Sunteți pe pagina 1din 22

Amazon Case Study: Development of a new framework for characterising dangerous climate change impacts

AVOID:
Avoiding dangerous climate change
AVOID is a DECC/Defra funded research programme led by the Met Office in a consortium with the Walker Institute, Tyndall Centre and Grantham Institute

Author(s): Penny Boorman, Richard Betts, Debbie Hemming Institute: Met Office Hadley Centre Reviewer: Richard Betts Institute: Met Office Hadley Centre Date: 06/04/2010
- -

AVOID is an LWEC accredited activity

Key outcomes / non-technical summary This study includes the progress towards development of a framework which begins by defining a dangerous impact and produces a range of global climate change and associated emissions. This reverses the order of the more usual technique and avoids the resulting large uncertainty in the impact. The first outcome of this work is the construction of threshold-based dangerous climate impacts framework. The outcomes from the application of this framework to the Amazon region include: Selection of a dangerous impact in Amazonia of forest loss. Identification of a regional climate regime associated with this impact in terms of measures of precipitation. Preliminary assessment of global mean temperature rise which could lead to the impact: which is greater than 2 in the models sampled. C Proposed link to mitigation scenarios using established relationships between global mean temperatures and peak emissions.

This report should be referenced as Boorman P., Betts R., Hemming D., 2010: Amazon case study: Development of a new framework for characterising dangerous climate change impacts. Work stream 2, Report 10 of the AVOID programme (AV/WS1/D1/R10). Available online at www.avoid.uk.net

AV/WS2/D1/R10

AVOID WS2 Deliverable 1 Report 10: Amazon case study

This report represents the tenth output to be delivered under Deliverable 1 of Workstream 2.

Contents:
Abstract ..........................................................................................................................3
1. Introduction................................................................................................................3

1.1 Dangerous Climate Change.................................................................................................................4

2 Methodology ...............................................................................................................5

2.1 Threshold-based Dangerous Climate Impacts Framework.................................................................6


2.1.1 Impact ..........................................................................................................................................6
2.1.2 Regional Climate Change .............................................................................................................6
2.1.3 Global Climate Change.................................................................................................................6
2.1.4 Greenhouse gas concentrations (and other forcings) .................................................................7
2.1.5 Emissions......................................................................................................................................7
Summary of framework ........................................................................................................................7

3 Case Study Application to Amazon .......................................................................8

3.1 Impact .................................................................................................................................................8


3.1.1 Metric from observational studies ..............................................................................................8
3.2 Regional Climate Change ....................................................................................................................9
3.2.1 Bioclimatic Zones .........................................................................................................................9
3.2.2 Observational and Model data ..................................................................................................10
3.3 Global Climate Change......................................................................................................................11
3.3.1 Significant features of changes between late 20th Century and late 21st Century ....................11
3.3.2 Range of global mean warming .................................................................................................14

4 Discussion on the case study .................................................................................17


5 Conclusions and Future Work.................................................................................18

Improving the Amazonia study...............................................................................................................18


Extending the Dangerous Climate Impacts methodology ......................................................................18

References ...................................................................................................................19
Acknowledgements .....................................................................................................21

AV/WS2/D1/R10

Abstract
'Dangerous climate change' has many definitions across social and natural science disciplines. One possible definition refers to major changes to key components of the Earth System which can result in accelerated or irreversible climate change. One example is the loss of large areas of forest which may result in increased carbon emissions and reduction of a carbon sink. Hence the impacts of climate change at regional scales are a crucial element of understanding dangerous climate change. If dangerous climate change is to be avoided, regional changes leading to these major impacts need to be understood. Uncertainties also need to be considered: those in the regional climate changes that would cause significant impacts; and also in the likelihood of particular changes occurring as a consequence of specific levels of global warming and greenhouse gas emissions. As part of this project we present progress towards building a framework to quantify the uncertainties in 'dangerous' regional impacts and to assess the range of emissions trajectories that would risk causing these regional changes. We begin by defining what changes are considered to be dangerous impacts within a specific region. In the case of Amazonia we define loss of forest cover as a dangerous impact. We quantify this using metrics that link the dangerous impact to observational studies of environmental variables. Here we use a 'bioclimatic zone' established from observational evidence to indicate when the forest loss could occur. Levels of annual precipitation and cumulative water deficit are used to define the thresholds in this region where a transition from forest to savanna is observed. Ensembles of climate models can then be used with these metrics to look at future changes and uncertainties, and assess when dangerous regional impacts could be realised. We investigate when these thresholds occur using ensembles of Met Office Hadley Centre global climate models (variants of the HadCM3 model from the Quantifying Uncertainty in Model Predictions project, QUMP), and the models from the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4), to span a range of scenarios and model structures Using this threshold-based dangerous impacts framework we present the findings of the Amazon case study. We find that some climate simulations have present-day climatologies which are offset from the observational values, this bias is corrected using observations. Linear approximations are used to extrapolate data past the dangerous thresholds to find the temperatures at which the dangerous impact occurs in each model. Hence a range of global mean temperature is found which is presented in terms of global warming thresholds. Noting this case-study is a first attempt to show the frameworks feasibility rather than giving definitive values, we find that the corrected data shows: some models breaching the dangerous zone by a level of global mean warming of 3 from pre-industrial values, given that the range of values found is much C larger; for a lower bound the dangerous impact does not occur at 2 but no upper C, bound is given for a temperature said to be a level of warming very likely to cause the impact.

