High-Resolution Modelling of Climate Variability

Abstract

Most climate models have a 1◦ (100 km) horizontal resolution. This resolution is too coarse to resolve mesoscale processes in the ocean: ocean eddies. An ocean eddy is characterised by a swirling and turbulent fluid. To resolve ocean eddies within a climate model, a horizontal resolution of 0.1◦ (10 km) is required. Climate models with a horizontal resolution of 1◦ and 0.1◦ are referred to as low-resolution models and high-resolution models, respectively. Ocean eddies are relevant to the ocean circulation as they stir the (upper) ocean and contribute to the transport of heat and salt. Model biases are reduced in climate simulations in which ocean eddies are explicitly resolved. In low-resolution climate models, eddy-related processes (transport and mixing) are parameterised at the cost of losing eddy characteristics within the model. As a result, the ocean circulation appears to be laminar in low-resolution climate models, sometimes referred to as the ‘honey ocean’. In this thesis, model output (300 years) of a high-resolution version of the Community Earth System Model (CESM) is analysed. We explore the following research questions in this thesis: 1) Is a high-resolution version of the CESM capable of capturing climate variability (sub-annual - multidecadal) as seen in observations? 2) Are sea-level projections different between the high-resolution CESM and low-resolution CESM? For the first research question, the analysis is restricted to the Caribbean Sea and the Southern Ocean. The simulated sub-annual ocean variability matches well with observations in the Caribbean Sea. This sub-annual variability is related to ocean eddies. Apart from sub-annual variability, multidecadal variability is found in the Caribbean Sea and surroundings in the high-resolution CESM. This multidecadal variability is induced by ocean eddies in the Southern Ocean and this variability propagates through the entire ocean circulation. From observations, it is known that multidecadal variability exists in the Southern Ocean. However, these observational records are too short (about 30 years) to verify the simulated multidecadal variability in the high-resolution CESM. The low-resolution version of the CESM does not represent ocean eddies, hence, the sub-annual and multidecadal variability is not resolved by this model. In addition to climate variability, model biases are reduced in the high-resolution CESM compared to the low-resolution CESM. For example, the ocean temperature and sea-ice concentration of the Southern Ocean are realistically resolved in the high-resolution CESM. The ocean temperature distribution of the Southern Ocean controls the amount of mass loss (through basal melt) of the Antarctic ice sheet. In an idealised forcing scenario in which atmospheric CO2 increases over time, the oceanic temperature of the Southern Ocean increases much slower in the high-resolution CESM compared to the low-resolution CESM. This slower temperature increase in the high-resolution CESM is related to ocean eddies. Consequently, the projected global mean sea-level rise is 25% lower in the high-resolution CESM with respect to the low-resolution CESM. Moreover, ocean eddies can affect regional sea-level projections. This demonstrates that both global and regional sea-level projections strongly differ between the high-resolution CESM and low-resolution CESM.