3.2 Implementing the experimental method with SimClimat: examples

3.2.1 Role of human activities in the observed recent global warming

The experimental method begins as usual with an observation, a question and a hypothesis.

• Observation: We observe that the Earth has warmed by about 1°C during the past 150 years.
• Question: How can we explain this warming?
• Hypothesis: The global warming is mainly caused by the increase in the concentration of greenhouse gases emitted by human activities, in particular $CO_{2}$ whose concentration has increased from 280 ppm to 405 ppm during the past 150 years.

In the case of the experimental method with numerical modeling, some additional steps are necessary before carrying out the experiments.

• Model choice: The model must be based on general physical equations and not on the above-mentioned observation or hypothesis. Otherwise, this is circular reasoning! In SimClimat's equations, nowhere is it written that a 125 ppm increase in $CO_{2}$ concentration induces a 1°C increase in global temperature. The equations just “say” that the $CO_{2}$ acts on the greenhouse effect, and that the greenhouse effect acts on the planet's radiative balance and therefore on the global temperature, with a lot of possible feedbacks that can modify the results (figure 9).
• Control experiment: The control experiment allows us to check the realism of the model compared to observations. Here, we perform a simulation starting from the pre-industrial era, lasting 250 years, with anthropogenic emissions of 2.5 GtC/year that lead the $CO_{2}$ concentration to increase up to the present-day concentration.
• Model validation: We check that at the end of the simulation, the temperature has increased by 1°C, consistent with observations (figure 10, red). Note that with SimClimat, we cannot easily prescribe time-evolving anthropogenic $CO_{2}$ emissions that would follow a realistic scenario. In these simulations, only the start and end of the simulation are analyzed.

Then the experimental method continues as usual with experience, result and conclusion.

• Sensitivity experiment: We run the same simulation as for control, but the $CO_{2}$ concentration remains constant.
• Result: We find that if the concentration of $CO_{2}$ remains constant, the overall temperature does not increase (figure 10, blue).
• Conclusion: We conclude that the observed global warming is caused by the increase in $CO_{2}$ concentration.

Figure 10: Screenshot of the results for a pre-industrial simulation with constant $CO_{2}$ concentration (blue) and with anthropogenic emissions that lead to the current $CO_{2}$ concentration (red). The green simulation is identical to the red one, except that the water vapor feedback has been disconnected by keeping the water vapor concentration constant. Note that for the $CO_{2}$ concentration, the green curve is hiden by the red curve.

3.2.2 Climate feedbacks at play in the recent global warming

We demonstrate in the previous section that the global warming is caused mainly by the increase in the $CO_{2}$ concentration. Does $CO_{2}$ act directly on the greenhouse effect? Or are there any amplifying feedbacks? We show here how to implement the experimental method with SimClimat to quantify the role of the water vapor feedback.

• Observation: The gas that contributes most to the natural greenhouse effect is water vapor.
• Question: Does water vapor play any role in global warming?
• Hypothesis: As the temperature increases, the humidity in the atmosphere also increases (according to the Clausius-Clapeyron relationship). In turn, the enhanced greenhouse effect associated to the water vapor leads to increased temperature.
• Model choice: SimClimat, whose representation of water vapor is based on physical equations.
• Control experiment: We run a 250-year simulation from the pre-industrial world to present-day, with anthropogenic emissions of 2.5 GtC/year that lead the $CO_{2}$ concentration to increase up to the present-day concentration (figure 10, red).
• Model validation: We check that at the end of the simulation, the temperature has increased by 1°C, consistent with observations.
• Sensitivity experiment: We run the same simulation as for the control, but we unplug the water vapor feedback by keeping the water vapor concentration constant.
• Result: We find that if the $H_{2}O$ concentration remains constant, the overall temperature increases less: 0.6°C only instead of 1°C (figure 10, green).
• Conclusion: We conclude that water vapor is involved in a positive feedback that contributes 40% to global warming.

Similarly, the role of other climate feedbacks can be highlighted by SimClimat. For example, by unplugging the surface albedo feedback, we can see that this feedback is positive but remains rather weak at short time scales. Finally, by unplugging the role of the ocean or vegetation in the carbon cycle, we can see that the increase in temperature is stronger. The concentration of $CO_{2}$ also increases faster. This shows that the ocean and vegetation partially mop up $CO_{2}$ human emissions, by about half.

3.2.3 Mechanisms at play in glacial-interglacial variations

Glacial-interglacial variations are characterized by large variations in temperature, in ice sheet extent, and in sea level, which can be observed in various paleoclimate records
([Masson-Delmotte and Chapellaz, 2002,Masson-Delmotte et al., 2015]). 21,000 years ago, the Earth underwent the last glacial maximum. The overall temperature was 5°C colder, a polar cap covered all of Northern Europe, and the sea level was 130 m lower. For 10,000 years, we have been in an interglacial period. There is an inter-glacial period every 100,000 years (Figure 11).

Figure 11: Variations in temperature and $CO_{2}$ concentration recorded in Vostok ice core in Antarctica.

Here we propose to implement the experimental method in three steps to understand what causes glacial-interglacial variations.

