Arctic sea level and geostrophic flow: from the development of a satellite-based pan-Arctic dataset to the study of the seasonality from satellite remote sensing and model output

Abstract

The geographical and environmental conditions of the Arctic region make it one of the most remote, harsh and challenging regions on Earth to observe. Indeed, being covered by sea ice all year round, the Arctic Ocean is the least observed and therefore the most unknown component of the whole Arctic climate system. In particular, the sea ice cover hampers both the access by ships and the observations via remote sensing, thereby hindering long-term monitoring. Yet, the Arctic Ocean and the neighbouring Nordic Seas, play a crucial role in the cycling of fresh water and heat across the global ocean. Changes in the Arctic Ocean circulation have been induced by global warming, that modified the atmosphere-ice-ocean interface. Up to recent times, changes in the circulation in the ice-covered Arctic have been observed mainly via in-situ data (e.g., ship observations and moorings), which are sparse and mostly cover short periods of time. Strong seasonal biases make it difficult to integrate in-situ observations to provide a large-scale perspective on the variability and long-term changes. Since 2010, with the launch of the CryoSat-2 satellite altimetry mission and the development of techniques to process altimetry data from cracks in the ice, a new opportunity to observe the Arctic Ocean sea surface height and geostrophic surface circulation up to 88°N has opened. The overarching goal of my thesis was to investigate the state-of-the-art of satellite altimetry in the Arctic Ocean, and use altimetry data in combination with model simulations to provide a basin-scale assessment of seasonality of the Arctic sea surface height and geostrophic surface circulation, and its drivers. Despite the availability of re-processed altimetry data in ice-covered regions, only few experimental gridded, multi-year datasets are available to date. Moreover, these products have been scarcely evaluated in terms of ocean velocity and it is not yet clear how they compare to each other. I therefore used observations from the ice-covered Arctic newly processed at the Alfred Wegener Institute, in combination with observations from the ice-free Arctic, to develop a new quality-controlled gridded pan-Arctic dataset of sea surface height and geostrophic surface velocity. The dataset, based on CryoSat-2 observations, extends up to 88°N and covers a period of 10 years (2011-2020). Both the sea surface height and the geostrophic velocity fields were evaluated in the ice-covered Arctic Ocean by comparison with in-situ data. Additionally, sea surface height was compared over the entire Arctic with an independent satellite altimetry product to evaluate the impact of different methodologies (source data, corrections, gridding) on the final product. This comparison showed that, while different methodologies do not prevent a generally good agreement between monthly fields (correlation coefficient higher than 0.7 over 85\% of the domain), local differences between the two datasets can be attributed to different corrections applied. Results of the evaluation of the sea surface height fields against distributed hydrographic profiles data demonstrate that the mean field is consistent with known large-scale circulation patterns, and that these are also preserved in the transition between ice-covered and ice-free areas. Furthermore, monthly time series of sea surface height are compared to the sum of steric plus bottom pressure equivalent height from mooring data, showing a relatively good agreement (correlation coefficients larger than 0.5, with p-value lower than 0.06), with differences on a month-to-month basis due to the different sampling of mesoscale activity. Geostrophic velocity derived from the altimetry dataset were compared to near-surface velocity from a total of 26 moorings. Among these, two mooring arrays located across the Fram Strait and the Laptev Sea continental slope observe the structure and variability of the Arctic Boundary Current at two locations. Comparison to data from the mooring arrays showed that the highest correlation was achieved when both satellite and in-situ data are averaged over 50-60 km across the slope current and intra-seasonal frequencies are removed. This allowed to establish that the altimetry dataset is able to resolve Arctic slope current variability at seasonal and longer time scales, over across-current scales of about 50-60 km. Having demonstrated the capability of the altimetry dataset to resolve the Arctic slope currents seasonal variability, my second objective was to provide a basin-wide assessment of the large-scale variability and its drivers. This was achieved in the second part of my thesis using the gridded altimetry dataset in combination with model simulations. Large-scale seasonal patterns in both the sea surface height and geostrophic velocity fields were identified. These manifest in the form of out of phase sea level anomalies between the Eurasian shelf seas and the central deep basin, accompanied by a modulation of the geostrophic currents at the shelf break. As a result, slope currents are stronger in winter and weaker in summer along the entire Eurasian continental slope from the southern Norwegian coasts to the western Laptev Sea. By separating mass-related and density-related contributions to sea surface height seasonality in the model simulations, I was able to attribute the large scale pattern of shelf-basin decoupling to the mass contribution, therefore to cross-slope water mass transport. I then investigated the mechanisms regulating the cross-slope transport. While the wind field is in agreement with shoreward Ekman transport in winter and offshore transport in summer, a quantitative analysis of the equivalent height change over the shelf seas showed that this exceeds the observed sea surface height seasonal changes by one order of magnitude. The cross-slope transport is found to match the observed sea surface height seasonal changes when an additional compensatory cross-slope transport at depth, in geostrophic balance, is considered. Although the large-scale seasonal pattern was attributed in large part to mass-related variability, density-related changes were as prominent in confined regions, among them the Laptev Sea continental slope. By comparing the seasonal variability in the simulated vertical sections of velocity and density there, with the large-scale density-related and mass-related geostrophic velocity anomalies, I demonstrate how the latter can be complementary in identifying the character (barotropic mass-driven or baroclinic density-driven) and the origin of a shelf break current variability. In conclusion the work presented in this thesis demonstrated the capability of satellite altimetry to provide a basin-wide assessment of the Arctic large-scale seasonal patterns of sea surface height and circulation, proving in particular to be a valid tool to investigate the drivers of the slope current seasonality.