MAGICC

(Model for the Assessment of Greenhouse-gas Induced Climate Change)

User's Guide and Scientific Reference Manual

CONTENTS

1.  Introduction

2.  A Teaching Example
     2.1 Background
     2.2 Selecting the Policy and Reference emissions scenarios
     2.3 Editing the gas cycle model parameters
     2.4 Editing the climate and sea-level model parameters
     2.5 Running MAGICC and viewing tabulated results
     2.6 Viewing the graphical output
           2.6.1 Gas Emissions
           2.6.2 Gas Concentrations
           2.6.3 Radiative Forcing
           2.6.4 Temperature and Sea Level Change
     2.7 Greenland Ice-melt

3.  Additional Operational Items
     3.1 Editing the emissions library
     3.2 Gas lifetimes and radiative forcing sensitivities
     3.3 Gas cycle model options
     3.4 Changing climate model parameters
     3.5 Stratospheric water vapour forcing

4.  Scientific Details
     4.1 Introduction
     4.2 Carbon Dioxide
     4.3 Methane
     4.4 Nitrous Oxide
     4.5 Halocarbons
     4.6 Sulphate Aerosols
     4.7 The Climate Model
     4.8 Ice-melt Models

5.  Acknowledgments

6.  References

1. INTRODUCTION

• To compare the implications of a "policy" emissions scenario with a "reference" scenario, in order to see how effective a particular emissions control policy is relative to a base-line "no policy" case.
• To determine the sensitivity of results to model parameter changes. The latter may be done by comparing results from a given emissions scenario for a user-specified set of parameters with those for a default¹ set of parameters.

The input emissions scenarios require values to be specified at 11 discrete dates between 1990 and 2100 (inclusive) for the following: fossil CO2, net land-use-change CO2, CH4, N2O, CO, NOx, VOCs, CFC11, CFC12, HCFC22, HFCl34a⁴ and SO2. The primary inputs to MAGICC are the user-selected policy and reference emissions scenarios. These are selected from an emissions scenario library (both scenarios may be the same if desired). The next step is to select either default or user-specified model parameters for the gas cycle, climate model and sea level (ice-melt) model parameters.

MAGICC executes four complete model runs over 1765-2100. To explain what these are and how they arise, consider the two initial question types. First, if a policy analyst wished to evaluate the climate effects of a particular emissions policy, he or she would choose to compare a Policy emissions scenario with a background Reference emissions scenario. For climate and sea level model parameters, it would be sufficient to use only the current best guess¹ set of values by choosing the Default¹ model parameter options. The two sets of results could be labeled PD and RD.

Alternatively, a scientist or educator may be interested in examining the sensitivity of the various models that comprise MAGICC to model parameter assumptions. In this case, he or she could choose the same emissions scenario for the policy and reference case, and then select a specific set of model parameters (User values) that differed from the default values. If R is used for the single emissions scenario, then the results could be labeled RU and RD, where RD is the same as in the first example. In general, one can combine these two types of application and produce four output data sets, PD, RD, PU and RU.

In addition to running these four cases, MAGICC estimates uncertainty ranges due to model parameter uncertainties relative to the Default model parameter set. To do this, each run calculates four sets of CO2 concentrations and four sets of CH4 concentrations (over 1990-2100) corresponding to low, default (best-guess), high, and user-specified model parameter sets. For gas concentrations, there are therefore two primary sets of model-generated data for each emissions scenario, corresponding to user-selected gas-cycle model parameters (in RU and PU) and default model parameters (in RD and PD). Both concentration data sets are used to force the climate model, which is an upwelling-diffusion energy-balance climate model. In each of the four climate model simulations, the climate and sea level models are run three times corresponding to low, mid¹ or user, and high model parameters values. Table 1 give a summary of the runs.

Notes:

(1) L, M, H. U denote Low, Mid (i.e., best guess or default), High and User-selected model parameters

(2) In the Default cases, the M values are computed twice for CO2 and CH4 (i.e., the U mode in the User case is replaced by M)

(3) L, M and H for the temp/mean-sea-level models correspond to climate sensitivities of 1.5 deg. C, 2.5 deg. C and 4.5 deg. C, best-guess values for other upwelling-diffusion model parameters, and values for all ice-melt model parameters (bar the initial "small" glacier mass) that produce low, mid and high ice-melt (i.e. low climate sensitivity goes with low ice melt, etc.)

(4) For the User case, the best-guess climate and sea-level model parameter results are only produced if specifically selected by using the appropriate de bult options.

(5) Only those items in bold are shown in the graphical displays.

Input and output are displayed both graphically and in tabular form. Convenient input and output summary tables may be viewed on screen (and printed directly from the screen) for the user model-parameter cases, RU and PU (default model values will, of course, be shown if default model parameter selections are made initially). More detailed output for all four cases may be viewed or printed from DOS in the files MAGOUTRD.DAT, MAGOUTRU.DAT, MAGOUTPD.DAT and MAGOUTPU.DAT. Graphical displays are given for input emissions, and output concentrations, temperature, sea level, and radiative forcing (gas by gas). For CO2 and CH4 concentration, uncertainties are indicated by showing the low and high model-parameter cases. Uncertainties for temperature and sea level are illustrated similarly (for the default concentration case only).

