Table of contents key findings

Download 0.7 Mb.
Size0.7 Mb.
1   2   3   4   5   6   7   8   9   ...   19

3.3 Land Use Change

Charcoal records from various parts of the world suggest that climate variability for the past two millennia has been the primary driver of variations in large scale biomass burning, but that human influences became a significant factor after 1750. Model studies suggest that the net global albedo effect of such human influences throughout history may have warmed global climates by 0.008˚C, and hence is negligible. Since 1860, biomass burning appears to have caused a slight global cooling of -0.05˚C. Studies that include other biogeophysical effects of land use change (e.g., impacts on the local hydrological cycle) suggest a somewhat larger cooling of about -0.2˚C. Regional effects, however, may be much larger, particularly in eastern Europe and over India. Over the United States, the net effect of land use change appears to have been a weak warming, although there were large positive effects in some regions, and large negative ones in others. Canadian studies also suggest that reduced summer fallowing in the Prairies has likely increased surface albedo in summer, contributing to a local cooling and reduced daily maximum temperature and diurnal range. Surface energy budget observations also suggest that cloud feedbacks due to increased evapotranspiration may further reduce albedo from vegetated agricultural lands. Projections for global land use change suggest it will continue to have a modest cooling effect over the next century on the order of -0.14˚C177-186.

3.4 Natural Forcings
Satellite data collected over the past few sunspot cycles suggest these cycles contribute about 0.2˚C to global decadal climate variability. Sunspot activity can also explain about 80% of total solar insolation variability on longer time scales. However, high sunspot activity like that observed in recent decades has occurred only 2-3% of the time during the past 7000 years, and hence is very unusual within millennial time scales. The impact of the longer term changes in solar forcing on climate remains somewhat uncertain and controversial. Models and data both suggest that the magnitude of change in the ultraviolet and extreme ultraviolet region of the solar spectrum is much larger than that in the visible portion. For example, one model study estimates that, since the Maunder Minimum several centuries ago (when the sunspot cycle virtually disappeared), the extreme ultraviolet radiation has increased by 30% (mostly during the early 20th century). In contrast, ultraviolet intensity has only increased by 4% and total irradiance by about 0.03%. Since the ultraviolet intensity has large impacts on ozone chemistry and air circulation in the stratosphere, the related effects on climate become complex. Some analysts suggest that solar forcing could explain as much as 50% of the climate change that occurred during the past century. Others challenge this, and argue that there has been little net change in total solar forcing over the past three centuries. However, there is now general agreement, based on both observational evidence and model simulations, that the global warming of the past 30 years is caused by anthropogenic forcing factors, not solar forcing. Over the next few decades, the current high level of sunspot activities is projected to weaken. While this could cause a global cooling effect of a much as -0.2˚C, the concurrent warmer effect of rising greenhouse gas concentrations are projected to be much greater187-194.
Other natural factors also affect net global forcing. For example, major volcanic eruptions can have important short term influences on climate by injecting large volumes of aerosols into the stratosphere. Recent studies imply that related effects on stratospheric transparency may also have a quasi-80 year cycle. However, there is no evidence of a long term trend in volcanic forcing. Another natural forcing is that due to atmospheric mineral dust loading. Like volcanic aerosols, dust makes the atmosphere less transparent and thus contributes to a ‘solar dimming’ at the surface. Paleoclimate studies indicate that dust loading is largely a climate system feedback linked to aridity and wind intensity, rather than a primary forcing. Recent model simulations indicate that higher dust concentrations in the atmosphere in pre-industrial times may have caused a cooling of about -0.2˚C relative to today. On the other hand, lower dust loading projected for a 2xCO2 climate could add 0.06˚C to any CO2 induced warming. Some studies suggest that, although there have been recent increases in mineral dust loading in regions such as the tropical Atlantic Ocean (associated with the desiccation of parts of northern Africa), there has been a general global decline in dust loading since 1990. This may have been a small contributing factor in the rise in global temperatures. It has also been hypothesized that changes in cosmic radiation can alter atmospheric ion concentrations and thus influence cloud formation processes and climate. Recent observations over England provide some tentative support for higher cloud cover and more diffuse radiation during days with high cosmic radiation. However, recent comparisons of observed global cloud cover changes with concurrent changes in cosmic radiation during a full 11-year solar cycle do not support a major cosmic radiation role. Finally, periodic changes in Earth’s geomagnetic activity appear to be causing a 60 year cycle in its rate of rotation. This may thus also be a small contributor to long term natural climate variability195-202.

