APPLICATION OF SMALL BASELINE SUBSET (SBAS) TIME-SERIES ANALYSIS FOR LANDSLIDE DETECTION IN HAWAIʻI

Abstract

The occurrence of landslides or rockfalls onto public roadways can result in a wide variety of issues for commuters and government agencies affiliated with the roadway systems. Under the best circumstances, landslide or rockfall debris on a roadway will cause commuter delays and result in minimal costs associated with remedial efforts. Under worst case circumstances, these occurrences can result in property and roadway infrastructure damage, injury, and even loss of life. The Hawaiʻi Department of Transportation (HDOT) is proactively attempting to identify areas near public roadways that may pose potential concern for landslide or rockfall activity. As part of these efforts, they have requested research be conducted involving the use of satellite synthetic aperture radar (SAR) imagery to conduct interferometry (InSAR) for two locations: Hawaiʻi Route 19 between mileposts 10 and 30; and Hawaiʻi Route 360 between mileposts 0 and 35. The Small Baseline Subset (SBAS) InSAR approach was determined to be the best method for the areas of HDOT interest. An extensive search of SAR imagery for the Hawaiian Islands proved that only limited datasets are publicly available for the areas of interest. To provide a proof of concept for implementation of SBAS analysis in the Hawaiian Islands, a case study of the Waiʻōmaʻo landslide located in Pālolo Valley on the Island of Oʻahu was conducted to compare available inclinometer data with SBAS datasets. Two-dimensional decomposition was implemented for opposite orbits of descending and ascending Sentinel-1 datasets. Decomposition was conducted to compare and review with in situ Waiʻōmaʻo landslide inclinometer data. Overall, the case study results show that when both ascending and descending datasets are used to derive displacements that are resolved in the direction of movement, InSAR analysis can effectively capture inclinometer trends in areas experiencing relatively small displacements over time (< 30 mm/year) but the accuracy diminished in fast moving slides (≥ 270 mm/year). Following the case study, the SBAS method was then applied to the Hawaiʻi Island and Maui study areas. Application of the InSAR technique was limited by the following: SAR data availability; geographic positioning and radar shadowing; and amount of vegetation in the research areas. The Hawaiʻi Island study area displacements were decomposed in a direction perpendicular to the slope contours to estimate the true displacements. Review of two-dimensional decomposition results was conducted for the potential of landslides and rockfalls in Hawaiʻi Island. Of the areas, minimal displacements were measured. Obtaining SBAS descending and ascending opposite orbits for Maui proved unsuccessful due to radar shadowing caused by Haleakalā shield volcano. SBAS processing was only successful for the descending orbit for Maui.