• Analysis of the 2015 Sagavanirktok River flood: associated permafrost degradation using InSAR and change detection techniques

      McClernan, Mark Timothy; Meyer, Franz; Zwieback, Simon; Minter, Clifton (2020-08)
      In 2015, the Sagavanirktok River experienced a sequence of high, early-winter temperatures that lead to a buildup of aufeis. The buildup displaced the spring runoff causing widespread flooding. Flood waters inundated the surrounding tundra introducing heat into ground ice-baring soils. The Sagavanirktok River flood was caused by an extensive ice dam that developed the previous winter. The first flooding pulse started in April 2015, when an aufeis obstruction diverted river water to the surface. The obstruction caused flooding along 24 km of the Dalton Highway and its surroundings, necessitating a prolonged highway closure and emergency repairs. A second flooding pulse was caused by annual spring runoff in May 2015, which was driven by rapid snowmelt due to warm seasonal temperatures. The washed-out highway had to be closed again. Field investigations showed that thermal erosion of ice wedges in the tundra adjacent to the Dalton Highway caused local subsidence by several meters. However, the full environmental impact of the flood has not yet been quantified regionally or temporally. Thermokarst formation, can cause rapid ecological and environmental changes. Thawing of permafrost can lead to terrain instability as the melting of ground ice induces subsidence and loss of soil strength. The processes involved in permafrost degradation are complex, as is predicting terrain stability and the associated impacts to permafrost surrounding infrastructure. The immediate impact of the 2015 Sagavanirktok River flood is evident, which caused rapid terrain collapse in the vicinity of the Dalton Highway and the Trans-Alaska Pipeline near Deadhorse, North Slope Borough, Alaska. Thermal degradation of permafrost can be expressed as the change in the surfacemicrotopography over several years following a flood. Change detection, digital elevation model differencing, and InSAR were employed within the area of interest to understand the extent of the flood and deformation within inundated areas. To determine the likely impacted areas within the area of interest and expanse of the flood, an unsupervised change detection technique of high resolution TerraSAR-X and Sentinel-1 amplitude images was utilized. The topographic deformation analysis to determine the motion on the ground surface used a short baseline subset InSAR analysis of Sentinel-1 data during the summer season following the Sagavanirktok River flooding events. Additional deformation analysis was conducted with ALOS-2 data for annual comparison of the 2015 to 2019 summers. TanDEM-X digital elevation model differencing compared surface models generated from before and after the Sagavanirktok River flood. Elevation model differencing would identify the absolute change between the acquisition time of the surface models. A joint data analysis between deformation and differenced elevation models analyzed the contrast within inundated and flood-unaffected areas; thus, the changes and impact to the permafrost following the 2015 Sagavanirktok River flood. The Sagavanirktok River flood highlights the vulnerability of ice-rich permafrost to flooding. A change in the vicinity of the Sagavanirktok River Delta to the hydrological cycle led to widespread increases in terrain instability. Analysis of summer season deformation data suggested inundated permafrost areas showed lower seasonal deformation in years following the flood. Analysis of annual deformation shows permafrost subsidence intensified in inundated areas in the years following the flood. Digital elevation model differencing produced a statistically ambiguous result. This research illustrates the value of combining TerraSAR-X, TanDEM-X, Sentinel 1, and ALOS-2 microwave remote sensing missions for evaluating widespread surface changes in arctic environments. However, annual deformation data proved the most usable tool in observing the changing permafrost ecosystems around the Sagavanirktok River.
    • Multiresolution digital soil mapping of permafrost soils using a random forest classifier: an investigation along the Dalton Highway corridor, Alaska

      Paul, Joshua D.; Ping, Chien-Lu; Prakash, Anupma; Rossello, Jordi Cristobal; Libohova, Zamir (2018-12)
      In order to complete soil inventories in the remote permafrost zones of Alaska, there is a need to develop efficient digital soil mapping tools that can be applied over large areas using a minimum of ground truth data. This investigation first used a random forest classifier to test combinations of environmental input data at multiple resolutions (10m, 30m, and 100m). Five tiers of soil taxonomic units were predicted: Order, Suborder, Great Group, "Series Concept", and Particle Size Class. Model outputs are compared quantitatively via estimated out-of-bag accuracy, and qualitatively via visual inspection by soil scientists. Estimated out-of-bag accuracy ranged from ~45% to ~75%, with results improving when fewer classes were modeled. Model runs at 10m and 30m resolution performed comparably, with 100m resolution performing ~5-10% worse in most cases. Increasing the number of trees used, including categorical environmental input data (e.g. landforms), and replacement of environmental covariates with principal component analysis (PCA) bands did not significantly improve model performance. The random forest classifier was then used in a digital soil mapping pilot study along the Dalton Highway in northern Alaska. Parameters suggested in the initial study were used to predict multiple soil taxonomic classes from a basic collection of environmental covariates generated using high resolution (10m) satellite images and sparsely sampled pedon data. Covariates included maximum curvature, multiresolution valley bottom flatness, normalized height, potential incoming solar radiation, slope, terrain ruggedness index, and modified soil and vegetation index. Five tiers of soil taxonomic units were predicted: Order, Suborder, Great Group, "Series Concept", and Particle Size Class. Model outputs are compared quantitatively via estimated out-of-bag accuracy. Estimated out-of-bag accuracy ranged from ~45% to ~75%, with results improving when fewer classes were modeled. We suggest future research into optimized sampling to ensure an adequate distribution of samples across the feature space, and the incorporation of expert knowledge into accuracy assessments. Overall, digital soil mapping with random forest classifiers appears to be a promising method for completing the soil survey of Alaska.