Mike Hulme, Sarah C.B. Raper
Climatic Research Unit, School of Environmental Studies
University of East Anglia, Norwich, UK
Tom M.L. Wigley
Office for Interdisciplinary Earth Studies, Boulder, CO, USA
ESCAPE (the Evaluation of Strategies to address Climate change by Adapting to and Preventing Emissions) is an integrated climate change assessment model constructed between 1990 and 1992 for DGXI of the Commission of the European Community by a consortium of research institutes headed by the Climatic Research Unit (CRU). It has been designed to enable the user to generate future scenarios of greenhouse gas emissions (through an energy-economic model), examine their impact on global climate and sea level (through two independent global climate models), and illustrate some of the consequences of this global climate change at a regional scale for the European Community (through a regional climate scenario generator and impact models). We provide a very brief overview of the ESCAPE model which, although innovative, suffers from a number of major limitations. Subsequent work in the CRU has concentrated on improvements to the global climate module and work has also just commenced on an improved regional climate scenario generating module. These improvements will lead to a new integrated climate change assessment model, MAGICC (Model
_______________
*The work described here has been the result of thousands of person hours funded by a range of sponsors. Contracts from the US Department of Energy, the UK Department of the Environment, and DGXI and DGXII of the Commission of the European Community in particular are acknowledged. The current version of MAGICC owes much to Tom Wigley, but important contributions have been made by Sarah Raper, Tim Osborn, Mike Hulme, Tao Jiang, and Mike Salmon (software design). ESCAPE is the result of work undertaken by many different institutes and individuals, too numerous to mention here. Tom Wigley, Richard Warrick, Jan Rotmans, and Martin Parry played particularly important roles, however, in establishing the project concept.
for the Assessment of Greenhouse gas Induced Climate Change) which can easily be incorporated into new larger integrated frameworks developed by other institutes.
Predicting future global climate change requires an interdisciplinary perspective which encompasses the physical, social, and political sciences. Policy makers, charged both with identifying possible national response strategies to climate change and with negotiating international conventions and protocols, need tools which enable them to estimate the implications for climate and climate change impacts of a wide range of policy options and which can provide a concise overview of the uncertainties surrounding global climate change (Dowlatabadi and Morgan, 1993). A prototype tool has recently been developed for the Directorate General for Environment, Nuclear Safety and Consumer Protection (DGXI) of the Commission of the European Community. This model framework is called ESCAPE: the Evaluation of Strategies to address Climate change by Adapting to and Preventing Emissions. The model was developed over an 18 month period extending from November 1990 to May 1992, with contributions from over 15 institutions in Europe and North America and coordinated by the Climatic Research Unit at the University of East Anglia, UK. The interactive computerized framework allows one to explore the implications of different climate-related policies both for global-mean climate and for indicators of the economic and environmental impact of climate change within Europe. The model provides a clear assessment of the scientific uncertainties surrounding the prediction of future climate change and its impacts. ESCAPE is well documented; both a User Manual (CRU, 1992a) and a Scientific Description (CRU, 1992b) of the model are available on request. A paper summarizing the ESCAPE framework, including some illustrations of analyses undertaken with this model, has been prepared (Rotmans et al., 1994).
ESCAPE represents a pioneering attempt to provide an integrated framework to address global climate change. A number of major limitations to the model are recognizable. This paper is divided into two parts. First, a very brief overview of ESCAPE is presented and some of the limitations of the model for the purpose of integrated assessments are identified. Second, we report on more recent developments undertaken within the Climatic Research Unit to improve the global and regional climate modules which, when taken together, are now referred to as MAGICC (Model for the
Assessment of Greenhouse gas Induced Climate Change). The evolution of models which led to the creation of ESCAPE and MAGICC is summarized in Figure 1.
2.1. Methodology
ESCAPE consists of a suite of linked models (modules) which enables scenarios of greenhouse gas emissions to be constructed and their impact on
global and regional climate and sea level and sectors of the European economy to be assessed. Each module in turn consists of a number of coupled sub-models which are reformulated, simpler versions of more elaborate multi-dimensional models which have previously been constructed and described in the literature. In this way "state-of-the-art" science is captured within the ESCAPE framework. ESCAPE comprises four basic modules, which are shown in Figure 2:
The four basic modules are linked, but not fully integrated. They differ in complexity, spatial resolution, aggregation level, time step, etc. The default time step for ESCAPE is five years and the projection period is 1990 to 2100. The model users can modify the projection period and the time step according to their own purposes.