AV/WS2/D1/R10

1. Introduction
1.1 Dangerous Climate Change
One of the United Nations Framework Convention on Climate Changes (UNFCCC) objectives requires the stabilization of greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system (UNFCCC, 2000). The achieved stabilisation level would be sufficient to allow ecosystems to adapt; to ensure food production and allow economic development to proceed in a sustainable manner. However, what constituted dangerous was not expanded upon. It has been noted that dangerous is a value judgement, for society to decide with science providing input to this decision deciding what constitutes dangerous for whomby when (Pachauri, 2006). One approach to understanding what constitutes dangerous can be illustrated by the five reasons for concern in the Intergovernmental Panel on Climate Change Third Assessment Report (IPCC TAR). This has become known as the burning embers diagram, and indicates five categories of key vulnerabilities and the global mean temperature changes at which particular damages may occur (figure 6-3, IPCC, 2001). The first objective of the Avoiding Dangerous Climate Change conference, held at the United Kingdom Met Office in 2005, was [f]or different levels of climate change what are the key impacts (Tirpak et al, 2005). While defining dangerous in terms of key vulnerabilities are described by Schneider and Lane (2006) as normative judgements for decision makers, natural scientists can provide information on systemic or natural thresholds or limits. Large-scale components of the Earth system may pass a threshold or tipping point beyond which a small perturbation can alter the state of that system (Lenton et al., 2008). Malhi et al (2009) describes these tipping points in the context of climate change as levels of change at which, though a small additional change in forcing, could lead to a component of the Earths climate system moving towards a new (and often undesirable) state. Lenton et al., (2008) describe a number of potential policy-relevant tipping elements of the Earth system, which include the Amazon rainforest. Subsequently Rockstrm et al (2009a) have included climate change as one of their nine key planetary boundaries which they suggest need to be observed to remain in a safe operating space for human beings. Climate change is one of three boundaries which they believe to have been transgressed, although other commentators have questioned their chosen description of a dangerous threshold for this phenomenon (Allen, 2009). One of the aims of the AVOID project is to provide core research for understanding dangerous climate change and its implications (AVOID, 2010). This work contributes to that aim. It explores the feasibility of using a framework for dangerous climate impacts which is focused on assessing the range of climate change and greenhouse gas (GHG) concentrations which may lead to a particular dangerous climate impact. Here deforestation in the Amazon is used as a case study.

AV/WS2/D1/R10

2 Methodology
One method of investigating dangerous climate impacts is to use numerical models of the Earth-system, using specified greenhouse gas concentrations and other forcings from selected future socio-economic emissions scenarios, and produce simulations of future climate. Results can be scaled to quantify effects of other emissions scenarios (eg Hulme et al, 2002) or multiple simulations using several scenarios can be run (eg Murphy et al, 2009). In this way a range of climate change responses can be obtained. These in turn may be used to derive a range of impacts for particular regions and systems (as eg Schneider and Lane, 2006, figure 2.4). The advantages of this method are: no a priori consensus on what constitutes a dangerous impact is needed; the result of this method is a large range of potential impacts mitigation policy decisions can identify a particular emissions scenario or scenarios and see the potential range of effects of each An alternative method is to start from a consideration of what key impacts would be considered dangerous in a particular location or context, and deduce an associated dangerous regional change, and in turn global change, associated with these. A bottom-up reversed methodology would allow the likelihood of a particular dangerous climate impact to be assessed, with uncertainty at each stage cascading to reach a particular set of emissions. Here the objective could be to provide a range of emissions which would follow the IPCC definition from very unlikely- defined as a 90% chance of avoidance to very likely or 90% chance of occurrence, illustrated in figure 2.1. The advantages of the reversed methodology are: Reduction of the uncertainty in the impact for adaptation policy decisions Identification of the range of potential emissions producing particular impacts

very likely to lead to impact

impact
impact
re cli gio ch ma nal an te ge g cli loba ch ma l an te ge

G co HG nc s

very unlikely
unlikely to lead to impact
impact

Figure 2.1 Schematic illustrating the dangerous climate impacts methodology

em

iss i on s

AV/WS2/D1/R10

2.1 Threshold-based Dangerous Climate Impacts Framework


2.1.1 Impact
A first stage to this methodology is to identify particular regions and systems where dangerous climate impacts might occur. The impact can be a complicated variable, which reaches a particular severity and is a function of location and time, and something which can be observed and quantified. The steps involved in this phase would cover: Expert identification of particular impact deemed to be dangerous Identify observational needs Identify model variable(s) that corresponds to observation, either those which are directly comparable, or by finding a proxy Uncertainties at this stage are by definition zero within this framework the threshold or range of impact is part of the process of defining dangerous. However it is possible that due to uncertainties or inhomogeneities in the observations for instance limited coverage or observation technique, observational datasets may need careful interpretation, and this could be explored using multiple observational datasets where available.

