Kooiti Masuda
[The following is the author's affiliation as of 2007]
Frontier Research Center for Global Change,
Japan Agency for Marine-Earth Science and Technology.
3173-25 Showa-machi, Kanazawa-ku, Yokohama 236-0001, Japan.
e-mail: masuda(at)jamstec.go.jp
This paper was published as follows,
and its copyright is held by the Japan Association for Quaternary Research
(http://www.quatenary.jp).
Please mention the publication (rather than this web page)
if you would like to cite this article.
(This article is written in English, though most of the papers
in the regular issues of the same journal are in Japanese.)
The issue of what the modern world call `global warming' is the climate change primarily due to carbon dioxide released by burning of fossil fuel. Carbon dioxide has a capability to absorb and emit infra-red radiation, and thus act as a forcing to the climate system. It is expected that climate will warm responding to the forcing. The temperature response will be delayed by several decades because of the heat capacity of the upper ocean, and the sea-level response will be delayed by a millennia because of the response times of the whole ocean and the continental ice sheets. As the climate warms, both precipitation and evaporation will increase in the global average sense, but the increase is not so strong as the increase of saturation specific humidity. On the other hand, local and short-term extreme values of precipitation rate will surely increase. Distribution of precipitation will be more heterogeneous. Paleoclimatological evidence suggests that there are possibilities of rapid climate changes which are not well represented by current numerical climate models used in projection experiments. We should not deny the possibilities of such `surprises', but we are very uncertain about their probability. It would be wise approach to primarily prepare for the climate responses as projected by an ensemble of current state-of-art simulation models, and to secondarily prepare for various uncertain events which include climate `surprises'.
global warming, greenhouse effect, response to radiative forcing, time constant, abrupt climate change
In this article, I review the current understanding about how the climate is likely to change in the coming century, and how the hydrological cycle is likely to change accordingly.
The issue which is recently called `global warming' can be summarized as follows. (Though atmospheric constituents other than carbon dioxide (CO2) also contribute to the issue, their contributions are not mentioned here for simplicity of argument.)
There are many books on the issue. Pittock (2005) is one of the recommendable introductory books on the whole issue.
In view of the Quaternary history of climate (e.g. Part 3 of Lovejoy and Hannah, 2005; Oldfield 2005), the rate of expected change is not exceptional, though the average warmth of the coming century may be unprecedented within the Quaternary. It suggests that global warming by itself may not be a great threat to the global biosphere and the human society. However, human activity has concurrently made extensive land use change, which has severely reduced adaptability of plants and animals by way of migration (Lovejoy and Hannah, 2005). Also, adaptability of humans by way of migration has also been reduced mainly because of increased population and tightened property rights on land. I think that we should view the current global environmental issue as climate change compounded with land use change, and that this is a serious issue with respect to sustainability of the biosphere and also to sustainability of the human society which depend on the biosphere.
Anthropogenic CO2 can be regarded as forcing to the climate system. If we can assume that the climate changes responding to the CO2 forcing, and that the way of response is smooth (though not necessarily linear), we can make quantitative projection of the climatic changes, for globally averaged quantities at least. There is large uncertainty in the estimation of future social factors which determines CO2 emissions. (Therefore we usually use the term `projection' in this context rather than `prediction'.) There are also uncertainties in natural processes, for example, feedback from ecosystems to atmospheric CO2 by way of increase of soil respiration. Nevertheless, this assumption enables us to explicitly state the probable range of future climate. The capacity of the scientists to make such projections have increased significantly during the recent three decades.
On the other hand, the climate system is a non-linear system which includes many feedbacks. Moreover, some of the feedbacks are not yet taken into models. Paleoclimate reveals cases of abrupt climatic changes, such as Dansgaard-Oeschger oscillations which occurred during the last glacial period (National Research Council, 2002; Cox, 2005). We cannot deny that unexpected abrupt changes may occur in the future as well. Oeschger called them `climate surprises' (Cox, 2005). Some people claim that reduction of CO2 is worthless because such abrupt changes may occur without human intervention. But there is another view that the forcing by anthropogenic CO2 will increase the probability of abrupt changes. The situation may be something like tilting balance scales, according to this view (see discussions in National Research Council, 2002, and Schellnhuber, 2006). We cannot objectively determine which view is correct. I subjectively favor the latter view, but not so strongly.
Currently, projections of global warming are mainly based on time-dependent simulations with comprehensive numerical models of the climate system (atmosphere, ocean and land surface combined) and several scenarios of emission of CO2 etc. Comprehensive reviews are made by Intergovernmental Panel on Climate Change (IPCC, e.g. Watson et al., 2001). Recent progress in Japan is reviewed by Koike (2006).
But, my focus here will be on more idealized (less realistic) studies which are useful for understanding of projections.
