A sensitivity study to global desertification in cold and warm climates: results from the IPSL OAGCM model
- First Online:
- 739 Downloads
Many simulations have been devoted to study the impact of global desertification on climate, but very few have quantified this impact in very different climate contexts. Here, the climatic impacts of large-scale global desertification in warm (2100 under the SRES A2 scenario forcing), modern and cold (Last Glacial Maximum, 21 thousand years ago) climates are assessed by using the IPSL OAGCM. For each climate, two simulations have been performed, one in which the continents are covered by modern vegetation, the other in which global vegetation is changed to desert i.e. bare soil. The comparison between desert and present vegetation worlds reveals that the prevailing signal in terms of surface energy budget is dominated by the reduction of upward latent heat transfer. Replacing the vegetation by bare soil has similar impacts on surface air temperature South of 20°N in all three climatic contexts, with a warming over tropical forests and a slight cooling over semi-arid and arid areas, and these temperature changes are of the same order of magnitude. North of 20°N, the difference between the temperatures simulated with present day vegetation and in a desert world is mainly due to the change in net radiation related to the modulation of the snow albedo by vegetation, which is obviously absent in the desert world simulations. The enhanced albedo in the desert world simulations induces a large temperature decrease, especially during summer in the cold and modern climatic contexts, whereas the largest difference occurs during winter in the warm climate. This temperature difference requires a larger heat transport to the northern high latitudes. Part of this heat transport increase is achieved through an intensification of the Atlantic Meridional Overturning Circulation. This intensification reduces the sea-ice extent and causes a warming over the North Atlantic and Arctic oceans in the warm climate context. In contrast, the large cooling North of 20°N in both the modern and cold climate contexts induces an increase in sea-ice extent.
KeywordsDesert world Cold and warm climate Vegetation-climate feedback Thermohaline circulation
Changes in land cover affect global climate via feedbacks between vegetation and the atmosphere. These feedbacks directly modify near-surface energy, moisture, and momentum fluxes via e.g. changes in albedo, leaf area and surface roughness. The presence of vegetation has an impact on the Earth’s surface albedo not only through its own albedo, which can contrast with bare soil albedo, but also because high vegetation such as forests can mask snow and therefore limit the increase in albedo related to a recent snowfall. The hydrological cycle is also affected, in many ways, by vegetation, since important variables of this cycle, such as evapotranspiration, is dependent on the plant type. For instance, anthropogenic deforestation in the Tropics during the last few decades is known to have resulted in reduced evaporation, increased surface temperature (e.g. Dickinson and Henderson-Sellers 1988; Lean and Warrilow 1989; Gash et al. 1996; Lean and Rowntree 1997; DeFries et al. 2002; Feddema et al. 2005; Davin and de Noblet-Ducoudré 2010), and rising river runoff (Piao et al. 2007).
Studying the impact of changes in vegetation and desertification is therefore relevant for future climate, as it is expected that the increase in global population will significantly alter the use of the Earth’s surface in the next centuries (e.g. Ramankutty and Foley 1999; Goldewijk 2001). Furthermore, it has recently been shown that the acceleration of the global warming predicted for the twentyfirst century could induce a die-back of the Amazonian forest as a consequence of reduced precipitation in this region (e.g. Betts et al. 2007). It is also important for our understanding of past climate changes, as the land surface characteristics changed along with climate. For instance, vegetation reconstructions for the last glacial maximum (LGM) show that forests were absent North of 55°N (Bigelow et al. 2003) and tropical forest cover decreased in Asia, Africa, and Australia, whereas the state of the tropical forest in South America is still debated (Harrison and Prentice 2003).
Past studies have investigated the biophysical effects of deforestation in specific modern climatic zones. For example, Henderson-Sellers et al. (1993) showed that deforestation of the Amazonian and South East Asian forests causes temperature reductions over these regions, because of the albedo increase due to deforestation, as well as because of a reduction in local evaporation. They also noted some remote effects of this tropical deforestation such as, for instance, a precipitation increase to the South of the deforested area. Using an AGCM and a scenario in which all tropical rain forests are entirely replaced by grassland, Sud et al. (1996) also found that evapotranspiration decreases but that the outgoing longwave radiation and sensible heat flux increase, which eventually result in a warmer and drier planetary boundary layer. Lean et al. (1997) showed that decreases in roughness associated with deforestation act to reduce evaporation. More recently, Costa and Foley (2000) studied the impact of deforestation in a context of rising atmospheric CO2 concentration. They found that both deforestation, via the associated decrease in evaporation, and CO2 increase, via its radiative forcing, tend to increase surface temperature. Finally, Werth and Avissar (2004) demonstrated that a land surface change in a tropical area can cause geopotential surfaces in the upper troposphere to subside, and that these geopotential changes can spread throughout the Tropics and into the midlatitudes. To summarise, the most recent studies about tropical deforestation show that it induces a warming over the deforested area via a decrease in evaporation.