 

Step 1: role of orbital parameters

• Observation: The time scales of temperature variations during inter-glacial variations are of the same order of magnitude as those of orbital parameters: obliquity (about 40,000 years), precession (about 20,000 years), eccentricity (about 400,000 years).
• Question: Can variations in orbital parameters lead to temperature variations consistent with those observed during glacial-interglacial cycles (i. e., 5°C)?
• Hypothesis: Yes. Let's take the obliquity as an example.
• Model choice: SimClimat, in which the effect of orbital parameters is described by physical equations.
• Control experiment: A simulation of 100,000 years is carried out starting from the pre-industrial world, all parameters being left at their default values. A sufficiently long simulation is necessary so that the ice sheet have time to reach equilibrium (figure 12, red).
• Model validation: The temperature remains constant at a value consistent with the observed global temperature.
• Sensitivity experiment: The simulation is the same as the control, but with the obliquity at its minimum value (figure 12, blue).
• Result: The temperature decreases by several °C. There is also a large increase in the ice sheet extent, and a decrease in the sea level of the same order of magnitude as observed for the glacial period.
• Conclusion: We conclude that obliquity variations can lead to temperature variations consistent with those observed during glacial-interglacial cycles.

The same approach can be applied to the other orbital parameters.

Figure 12: Screenshot of the results of a pre-industrial control simulation of 100,000 years (red), with minimal obliquity (blue), with minimal obliquity and constant albedo (green) and with minimal obliquity and the $CO_{2}$ solubility in the ocean that does not depend on temperature (purple). Note that in panels where the green and purple curves are absent, they are actually hidden by the red curve.

 

Step 2: role of summer insolation in polar regions

• Observation: When we modify the orbital parameters, we do not modify the global-mean, annual-mean incoming solar energy. Orbital parameters only change the distribution of incoming energy as a function of latitude and season.
• Question: How can orbital parameters change the global temperature?
• Hypothesis: By acting on the incoming energy in the polar regions in summer, the orbital parameters modulate ice sheet melting. In turn, the extent of polar ice sheets influences the planetary albedo and thus its temperature.
• Model Choice: SimClimat.
• Control experiment: The previous 100,000 year-long experiment with minimal obliquity (figure 12, blue).
• Model validation: The temperature decreases consistently with a glacial period.
• Sensitivity experiment: The simulation is the same as the control experiment, but the albedo feedback is “unplugged” by setting the albedo to a constant, pre-industrial value (figure 12, green).
• Result: The temperature remains constant.
• Conclusion: We conclude that the modification of the albedo is responsible for the modification of the temperature when the obliquity decreases. As the obliquity decreases, the sun's rays arrive more inclined in boreal polar regions in summer. This prevents ice sheet melting, and thus promotes its extension. This increases the planetary albedo and therefore decreases the temperature.

The same mechanism applies to other orbital parameters. The obliquity is the easiest parameter to understand: if the polar axis is more inclined, in boreal summer the sun rays hit more perpendicularly the Northern polar regions. It favors the melting of the ice sheet. Precession acts on the season for which the Earth is closest to the sun. Presently, the Earth is closest to the sun in boreal winter. If, on the contrary, the Earth is closer to the sun in boreal summer, then the Northern polar ice sheet receives more energy in summer, which favors its melting. Eccentricity is the most complex parameter because its effect depends on precession. For the present precession where the Earth is furthest from the sun in boreal summer, if the orbit becomes more eccentric, the Earth will be even further away from the sun in summer. The Northern polar ice sheet will then receive less energy in summer which favors its extension.

Note that what is important here is the energy received by the Northern polar ice sheet and not the Southern polar ice sheet (i.e. Antarctica). This is because the Northern polar ice sheet is free to extend over Europe, Siberia, North America. On the contrary, the Southern polar ice sheet is limited to the Antarctic continent and can not extend over the Southern Ocean.

 

Step 3: Why does the $CO_{2}$ concentration decreases during the glacial period?

Air bubbles trapped in ice cores show that changes in $CO_{2}$ concentration co-vary with temperature during glacial-interglacial variations (Figure 11). Why?

• Observation: When the temperature decreases, the $CO_{2}$ concentration decreases. At the last glacial maximum, the $CO_{2}$ concentration was 100 ppm lower while the global temperature was 5°C lower.
• Question: How can we explain this decrease in $CO_{2}$ concentration?
• Hypothesis: When the oceans are colder, the $CO_{2}$ solubilizes more easily.
• Model choice: SimClimat.
• Control Experience: The previous 100,000 year-long experiment with minimal obliquity (figure 12, blue).
• Model validation: The $CO_{2}$ concentration simulated by SimClimat decreases as temperature decreases, down to values of the same order of magnitude as those observed for the last glacial maximum.
• Sensitivity experiment: The simulation is the same as the control simulation, but the $CO_{2}$ solubility is set to a constant value whatever the temperature (figure 12, purple).
• Result: The $CO_{2}$ concentration remains constant. In addition, the cooling is reduced.
• Conclusion: The colder the oceans, the higher the $CO_{2}$ solubility. A larger fraction of the atmospheric $CO_{2}$ is thus dissolved into the ocean. Therefore the atmospheric $CO_{2}$ concentration decreases. Since $CO_{2}$ is a greenhouse gas, decreasing the atmospheric $CO_{2}$ concentration amplifies the cooling: it is a positive feedback.