Energy balance and ocean Interactions in idealized future projections based on reduced-complexity climate models

Pfister, Patrik L. (2017). Energy balance and ocean Interactions in idealized future projections based on reduced-complexity climate models (Unpublished). (Dissertation, Universität Bern, Philosophisch–naturwissenschaftliche Fakultät, Physikalisches Institut, Abteilung für Klima– und Umweltphysik)

Text pfister17phd.pdf - Other Restricted to registered users only Available under License BORIS Standard License. Download (36MB)

Fossil fuel burning and deforestation over the industrial period have increased the atmospheric concentration of carbon dioxide (CO2) to a level unprecedented in the past hundred thousands of years. In conjunction with human-made emissions of other greenhouse gases, this has caused an increased trapping of longwave radiation in the atmosphere (anthropogenic greenhouse effect). This energy imbalance has caused global warming, although most of the excess heat has been absorbed by the ocean and some also by the cryosphere (ice and snow). Consequently, the thermal expansion of the ocean and the melting of ice sheets and glaciers have caused sea-level rise. CO2 emissions also cause ocean acidification and further impacts on the Earth System.
Knowledge of these past changes is based on independent evidence from direct observations and climate archives such as tree rings and ice cores. However, global CO2 emissions have rapidly increased during the recent decades; and even if we were able to suddenly stop emissions, global warming may continue for decades to centuries and sea-level rise even for millennia, due to the inertia of the Earth System and the long atmospheric lifetime of CO2. To assess these future climate change commitments, Earth System models are required. These models can reproduce past Earth System changes based on physical and biogeochemical calculations, and project future changes based on scenarios of how emissions of CO2 and other anthropogenic forcings may change in the future.
In this thesis, we carry out such future projections using an Earth System model of intermediate complexity (EMIC), the Bern3D-LPX model. The computational efficiency of this model allows us to carry out multi-millennial projections, as well as large numbers of simulations to assess the model’s sensitivity to configuration and parameter changes. Idealized future scenarios are used to enable intercomparison with other EMICs and more complex models. Our analysis mainly focuses on the physical response of the ocean-atmosphere system to changing CO2 concentrations. For this purpose, global and regional energy balance models are used as diagnostic tools to understand physical differences between models, or between different configurations of the Bern3D-LPX model.
The main content chapters of this thesis can be roughly divided into two parts, both based on idealized future projections. In the first part (Chapters 3 and 4), the degree of idealization is lower, and quantitative estimates of future Earth System changes in response to human-made CO2 emissions are presented. In the second part (Chapters 5 to 8), fully idealized scenarios are used to gain process-oriented insight into the physical model response to CO2 forcing.
The Introduction of this thesis gives a broad overview of the aspects of climate change research indicated by the thesis title: Earth’s energy balance and its interactions with the ocean, idealized projections of future climate change, and the reduced-complexity climate models we use to carry out such projections. Section 1.1 mainly summarizes well-established knowledge of the energy balance and the human influence on the climate system, based mostly on results presented in the fifth Assessment Report (AR5) of the Intergovernmental Panel on Climate Change (IPCC). Subsequent sections contain more specific information, including a description of the idealized scenarios, climate change metrics, and models used throughout this thesis. The most widely used metric is the equilibrium climate sensitivity (ECS), defined as the equilibrium warming in response to a CO2 doubling. Ocean interactions are introduced last, as they require an understanding of the aforementioned processes. These sections also include reviews on important literature beyond AR5. More specific literature reviews are given in the separate introductions to each of the main content chapters.
Chapter 2 contains a brief description of the Bern3D-LPX model, which is the main EMIC we use throughout this thesis. In addition, differences in the physical response between the current model version and an older version used in previous studies are illustrated and discussed.
In Chapter 3, we quantify the Earth System commitments caused by delaying CO2 emission reductions. We use the Bern3D-LPX model to estimate by how much future warming increases if reductions are delayed by 10 years. This additional warming per decade of delay is the Mitigation Delay Sensitivity (MDS). In agreement with an earlier analytical calculation, our model results estimate an MDS of 0.3–0.7 ◦C per decade, depending on the achievable rate of emission reductions (0.5–5% per year). This estimate quantifies the peak committed warming, which is reached decades to centuries after emission reductions start. As long as emission reductions are delayed, this committed warming thus increases much faster than the observed warming of 0.08–0.14 ◦C per decade, as illustrated by an additional Figure in Section 3.3. We extend the concept of MDS to steric sea level rise (SSLR) and two metrics of ocean acidification. The simulated MDS of millennial SSLR is also much higher than the observed SSLR, even more so than for temperature due to the inertia of ocean heat uptake. One of the ocean acidification metrics is the extent of surface ocean areas with ideal chemical conditions for coral reef growth. A near-complete loss of such areas becomes imminent if emission reductions are delayed by few years to decades, again depending on the achievable reduction rate. In addition to the reduction rate, MDS also scales with the prior rate at which emissions increase, and with ECS (only for temperature and SSLR). In summary, policy decisions today and during the next few decades largely determine the future state of the Earth System.