First we consider `steady-state response' of the climate to different CO2 concentrations. That is, difference of long-term average conditions of climate between alternative worlds in which a certain constant CO2 concentration prevails for each. The accumulated knowledge based on various numerical modeling studies can be summarized as follows. The response of global mean surface air temperature is roughly proportional to the logarithm of CO2 concentration. In other words, temperature increases by a constant amount each time when CO2 doubles. It is likely that the value of the constant is between 1.5 and 4.5 degrees Celsius. It was suggested by National Academy of Sciences (1979) with the few evidence available then, and seemingly right in the light of more information (Annan and Hargreaves, 2006).
Next, we have to consider that the actual climate is in transient, rather than steady, states. Compared with the evolution of hypothetical steady-state response to the contemporary CO2 concentration, the actual response will be delayed. The way of delay may be approximated, as valid in linear systems, by the exponential function e{-t / τ} of time t, where τ is the time constant. One time constant is not enough, however. When we consider the surface air temperature, the sea surface temperature, or phenomena in the atmosphere, the time scale of the delay is several decades, because the evolution of atmospheric temperature is likely to accompany that of the upper layer (about 500 m deep) of the ocean. It was demonstrated by mechanistic model experiments by Spelman and Manabe (1984). (If we also consider spatial inhomogeneity, the delay will appear smaller in the continental interior than over the ocean.) On the other hand, when we consider the sea level, the time scale of the delay is centuries to a millennium, because it is mainly determined by warming-up of the whole ocean (about 4000 m deep) and by the mass balance of ice sheets.
Some of `climate surprises' are such events that no one has yet imagined. But some are already envisioned as cause-and-effect chains, though we cannot yet give quantitative probability (and I subjectively consider that the probability is low). Here are some examples of the latter.
Some of the suggestions are based on reconstruction studies of the past climate, and others are based on theoretical studies of processes which are not yet fully taken into account in the projection experiments.
It is considered that the thermohaline circulation of the deep ocean may have a few different regimes, and some gradual change of forcing may kick it from one regime to another. For example, global warming likely leads to increase of precipitation of high latitude in the North Atlantic, which may act to reduce subsidence of seawater there. As a result, the overturning circulation of the deep ocean may stop, which may cause cooling of Europe despite of global warming. Many discussions on this subject are found in the conference proceedings of Schellnhuber (2006). Different types of regime changes can be supposed if we also consider the subsidence near Antarctica (Maslin, 2004).
The time scale of the change (as `response') of continental ice sheets will be millennia if it is governed by mass balance at the surface. But, it is suggested that it can be much faster if it is governed by dynamics. Global warming may make the ice shelves (the part of the ice sheet which is grounded on the sea floor) unstable. If the ice shelves are broken, the upstream ice flow may accelerate. Some scientists suggest that sea level rise of several meters may occur in centuries by this mechanism. Reviews on this topic are found in Alley et al. (2005) and Schellnhuber (2006).
Methane hydrate is a kind of ice which holds molecules of methane gas. It exists under the sea floor as well as in the permafrost. If it is broken down, methane will escape to the atmosphere. Since methane is another absorber and emitter of infra-red radiation, it will accelerate global warming. It is likely that methane hydrate in continental permafrost will decompose due to warming. This process should now be considered as a kind of `response', though it has not yet been incorporated into the standard scheme of projection.
On the other hand, decomposition of methane hydrate under the sea floor will generally be suppressed by global warming, since the increase of the mass of sea water due to melting of continental ice will increase pressure at the sea floor. But, there is a suggestion that decomposition may be enhanced in some setting. In the region relatively near the ice sheets which will be melting, the sea floor will rise as isostatic adjustment to the lost weight of ice, and thus the pressure at the sea floor will be reduced (Maslin, 2002).
A simulation of the 21st century climate coupled with vegetation dynamics with a numerical model called ``HadCM3LC'' shows that the climate in the Amazon river basin becomes so dry that it cannot support forests (Cox et al., 2004) . Such `dieback' of forests also means CO2emission which accelerates global warming. The sensitivity of vegetation to global warming in HadCM3LC is particularly large among other models according to an intercomparison study (Friedlingstein et al., 2006). Maybe the Amazonian `dieback' is peculiarity of that model, but the possibility of large change of ecosystems is not ruled out.
Most of current numerical projections shows that the warmer climate of the late 21st century still have the irregular oscillation of the tropical ocean-atmosphere system known as ENSO, though the average state may be shifted a little towards El Niño side. On the other hand, paleoclimatologists show that ENSO was not found or much weaker than today in the early Holocene (as reviewed by Gagan et al., 2004, also by Oldfield, 2005). It is likely that the warmth of early Holocene is limited in the northern hemisphere and explicable by the orbital parameters of the earth. Therefore warmer climate does not logically imply halt of ENSO. As a `surprise' scenario, however, it is imaginable that the state of the tropics is fixed at either El Niño or La Niña conditions.