The impact of changes of vegetation in the LGM climate has mainly been studied through sensitivity studies comparing Atmospheric General Circulation Models forced either by a modern (actual of potential vegetation) or a reconstructed LGM vegetation (Crowley and Baum 1997; Kubatzki and Claussen 1998; Levis et al. 1999; Wyputta and McAvaney 2001; Harrison and Prentice 2003; Ramstein et al. 2007). These studies have therefore not focused on as large a change as global deforestation and climate differences related to these two different vegetation covers appear to be of second order, but not negligible, compared to the impact of other changes in boundary conditions relevant for the LGM, i. e. northern hemisphere ice-sheets and decrease in atmospheric CO2 concentrations. However, vegetation changes can be important regionally. Indeed, using HadSM3 coupled to the dynamical vegetation model TRIFFID to simulate LGM climate, Crucifix and Hewitt (2005) demonstrated that over Eurasia, particularly over Siberia and the Tibetan plateau, the response of the biosphere substantially enhances the glacial cooling through a positive feedback loop between vegetation, temperature, and snow-cover. They also find that in central Africa, the decrease in tree fraction results in a reduction in precipitation. The atmosphere dynamics, and more specifically the Asian summer monsoon system, are significantly altered by remote changes in vegetation: the coolings in Siberia and Tibet act in concert to shift the summer subtropical front southwards and to weaken the easterly tropical jet and the associated momentum transport. Thus, climatic changes related to the use of a more realistic vegetation cover in LGM climate simulations can be regionally important and suggest that changes in vegetation should be taken into account for detailed model-data comparisons. However, the LGM vegetation is not well known everywhere, which makes the assessment of the impact of LGM vegetation difficult. Here, by choosing an idealised set-up, i.e. global desertification, we aim at examining the impact of one of the largest possible changes in surface characteristics.
Fraedrich et al. (1999) and Kleidon et al. (2000) use an AGCM with modern boundary conditions and study the impact of a “green” and a “desert” world. They find that a desert world is associated with a less intense hydrological cycle, and that surface temperatures are generally higher due to decreased evapotranspiration. They noted, however, that this sensitivity could be dependent on the climatic context. Furthermore, oceanic feedbacks were missing in all of above studies. One study of the atmosphere–ocean response to changes in vegetation is that of Renssen et al. (2003), who showed that large-scale changes in preindustrial forest cover may lead to a non-linear response of the ocean thermohaline circulation. They analyzed the transient response of the atmosphere–ocean system to a deforestation and showed that the initial cooling due to the albedo decrease is further amplified by sea ice formation at high latitudes in a first step, and in a second step to a southward shift of the North Atlantic convection sites, which further acts to cool the northern hemisphere high latitudes by decreasing the North Atlantic northward heat transport. This then feedbacks on the sea-ice cover and high latitude temperatures, which cool enough to induce year-round snow cover over North America and northern Scandinavia.
In the case of preindustrial climate it is clear that there is an important feedback between the vegetation and the thermohaline circulation (Renssen et al. 2003) but this feedback could be negligible for other climatic contexts, especially as far as the slowdown of the thermohaline circulation is concerned. On the one hand, in the case of a warmer world, there is some speculation that a shutdown or a slowdown of the thermohaline circulation could occur, via a warmer North Atlantic and Arctic sea surface temperatures (IPCC 2007). On the other hand, for the LGM, the deep water formation in the North Atlantic was shown to be less intense than today (Duplessy et al. 1980; Boyle and Keigwin 1987) and the Atlantic meridional overturning circulation cell to be limited in the vertical (McManus et al. 2004; Piotrowski et al. 2005; Lynch-Stieglitz et al. 2007). Here we will study whether these first order responses in AMOC could be modified by vegetation changes.
What are the impacts of vegetation on the atmospheric energy balance?
What are the impacts of global desertification on the hydrological cycle?
What are its consequences for the ocean dynamics?
Finally we aim at analyzing the similarities and differences on a large spectrum of climatic conditions.