3.5 Net Forcings
The significant improvement in quantifying the radiative forcing effects of aerosols in recent years has helped to substantially reduce the uncertainty in estimates of total anthropogenic influences on the climate system since pre-industrial times. Solar and volcanic forcings appear to have dominated climate fluctuations prior to the 20th century, particularly during events such as the Maunder Minimum. However, anthropogenic forcings are now estimated to be a factor of ten greater than that due to the combined effect of all natural forcings. The gradual increase in human-induced atmospheric aerosol loading during the past century has helped to reduce the amount of solar radiation absorbed by the climate system, thus partially offsetting the warming effects of rising greenhouse gas concentrations. However, such offsets are now waning, particularly in industrialized countries where pollution control measures have significantly reduced aerosol emissions, and are projected to further weaken on a global scale in future decades. One estimate suggests that, since 1986, the combined effects of increased solar insolation due to a more transparent atmosphere and the concurrent rise in greenhouse gas forcing may have enhanced net surface radiation over global land areas by as much as 2 W/m2. Looking into the future, business-as-usual emission scenarios suggest that, by 2050, radiative forcings from relatively small increases in short-lived greenhouse gases such as methane and ozone will largely offset any remaining direct and indirect forcing effects of sulphate aerosols. Hence, by then, radiative forcing will be dominated by changes in well-mixed greenhouse gas concentrations and any additional heating effects from rising black carbon aerosol concentrations148,203-208.
In cold, snow covered regions such as that of the Arctic, climate response to radiative forcing is also very sensitive to the season and to the nature of the forcing. In winter, for example, temperatures in the Arctic are much more sensitive to changes in ozone and aerosol concentrations than to that of well-mixed greenhouse gases. Furthermore, in all but summer seasons, persistent changes in hemispheric atmospheric circulation (and hence the transport of air masses into the region from elsewhere) has a much larger impact on regional Arctic climates than local radiative forcings209-210.

4.0 Models
4.1 Climate Processes and Model Development
An enhanced understanding of the biogeochemical and dynamical processes that affect the climate system and its internal feedbacks is essential to improve the ability of climate models to accurately simulate the climate system, attribute past changes in climate to specific causes and increase the confidence in projections of climate system response to future changes in radiative forcing. Following are some of the recent developments that have helped do so.
4.1.1 Atmospheric Processes

Inadequate understanding and parameterization of atmospheric hydrological processes are primary reasons for uncertainty in model simulations of climate system behaviour. In particular, significant concerns remain about the ability of models to replicate the positive climate system-water vapour feedback. For example, models still appear to simulate an atmospheric liquid water path that is much larger and extends to higher altitudes than observed. On average, models overestimate the amount of water vapour in the mid troposphere and underestimate it in the lower troposphere. However, over the tropics, they show the relative humidity in the upper part of the troposphere decreasing as the climate warms, consistent with satellite and direct observations. While models disagree on the magnitude of the water vapour feedback on its own, they now agree quite well on the combined role of the water-vapour-lapse rate feedback within the climate system, including its strong positive nature throughout the troposphere. Recent estimates suggest that the water vapour feedback could add about 2 W/m2 to net radiative forcing for each degree of rise in average global surface temperatures211-218.

Observational data provide support for such a strong positive water vapour feedback. Satellite data collected between 1996 and 2003, for example, indicate that globally and yearly averaged total column precipitable water content increased by about 0.19 g/cm2 per degree of warming in near surface temperatures, with the response in the tropics being about 50% greater. Most of the increase occurred over ocean areas. Results, averaged globally and throughout the troposphere, are consistent with a constant relative humidity response to temperature change. Comparison of surface humidity trends with model simulations also implies constant relative humidity. However, this relationship differs from region to region. Analyses of tropospheric trends over North America, for example, indicates that the enhanced warming of the lower troposphere that has occurred since 1979 has not been accompanied by a similar increase in precipitable water vapour and total preciptable water content. That is, relative humidity appears to have decreased in this region219-221.