The different levels of spatial aggregation in the modules results in an "hourglass" structure to the model. The front-end of ESCAPE, the IMAGE-Emissions module, calculates emissions for four major world regions: the EC, the rest of the OECD, former centrally planned countries of Europe and Asia (CPC), and the rest of the world (ROW). The core part of ESCAPE, the IMAGE and STAGGER climate modules, uses global emissions projections to calculate global-mean changes in atmospheric concentrations, temperature and sea level. The final module, CLIMAPS, assesses the impacts of climate change for Europe at a resolution of 0.5° latitude by 1.0° longitude.
2.2. Uncertainties
Predicting future global climate change and its consequences for human society is beset with many uncertainties. These uncertainties may be separated into "scientific uncertainties" (some of which may be narrowed as a result of
further scientific research) and "economic uncertainties" (those which result from future geopolitical, socioeconomic, and demographic evolution and which are inherently "unknowable").
Scientific uncertainties include, for example, incomplete knowledge about the magnitude of the sources and sinks of the various greenhouse gases. For CO2 emissions, the contribution from fossil fuel combustion is well known (with an uncertainty of about 5%), but emissions from land use changes remain poorly known (uncertainties in the order of 50%, Leggett et al., 1992). With respect to the oceanic and terrestrial carbon sinks, the likely errors are of the order of 100%. The only well known component of the global carbon budget is the past atmospheric concentration of carbon dioxide. The gas-cycle uncertainties with respect to methane are even larger (although, since methane is not the dominant greenhouse gas, these are not so important with regard to actual climate change). While the overall magnitude of global methane emissions is reasonably well-known, estimates of methane emissions from individual sources are highly uncertain (Rotmans et al., 1992).
Another important source of uncertainty originates from our deficient knowledge of the key physiological, chemical, and biological processes within the climate system itself. Illustrative of this is the inadequate understanding of the many potential feedback responses to increasing atmospheric CO2 and rising temperatures (Vloedbeld and Leemans, 1993). Feedback processes can either amplify (positive feedback) or damp (negative feedback) the response of the climate system to anthropogenic greenhouse gas emissions and can be separated into geophysical and biogeochemical feedbacks.
The former feedbacks (for example, those due to water vapor and sea-ice) are caused by physical processes in the atmosphere-ocean-cryosphere system which directly affect the response of climate to radiative forcing. These determine the value of the climate sensitivity (defined as the equilibrium warming of global-mean surface air temperature resulting from a doubling of CO2 concentration) which is still poorly known (it probably ranges from 1.5 to 4.5°C). Reducing uncertainties in this parameter should have high priority, although the best strategy for achieving this is far from clear. The latter feedbacks (for example, the CO2-fertilization feedback) may affect the concentrations of the greenhouse gases themselves and thus the radiative forcing, and may also alter the response of the climate system to any given radiative forcing. New biogeochemical feedbacks continue to be identified and/or quantified, as witnessed by the negative forcing roles played by stratospheric ozone and fossil fuel related emissions of sulphur dioxide (Isaksen et al., 1992).
Despite these major gaps in our scientific knowledge of the response of the climate system to forcing, the most precarious scientific aspect of assessing the significance of global climate change is the estimation of its ecological
and socioeconomic impact. In many cases, even the direction of the impact cannot be stated confidently (Tegart et al., 1990). It is clear, therefore, that our knowledge of the phenomenon of climate change is far from complete. In view of these uncertainties, the interpretative and instructive value of an integrated model such as ESCAPE is far more important than its predictive capability, which is limited by the incomplete knowledge upon which it is constructed. The ESCAPE analysis was, however, the first major attempt to synthesize current scientific understanding of the causes and impacts of global climate change due to the enhanced greenhouse effect into one integrated framework.
2.3. Deficiencies and limitations of ESCAPE
The modeling framework of ESCAPE is a linked set of modules, rather than a single fully integrated model. Consequently, validation of the individual sub-models of ESCAPE is more important than an overall validation of the model. All the sub-models described here have been extensively tested and results from most of them have appeared in the scientific literature. Some of the outstanding limitations of the individual models are identified below. Some limitations of the end-use energy model are:
The land-use model within ESCAPE has a number of structural limitations:
With regard to the core climate modules IMAGE and STAGGER:
With respect to the CLIMAPS module:
Some of these limitations have been addressed in IMAGE 2 (which is a recently completed major revision by RIVM of the IMAGE 1 model) and through further developments which have been undertaken within the Climatic Research Unit (and which are now referred to as the MAGICC model). Other limitations represent more fundamental difficulties faced by all integrated modeling initiatives.
