Slope Stability Problems Induced by Human Modification of the Soil Covered Hill Slopes of Oʻahu, Hawaiʻi

Abstract

This dissertation attempts to study slope stability problems induced by human intervention on the Island of Oahu. The selected area for the study is the eastern half of Oahu where much of the slope stability problems have occurred. As a background to the problem, the factors that cause unstable slopes are examined as they occur in nature and as they are caused by human intervention in various parts of the world. Attention is then directed to the landslides which have occurred on Oahu, particularly to 23 cases in the study area. The study was undertaken with a view to develop a suitable empirical model in slope stability. To achieve this objective, the variable factors which contribute towards unstable slope conditions were isolated and evaluated. These factors were determined as slope, soil, geology, rainfall and proximity to streams or sea. With three subdivisions in the soil factor, i.e., soil depth, soil type and shrink/swell the total number of variable factors determined as causing unstable slope conditions were determined as seven in number. However, after a multiple regression program was run, it was found that geology and shrink/swell contributed very little to the model over and above the contributions of the other variable factors and therefore, they were dropped from the model. Each variable was divided into categories and these were ranked according to their landsliding potential. A numerical value was allocated to each category based on the frequency of sliding units in each. The total number of sliding units in the 23 cases of landsliding were 142. An equal number of random units were obtained from the study area and they were also evaluated according to their categories in the seven variables. A multiple regression program was written with a sub program where each sliding unit value was entered as "1" and a random unit value was entered as "0". The percentage probability of sliding was obtained from the computed values and a sliding probability scale was constructed with numerical values 0 to 6. The 0 denotes stable slopes and the 6 denotes slopes with 100 percent sliding probability. The study area was covered with a grid system, each grid one half inch square on the Topographic Quadrangles of Oahu of the 1:24,000 scale. Data for each grid square was assembled as for the studied and random units and these were programmed using the regression equation to obtain computed values. According to the computed values, the sliding probability values were obtained for each square from the constructed scale and these probability values were incorporated in a landslide probability map covering the study area. The empirical model is intended to help minimize slope stability problems by affording a method to evaluate slopes before modification. The landslide probability map is intended to help select sites which are least susceptible to slope stability problems. Both model and map are of little use in minimizing potential slope stability problems if a suitable Grading Code is not in force. The Hawaii Grading Ordinance, though fairly comprehensive is found to lack certain restrictions which need inclusion in the Ordinance if future slope stability problems are to be avoided. In the light of the cases of studied landslides, these restrictions are enumerated and are suggested for inclusion.