Clouds also have an important role in the climate system, both as a component of the hydrological cycle and in their effects on the flux of short wave solar and long wave heat energy through the atmosphere. Hence, proper description of their properties and feedback processes is also key to realistic model simulation. These properties and processes vary in space and time. In most regions of the world, the presence of low clouds has a net cooling effect, since their role in reflecting incoming sunlight in general exceeds that of absorbing outgoing heat radiation. Hence, increasing low cloud in response to warming would function as a negative feedback. However, in the Arctic, increasing convective clouds and humidity in response to melting ice cover would likely provide a strong positive feedback that could eventually help keep Arctic waters ice free as it did in past eons, even during winters. Although the description of clouds within models has improved significantly, the regional and global strengths of the cloud feedback still remain the largest contributor to uncertainty in the projections of climate system response to future external forcing211,218,222-223.

Analyses of observed climate data suggest that, globally, precipitation increases about 2.3% per degree of warming. Precipitation in turn affects the hydrological cycle and ecosystem properties, which can feed back into temperature changes, particularly at the local and regional scales. The regional pattern of this relationship is complex, particularly in high latitudes. In low latitudes, there is a strong linkage of the precipitation-temperature relationship to El Niño-Southern Oscillation (ENSO) behavior in the tropics. There is now also evidence from both observations and models that the hydrological processes in response to warming affect precipitation north of the equator differently than in the Southern Hemisphere. For example, in the Northern Hemisphere, average tropical precipitation increases in summer but decreases in winter, while the reverse happens south of the equator. This increases the amplitude of the dry-wet tropical rain cycle and the seasonal differences in rain across the equator224-226.

There are also other atmospheric feedbacks that are important. Recent studies, for example, confirm the strong linkage between changes in surface and tropospheric climates with that of the stratosphere. This emphasizes the importance of modeling related processes that link these components in adequate detail. Warmer climates alter the life cycle and distribution of tropospheric ozone, causing reductions in non-polluted areas of the extratropics and increases in the tropics and heavily polluted areas. Since ozone is a greenhouse gas, this affects regional and global radiative forcing. Some suggest that the increased poleward transport of latent heat by transient eddies over oceans and atmospheric storms may also be an important regional feedback that, among other things, contributes to the amplification of changes in climate over the Arctic relative to global changes227-229.

4.1.2 Land Processes

In general, most vegetation models still replicate fundamental aspects of land surface systems poorly and fail basic conservation tests for water and energy budgets. Researchers caution that these challenges result in large differences in the performances of dynamic vegetation models that must be resolved before important vegetation-climate feedbacks, from the individual leaf to global scales, can be properly represented in earth system models. Few models, for example, include the direct effects of CO2 enhancement on plant physiology, evapotranspiration and the local hydrological cycle. This effect is likely to make some regions, like the Amazon, become drier, but make others like the Arctic and east Asia become wetter and greener. Canopy nitrogen also plays an important role in forest carbon fluxes and is closely linked to albedo factors in temperate and boreal regions, but is not yet included in most related models. As noted in section 2, models that do include nitrogen processes interactively generally show weaker CO2 fertilization effects than those that do not. Hence, most studies to date appear likely to have overestimated direct CO2 enhancement of photosynthesis230-237.

Rising temperatures also significantly increase CO2 release through ecosystem respiration, fire loss and permafrost decay. A study using simulations with multiple coupled climate model, with and without this feedback, imply that the response of the carbon cycle to future climate change could add between 20 and 200 ppm to atmospheric CO2 concentrations projected for 2100. There is also evidence from paleo-climate data of a positive carbon cycle-climate system feedback. This feedback alone may be enough to increase the projected rise in temperature expected by 2100 by as much as an additional 1.5°C. In addition, decaying permafrost may reintroduce large amounts of old carbon stored within it for millennia back into atmosphere238-241.