3.1. The global climate model
MAGICC (Model for the Assessment of Greenhouse gas Induced Climate Change) provides internally-consistent estimates of global-mean temperature and sea level change between 1990 and 2100 resulting from scenarios of anthropogenic emissions, viz. CO2, CH4, N2O, the halocarbons, and SO2. So that comparisons can be made, the user must specify two emissions scenarios. These are called the Reference scenario and the Policy scenario. The global-mean temperature projections may be used to scale General Circulation Model (GCM) results in order to produce regional scenarios of climate change in Europe (as in ESCAPE), or globally and for other regions (as in SCENGEN, see below).
MAGICC is an integrated model constructed of individual model components that are highly parameterized simple models which, nevertheless, capture the essential features of more complex models in the different fields of research. (Such models are often referred to as "reduced-form" models.) The integrated model is therefore computationally highly efficient, and takes only a few seconds to run on a personal computer. The model is designed to allow users to alter key model parameters in order to easily conduct sensitivity analyses.
One of the model components is an upwelling-diffusion climate model (Wigley and Raper, 1987) as used by the Intergovernmental Panel on Climate Change (Houghton et al., 1990; referred to below as IPCC90). The sea level rise models are also updated versions of the models used by IPCC90 (which we developed in conjunction with Warrick and Oerlemans, 1990), with improvements described in Raper et al. (1994) and Wigley and Raper (1993). The projections made in the IPCC90 report have subsequently been updated by Wigley and Raper (1992) incorporating new scientific results and the new set of greenhouse gas (GHG) emission scenarios produced by the IPCC in 1992 (Leggett et al., 1992). In order to do this, it was necessary to develop the gas-cycle models which are incorporated into MAGICC. These, and the other model components, are described briefly below.
A schematic representation of the integrated model is given in < a href = "MH1994Fig3.gif">Figure 3. Specified GHG emission scenarios are first used as input to gas-cycle models to obtain estimates of future atmospheric concentrations. The concentration projections for the individual gases may be examined graphically during the interactive session.
The carbon cycle model (Wigley, 1993) is based on a convolution ocean model (Maier-Reimer and Hasselmann, 1987) that allows the efficiency of atmospheric-to-ocean gas exchange to be varied. The terrestrial component of the carbon cycle model incorporates a CO2 fertilization feedback which enables the contemporary carbon budget to be altered to agree with observations and account for uncertainty in these observations (past emissions due to deforestation are particularly uncertain). The methane model (Osborn and Wigley, 1994) expresses the atmospheric lifetime of methane as a function of atmospheric composition. The effects of uncertainties in the present lifetime of methane and/or uncertainties in future lifetime changes may be explored. Also, the contribution to radiative forcing from stratospheric water vapor produced through the oxidation of methane may be included. The lifetime of nitrous oxide and the halocarbons can also be specified. Stratospheric ozone depletion feedback may be included. Default values for all these parameters are, however, provided.
A range of values for the negative forcing due to sulfate aerosols may be used to account for uncertainties in this factor. Because of the very short atmospheric lifetime of sulfate aerosols, the radiative forcing is expressed as a function of SO2 emissions, so that a gas-cycle model is not needed. For all gases, except SO2, the atmospheric concentrations resulting from the gas cycle models are converted to radiative forcing using radiative transfer model results following Shine et al. (1990). MAGICC allows the radiative forcing contributions of the various gases to be graphically compared.
The resulting total global-mean radiative forcing (partitioned by land/ocean and by hemisphere for the aerosol component) drives the upwelling-diffusion energy balance climate model. This model calculates hemispheric-mean vertical ocean temperature change profiles (40 levels from the surface to 4000 m) from 1765-2100. There are four model parameters which determine the rate at which heat is taken up by the ocean. However, the single most important parameter in determining uncertainty in future temperature and sea level is the climate sensitivity. The climate sensitivity may be defined as the equilibrium surface warming for a doubling of atmospheric carbon dioxide ([Delta]T2x) The IPCC best estimate of [Delta]T2x is 2.5°C with a range of 1.5 to 4.5°C. MAGICC runs the climate model six times for each scenario. All three values of [Delta]T2x are used with both best guess and user specified values for the other parameters. This generates a range of uncertainty for the temperature projections.
