Numerical Modelling of Arctic Costal Erosion

Abstract

Arctic coastal erosion is a growing concern as global climate change affects the planet. Unlike low-latitude and mid-latitude coastal areas, the sediments along Arctic coastlines are often frozen, even during summer. It is essential to consider both thermal and mechanical factors when analysing Arctic coastal erosion. Two major erosion mechanisms in the Arctic have been identified: thermodenudation and thermoabrasion. Thermodenudation refers to the thawing and melting of the frozen soil, while thermoabrasion refers to the mechanical action of waves and ice against the coast. Various numerical models have been developed in recent years to describe these mechanisms, but one limitation of such models is the difficulty of including hydrodynamic forces. Moreover, the available coastal erosion models developed for warmer climates cannot be applied to Arctic coastal erosion, where permafrost is a significant environmental parameter. To address these challenges, a new methodology has been proposed in this thesis that allows using models designed for warmer climates to simulate Arctic coastal erosion. A modular approach is adopted where the processes related to permafrost thawing and Arctic coastal erosion are represented as submodules. The open-source software XBeach is used to simulate the waves, sediment transport, and morphological changes in the nearshore. Different submodules are developed to simulate the processes unique to Arctic coasts, such as thawing-freezing, slumping, wave-cut niche, bluff failure, and more. These submodules are coupled with XBeach to enable concurrent simulation of the thermodenudation and thermoabrasion Arctic coastal erosion. Since 2012, field investigations have been conducted at an Arctic coast named Baydaratskya Bay in the Kara Sea, Russia, as part of the Sustainable Arctic Marine and Coastal Technology (SAMCoT) project. These investigations are carried out annually in the summer by the Lomonosov Moscow State University (MSU) to collect data from the study area. The measurement of coastal profiles during these investigations are used in this study to calibrate and validate numerical models. The study area is unique since the two erosion mechanisms - thermoabrasion and thermodenudation - are both active. The backbone of this thesis is based on the reports, in situ tests, measurements, and observations made during the field investigations carried out over the years. The investigations of the study area are critical to understanding the complex coastal dynamics in the region, and it provides an invaluable resource for working on issues related to climate change and coastal erosion. Some of the model’s input parameters are calibrated using field measurements and the model is then validated by another set of mutually exclusive field measurements under different morphological conditions from the study area. The sensitivity analysis of the model reveals that nearshore waves are a significant driver of erosion, and the inclusion of nearshore hydrodynamics and sediment transport is essential for accurately modeling the erosion mechanism. This finding highlights the importance of considering multiple physical processes and their interactive nature when modeling Arctic coastal erosion. In data-poor environments such as the Arctic coastal regions, developing a deterministic model to describe the processes and predict the outcomes of coastal erosion can be challenging. Deterministic models rely on precise and accurate input data and may not capture the full range of uncertainties associated with the system being studied. A probabilistic model based on Monte Carlo simulation is developed by assuming a probabilistic distribution of the input parameters. The probabilistic approach is demonstrated using an earlier Arctic coastal erosion model version. Crest retreat due to thermoabrasion on the coastal profile is simulated, and the probability of the bluff collapsing during a storm is estimated. One of the challenges in studying Arctic coastal erosion is the lack of temperature measurements of permafrost. The temporal and spatial resolutions of the temperature observations are relatively high, making it difficult to analyse the data. Therefore, in this study, a data-driven model is developed to interpolate, hindcast, and forecast temperature measurements within the active layer and shallow permafrost when the temperature measurements at the surface or near the surface are available. The temperature variations along the year are periodic, and hence attempts are made to express the seasonal variations with a combination of periodic functions (Fourier components), which are then used as boundary conditions to reach the analytical solutions.