Rosser BJ, Carey JM, Jones KE, Townsend DB, Crouch JF. 2020. Landslides and debris flows in the Tasman District caused by ex-Tropical Cyclone Gita, 20-21 February, 2018. Lower Hutt (NZ): GNS Science. 52 p. (GNS Science report; 2019/34). doi:10.21420/HN7F-K408.
Ex-Tropical Cyclone Gita hit New Zealand on 20–21 February 2018, bringing heavy rain and high winds to much of the country. The heaviest rainfall occurred in the Tasman District where more than 230 mm of rain fell in 18 hours triggering extensive shallow landslides and debris flows. The Takaka Hill Road was closed by 16 slips, which impacted the road for the following year. The most intense rainfall (78 mm/hr) occurred over a 3-hour period between 3 pm and 6 pm near Marahau and it was during this period that most of the landslides were triggered. Annual recurrence intervals (ARI) for the rainfall in this area may have exceeded 200 years, however the heaviest rainfall was isolated and for most rain gauges the ARI was <10 years. An aerial reconnaissance was flown by GNS Science and Tasman District Council staff on 7 March 2018. A regional-scale assessment was undertaken to map the landslides triggered by ex-Tropical Cyclone Gita by differencing of the pre- and post-event, 10-m resolution Sentinel satellite imagery. The extreme rainfall triggered more than 2000 landslides, in two distinct areas: 1) Aorere-Kaituna River area, near Collingwood (417 landslides) and 2) Marahau-Riwaka area northwest of Motueka (1638 landslides). The highest density of landsliding occurred along the northern end of the Arthur Range (Takaka Hill). Debris flows were initiated in several of the receiving catchments, particularly in the Riwaka-Marahau area, where significant debris flows occurred in the Otuwhero, Marahau, Riwaka, Rocky and the Shaggery rivers. The landslide distribution was overlaid onto attribute layers containing key site characteristics (rainfall, geology, slope angle, slope aspect, vegetation type) to assess their potential controlling influence. Melton ratios, a measure of the ability of catchments to generate debris flows that has been used to identify debris flow prone catchments, were calculated for each catchment in the area of heaviest rainfall. The distribution of landslides was strongly influenced by the underlying geology, with 80 % of the landslides occurring on Separation Point Granite which underlies only 23 % of the area where landslides occurred. Rocks of the Separation Point Granite suite are often deeply weathered and highly erodible. The rainfall threshold for triggering debris flows in Separation Point Granite was about 200 mm/18 hrs, or 120 mm/3 hrs. This equates to rainfall intensities >10 mm/hr for longer durations (18 hrs) or 40 mm/hr for short durations (<3 hrs). Analysis of the landslide distribution in relation to the rainfall estimates derived from rain radar identified several key issues with the rain radar data, mostly associated with the distance of the site from the Wellington radar station. Firstly, because of the distance from the radar station to the site, the rainfall was measured at an altitude of 3-4 km. With northerly winds of 110-130 km/hr, it was estimated that the rain fell on the ground roughly 10 km south of where is was measured in the air. Secondly, the radar had difficulty ‘seeing through’ the heavy rain at Abel Tasman National Park and underestimated the rainfall in the Aorere-Kaituna River area. Ideally, when using rain radar to investigate rainfall induced landslides, it should be corrected to the local rain-gauge network, corrected for wind, and used with caution. (auth)