One of the contributing factors to future sea level rise is the thermal expansion of the ocean. Thermal expansion is calculated from the vertical ocean temperature change profiles predicted by the climate model. As
with surface temperature, results depend on the climate model parameter settings.
The contribution of the melting of land ice to sea level rise is also considered. Greenland, Antarctica and other small glaciers are treated separately. The Greenland and Antarctic melt models are expressed simply as a sensitivity to temperature change. Uncertainties due to their pre-industrial state (were Greenland and Antarctica in equilibrium in 1765?) and to the regional temperature changes of Greenland and Antarctica relative to global-mean temperature changes can also be explored. A parameter for the possible increased discharge from the West Antarctic Ice Sheet is included. However, the default parameters for Antarctica give a negative contribution to sea level rise because increased accumulation is expected to dominate the other responses to increased temperatures over the next century. The small glacier model includes uncertainty in the initial (1765) ice volume and keeps track of the volume of ice available for melting. In the case of this model, the rate of melting is determined both by a temperature sensitivity parameter and by a collective glacier response time.
To express the large range of uncertainty in the sea level rise projections, high, medium, and low ice melt parameter settings are run with the high, medium, and low temperature projections (due to [Delta]T2x uncertainty), respectively. Finally, a comprehensive set of results, both tabular and graphic, may be examined on screen or as hard copy.
To illustrate the use of MAGICC, Figures 4 to 7 and Table 1 result from choosing the central 1992 IPCC emissions scenario (IS92a) as the Reference Scenario and a Policy Scenario which assumes stabilization of global fossil CO2 emissions at 1990 levels by 2000 (CO2STAB).1 For simplicity, default parameter values are used throughout. However, a range of uncertainty is automatically explored by MAGICC. The resulting best guess temperature and sea level changes, from 1990 to 2100, are 2.47°C and 45.4 cm for IS92a and 1.42°C and 31.0 cm for CO2STAB.
The Climatic Research Unit is continually updating MAGICC. Recent changes to the carbon cycle model (Wigley, 1993) are to be added shortly. These will affect the temperature and sea level predictions. For example, for IS92a, the best guess 2100 atmospheric CO2 concentration is revised downwards from about 740 ppmv to about 680 ppmv. Preliminary results indicate
_______________
1It should be noted that in this scenario no change was made to the SO2 emissions. This is unrealistic since if CO2 emissions were stabilized, then a reduction in SO2 emissions would also occur. The net forcing of the CO2STAB scenario may therefore be slightly too small.
that there will be a decrease in the corresponding best guess temperature projection from 2.5°C to 2.1°C.
3.2. The regional climate change scenario generator (SCENGEN)
The current version of MAGICC uses CLIMAPS-Europe (Version 3.2a, January 1994) as its regional scenario generator and impact module. This is a slightly adapted version of the CLIMAPS module (Version 3.1, June 1992) used in ESCAPE Version l.la, and models only Europe. The regionalization algorithm in CLIMAPS is a development of that first proposed by Santer et
al. (1990). Grid point changes in climate for a set of GCM experiments are first normalized by the GCM climate sensitivity (or some equivalent global-mean temperature change for transient results from coupled O/AGCMs) and then averaged to obtain composite patterns of monthly-mean changes of precipitation, temperature, etc., per °C of global-mean warming. These normalized patterns are then scaled up by the transient global-mean temperature change for any selected future year obtained from the upwelling-diffusion model. It is not possible to fully justify this procedure in the space available here, except to note that the composite model control climatologies
obtained in this way validates better for global and regional precipitation patterns than any individual model used in the compositing.
Future versions of MAGICC will be able to be used in conjunction with SCENGEN, a new global and regional scenario generator which is currently under development in the Climatic Research Unit. It has been designed to allow the user full scope to generate global and regional scenarios of climate change based on a wide range of GCM experimental results of his/her own choosing. Options exist to select scenarios based either on single GCMs or on groups of GCMs. The scenarios may be presented simply as change fields for a given global-mean warming or else added to a baseline climatology to generate an actual future climatology. In Version 1.0 of SCENGEN, global scenarios at 5° resolution and regional scenarios for North America
at 1° resolution can be generated. Future versions will add high resolution windows for Europe, East Asia, and Australasia. Scenarios can be viewed on screen and then dumped as ASCII data files for input into other analyses.
As a stand-alone module, SCENGEN can be driven by only one global-mean temperature projection for which SCENGEN can produce a range of regional patterns reflecting inter-GCM differences in the spatial patterns of climate change. SCENGEN has been designed, however, to be used within MAGICC. The entire MAGICC framework will therefore offer the user complete flexibility about the choice of emissions scenario, a range of global-mean temperature projections and then a choice about the origin of the global or regional climate change scenario generated.