Global surface albedo feedbacks as simulated by AR4 models (which do not include dynamic ecosystems feedbacks) are also moderately positive, adding about 0.3 W/m2 of radiative forcing for each degree of warming. It is, therefore, likely second only to water vapour feedbacks in importance. It is strongly positive in polar and mid latitude regions where retreating sea ice and snow cover can dramatically change surface properties, particularly in transitional seasons. Although its magnitude also remains uncertain, it may add a further 1 to 2˚C to regional warming in the Arctic over the next century. This surface albedo amplification may be partially offset by negative regional non-surface albedo effects, such as that due to changes in Arctic cloud. On time scales of many centuries, retreating ice sheets on Greenland may add to this albedo effect. Adding albedo changes due to ecosystem response will likely weaken the magnitude of net global surface albedo feedback211,236,242-245.

4.2.3 Ocean Processes

A broad range of recent studies have explored how the circulation system of the North Atlantic is affected by long-term feedbacks within other parts of the climate system, and how it in turn affects other parts of the system. Results suggest that salinity anomalies due to El Niño events as well as shifts to long-term El Niño-like global pressure patterns will likely not have a major impact on deep water formation and related thermohaline circulation changes in the North Atlantic. The influx of freshwater from melting polar glaciers and ice sheets also has only a minor impact on the initial slowdown of the Atlantic thermohaline circulation system, although very large flux increases could be important during its projected recovery centuries later. There will also be related changes in atmospheric circulation. Reduced Atlantic circulation, on the other hand, is likely to reduce CO2 uptake in the North Atlantic. The concurrent changes in regional climates induced by the weakened ocean circulation are expected to create a modest mid-latitude land carbon sink that partially offsets the reduced ocean carbon sink. However, the net effect, on millennial time scales, is projected to be a small net positive feedback243,246-248.

Poleward intensification of Southern Hemisphere westerlies under warmer climates is also likely to keep deep water formation in Southern Ocean active, despite other processes that tend to enhance stratification and reduce circulation. This would likely result in increased carbon sinks in that region that thus provide a negative climate system feedback. In Arctic regions, enhanced winds due to increased storm activity may cause a similar negative feedback. Not all models can replicate this feedback249-250.
4.2 Model Performance

      1. Climate sensitivity

Climate sensitivity is a term used by the research community to describe how much the average global surface temperature would increase once the global climate system has reached an equilibrium response to enhanced radiative forcing equivalent to a doubling of atmospheric CO2 concentrations.
Over the past few decades, a number of national and international assessments have consistently proposed an uncertainty range for climate sensitivity of between 1.5 and 4.5˚C. Many recent studies have attempted to improve upon these estimates. Some of these studies have demonstrated that, as more feedbacks are considered in the assessments, the upper end of the plausible sensitivity range increases. Several that use past observational and paleo data for example suggest a 90 to 95% confidence range for climate sensitivity of between 1.5˚C and up to 8.9˚C. Model simulations of the climate system suggest a somewhat smaller range, although most simulate sensitivities in excess of 2˚C. Model treatment of water vapour and cloud feedbacks, as well as different adjustment rates to radiative forcings, appear to be prime reasons for differences between models. Models that have the best performance in simulating current climates and display weaker seasonal surface air temperature variations also appear to have higher sensitivities. While most experts agree that the probability that climate sensitivity exceeds 7˚C is small, there is much greater uncertainty about the high end of the range than the low end. This is because uncertainties in the various feedbacks that may be involved, when compounded, amplify the variability in projections for the upper threshold much more than at the lower threshold. Some argue that these uncertainties are biased towards underestimating the upper end of the range. Incorporating ecosystem feedbacks in model simulations, for example, could add an additional 1.5˚C to the upper threshold but this feedback has not yet been included in many of the sensitivity assessments. A few studies have suggested sensitivities of as low as 1.1˚C per doubling, but others have challenged those results, claiming that they fail to recognize the complexities of the climate system251-275.