In Chapter 4, we present an international collaboration of modeling groups assessing the 10,000-year global mean sea-level rise (GMSLR) in response to human-made CO2 emissions. Two versions of both the Bern3D-LPX and UVic EMICs were forced by idealized emission scenarios with total cumulative emissions of 0–5120 GtC after the present reference period (1980–2004). This yielded four-EMIC mean estimates of the thermal expansion contribution to GMSLR, as well as estimates of global warming. Using these warming estimates in conjunction with independent ice-sheet models, the GMSLR contributions from melting of the Greenland and Antarctic ice sheet as well as glaciers were projected. Summing all contributions, GMSLR amounts to ∼ 25–52 m for emissions of 1280–5120 GtC after present, corresponding to a warming of ∼ 2–7.5 ◦C after present (add ∼ 0.7 ◦C for warming since preindustrial). GMSLR in low-emission scenarios is highly non-linear: The human carbon footprint of about 470 GtC by year 2000 has already committed Earth to a GMSLR of ∼ 1.7 m, but the release of another 470 GtC will result in a further committed rise of ∼ 9 m (mainly due to Antarctic ice sheet melting). The 10,000-year time scale of these estimates is much longer than in previous GMSLR assessments, fully capturing the slow response of oceans and ice sheets. The projected GMSLR is put into context by comparing it to GMSLR over the past 20,000 years. The uncertainty with respect to ECS is assessed based on additional Bern3D-LPX simulations, and regional sea-level rise and its impacts for coastal megacities are also discussed. The study concludes that policy decisions should not only be influenced by the common end-of-century perspective, but also take into account the GMSLR and related impacts of today’s emissions over the next ten thousand years and beyond.
Chapter 5 is a brief introduction to the Carbon Dioxide Removal - Intercomparison Project (CDR-MIP), including a published meeting report from the first CDR-MIP workshop. Carbon Dioxide Removal (CDR) will be necessary to limit future warming to less than 1.5 ◦C, and probably also for 2 ◦C. Therefore, CDR-MIP simulations will assess the Earth System response to CDR and the efficacy of different CDR methods.
In Chapter 6, we combine a multi-model analysis of a published ensemble of EMICs with new simulations from the Bern3D-LPX model, to investigate whether the ECS is state-dependent in those models. Although the ECS is defined as the equilibrium warming due to CO2 doubling, it is commonly estimated by halving the warming due to CO2 quadrupling. Such scaling is inappropriate if the warming is non-linear with forcing, which is referred to as state-dependence. Most EMICs exhibit slight to substantial state-dependence. In 12 out of 15 EMICs, ECS is lower in response to the higher forcing, in contrast to the opposite behavior reported from more comprehensive climate models. In the Bern3D-LPX model, we show that the state-dependence is mainly due to the sea ice-albedo feedback in the Southern Ocean. Furthermore, ocean mixing adjustments delay the emergence of state-dependence by multiple centuries, highlighting the need for long model integrations.
Chapter 7 is a rather technical analysis of the time-dependence in the relationship between transient warming and ocean heat uptake (OHU) in the Bern3D-LPX model. In other models, this time-dependence has previously been attributed to a non-unitary OHU efficacy ε. ε quantifies that the cooling caused by a unit OHU is stronger than the warming caused by a unit CO2 forcing, due to the high-latitude focus of OHU. We extend an existing diagnostic regional EBM framework by including time-dependent local feedbacks, for example caused by retreating sea-ice regions due to warming. This yields a regional decomposition of ε, for which we present a revised physical interpretation that is also relevant for other climate models. Non-unitary ε was previously explained by changes in cloud feedbacks in comprehensive climate models, but we demonstrate that it can also be caused by a strong ice-albedo feedback as simulated by the Bern3D-LPX model. The extended efficacy decomposition reveals that this is mainly reflected in evolving warming patterns rather than changing local feedbacks. We back up this finding by performing a secondary efficacy decomposition and employing a corresponding version of the regional EBM, using bottom-of-atmosphere rather than top-of-atmosphere heat uptake patterns. Possible implications for other climate models are thoroughly discussed, and a process-oriented generalization of global diagnostic EBMs is proposed.
In Chapter 8, we analyze physical contributions to the spread in the realized warming fraction f∆T , in three published model ensembles and additional Bern3D-LPX simulations. f∆T is the fraction of equilibrium warming that is presently realized, diagnosed here at the time of CO2 doubling. It measures the degree of physical equilibration of the Earth System, which was shown to largely determine whether global warming continues after CO2 emissions cease. We analyze published output from an ensemble of comprehensive Earth System models (ESMs), an ensemble of EMICs, and a constrained parameter ensemble from an older Bern3D-LPX version. In all ensembles, f∆T decreases with increasing ECS, in agreement with earlier studies. Ensemble-mean f∆T is lower for the EMICs than for the ESMs, as pointed out earlier. However, we show that this difference is not inherent to the different model complexities, as it is partly due to the lower ECS of the EMIC ensemble, which is tuneable as demonstrated in the Bern3D-LPX. The remaining f∆T difference is mainly related to the OHU efficacy ε, which is unitary in most EMICs but larger in ESMs. In contrast to ECS, tuning ε is not straightforward. Building on Chapter 7, we vary ECS and ocean parameters in Bern3D-LPX to show that ε depends mainly on the availability and melting rate of Southern Ocean sea ice, but this finding may be model-specific. Building on Chapter 6, we show that ECS state-dependence needs to be taken into account to accurately estimate f∆T and ε. We disscuss the implications of these findings for EMICs and for long simulations of ESMs.
Finally, an outlook is given in Chapter 9. The appendices contain an earlier paleoceanographic application of the Bern3D model, a derivation and brief description of an error correction in the physical model core, and a german plain language summary of Chapter 3 intended for communication and outreach.

Interest & Impact

Downloads

2 since deposited on 22 Feb 2024
2 in the past 12 months

Citations

5 Citations in Web of Science ®
6 Citations in Scopus

Search

in Google Scholar™

Services

Actions (login required)

Edit item

Actions (login required)

Edit item