Table 1. Example model parameter settings for MAGICC.
Policy scenario: CO2STAB
Reference scenario: IS92A
Gas Cycle Parameters (default)
Sulfate Forcing: medium
Climate Parameters (default)
Sensitivity (deltaT2x): 2.500 °C Diffusivity (K): 1.000 cm2/sec Mixed Layer Depth (h): 90.000 m Sinking Water DeltaT Factor (pi): 0.200 Upwelling Rate (W): 4.000 m/yr
Sea Level Parameters (default)
Small Glacier Sensitivity: 0.250 °C Small Glacier Response Time: 0.250 years Greenland Sensitivity: 0.030 cm/yr/°C Antarctic B1: -0.030 cm/yr/°C Antarctic B2: 0.010 cm/yr/°C Antarctic Constant: 0.000 Initial Small Glacier Mass: 50.000 cm (sea-level equivalent)
Two spatial scales of display will be used, a global resolution of 5° latitude/longitude and a series of pre-defined regional windows at 1° latitude/longitude resolution. At a global scale, mean monthly, seasonal and annual precipitation and mean surface air temperature change fields at 5° resolution will be displayed. These change fields can be determined from a suite of eleven (currently) GCM experiments (eight equilibrium and three
transient), either based on individual GCMs or on user-selected composites. The minimum number of GCMs for a model-composite scenario will be five. Mean global pattern correlation coefficients for precipitation will be displayed to guide the user's GCM selection and these coefficients will be used as weights in the resulting precipitation composite. Composite scenarios will also display a 90% confidence interval around the composite "central estimate" based on inter-GCM pattern differences (Santer et al., 1990). The change fields can be added to a global baseline climatology to generate "actual" climatologies for future years. These global baselines will exist for land areas only and will be based on the period 1961-1990.
In Version 1.0 of SCENGEN, only a North American high resolution window will be active. Options will be identical as for the global fields above, except that a wider range of variables can be selected. In addition to mean temperature and precipitation, maximum and minimum temperature, mean 10m wind speed, total cloud amount, and specific humidity will be displayed. The GCM change fields are stored at 5° resolution. For high resolution windows, these will be interpolated down to 1° resolution. The baseline climatologies will be held at 1° resolution. For any displayed scenario, the user will be able to request an output file to be created under the Operating System which will contain the actual scenario data- these files can then be used an inputs for other analyses.
A number of integrated assessments of global climate change are now being undertaken and a variety of model tools exist which address the problem of global climate change. There are a large number of questions which need to be answered at the outset of an integrated assessment modeling exercise. Three of the more important ones are:
There are no "right" answers to these questions. A diversity of approaches - and hence models and hence integrated assessments - is likely to be the most beneficial strategy to follow at the present time. In this paper we have discussed briefly one such integrated assessment model (ESCAPE) and have described recent developments concerning two components of the problem (global and regional climate change) that any integrated model needs to incorporate.
Climatic Research Unit (CRU), 1992a, Development of a Framework for the Evaluation of Policy Options to Deal with the Greenhouse Effect: A User Manual for the ESCAPE Software: Version 1.1, Climatic Research Unit, Norwich, UK.
Climatic Research Unit (CRU), 1992b, Development of a Framework for the Evaluation of Policy Options to Deal with the Greenhouse Effect: A Scientific Description of the ESCAPE Model, Version 1.1, Climatic Research Unit, Norwich, UK.
Dowlatabadi, H. and Morgan, M.G., 1993, Integrated Assessment of Climate Change, Science 259:1813, 1932.
Environmental Resources Limited (ERL), 1992, Development of a Framework for the Evaluation of Policy Options to Deal with the Greenhouse Effect: Economic Evaluation of Impacts and Adaptive Measures in the European Community, May 1992, London, UK.
Hope, C., Anderson, T., and Wenman, P., 1993, Policy Analysis of the Greenhouse Effect: An Application of the PAGE Model, Energy Policy 21:327-337.
Houghton, J.T., Callander, B.A., and Varney, S.K., eds., 1992, Climate Change 1992: The Supplementary Report to the IPCC Scientific Assessment, Cambridge University Press, Cambridge, UK.
Houghton, J.T., Jenkins, G.J., and Ephraums, J.T., eds., 1990, Climate Change: The IPCC Scientific Assessment, Cambridge University Press, Cambridge, UK.