2.1.2 Regional Climate Change


Climate models can be used to identify regional changes associated with a particular dangerous impact. This can be through Regional Climate Models (RCMs) directly, through statistical downscaling of Global Climate Models (GCMs) which can represent regional features below the resolution of the model, or if the scale of the impact is large enough, sub-domains of GCMs. When using climate models to simulate future changes, some dangerous impacts can be identified directly in the model. These include irreversible or accelerated changes in the climate system, for example forest cover in a model with interactive vegetation. Others can be identified indirectly using a proxy, such as biodiversity changes through a particular threshold of climate variables. Here the steps would be: Obtain suitable regional climate model data, or a sub-domain of a global climate model Use baseline to compare model data with available observations Identify suitable transient model data continuous output over the twenty-first century (and potentially beyond) Uncertainties in model structure and initial conditions are explored using ensembles of model runs. Model biases can be ameliorated using bias correction techniques. Uncertainties are passed to the next stage using the range of response of the regional climate state.

2.1.3 Global Climate Change


Once a regional climate state is identified, a global mean variable is needed to progress towards an understanding of global climate change which ideally would encompass many future scenarios. Where a range of driving scenarios are not available, pattern scaling could be used for this task. The steps involved in this stage include: Choose global variable to assess climate change as multi-year annual mean Identify suitable global model fields related to the regional changes Find global parameters suitable to derive global forcings/emissions
6

AV/WS2/D1/R10
Uncertainties in forcing scenario should be considered, ideally encompassing IPCC SRES and mitigation scenarios available. Uncertainties are cascaded to the next stage using the range of global mean climate state found. This is the current extent of the framework, with the next two sections outlining plans to take this forward.

2.1.4 Greenhouse gas concentrations (and other forcings)


Identifying a range of global mean response of the climate system, either the prescribed concentrations used within global climate models may be used directly, or a simple model using the energy balance of the climate system may be used to deduce the atmospheric greenhouse gas concentrations which result in that climate state. Uncertainties in GCM GHG concentrations will be inherent in the range of forcing scenarios plus further evaluated if results from GCMs with carbon cycle are used, which diagnose carbon concentrations. Using energy-balance models uncertainties could be quantified using a range of climate sensitivities and aerosol forcing. In some cases knowledge of the forcings would be valuable, for example, where feedbacks on vegetation are not implicit in the model. In other cases it would be possible to remove this step and proceed directly from global climate change to evaluation of emissions.

2.1.5 Emissions
The final stage of the framework could lead to a time independent range of cumulative emissions which could give rise to the impact initially defined. Alternatively, through analysis such as Lowe (2009) relationships between maximum global temperature rise and peak year of emissions for various mitigation scenarios have been quantified using a simple climate model. Hence a range of global temperatures from the third stage of this framework could enable very unlikely 10% chance of occurrence and median emissions to be enumerated for multiple mitigation scenarios

Summary of framework
The framework is summarised in table 2.1, with the extent of the current methodology highlighted. It should be noted that the reversal of the standard methodology is that of the order of steps from emissions to impact, rather than the steps themselves being inverse. This means, for instance, that the climate models are not initialised with a particular impact to see the climate response; rather existing modelling is utilised to provide input to the process. The use of extensive model results already available means the framework can be flexible in the tools it uses, depending on the nature of the impact.
DANGEROUS IMPACT Variable Complex impact variable REGIONAL
CHANGE

LARGE SCALE
CHANGE

CO2
CONCENTRATION

EMISSIONS

Simplified impact Temperature CO2 equivalent variable Scale Regional Regional Global Global Global fn f(x,t) f(x,t) f(t,f) f(t,f) f(t,f) Table 2.1 Summary of the information needed at each stage of the dangerous impacts methodology (where x spatial position; t timeperiod; f radiative forcing scenario)

AV/WS2/D1/R10
Of the five reasons for concern, the five key vulnerability groups from IPCC TAR, the fifth risks from future large-scale discontinuities includes large scale and high impact events (IPCC, 2001). These include: the North Atlantic meridional overturning circulation (MOC) collapse (eg Challenor et al, 2006); ice sheet melting (eg Lowe et al, 2006); and Amazon dieback (eg Cox et al, 2004). Here Amazonia is used as the first case-study for this threshold-based dangerous impacts methodology.