We have performed numerical experiments using the IPSL_CM4 coupled ocean–atmosphere-sea-ice general circulation model (Marti et al. 2010). The atmospheric model LMDz (Hourdin et al. 2006) and the land-surface model ORCHIDEE (Krinner et al. 2005) are run with a horizontal resolution of 96 points in longitude, 72 points in latitude (3.75° × 2.5°) and 19 vertical levels for the atmosphere. The ocean component, ORCA/OPA (Madec et al. 1998) is run with 182 points in longitude, 149 points in latitude and 31 vertical levels with the highest resolution (10 m) in the upper 150 m, the horizontal mesh is curvilinear and orthogonal on the sphere, the grid spacing is about 2°, with a refined resolution of ~0.5° near the Equator. The sea-ice model LIM2 (Fichefet and Morales Maqueda 1997), which computes ice thermodynamics and dynamics, is included in the ocean–atmosphere model. The models are coupled using the OASIS coupler (Valcke et al. 2004), which interpolates the net surface heat flux, the net water flux, solar flux, and the wind stresses from the atmosphere model grid to the ocean model grid, and the SST and the sea-ice cover from the ocean model grid to the atmosphere model grid. The models exchange the time-averaged files once a day.
Summary of the boundary conditions used for the numerical experiments analyzed in the present study
Albedo computation based on
Modern experiment versus data
Trenberth et al. (2007)
3 Results: sensitivity to global desertification in cold and warm climates
In this section, we discuss the results of the six experiments (DW vs. CS in the three climatic contexts), focusing on the global energy and water balance (issues 1 and 2), and on the dominant atmospheric and ocean circulation patterns (issue 3).
3.1 Global water and energy balance
Along with the changes in latent heat transfer, the upward shortwave anomalies at the surface 20, 34, 33% (for the warm, modern and LGM climates, respectively) appear to be related to the increased albedo anomalies (11, 31 and 30%). The values of the differences for the outgoing short wave radiation and the albedo do not exactly correspond to each other because of differences in incoming short wave fluxes. These differences are due to the influence of changes in the atmospheric water vapor and cloudiness which are reduced by about 42, 19 and 17%. Along with a reduction in evaporation in the DW experiments and a colder ground temperature (from 0.2 to 3.2°C), the upward longwave radiation decreases, as well as the downward longwave radiation. Because of the reductions of the atmospheric water content, the total continental longwave radiation budget at the surface tends to cool the DW climate, as does the shortwave budget, but with a smaller magnitude in both modern and cold climates. In contrast, the future warm climate, in which the difference in the atmospheric water vapor is maximum (42%, while it does not exceed 20% in both of the present and LGM climates), the long wave effect on the surface land temperature becomes larger than the short wave effect.
For all climatic contexts, a desert world results in hydrological cycle weakening: precipitation, evaporation and atmospheric moisture all decrease, except for the runoff which increases (Fig. 3b). The land evapotranspiration (Ec) decreases by about 49, 50 and 51% in the warm, modern and LGM climates respectively. This induces a ~25% decrease in land precipitation in all three climatic contexts. Consequently, the runoff is increased about 9, 11 and 15% in the warm, modern and cold climate respectively. This amplification is larger in the low atmospheric CO2 case (cold climate), probably due to a stronger stomatal conductance effect (Gedney et al. 2006; Betts et al. 2007; Alkama et al. 2010). Indeed, for a low atmospheric CO2, plants’ stomata open more or longer to absorb the CO2 they need for their growth but this adaptation of the stomata to lower CO2 conditions also results in a larger transpiration of water than for higher CO2. This therefore acts to reduce runoff in the CS simulations, and the reduction is larger for lower atmospheric CO2 values. In the DW simulations, this effect is not active anymore, so, compared to the CS runs, the runoff is larger, and the increase is larger for lower values of CO2.
Finally, global desertification also affects the hydrological cycle over the oceans. The evaporation over the ocean (Eo) decreases by about 4% in all runs. This contributes to a decrease in precipitation over the oceans, largest for the LGM (7%) compared to the modern (6%) and warm (5%) climates. More details concerning the ocean heat and hydrological budgets and their consequences on modifying ocean circulation are given in Sect. 3.3.
3.2 Spatial patterns in the response to desertification
3.3 Ocean circulation and feedback
Freshwater input over Atlantic and Arctic oceans 45–90°N in different simulations
River flow (1,000 m3/s)
Calving (1,000 m3/s)
Atmosphere to ocean (1,000 m3/s)
Total freshwater (1,000 m3/s)
Maximum AMOC (Sv)
This drying of the North Atlantic and Nordic Seas should favour an intensification of the AMOC. Indeed, the DW simulations produce an increase in the meridional overturning cell. This increase in intensity and in the cell’s vertical size is larger in the cold climates with a maximum of 12 Sv (1 Sv = 106 m3/s) in the LGM and modern climates, while it is only about 3 Sv in the warm climate. The North Atlantic overturning maximum anomaly is located at 500, 2000, 2,500 m in the warm, modern and LGM climates, respectively (Fig. 9).
