Modification and Statistical Analysis of the Colorado Rockfall Hazard Rating System

摘要

The purpose of this study was to improve the current Rockfall Hazard Rating System (RHRS) in use by the Colorado Department of Transportation (CDOT) by adding several climatic and geological factors recognized in technical literature that contribute to rockfall. Once the system was improved, 200 slopes were rated within the Colorado Front Range: 106 crystalline cut slopes (a crystalline rock cut); 51 crystalline total slopes (both a cut and the natural slope above); 35 block-in-matrix slopes; and 8 sedimentary slopes. The resulting data for each slope type was analyzed using univariate least squares regression, multivariate ordinal logistic regression, and multivariate stepwise regression to identify and rank the dominating factors that contribute to rockfall. The rank of the new scores for the 200 slopes was compared to the rank of the original scores to ensure that the modifications allow for a better prediction of rockfall potential. There were not enough sedimentary slopes rated to accurately assess the factors that control rockfall for these slope types. However, the results from the least squares regression illustrated that several of the parameters added to the new RHRS could be used to predict the total hazard score for the remaining slope types. Analysis of these parameters using logistic regression resulted in the following parameters having the most influence on rockfall hazard for each slope type (ranked in order): for crystalline cut slopes discontinuity aperture, ditch catchment, and rock character; for crystalline total slopes launching features, block size/volume, discontinuity persistence and orientation; and for block-in-matrix slopes block size, vegetation, and slope aspect. The stepwise regression produced equations in which the total hazard scores can be estimated by scoring the slope angle, launching features, overhang, and persistence and orientation for crystalline total slopes; and by scoring the slope aspect, block size, and vegetation for the block-in-matrix slopes. A succinct equation could not be produced for the crystalline cut slopes. Finally, comparison of the rank of the 200 slopes using both the modified and the original RHRS illustrated that there is no relationship between the two systems. In fact, the modified version allows for a larger spread of scores, and slopes that are characterized by a high rockfall potential are more easily identified.