3 Case Study Application to Amazon


3.1 Impact
3.1.1 Metric from observational studies
The climate of the Amazon region described in (Malhi et al, 2009) indicates that the lowland forests have a mean annual temperature of 26 with little spatial variability, C and a mean annual precipitation of around 2400mm, which varies from over 3000mm to less than 1500mm where the forest becomes savanna. Climate Models can be assessed by their representation of present day climatology, or historical data where known. Some climate models are known to be too dry in the Amazon region and often exclude feedbacks such as those from land-use changes and forest fires (Cox et al, 2004). However, the version of the Hadley Centre HadCM3 model used by Cox et al (2004) showed warming and drying of Amazonia driven by climate change when compared to a simulation which omitted climate change forcings. These results indicated that Amazon collapse may occur at a level of global climate change in the range 2-3 The mean global temperature of the IPCC AR4 Atmosphere-Ocean Global C. Climate Models reaches this range of temperatures by the end of the 21st Century in all but one of the six SRES scenarios (IPCC, 2007, p71) and with general agreement that a temperature rise will occur over the South American region (IPCC, 2007, p75). From observed datasets, temperatures in lowland tropical regions have seen an upwards trend in recent data of 0.25 decade-1, while no overall regional trend in precipitation C has been observed (Malhi et al, 2009). Given that, as described above, an impact can be a complicated variable which can be observed and quantified in reality, a workshop was held bringing together experts in both Amazon observational studies and climate modelling of the region, to obtain an understanding of what impacts may constitute dangerous there. The experts convened included those involved in observational campaigns to look at carbon levels, the effects of altitude and incline, fires and biodiversity changes. The modelling experts have been involved in regional Amazonian and tropical studies, studying carbon-cycle feedbacks and uncertainty. The intended objective was definitions of dangerous impacts in Amazonia and associated regional climate change. Three impacts were considered loss of biodiversity/ species in the region, loss of forest, and a switch from carbon sink to source of the region. Once identified, avenues for quantification of climate change in the region associated with particular impacts were also established. One important impact identified is the loss of forest from Amazonia. The Amazonian forest is currently a sink for carbon dioxide, and the loss of forest cover would impact on
8

AV/WS2/D1/R10
the amount of carbon dioxide removed from the atmosphere by the biosphere. Hence the transition from forest to savanna is an important effect, and it is this impact we examine here.

3.2 Regional Climate Change


3.2.1 Bioclimatic Zones
Typifying particular attributes of an individual area gives a classification of zones which can be used to generalise about the climate and biodiversity of a particular place. Such classification includes Holdridge lifezones (Holdridge, 1967) and Koppen climate states (Kaye, 2007). This idea of a bioclimatic zone can be used to look at changes seen under a changed climate in regions such as Amazonia (Good et al., submitted). Defining zones where tipping points occur in the climate system is one way of looking at regional climate changes associated with a dangerous impact. Malhi et al (2009) uses observational evidence of bioclimatic zones in recent data to asses if a tipping point in Amazonia is crossed by projections from climate models in the twenty-first century. The tipping point in this region suggests a level at which there is a transition from forest to savanna (here meaning grasslands and shrublands). Malhi et al identify this using current precipitation data from TRMM (Tropical Rainfall Measuring Mission) satellite data and vegetation cover classification from the Tropical Ecosystem Environments by Observation maps. This work follows Malhi et al (2009) using two precipitation related measures as thresholds: i) Annual accumulated precipitation to represent overall water supply ii) Cumulative Water Deficit (CWD), used to represent the accumulated water stress in the dry season, defined by Arago et al (2007) as: WDn = WDn-1 + (Pn En) ; Max(WD) = 0; WD0 = WD12 ; CWD = Min (WDi, i=1 to 12) where Pn - En is precipitation minus evapotranspiration, the monthly change in water deficit, and Max(WD) is assumed to be zero (ie the soil is saturated) in the wettest month. E is fixed at 3.33 mm day-1, using an approximation that a moist tropical canopy transpires ~100 mm month1. The location in climate space of Malhi et als zones are broadly these:
Annual Precipitation > 1500mm and -200mm < CWD evergreen forests predominate Annual Precipitation > 1500mm and -400mm < CWD <-200mm seasonal forests predominate Annual Precipitation <1500mm and CWD < -400mm savannas predominate

The thresholds used to represent the transition are CWD -300mm and annual precipitation 1500mm, although in reality there is a more gradual transition. This is perhaps because very localized conditions are not taken into account, or it may be due to data limitations (Malhi et al, 2009). The transition is analogous to the zone of uncertainty around a threshold in Rockstrm et al (figure 2a, 2009b), where a threshold effect is thought to exist but the precise position is uncertain. This uncertainty could be explored further in future work.

AV/WS2/D1/R10
Using these thresholds Malhi et al (2009) investigated the crossing of these tipping points in the future using data from models included in the IPCC AR4 (IPCC, 2007). Observational datasets show the Eastern Amazon (EA, defined 60to 48 W; 3 to N 12 to be more vulnerable than Western Amazon (WA, defined as 72to 60 3 S) W; N to 12 with EA drier and experiencing seasonal water stress (figure 1a in Malhi et al, S), 2009). Hence, in choosing the most vulnerable component in a region to define the dangerous climate impact, here the transition from forest to savanna in Eastern Amazon is used. This work is built upon in this study to investigate the global temperatures at which these thresholds are reached.

3.2.2 Observational and Model data

Climate Research Unit observed precipitation To obtain a present day climatology for the Amazon, here one version of the Climate Research Unit (CRU) precipitation dataset of Hulme is used (an outline of the technique can be found in Hulme et al, 1998). This is a monthly mean gridded dataset of resolution 2.5latitude by 3.75longitude. Where no observations are located within a gridbox, this is left as missing data no interpolation is attempted. Here the period 1970-1999 is used as a late twentieth century observed present-day baseline, understanding that some missing data means the length varies in terms of number of years used for each month. Other datasets which interpolate existing data to give a uniform gridded field could also be used. The variation in observed climatology between different datasets could also be explored. IPCC AR4 global climate model ensemble The Amazon region is of sufficient areal extent that a sub-domain of global climate models can be used in this case study. To obtain a spread of models with different formulations, but with necessary data available for this study, a sub-set of models used in the IPCC AR4 from a range of institutions was used (IPCC, 2007). These models include a range of horizontal resolutions and physical representations. Some modelled precipitation, when compared to observations over the same region, show differences from the present day climatology (Malhi et al., 2009). This bias is corrected for using observations, both in the late twentieth century and in the future, using an offset to twentieth century values.
P21(m,i)=( 1+ ((PGCM21(m,i)-PGCM20(m,i)) / PGCM20(m,i)) ) x POBS20(m) (eq 3.1)
where m=month, i=model; 20/21 refers to the century, and GCM/OBS to model/observed data

While the twenty-first century correction is justified by Malhi et al by the existence of twentieth century biases, it should be noted that the bias correction is likely to alter the climate change signal. In future work the method of bias correction and correlation of the correction with the climate change signal are important uncertainties to quantify.

10

AV/WS2/D1/R10
In addition to the present-day baseline 1970-1999, twenty-first century values are calculated as thirty year means, offset by ten years (hence 1980-2009, 1990-2019, 2000-2029 etc.,). Data from two IPCC SRES forcing scenarios (IPCC, 2007, p18) for the AR4 data are used, the A2 and A1B scenarios are used. The pre-industrial model data values use model output from historical runs covering 1861-1890. These have historical forcing data rather than being simulations with true pre-industrial forcings (ie often defined as before any anthropogenic emissions of greenhouse gases from the industrial revolution), but the difference between the late nineteenth century and earlier forcing is assumed to be small. As the AR4 models historical forcing runs begin in different years, the lengths of the pre-industrial era climatologies vary in length. Met Office Hadley Centre Quantifying Uncertainty in Model Prediction (QUMP) GCM ensembles The QUMP ensembles of model runs are based on Met Office Hadley Centre HadCM3 climate model (Gordon et al, 2000). To explore model structure, physical parameters in the model are varied in accordance with valid ranges defined by experts to produce a large number of variants on the standard version of the model (as in Murphy et al., 2007). The same bias correction (equation 3.1 above) using observed precipitation dataset is used In the QUMP data, two ensembles which use forcing scenarios A1FI and A1B are used here, the latter enabling a direct comparison with the IPCC AR4 datasets. Again thirty year periods are used to obtain climatologies, and historical forcing simulations are used as a pre-industrial baseline. All data are approximated as 30-day months.

3.3 Global Climate Change


With the model diagnostics for precipitation and derived values of CWD, we look here at the uncertainty in the global mean temperature associated with changes in these two diagnostics. The regional climate change, which was identified to cause the dangerous climate impact of deforestation in the Eastern Amazon, is used.

3.3.1 Significant features of changes between late 20th Century and late 21st Century
To see how the bias correction affects the range of values spanned by the end of the twenty-first century in the IPCC AR4 models, figure 3.1 shows precipitation (left) and CWD (right) for both A1B and A2 scenarios. The uncorrected values (upper panels) show that the majority of models fall into the savanna tipping point zone (shaded area) for both precipitation and CWD. The corrected values (lower panels) show fewer models reaching the dangerous impact level by the end of the twenty-first century.

11

AV/WS2/D1/R10

Figure 3.1: 30 year mean impacts variable (2070-2100) vs global mean temperature change (2070 2100 minus pre-industrial) for AR4 uncorrected (upper) and corrected (lower) precipitation (left) and CWD (right) for A2 (blue) & A1B (purple)

To evaluate what climate change trends exist in the models, figure 3.2 shows the changes at the end of the twenty-first century from the twentieth century values. For both precipitation (left) and CWD (right) it seems that most of the models do underpredict rainfall, looking at the uncorrected values (upper panels). Hence some are already in the savanna regime. Most of those that simulate the late twentieth century EA precipitation well, do not show a large trend over the twenty-first century. When corrected (lower panels) there are more apparent trends (whether the trends are significantly different before and after correction would indicate to what extent the correction was affecting the climate change signal). The variation in temperature change from pre-industrial to twentieth century values is apparent in the different models; this may be due to the short period taken for the pre-industrial baseline if a long control pre-industrial run was used in preference to this nineteenth century baseline, it may reduce the variability seen in the shorter period and act to make the values seen in different models more similar. To illustrate the trends in the model ensembles, the two values, from late twentieth and twenty-first centuries are used to linearly extrapolate the models that do not reach the dangerous threshold before the end of the model run (figure 3.3). In both precipitation (upper panels) and CWD (lower panels) for some IPCC AR4 models (left), the threshold is never reached, or reached at very high temperature. Therefore a difference in response is apparent within the AR4 ensemble, and also between the AR4 (left) and QUMP (right) ensembles.

12

AV/WS2/D1/R10

Figure 3.2: 30 year mean impacts variable vs global mean temperature change from pre-industrial for two periods (1970-1999 and 2070-2099) for AR4 uncorrected (upper) and corrected (lower) precipitation (left) CWD (right) for A2 (blue) A1B (purple)

Figure 3.3: 30 year mean impacts variable vs global mean temperature change from pre-industrial for two periods (1970-1999 and 2070-2099 extrapolated to show when the dangerous thresholds are reached. Corrected AR4 (left) and QUMP (right) precipitation (upper panels) and CWD (lower th st panels) for A1B using late 20 & late 21 century values (purple lines) and linear extrapolation (brown dotted lines)

13

AV/WS2/D1/R10

3.3.2 Range of global mean warming


Noting the caveats on the change between the present-day baseline and the future, the pathway the model takes through the twenty-first century is important to see if the final range of temperature changes adequately describes the natural variability and to evaluate whether a linear extrapolation of values is indeed a reasonable approximation. Figure 3.4 shows the path of the individual models for the AR4 ensembles using uncorrected data. Natural variability will mean some fluctuations will be evident and the length of climatology used (thirty years here, but ten was also considered, unshown) will obviously influence the variability of the data. Through the twenty-first century some models do indicate a non-linear path dependency, although other indicate a more linear path. Given the differences it is not a straightforward decision how to extrapolate models to quantify when they pass thresholds, linear extrapolation is used as a simple first-order approximation here.

Figure 3.4: 30 year mean impacts variable vs global mean temperature change from pre-industrial for periods every ten years from 1970-2000 to 2070-2100 for AR4 uncorrected (upper) and corrected (lower) precipitation (left) CWD (right) for A2 (blue) & A1B (purple)

The above indicate a range of global mean temperatures at which the dangerous regional thresholds are passed, the following attempts to encapsulate this in a form which is useful to illustrate those levels of global mean temperatures. Figure 3.5 shows that, uncorrected, the two ensembles show similar threshold temperatures where some members breach the dangerous zone. Indeed, uncorrected, some models have crossed the threshold at 1 indicating that the lower limit may have already been C, passed. As many models do not reach 4 a simple linear extrapolation is employed to C give the values of the two precipitation measures at that temperature (figure 3.5, two lowest panels). In figure 3.6 the values of both impact variables, corrected for biases from twentieth century values, is also shown for levels of global mean temperature rise from preindustrial baseline from 1 to 4 The spread of both ensembles increases as the C. threshold temperatures increase, with very few AR4 models reaching 4 by the end of C the twenty-first century. It appears that two models in the AR4 run reach the dangerous regional warming prior to any in the QUMP ensemble. The bias corrected data indicates that the dangerous impact does not occur at 2 but no upper bound is noted where C, there may be said to have a level of warming definitely likely to cause the impact.

14

AV/WS2/D1/R10

Figure 3.5: Uncorrected model precipitation and CWD values at threshold temperatures of mean global temperature rise up to 4 plus all models (extrapolated where necessary) at a global C, mean warming of 4 AR4 (left A2 (blue) & A1B (purple)) QUMP (right - A1B (purple) & A1FI C: (brown))

15

AV/WS2/D1/R10

Figure 3.6: as figure 3.5 but with corrected model values

16

AV/WS2/D1/R10

4 Discussion on the case study


The Amazon case study is presented as a proof of concept of the threshold-based dangerous impacts methodology rather than as definitive results of a range of mean global temperature change necessary to cause the Amazon rainforest to dieback. To explore the uncertainties further the method of bias correction, and differences observational datasets could have would need to be considered. An expanded set of model results which included those with a carbon-cycle, giving some quantification of uncertainties in feedbacks would also be beneficial. This case study does not consider all of the possible human influences which could be nucleation points for the tipping point such as land use changes or fires (as Malhi et al describe, 2009) which are not included in the climate simulations used. Some choices in model baselines and length of datasets could also be explored, such as the difference between using a pre-industrial baseline of 1861-1890 historical emissions and a long pre-industrial control run. The late nineteenth century values used above may mean the temperature change is (very slightly) underestimated for some models. Indeed no attempt is made to distinguish between models in terms of weighting them taking into account how realistic their representations of reality are this could be done in an objective way comparing simulations of the present day with observations, although limitations on datasets mean this may not be straightforward in all cases. This analysis is also useful in comparing models with each other or benchmarking towards a better understanding of the physical and dynamical processes involved.
DANGEROUS IMPACT Variable Complex impact variable Regional f(x,t) REGIONAL CHANGE LARGE SCALE
CHANGE

CO2
CONC.