Intergovernmental Panel on Climate Change (IPCC), 1991, Climate Change: The IPCC Response Strategies, WMO/UNEP, Geneva, Switzerland.
Isaksen, I.S.A., Ramaswamy, V., Rodhe, H., and Wigley, T.M.L., 1992, Radiative Forcing of Climate, pp. 47-68 in J.T. Houghton, B.A. Callander, and S.K. Varney, eds., Climate Change 1992: The Supplementary Report to the IPCC Scientific Assessment, Cambridge University Press, Cambridge, UK.
Leggett, J., Pepper, W.J., and Swart, R.J., 1992, Emissions Scenarios for the IPCC: An Update, pp. 69-96 in J.T. Houghton, B.A. Callander, and S.K. Varney, eds., Climate Change 1992: The Supplementary Report to the IPCC Scientific Assessment, Cambridge University Press, Cambridge, UK.
Maier-Reimer, E. and Hasselmann, K., 1987, Transport and Storage of Carbon Dioxide in the Ocean: An Inorganic Ocean-Circulation Carbon Cycle Model, Climate Dynamics 2:63-90.
Osborn, T.J. and Wigley, T.M.L., 1994, A Simple Model for Estimating Methane Concentration and Lifetime Variations, Climate Dynamics 9:181-193.
Raper, S.C.B., Wigley, T.M.L., and Warrick, R.A., 1994, Global Sea Level Rise: Past and Future, in J.D. Milliman, ed., Proceedings of the SCOPE Workshop on Rising Sea Level and Subsiding Coastal Areas, John Wiley and Sons, Chichester, UK, in press.
Rotmans, J., Den Elzen, M.G.J., Krol, M.S., Swart, R.J., and Van der Woerd, H.J., 1992, Stabilizing Atmospheric Concentrations: Towards International Methane Control, Ambio 21:404-413.
Rotmans, J., Hulme, M., and Downing, T.E., 1994, Climate Change Implications for Europe: An Application of the ESCAPE Model, Global Environmental Change, in press.
Santer, B.D., Wigley, T.M.L., Schlesinger, M.E., and Mitchell, J.F.B., 1990, Developing Climate Scenarios from Equilibrium GCM Results, Max Planck Institute for Meteorology Report No. 47, Hamburg, Germany.
Shine, K.P., Derwent, R.G., Wuebbles, D.J., and Morcrette, J.-J., 1990, Radiative Forcing of Climate, pp. 41-68 in J.T. Houghton, G.J. Jenkins, and J.J. Ephraums, eds., Climate Change: The IPCC Scientific Assessment, Cambridge University Press, Cambridge, UK.
Tegart, W.T., Sheldon, G.W., and Griffiths, D.C., eds., 1990, Climate Change: The IPCC Impacts Assessment, Australian Government Publishing Service, Canberra, Australia.
Vloedbeld, M. and Leemans, R., 1993, Quantifying Feedback Processes in the Response of the Terrestrial Carbon Cycle to Global Change: The Modeling Approach of IMAGE 2, Water, Air and Soil Pollution 70:615-628.
Warrick, R.A. and Oerlemans, H., 1990, Sea Level Rise', pp. 257-282 in J.T. Houghton, G.J. Jenkins, and J.J. Ephraums, eds., Climate Change: The IPCC Scientific Assessment, Cambridge University Press, Cambridge, UK.
Wigley, T.M.L., 1993, Balancing the Carbon Budget: Implications for Projections of Future Carbon Dioxide Concentration Changes, Tellus 45B:409-425.
Wigley, T.M.L., Holt, T., and Raper, S.C.B., 1991, STUGE: An Interactive Greenhouse Model, User's Manual, Climatic Research Unit, Norwich, UK.
Wigley, T.M.L. and Raper, S.C.B., 1987, Thermal Expansion of Sea Water Associated with Global Warming, Nature 330:127-131.
Wigley, T.M.L. and Raper, S.C.B., 1992, Implications for Climate and Sea Level Rise of Revised IPCC Emissions Scenarios, Nature 357:293-300.
Wigley, T.M.L. and Raper, S.C.B., 1993, Future Changes in Global Mean Temperature and Thermal-expansion-related Sea Level Rise, pp. 111-133 in R.A. Warrick, E.M. Barrow, and T.M.L. Wigley, eds., Climate and Sea Level Change: Observations, Projections and Implications, Cambridge University Press, Cambridge.
