ground surface and the velocity (Vx,Vy,Vz) at that location, the time passed since the initiation of the rockfall, the aspect (aspect is the dip direction of the Tin triangle), the slope, the friction angle and the restitution coefficients of the underlying slope and comments for the present state of the boulder. ● A 3D polyline shapefile containing all the rockfalls simulated. The length and the duration of each rockfall are given in the attribute table. ● A table with statistics for each rockfall (e.g. total horizontal and vertical movement, total duration and length, maximum velocity and height DZ above ground) as well as for the whole set of the rockfalls (e.g. minimum and maximum length, duration and DZ). ● A 3D real time animation of the simulated rockfalls. 2.2.2 Initial Conditions Each rock boulder starts its rockfall from a point (X0, Y0, Z0) within the source area and upon or above the ground surface, having a translational initial velocity V0(V0X, V0Y, V0Z). Setting an initial velocity greater than zero also assists in the execution of a seismic hazard assessment. 2.2.3 Bouncing/Impact When the velocity V0 is greater than Vmin, then the rock boulder follows a parabolic path through air, described by simple kinematic laws (Charalambous, 2006), until it hits the ground surface. Impact with the ground is the most complex, uncertain and poorly understood process of a rockfall because the relationships linking the energy loss to a number of variables (the slope roughness, the geotechnical properties of the slope and the block’s shape and dynamics) are not clearly defined (Agliardi and Crosta, 2003). In the proposed simulation model, the impact is considered partially elastic, to a grade depending on the restitution coefficients. Though in 3D space, tangential (Rt) and normal (Rn) restitution coefficients are used. When impact occurs, energy is lost and the direction of motion of the rockfall changes depending on the slope’s 3D geometrical (slope, aspect) and mechanical properties (Rt, Rt). When the point of impact (Ximp, Yimp, Zimp) and the impact velocity (VXimp, VYimp, VZimp) are calculated, then the impact velocity is transformed from the Global Coordinate Cartesian System XYZ to a Local Coordinate Cartesian System. A Cartesian System UVW (Fig. 2), local to the TIN triangle (each triangle represents a slope) is used. The System UVW is defined by the impact point as the point of origin, the axis U tangential to the slope and towards the steepest downslope direction, axis W perpendicular to the slope and axis V tangential to the slope. The vector of the impact velocity in the UVW Cartesian system will be: