Climate Impact of Quadrupling CO2
An Overview of GFDL Climate Model Results
The following overview of GFDL climate model results is based on earlier versions of the GFDL climate models than are currently being used. Overviews of more GFDL climate model results can be found on our NOAA/GFDL Climate Research Highlights web page.
An overview of GFDL climate model results is presented from a series of experiments examining the possible climate impact of a quadrupling of atmospheric CO2. Much of the recent anthropogenic climate change research has been focused on the issues of climate change detection and projections of climate change over the next century. On the other hand, analyses of future emission scenarios in the IPCC and elsewhere indicate that on a multi-century time scale, CO2 levels are likely to rise well beyond a doubling unless very substantial emission reductions occur. Therefore, longer term aspects of climate change, based on higher-than-doubling CO2 levels, are becoming an increasing part of the debate. In this report, the possible climate impacts of a CO2 quadrupling are examined.
The results presented are based on several multi-century integrations of a GFDL global coupled ocean-atmosphere model. The model resolution is about 4 degrees latitude. In the quadrupled (4xCO2) experiment, CO2 increases at 1% per year (compounded) to a level four times that of the present climate. After this 140 year “ramp-up” period, the model is integrated for several more centuries to examine the long time response of the climate system, including the deep ocean. For comparison, results are also presented from the 2xCO2 experiment, in which CO2 increases by a factor or two in the first 70 years, and then is constant at 2xCO2 for several more centuries of model integration.
Before presenting the CO2 -quadrupling results, we first show in section 2.0 the results of a recent preliminary attempt to simulate the observed global-scale warming over the past century using the GFDL coupled climate model.
2.0 Simulation of global temperature changes since 1850
due to greenhouse gases and aerosols (Fig. 1)
The results presented in this report are crucially dependent upon the validity of the GFDL model’s climate sensitivity to increased CO2. The change in global surface air temperature predicted by the GFDL model is 3.7C for a doubling of CO2, which lies in the upper half of the range of 1.5 to 4.5C estimated by the Intergovernmental Panel on Climate Change (IPCC).
Figure 1 shows that despite the model’s relatively high climate sensitivity, its simulated global temperature change due to CO2 and aerosol forcing from 1850 to the present is roughly equivalent to the observed global temperature change. Although the aerosol forcing is highly uncertain and other climate forcings or natural variability may have contributed to the observed changes, the similarity between the modeled and observed global temperature curves suggests that the model’s
sensitivity is plausible in the light of currently available observations.
The results in Fig. 1 also indicate that the warming rate for the next 50 years is projected to be substantially higher than the rate simulated or observed for the past century. In fact, the anticipated relative future climate forcing contributions from anthropogenic sulfate aerosols and CO2 are such that the aerosols will become largely irrelevant on the multi-century time scale. However, it appears that aerosols will play an important role in the detection/attribution issues over the next several decades.
Maps of the projected increase of surface air temperature (in degrees Fahrenheit) from the 2xCO2 and 4xCO2 experiments with the GFDL climate model are shown in Fig. 2. The warming is projected to be particularly large over much of the mid-latitude continental regions, including North America and Asia. The temperature changes shown for the 4xCO2 experiment are almost as large as the difference between the present climate and that of the Late Cretaceous approximately 65-90 million years ago.
Along with this surface warming, sea ice coverage over the Arctic Ocean is projected to decrease substantially, as illustrated in Fig. 3. In this figure, mean sea ice thickness (in meters) in the model during late winter is shown for both the control and 4xCO2 experiments. The view is from above the North Pole; brown areas indicate the model’s land regions. During late summer (not shown), sea ice is virtually absent in the 4xCO2 experiment. See this research summary for more details on sea ice responses to increasing greenhouse gases.
In response to greenhouse gas warming, sea level is expected to rise due to the thermal expansion of sea water as the ocean warms. Because the deep ocean will warm much more slowly than the upper ocean, the thermally driven rise in sea level is expected to continue for centuries after atmospheric CO2 stops increasing. To illustrate, Fig. 4 shows the increase of global mean sea level in the GFDL 4xCO2 coupled climate model experiment. Even though CO2 no longer increases after year 140, sea level continues to rise steadily well beyond year 500. The final equilibrium sea level change in the model is 1.9 meters for a CO2 doubling (not shown) which is roughly the level attained in the CO2 quadrupling experiment after 500 years. The equilibrium rise for the quadrupling experiment has not yet been simulated.
The sea level rise projections in Fig. 4 are the expected changes due to thermal expansion of sea water alone, and do not include the effect of melted continental ice sheets. With the effect of ice sheets included, the total rise could be larger by a substantial factor. However, projections of the contribution of the ice sheets to future sea level rise are not presented here due to the difficulties of performing a credible calculation. For reference, the volume of ice in the Greenland and Antarctica ice sheets is equivalent to a sea level rise of 7 and 73 meters,
respectively (IPCC, 1995).
The sea level rise is not anticipated to be uniform over all regions of the globe due to the influence of ocean circulation changes, as well as land movements unrelated to global warming. A more realistic projection of the geographical distribution of sea level rise remains as a problem for future research. However, to crudely illustrate the effect of various hypothetical spatially uniform sea level rise scenarios, the red areas in Fig. 5 indicate regions of the southeastern United States that would be below sea level for regionally uniform rises of one, two, four, and
eight meters, respectively.
The global thermohaline circulation, sometimes referred to as the ocean’s “conveyor belt” is important because it is responsible for a large portion of the heat transport from the tropics to higher latitudes in the present climate. For example, the Gulf Stream in the Atlantic Ocean forms part of the global thermohaline circulation,
transporting warmer waters northward, thereby contributing to western Europe’s relatively mild climate for its latitude.
GFDL climate model simulations project that the global thermohaline circulation will decrease in intensity as greenhouse gas warming occurs, due to enhanced precipitation and runoff from the continents in high latitudes. In the 4xCO2 experiment, the thermohaline circulation essentially disappears in the GFDL model; in the 2xCO2 experiment, the thermohaline circulation initially weakens to less than half its original intensity, but eventually recovers to its initial strength after several centuries (Fig. 6). Further experiments
have shown that the rate of CO2 build-up has an important
effect on the evolution of the thermohaline circulation in CO2
doubling scenarios (Fig. 7). The faster the build-up of CO2, the greater the eventual reduction in the thermohaline circulation and the longer the delay in its recovery.
Paleoclimate records suggest that relatively large transitions of the ocean circulation have occurred in the past on time scales of a few decades. Whether such abrupt changes could occur in response to global warming is a subject of current scientific focus. Coupled climate models will be important tools for helping to address this issue.
The development and improvement of coupled ocean-atmosphere models necessary for studying these climate change problems has been a primary area of focus for the GFDL climate dynamics project. The task of successfully coupling together complex atmosphere and ocean models has proven to be a very challenging one for climate modelers and will continue to be an important research topic.
Soil moisture as simulated in climate models refers to the amount of moisture available over land areas for humidification of the atmosphere. A highly simplified parameterization of soil moisture is used in the present GFDL climate model. Nonetheless, the model simulates many of the observed large-scale climate features related to soil moisture content, such as major desert regions and moist temperate zones. Some persistent regional problems remain with these present-day simulations, including an excessively dry southeastern United States.
In response to increasing CO2, the GFDL model projects
substantial decreases in soil moisture over most mid-latitude continental areas during summer. For example, Fig. 8 shows the percent reduction in summer soil moisture at each model gridpoint over North America for the 2xCO2 and 4xCO2 experiments. Typical reductions are of the order of 50 percent in the 4xCO2 case. Such changes, if realized, could have a substantial impact on agricultural practices throughout many of the world’s important food-producing regions.
The development and application of more realistic soil moisture parameterizations in climate models is an area receiving current attention at GFDL through a research collaboration with U.S. Geological Survey scientists.
The heat index (also called “apparent temperature”) is a measure of the stress imposed on humans by elevated levels of atmospheric moisture. These conditions inhibit the ability of the body to dissipate heat, thereby causing discomfort. Specifically, for a given atmospheric temperature and moisture content, the heat index is the temperature that the body would “feel” if the moisture content were reduced to a predetermined reference amount. For example, an air temperature of 95F with a relative humidity of 60 percent translates to an apparent temperature (heat index) of 110F, thereby denoting the equivalent temperature that the body would feel with the reference
amount of moisture in the air.
Figure 9 shows time series of surface heat index values from three GFDL climate model experiments. The corresponding time series for surface air temperature are shown in Fig. 10. The black, blue, and red lines indicate results from the control, 2xCO2, and 4xCO2 experiments, respectively. The time series are plotted for a large region of the southern United States (79-97W, 31-40N). The differences between the control and enhanced CO2 heat index values in Fig. 9 result from increases in both air temperature
and moisture content.
It should be noted that temperatures predicted in the enhanced CO2 experiments can exceed the range for which the definition of a heat index is valid.
Possible changes in hurricane intensity or frequency in response to greenhouse gas warming are of concern because of the profound impact that these storms have even in the present climate. Similarly, El Nino has been linked to broad scale weather disruptions, such as flooding and droughts, over wide regions of the globe, and hence any change in its behavior in a warmer climate would be of concern.
The present-generation global climate models at GFDL simulate weak versions of both tropical storms and El Nino. These simulated phenomena do not appear to become either more frequent or significantly more intense in the present global models. However, it will be important to use higher resolution models with more realistic simulations of these phenomena to more reliably assess this issue. At GFDL, work is underway to develop higher resolution global coupled models for El Nino/climate change simulations.
Researchers at GFDL have used regional nested models — such the laboratory’s high-resolution Hurricane Prediction System — to explore the impact of greenhouse gas-induced climate warming on hurricane intensities. See the related web site for details.
For further information:
- Delworth, T. L., J. D. Mahlman, and T. R. Knutson, 1999: “Changes in heat index associated with CO2-induced global warming.” Climatic Change, 43(2), pp. 369-386.
J. M., R. J. Stouffer, R. T. Wetherald, S. Manabe, and V. Ramaswamy, 1997: “Transient response of a coupled model to estimated changes in greenhouse gas and sulfate concentrations.” Geophysical Research Letters, 24(11), pp. 1335-1338.
- Knutson, T. R., S. Manabe, and D. Gu, 1997: “Simulated ENSO in a global coupled ocean-atmosphere model: Multidecadal amplitude modulation and CO2 sensitivity”. Journal of Climate, 10(1), pp. 138-161.
- Manabe, S. and R. Stouffer, 1994: “Multiple century response of a coupled ocean-atmosphere model to an increase of atmospheric carbon dioxide,” Journal of Climate,
7, pp. 5-23.