EMISSIONS

Simplified impact Temperature CO2 equivalent variable Scale Regional Global Global Global fn f(x,t) f(t,f) f(t,f) f(t,f) Thresholds: Change from Precipitation Global annual Amazon -1 evergreen forest to <1500mm yr mean temperature case study CWD <-400mm savanna -1 yr Table 4.1 Summary of the threshold-based dangerous impact methodology for the Amazon region

Table 4.1 summarises the threshold-based dangerous impact methodology for Amazonia, with the next stage to address the relationship between the range of global mean temperatures produced with greenhouse gas concentrations and emissions. Before proceeding to describe the way these results can be carried forward, the results could also be presented in terms of the very likely and very unlikely large scale change by choosing appropriate thresholds in the global mean variable and with suitable extrapolation displaying the impacts which result. The nature of the extrapolation would depend on the pathways of individual model runs, and could involve non-linear functions, for instance, relating forcings to temperature rises. The final panels in figure 3. and 3.6 illustrates the concept in a simplistic way, using linear extrapolation of the two final thirty year periods before the threshold is crossed where individuals models do not reach a global mean temperature change of 4 by the end of the C twenty-first century. Note that 4 is chosen here for the interest in this threshold in C mitigation discourse rather than illustrative of very likely in this particular case study.
17

AV/WS2/D1/R10
To progress to the next stage of the framework, existing quantified relationships between mean temperature rise and emissions could be used. Figure 4.1 shows an example (from the conference presentation of Lowe et al., 2009) where emissions from the A1B scenario are followed until a peak emissions year, after which the emissions decrease by a particular percentage annually. The relationship is derived from multiple runs of a simple climate model, which gives an uncertainty range, the example in figure 4.1 shows this for the median case.

Figure 4.1: Relationship between maximum global temperature rise (from pre-industrial) and year of peak emissions for mitigation scenarios of 1 to 6% emissions decrease annually, for the median case (Lowe, 2009)

5 Conclusions and Future Work


The results of the Amazonia case-study give a range of global mean temperature which is presented in terms of global warming thresholds. With the caveat that this study is a proof of concept to show the frameworks feasibility rather than giving definitive values, we find that the bias corrected data illustrates two main points. Some models cross into the dangerous zone by a level of global mean warming of 3 from pre-industrial C values, although the range of values found is much larger. To indicate a lower bound, the dangerous impact does not occur at 2 but no upper bound, said to be a level of C, warming very likely to cause the impact, is given.

Improving the Amazonia study


New climate model simulations including carbon feedbacks and ensembles spanning a greater range of alternative, valid, model structures would improve this work. The observational datasets and how they are used to correct model biases are an area which needs to be investigated. How the bias correction affects the climate change signal could also be quantified. It may be possible in future to use probabilistic assessments of regional changes which would encompass additional sources of uncertainty. Other metrics or diagnostics could be used as alternatives to the above precipitation based bioclimatic zone, such as net primary productivity changes in Amazonian rainforest.

Extending the Dangerous Climate Impacts methodology


The threshold-based dangerous climate impacts framework has the potential to be a useful tool to provide scientific evidence in both adaptation and mitigation policy due to the uncertainty in the impact being minimised. Combining expert judgement on climate
18

AV/WS2/D1/R10
system thresholds from a real-world and modelling perspective can give a more robust definition of dangerous impacts from the point of view of the natural sciences. How to include the time lag in some impacts should be contained within the methodology. The methods used to correct and extrapolate data need careful consideration, as does the initial choice of data. Exploring the uncertainties in model formulation and socio economic scenarios could be done through scaling of existing simulations and potentially probabilistic approaches. Including committed climate change model simulations (such as Jones et al, 2009) may be a useful way forward. Developing the methodology to include emissions could potentially extend to quantifying how emissions are distributed.

References
Allen M., 2009, Tangible targets are critical commentary, Nature, Vol 3, October 2009, pp114-115 Arago L.E.O.C., Malhi Y., Roman-Cuesta R.M., Saatchi S., Anderson L.O. and Shimabukuro Y.E., 2007, Spatial patterns and fire response of recent Amazonian droughts GRL, Vol 34, pp AVOID, 2010, http://ensembles-eu.metoffice.com/avoid/ (last accessed 31/03/2010) Challenor P.G., Hankin R.K.S. and Marsh R., 2006 Towards the Probability of Rapid Climate Change in Avoiding Dangerous Climate Change, Schellnhuber, H.J., et al.(eds.), Cambridge University Press, New York, NY, pp55-63 Cox P.M., Betts, R.A., Jones, C.D., Spall, S.A. and Totterdell, I.J., 2000, Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model, Nature, Vol. 408 Good P., Jones C., Lowe J., Betts R., Booth B. and Huntingford C., submitted Quantifying environmental drivers of future tropical forest extent, submitted to Journal of Climate Gordon, C, C. Cooper, C A Senior, H Banks, J M Gregory, T C Johns, J F B Mitchell, and R A Wood, 2000, The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments, Climate Dynamics, 16, pp147-168 Holdridge LR, 1967, Life zone ecology. Tropical Science Center. San Jose, Costa Rica. Hulme, M., Osborn,T.J. and Johns, T.C., 1998, Precipitation sensitivity to global warming: Comparison of observations with HadCM2 simulations, Geophys. Res. Letts., 25, pp3379-3382 Hulme, M., G.J. Jenkins, X. Lu, J.R. Turnpenny, T.D. Mitchell, R.G. Jones, J. Lowe, J.M. Murphy, D. Hassell, P. Boorman, R. McDonald and S. Hill, 2002, Climate Change Scenarios for the United Kingdom: the UKCIP02 Scientific Report, Tyndall Centre, University of East Anglia IPCC, 2001, Climate Change 2001: Synthesis report, Watson RT and the Core Writing Team (eds) Cambridge University Press, Cambridge, United Kingdom, and New York, NY, USA IPCC, 2007, Climate Change 2007: The Physical Science Basis Summary for Policymakers, Technical Summary and Frequently Asked Questions, Solomon S, et al (eds) Cambridge University Press, Cambridge, United Kingdom, and New York, NY, USA

