Engineering geological assessment of selected landslide dams formed from the 1929 Murchison and 1968 Inangahua earthquakes
Degree GrantorUniversity of Canterbury
Degree NameMaster of Science
This study investigated the characteristics of 26 failed and non-failed landslide dams (Murchison dataset) formed in the northern part of the South Island, New Zealand, from the 1929 Murchison and 1968 Inangahua earthquakes. The dataset was compiled from a combination of engineering geological mapping, field investigations, aerial photography interpretation and a review of existing literature. Current analysis techniques have been applied to the investigated landslide dams to assess their ability to accurately predict the post-formation development (or ‘evolution’) of the dam overtime. This has allowed the recognition of a number of additional landslide dam attributes that influence long-term stability, allowing modification of the stability analysis techniques in current use. Dam, lake, catchment and landslide characteristics were collated and analysed for the Murchison dataset by distinguishing failed from non-failed landslide dams, and then assessing the parameters common to both. Parameters that influence the post-formation development of selected landslide dams in the dataset include the dam volume, catchment area above the point of blockage, average block size of material comprising the dam, slope angle of the downstream dam face, and rock mass anal material characteristics in the source area of the dam-forming landslide. The stability of the dams in the Murchison dataset was not significantly affected by rock type, landslide movement, or the state, distribution and style of the dam-forming landslide. Existing geomorphic indices were applied to selected dams in the dataset. The Impoundment, Blockage and Dimensionless Blockage Indices (Casagli and Ermini (1999); Ermini and Casagli (2003)) predicted the correct post-formation development for 58, 86, and 81% of the selected landslide dams in the Murchison dataset, respectively. Four landslide dams covering both failed and non-failed types were investigated in detail to assist with this analysis, two being stable dams impounding lakes, and two having failed ‘catastrophically’ post-formation. Detailed investigation was carried out on Lake Stanley landslide dam, which agrees with all three indices predicting post-formation development, and of Lake Matiri, Ram Creek and Rain Peak landslide dams for which the indices incorrectly predict their post formation development. This investigation has shown that the average block size (D₅₀) of the dam material strongly influences the post-formation development of the four dams studied in detail. Dams consisting of material. with larger(> 200mm) average block sizes correspond to stable dams; while those with small(< 100mm) average block size correspond to failed dams. Rainfall duration/ intensity and slope angle of the downstream dam face were also found to influence post-formation development of the dams. The recently formed and failed Poerua landslide dam on the West Coast of the South Island was included in the geomorphic index evaluation because of the excellent documentation available, together with the prediction of its long-term stability using the index approach. The D₅₀ of the material forming the landslide dam, and the landslide dam's basal (or footprint) length, were incorporated with the parameters used in the existing geomorphic indices to produce a new geomorphic index, the Modified Dimensionless Blockage Index (MDBI). This is defined as: [complicated equation here] where Ac is the catchment area (m²), At is the lake area (m²), Hd is the height of the dam (m), Lʄ is the length of the dam footprint (m), Vd is the volume of the dam (m³), and Vd is the volume of the mean block size forming the dam (cube of the D₅₀ expressed in m³). Calculated MDBI values for the two stable landslide dams (Lake Stanley and Lake Matiri) are less than 10 (8.90 and 6.94 respective]y), while those for three failed landslide dams (Rain Peak, Ram Creek and Poerua) are greater than 10 (10.75, 10.80 and 14.9, respective]y). This suggests that the MDBI can be tentative]y used as a tool in forecasting the post-formation development of a landslide dam, with MDBI values > 10 corresponding to catastrophic dam failure, and an MDBI value < 10 corresponding to probable longer-term stability. However, it is recommended that a wider landslide dam dataset be applied to the MDBI to further test its accuracy, and to refine the parameters used both for short-term stability assessment following impoundment, and for longer-term prediction of post-formation dam (and lake) development. Rainfall duration and maximum block size of the dam material also require further evaluation, and a refinement incorporating grading parameters (such as D₆₀/D₁₀) may provide a better estimation of the post-formation landslide dam development. It is clear from this study that the block size and grading of the landslide dam material (in particular matrix or block support) exert significant influence on dam longevity and evolution, and this is reflected in the substantial weighting given to D₅₀ in the MDBI.