Figure 5. Various estimates for global climate sensitivity to a doubling of CO2, derived from a broad range of sources of evidence and estimation methodologies. Horizontal lines below each estimate indicates the very likely range (>90% probability), horizontal bars indicate likely range (>66% probability), qnd the circle inside the bar denotes best estimate. The bottom graph indicates the composite estimate from all sources, and suggests climate sensitivity is likely between 2 and 4°C per CO2 doubling, and very likely between 1.5 and 5°C. (Knutti and Hegerl, 2008, ref. #275)

4.2.2 Model Evaluation
Recent assessments of climate model simulations against observations, using improved evaluation methodologies, indicate that the current generation of climate models do much better in replicating current and past climates than their predecessors. They are also much better at replicating the glaciations and deglaciation processes of the Earth’s climate in the more distant past. While there remains considerable scatter of model outputs for various climate parameters relative to observed values, ensembles of model simulations do quite well. Hence, there is now much less evidence of common biases. Much of the differences between individual models are centered around two major contributors to uncertainty – the role of aerosols within the climate system, and the sensitivity of the climate system to radiative forcing211,276-279.

The improvements in climate simulation can be credited both to enhanced computing power that allows inclusion of higher resolution and more complex processes involved in the climate system, and to improved formulation of those complex processes. In general, general circulation models capture the radiative forcing effects of changes in greenhouse gases well. However, when their performance is compared with line by line radiation models, there are still errors as well as differences between models, particularly with respect to short wave absorption. Models also replicate water vapour parameters well, including those for the tropical troposphere. However, globally, the simulated amount of water vapour in the free troposphere is too high and that in the boundary layer too low. In the upper troposphere, models capture the nature of the relationship of water vapour and cloud ice concentrations to sea surface temperatures, but disagree significantly on the magnitude of this relationship and on the amount of ice in clouds at this level. While many models still continue to have an atmosphere that is too transparent, and hence overestimate the incoming flux of shortwave solar radiation, some now show more realistic fluxes. Most models also poorly simulate cloud performance, particularly in tropical regions. Related uncertainties are the dominant reasons for differences between models in simulating climate280-287.

Surface temperatures simulated by models agree quite well with observations, at both global and regional scales. Models on average project a mean land-sea warming ratio in recent years of 1.51, very close to that observed. Likewise, when allowances are made for measurement errors and effects of volcanic eruptions, modeled trends and variability in ocean heat content agree well with those observed. The lag in sea surface temperature response to radiative forcing in the North Atlantic and the Southern Ocean relative to that for the western Pacific and Indian Oceans is also consistent with the observed lag. However, models in general still have a cold bias of between 1 and 4˚C in tropospheric temperatures relative to observations. They also significantly overestimate the recent warming over Antarctica283,288-291.

Recent model performance with respect to global precipitation characteristics generally shows much improvement as well, particularly with respect to large scale monsoonal circulation and precipitation patterns and the strong relationship between mean global precipitation and global temperatures. Globally, there are still too many rain days, and not enough heavy rain events. Recent observations also indicate that global precipitation and atmospheric water vapour content have both been increasing at a significantly higher rate than that projected by models. This may be related to model underestimation of wind speeds, which is an important factor in the temperature-hydrological cycle feedback. This also implies that future model projections may underestimate global rainfall response to warming temperatures. There are also some significant discrepancies with observations on regional details. Performance is poorest in the west Pacific and in some tropical regions that are, ironically, also particularly vulnerable to changes in rainfall characteristics. Northern Hemispheric monsoonal behavior is somewhat weaker than observed, and its southward retreat in the autumn is not well simulated. Sahelian droughts are also not well simulated, perhaps because of inadequate coupling between regional rainfall, land surface feedbacks and adjacent ocean sea surface temperatures. Much of the uncertainties in precipitation projections appear to originate with inadequate simulation of relevant tropical processes, particularly with respect to the variability of the Pacific Ocean climate290,292-299.

Current models also simulate temperature extremes of the past half century moderately well. Since the simulated increase in temperature extremes is significantly linked to trends in mean temperatures, anthropogenic forcing is implicated. However, model results do not agree well with observations with respect to precipitation extremes, although some experts caution that this could also be at least partly due to observational error300-301.

Newer models are much better than earlier generations at forecasting the behavior of the El Niño Southern Oscillation (ENSO) and the location of related eastern Pacific temperature changes. They replicate observed recent enhancement of El Niño intensities relative to those of La Nina events, but do much more poorly on related details, including the duration of the ENSO cycle. The maximum rainfall and cycle of the Asian monsoon are now also much better replicated. However, most models fail to capture the recent widening of the atmospheric circulation belt over the tropics and the associated changes in precipitation at its margins, or the change in circumpolar circulation in the Southern Hemisphere. The latter may be related to inadequate representation of the middle atmosphere and the ozone processes that occur there. In general, models also overestimate ocean heat uptake and the related ocean thermal expansion contribution to sea level rise, and thus the rate of increase of SSTs145,257,302-306.