19

AV/WS2/D1/R10

Jones, C., Lowe, J., Liddicoat, S. and Betts, R., 2009, Committed terrestrial ecosystem changes due to climate change, Nature Geoscience, 2, pp484-487 Kaye, N., 2007, Analysis of Koppen climate classification changes following a future climate change scenario, Met Office User Documentation Lenton TM, Held H, Kriegler E, Hall JW, Lucht W, Rahmstorf S, Schellnhuber HJ, 2008, Tipping elements in the Earths climate system, Proceedings of the National Academy of Sciences, 105, 6, pp1786-1793 Lowe, J., Gregory, J. M., Ridley, J., Huybrechts, P., Nicholls, R.J. and Collins, M., 2006, The Role of SeaLevel Rise and the Greenland Ice Sheet in Dangerous Climate Change: Implications for the Stabilisation of Climate in Avoiding Dangerous Climate Change, Schellnhuber, H.J., et al.(eds.), Cambridge University Press, New York, NY, pp29-36 Lowe , J., 2009, 4+C: the emissions reduction challenge, proceedings of 4degrees and beyond conference, http://www.eci.ox.ac.uk/4degrees/programme.php (last accessed 31/03/2010) Malhi, Y , Aragao, LEOC, Galbraith, D, Huntingford, C, Fisher, R, Zelazowski, P, Sitch, S, McSweeney, C, Meir, P 2009 Exploring the likelihood and mechanism of a climate-change-induced dieback of the Amazon rainforest Proceedings of the National Academy of Sciences, 106, 49, pp20610-20615 Murphy, J., M., Sexton, D.M.H., Barnett, D.N., Jones, G.S., Webb, M.J., Collins, M. and Stainforth, D.A., 2004, Nature, Vol 430, pp768-772 Murphy J. M., Sexton D. M. H., Jenkins G. J., Boorman P. M., Booth B. B. B., Brown C. C., Clark R. T., Collins M., Harris G. R., Kendon E. J., Betts R. A., Brown S. J., Howard, T., Humphrey K. A., McCarthy, M. P, McDonald R. E., Stephens, A., Wallace C., Warren, R, Wilby, R., Wood, R.A., 2009, UK Climate Projections Science Report: Climate change projections, Met Office Hadley Centre, Exeter, UK. Pachauri, R. 2006 Avoiding Dangerous Climate Change in: Avoiding Dangerous Climate Change, Schellnhuber, H.J., et al.(eds.), Cambridge University Press, New York, NY, pp3-5 Rockstrm J., Steffen, W., Noone, K., Persson, A., Chapin, F.S, Lambin, E.F., Lenton, T.M., Scheffer, M., Folke, C., Schellnhuber, H.J., Nykvist, B., de Wit C.A.., Hughes T., van der Leeuw, S., Rodhe, H., Srlin, S., Snyder, P.K., Costanza, R., Svedin, U.,Falkenmark, M., Karlberg, L., Corell, R.W., Fabry, V.J., Hansen, J., Walker, B., Liverman, D., Richardson, K., Crutzen, P. and Foley, J.A., 2009a, A safe operating space for humanity, Nature, Vol 461, pp472-475 Rockstrm, J., W. Steffen, K. Noone, . Persson, F. S. Chapin, III, E. Lambin, T. M. Lenton, M. Scheffer, C. Folke, H. Schellnhuber, B. Nykvist, C. A. De Wit, T. Hughes, S. van der Leeuw, H. Rodhe, S. Srlin, P. K. Snyder, R. Costanza, U. Svedin, M. Falkenmark, L. Karlberg, R. W. Corell, V. J. Fabry, J. Hansen, B. Walker, D. Liverman, K. Richardson, P. Crutzen, and J. Foley, 2009b,Planetary boundaries: exploring the safe operating space for humanity, Ecology and Society, Vol 14, No 2, 32, http://www.ecologyandsociety.org/vol14/iss2/art32/ (last accessed 30/03/2010) Schneider, S. H. and Lane, J. 2006 An Overview of Dangerous Climate Change in: Avoiding Dangerous Climate Change, Schellnhuber, H.J., et al. (eds.), Cambridge University Press, New York, NY, pp7-23

20

AV/WS2/D1/R10

Tirpak et al 2005 International Symposium on the Stabilisation of greenhouse gas concentrations Report of the International Scientific Steering Committee Met Office http://www.stabilisation2005.com/Steering_Commitee_Report.pdf (last accessed 30/03/2010) UNFCCC 2000 Climate Change Information Sheet 19 http://unfccc.int/cop3/fccc/climate/fact19.htm (last accessed 30/03/2010)

Acknowledgements
With thanks to: the July 2009 Amazon workshop contributors; Met Office Hadley Centre colleagues John Casear, Mike Sanderson and Ben Booth for, respectively, help with observed precipitation, AR4 and QUMP data.

21

S-ar putea să vă placă și