Changes in the hydrological cycle associated with global warming is less certain than the changes in temperature. Some broad features can be stated based on simple reasoning, however. I do not discuss `surprises' here but concentrate on `responses'.
The saturation specific humidity, that is the amount of water vapor that the air can contain, increase according to temperature. The relationship is like an exponential function. If the global mean air temperature increases, it is certain that the specific humidity just above the sea surface, which can be assumed as saturated, will increase accordingly. Also it is projected that the total water vapor content in the atmosphere will increase nearly proportional to the increase of the saturation specific humidity (i.e. retaining relative humidity). Note that water vapor is an important absorber and emitter of infra-red radiation. The above mechanism of increase of water vapor acts as a positive feedback to the change of temperature. This feedback is already incorporated in the models which are used for projection studies (see Soden and Held, 2006).
As the climate warms, global mean precipitation and evaporation will increase. (In this sense, it can be said that the hydrological cycle will be accelerated.) But the response of evaporation, or of global mean precipitation, to temperature will not be so strong as that of saturation specific humidity, because these variables are controlled by the surface energy balance as well. Accordingly, the recycling of water vapor in the atmosphere will decelerate. In other words, the mean residence time of water vapor will increase. (In this sense, it can be said that the hydrological cycle will be decelerated.)
On the other hand, local, short-term precipitation, which can be supported by horizontal advection of water vapor and is not restricted by the surface energy balance, may increase at a rate similar to that of saturation specific humidity. Therefore, precipitation tends to concentrate temporally as well as spatially (Kitoh, 2005). Roughly speaking, the pattern of rainfall in the temperate zone tends to resemble that in the tropics. This analogy is not accurate because the difference between the tropics and the temperate zone is caused not only by temperature but also by the rate of rotation of the earth projected to the local zenith.
Emori and Brown (2005) examined precipitation in warmer climate of the late 21st century and the present climate according to projections by many models. They notice that increase of precipitation in heavy rain events is more significant than the increase of mean precipitation.
Water available on land is important for human society, especially for agricultural production. We should rely on flow rather than stock of water in order to live sustainably. Thus the key variable is precipitation minus evaporation, which can be equated with runoff in the annual mean condition.
Some characteristics of the change seem favorable to us. The distribution of zones of atmospheric circulation will be largely intact. The amount of flow will increase somewhat, in the global sense.
But some characteristics seem hazardous. The place and time where precipitation concentrates is likely to be where there is already too much water, increasing flood hazards while not increasing usable water resource. On the other hand, precipitation may not increase over many continental regions other than a selected few. But, potential evaporation will increase according to temperature. Thus, available water on land will decrease there.
Nohara et al. (2006) examined river runoff in model projections. Averages of results from many models suggests that runoff generally decreases in continental Europe, western North America, temperate South America and southern Africa, while it increases in north cold regions and South and East Asia. Difference among model projections is still large, however.
As a matter of warning rather than projection, I think that we should care about possible droughts in the wheat and pastoral zones of North America and Australia, as well as floods in rice-producing regions of South and Southeast Asia.
Another simple consequence of warming is the fact that snow turns to rain. The change result in the change of seasonal patterns of runoff, river discharge and water resource availability. In very cold regions of the Antarctic and the Arctic, however, snow is likely to increase due to increased supply of water vapor.
Hosaka et al. (2005) examined snow mass in the projection by the model of the Meteorological Research Institute of Japan. According to the simulation, winter maximum snow mass will decrease in most of Europe, Canada and Japan, while increase in Siberia and Alaska. Even in the latter cold regions, however, snow mass which remains in summer will decrease.
Paleoclimatology, including studies of the Quaternary glacial cycles, was one of the most important motives that contributed to the progress of the present understanding of climate change (see Weart, 2003). Studies on abrupt climate changes remind us of the processes not yet fully incorporated in the current models for projection. In order to become more capable of prediction, it would be fruitful for modelers to collaborate more closely with Quaternary scientists who reconstruct the past environment.
Currently, our ability of outlook is limited in the sense that we do not yet know how much weight we should put on what we call `surprises' here. But we are relatively sure about what will happen if `surprises' do not happen. I think that it is wiser for us to have a two-tier strategy. We should prepare for probable climate changes based on projected `responses'. On the other hand, we should also prepare for more uncertain events including `climate surprises' as well as earthquakes, volcanic eruptions, etc.
This review article is essentially a record of my presentation at a symposium of JAQR held in 6 August 2006. Therefore it does not refer to the Fourth Assessment Report (AR4) of IPCC which is to be published in 2007.
http://www.jstage.jst.go.jp/browse/sola/), 1, 93 -- 96.