In general, models overestimate the Earth’s albedo in most regions, particular during the boreal summer. One causal factor may be excess modeled winter snow cover and delayed spring melt in the Northern Hemisphere relative to observations. However, there are also reservations about the quality of the observational data used to evaluate the model performance. In contrast, albedo is underestimated in the austral summer, and the seasonal patterns of albedo changes in the tropics are not well captured307-309.

Simulating climates in polar regions still appears to be a significant challenge. Most current models replicate the basic features of Arctic climates reasonably well, but all overestimate annual and seasonal precipitation in the west and central Canadian Arctic, and temperatures are generally too cold, especially over the Barents Sea. Poor model simulation of sea ice may be the underlying cause of these continuing discrepancies. While individual model simulations vary significantly, the ensemble of model results replicate Arctic sea ice conditions reasonably well. However, the ensemble fails to capture the multi-decadal variability in sea ice concentrations related to natural oscillations in atmospheric circulation. Ensemble results for Antarctica are also much improved. However, as with the Arctic, there remains large scatter in the results from individual models. Altough models now appear to capture snow fall amounts reasonably well, most simulate too large a warming for most of the region, but do not capture the large warming over the Antarctic Peninsula over the past half century. Key factors appear to be poor ability to replicate regional ice conditions and unrealistic simulation of water vapour feedbacks310-317.

Most models simulate the observed changes in circulation in the Southern Hemisphere since the mid-1970s quite well. These include the weaker mid-latitude and stronger high latitude westerlies and an intensified mid-latitude southern Ocean gyre. Best replication of these features is with models that include forcing effects of ozone depletion as well as greenhouse gases318.

The above summary suggests that current advanced climate system models are now able to replicate the climate system reasonably well, and are therefore useful for detecting and attributing past changes in observed climate, at global scales. That is as yet not generally the case at the regional scale, where changes are most relevant to society. For this, modellers will need better simulation of regional scale processes, better observations and a better understanding of how to attribute extreme events. One tool for doing so is a regional climate model, or RCM. While these models are driven by GCM outputs that may be biased, some studies suggest that nesting RCMs within GCMs can help to significantly reduce such biases. They also appear to carry residual biases in current climates forward into projections. Hence, those nested models that get the current climate right may also do well for future ones. RCM performance in simulating current climates at the regional scale indeed appears to be improving, although some still show significant biases in simulating current climates. For example, despite performing poorly in replicating regional cloud properties and surface albedo effects, simulation results with eight different RCMs during the intensive SHEBA observation program in the western Arctic Ocean in 1997-98 compared reasonably well with observations with respect to the regional details of surface radiation fluxes319-322.

Another approach to downscaling GCM outputs to more detailed regional projections uses statistical relationships between the broadscale atmospheric variables and surface climate details derived from observational data as a proxy for deriving similar surface details from the relatively low resolution GCM outputs. However, this approach also requires model inputs that are reasonably accurate. A third method is to use stochastic weather generators that simulate how day to day weather behaves under different climate regimes. These can be useful in generating extreme statistics for warmer climates. Tests suggest they do well for developing probabilities of precipitation extremes under warmer climates, but do less well for temperature extremes323-324.

Some researchers have noted that the relatively good agreement between model simulations and observations noted above is somewhat surprising, since few models include indirect aerosol forcing. They argue that either net total forcing from all aerosols must therefore be small or that other recent positive forcings within the climate system are underestimated. This implies that good performance of models may not be the sole criteria for establishing confidence in their ability to simulate future climates or attribute cause and effect. Despite this concern, results do indicate that any bias that might have been introduced by the minor tuning that is used in some models to improve performance would not be large enough to mask other major errors. Hence, the conclusion that recent climate trends cannot be explained by natural forcings alone remains robust325-326.

Download 0.7 Mb.

Share with your friends:
1   2   3   4   5   6   7   8   9   ...   19

The database is protected by copyright © 2023
send message

    Main page