Sammendrag
Hyperbolic Lagrangian Coherent Structures (LCS) are time-dependent manifolds that organize tracer patterns in chaotic flow systems. In two-dimensional flow systems, LCSs take the shape of one-dimensional curves which act as the locally most attracting or repelling structures over a finite time interval. LCSs yield a description of the flow field itself by defining transport barriers which attract or repel material, without allowing for propagation through them. Generally, Lagrangian descriptions are prone to large errors and uncertainty due to non-linear and turbulent oceanic and atmospheric flow fields. Although extensive studies on LCSs have previously been conducted, few studies investigate the implications of this inherent uncertainty in chaotic flow fields on LCSs. This study investigates the sensitivity of LCSs to uncertainty in the flow field of realistic Oceanic General Circulation Models. Two flow systems are considered: 1) A simplified controlled analytical double-gyre and 2) turbulent velocity data from the Barents-2.5 EPS model, simulating realistic ocean conditions in the Barents Sea and off the coast of northern Norway. The coastal region around the Lofoten-Vesterålen islands in northern Norway is chosen as the study region due to its ecological importance. The Barents-2.5 EPS includes 24 realizations of the same scenario, each with either perturbed initial conditions or forcing that can lead to large differences in the flow field. An ensemble of the double-gyre system is also produced by perturbing the dependent variables. I develop an LCS detection software utilizing the Finite-Time Lyapunov Exponent approach. The software and resulting LCSs are verified by computing LCSs in both flow systems. I then investigate whether these correspond to independently simulated particle trajectories to study their effect on material transport. Then, LCSs are computed in all ensemble members of the double-gyre ensemble and Barents-2.5 EPS. Following variations in the velocity fields between ensemble members, LCSs vary between ensemble members. Robust LCSs are LCSs predicted by the majority of ensemble members for a particular time. Averaging over ensemble members smooths out the LCSs, but a few clearly distinguishable LCSs are detected in the average, thus these are considered to be highly robust. These are most commonly formed in regions where the current is steered by geomorphological features, which are constant between ensemble members. For the double-gyre system, these include the system boundaries and the separation between the two gyres. For the Barents-2.5 EPS, these features include coastlines and bathymetry. LCSs are time-dependent and only valid for the time interval they are computed over. Flow structures in the real ocean can form and dissipate quickly, thus LCSs can do so just as quickly. Their persistence, i.e. existence over time, is therefore investigated to study whether LCSs exist long enough to have an influence on nearby material transport. Persistence can change depending on time-scales and has been studied over three time periods in the Barents-2.5 EPS: i) daily, ii) monthly and iii) seasonal. Daily variations in LCSs are investigated in the straits between islands in the domain. It is known that strong periodic currents form in these straits due to tidal flow. LCSs are revealed to form around these straits and swap east-west positions periodically. This periodicity is shown to be connected to the tidal phase, thus a daily periodic persistence dominated by tides is uncovered. Monthly persistence has been investigated for April 2022. I find that averaging over time heavily smooths out the LCS field, more than when averaging over ensemble members. This most likely happens because LCSs emerge, drift and decay. As a result, no distinguishable individual curves representing LCSs are detected in the time average. Instead, large high-value regions in the smoothed average reveals locations where LCSs frequently form over the time period. Similarly to robustness, these locations are also primarily dominated by geomorphology. It is also shown that there exists examples of robust features which are not persistent. To investigate seasonal variations, LCSs from April 2022 were compared to LCSs from October 2022. LCSs are shown to regularly form at different locations in the domain in the two seasons. In April, LCSs are frequently formed along the continental slope and northern and southern tips of Lofoten-Vesterålen. In October, LCSs tend to form along the coastline, whereas few LCSs form on top of the continental slope. This is most likely due to stronger and less topographically-steered currents in October. Furthermore, the main current flows closer to the coast in October, whereas there is an indication of more frequent small-scale flow structures forming and breaking off from the main current in April. Annual variability requires further investigation.