The upper panel of Fig. 12 shows that the atmospheric meridional heat transport decreases in response to the increase in oceanic meridional heat transport in the case of the warm climate. This type of effect has been depicted by Bjerknes (1964) and is therefore usually called the “Bjerknes compensation” effect (Shaffrey and Sutton, 2006). In order to have a perfect compensation, a constant radiative budget and no storage in any heat reservoirs are required. Even though these hypotheses are not verified here, we see that the atmospheric heat transport compensation seems to be total. This suggests that the increased albedo over continents (Siberia) via increasing snow cover is compensated by the decline of sea ice extent. In the case of the modern and cold climates (Fig. 12, middle and bottom panels), the Bjerknes compensation is far from being satisfied, showing that the climate system has largely modified its radiative budget and heat storage in response to the changes in the AMOC. These modifications in radiative budget are notably due to changes in sea-ice cover that modifies the albedo and therefore the radiative budget (Winton 2003) and to continental albedo change related to snow cover. In these climatic contexts, the difference in the radiative budget between the DW and CS is large enough to prevent the compensation between the anomalies in oceanic and atmospheric heat transports.
4 Summary and discussion
What are the impacts of vegetation on the energy balance of the three different climatic states?
What are the impacts of global desertification on the hydrological cycle?
What are the consequences for the ocean dynamics?
This study also highlights the importance of the coupling with the ocean. Up to now, most of our knowledge concerning the impact of land cover change on climate came from atmospheric-only models, assuming fixed oceanic conditions (e.g. Nobre et al. 1991; Fraedrich et al. 1999; Kleidon et al. 2000; Gedney and Valdes 2000; Chase et al. 2000; Voldoire 2006). Implicitly, this assumption was justified by the fact that the perturbation owing to land cover change is applied to the land and not to the ocean. However, our experiments show that taking the coupling with the ocean into account greatly affects the simulated response to desertification. First, as demonstrated by Renssen et al. (2003), we noted that the ocean surface responds to desertification by a cooling. But this cooling is not generalised and differs from one climate to another, especially over North Atlantic and Arctic. In the case of the modern and cold climate contexts, the more extensive sea-ice formation in the Arctic and/or North Atlantic due to the cooler temperatures at high latitudes induces an amplification of the AMOC and a southward shift of the convective sites in which the temperature slightly increases compared to CS runs. In all other surrounding regions, temperature decrease, especially over the regions where new sea ice appears. To compensate this large cooling over regions North of 40°N, both the oceanic and atmospheric meridional heat transports increase. In contrast, in the warm climate, the increased AMOC is largely due to the increased salinity over the Arctic and North Atlantic. This intensification of the AMOC induces an important strengthening in northward oceanic heat transport that is counterbalanced by the reduction of the meridional atmospheric heat transport.
Finally, does vegetation have a different impact for a warm, a modern and a cold climate context?
Supporting earlier hypothesis which studied the impact of deforestation (e.g. Pielke et al. 2002; Davin et al. 2007), we showed that desertification triggers two contrasting mechanisms: a radiative one (owing to surface albedo change) and a non radiative one (owing to change in evapotranspiration efficiency and roughness). We found that, replacing the vegetation by bare soil has roughly similar impacts South of 20°N in all three climatic contexts. This impact is mainly dominated by the non radiative forcing. In the other hand, the main differences concern the latitudes North of 20°N, where the snow cover and sea-ice extent play an important role in modifying the albedo and consequently the ocean and atmosphere dynamics. In both the modern and cold climatic contexts in which there is snow, global desertification causes an increase in albedo which enhances the upward shortwave radiation and thus cause a cooling (via radiative forcing). There is less snow in the warm climate compared to the modern and cold ones. Indeed, continental snow cover is only simulated in winter in the future climate runs, with a larger extent and albedo in DW compared to the CS warm climate. As consequence, North 20°N, in the future DW run, the seasonal cycle is more contrasted, with cooler winters and warmer springs, summers and autumns, especially over areas where there are forests in the corresponding control run. This warming is principally caused by the reduction in latent heat flux (i. e. non radiative forcing) and in the sea-ice extent. This sea ice extent decrease is due to an increase of the evaporation over the North Atlantic and the Arctic which acts to enhance the salinity (density) and consequently the AMOC, resulting to an intensification of oceanic northward heat transport.
This work has been supported by ANR-BLANC IDEGLACE and the RTRA STAE Toulouse, with computer time provided by CEA/CCRT. The authors wish to thank the three anonymous reviewers for their useful comments on this paper.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.