2. A TEACHING EXAMPLE

To understand the overall operation of MAGICC, the reader is referred to the accompanying flow chart (Fig. 1). To put flesh on this skeleton, we consider a specific example. This is a rather complex example, chosen to illustrate a reasonable selection both of MAGICC's features and of the questions that may be addressed using MAGICC. As noted in the Introduction, two different types of question may be addressed: what is the effect of a particular emissions control policy; and how do the results vary if model parameter values are altered? Both will be considered simultaneously here.

The first question is: how much climate and sea level change might be averted by a specific emissions control strategy? To answer this we compare results for a reference and a policy emissions scenario.

For the reference emissions scenario we choose the 1992 IPCC scenario, IS92a (Leggett et al., 1992). This is IPCC's central "existing policies" scenario; as such, it is often used as a reference case. For the policy scenario, we choose IS92d. This is also an "existing policies" scenario (Leggett et al., 1992): however, it is based on different background assumptions for population growth, economic growth, etc. that lead to lower overall emissions. Nevertheless, it can serve well as an example of what might be achieved if one assumed that IS92a background assumptions applied and that specific policies were introduced to reduce emissions of all gases¹. The emissions for these scenarios are shown in Tables 2 and 3.

As a second (multi-part) question we ask, how sensitive are the results to particular model parameter choices? The details of this question are shown by the user-defined set of parameters listed in Table 4. Because we have already selected different Policy and Reference scenarios, MAGICC will provide answers to this second question for both emissions cases.

• For CO2, we ask what effect the use of the most recent IPCC information on the carbon budget (Schimel et al., 1994) has on future CO2 concentrations and radiative forcing compared with using IPCC90 information (from Watson et al., 1990).

• For CH4, we investigate the effect of assuming a constant lifetime (of 11 years) compared with the range of uncertainty for the variable lifetime case.

• For N2O and CFC11, we compare results for IPCC94 and IPCC90 lifetimes.

• For SO2, we consider the effect of assuming a low value for the aerosol forcing.

• For the climate sensitivity we consider a value near the upper limit of the most likely range.

• For ice-melt, we look specifically at the contribution from Greenland using a high value for the mass-balance sensitivity.

Click for Table 4.

*Net land-use-change emissions averaged over the 1980s. This is a proxy for the strength of the CO2 fertilization effect--higher values require higher fertilization to balance the contemporary carbon budget. Further details are given in Wigley (1993) and in the text.

**See the following Technical Details section. The default model has a lifetime that varies with the concentrations of CH4, CO, NOx, and VOCs. Further details are given in Osborn and Wigley (1994).

2.2 Selecting the Policy and Reference emissions scenarios

The first step is to move into the MAGICC directory by typing CD\MAGICC and hitting the 'Enter' key. Typing MAGICC and 'Enter' will then display the first MAGICC screen. The choices then are: to enter the climate/sea level model part of MAGICC; to calculate Global Warming Potentials (GWPs); to view a help screen; to access the model documentation; or to quit. The GWP calculations in this version of MAGICC are out-of-date, and should not be used.

Use the mouse to move the arrow to the Climate Model button and then click on the left mouse button (or type 'C'). This will move you to the next (CLIMATE MODELLING) screen¹.

All choices are made through the CLIMATE MODELLING screen. The first task is to select and install a policy scenario and a reference scenario². These may be selected by clicking on the 'Change' buttons at the top of the screen (or keying 'A' or 'B'). Clicking on 'Change' for the policy scenario leads to the CHANGE POLICY SCENARIO screen. To select a new policy scenario, move the mouse arrow to anywhere on the appropriate line (IS92d in this case), and click. Clicking on Okay (or keying 'O') selects the chosen scenario and returns the user to the CLIMATE MODELLING screen. Clicking on Cancel ignores the selection and returns the user to the CLIMATE MODELLING screen retaining the original scenario selection. The Help button on this screen has not been activated and should be ignored. After returning to the CLIMATE MODELLING screen, repeat the above process for the reference scenario to install IS92a.

2.3 Editing the gas cycle model parameters

Back at the CLIMATE MODELLING screen, the next task is to set the appropriate gas-cycle model parameters (viz. lines 1,2,3 and 4 in Table 4). To do this, move the mouse and click on the User Defined button in the Gas Cycle Parameters panel (or type '2'). This brings up the GAS CYCLE MODEL PARAMETERS screen.

In the Carbon Cycle Model panel, we wish to select a user-defined value for the 1980s-mean value of net land-use-change emissions (Dn(1980s)) of 1.6 GtC/yr; i.e., the best-guess value recommended in the 1990 IPCC report (the current best guess value is 1.1 GtC/yr). First, click on the User button (or type 'S'). Then, using the move-right arrow on the keyboard, move the edit bar in the Dn(1980s) box to the right of the first decimal place, key 'Backspace' three times to delete the numbers to the left of the edit bar, and type in 1.6 (the box should now read 1.600).

Now move the mouse to the Methane Model panel and choose 'Constant' (or type 'O') on the CH4 lifetime row. Next, to select a fixed lifetime value, click on the edit button (or type #). This will bring up the GAS LIFETIMES AND DQ/DC screen. This screen allows one to change lifetimes and forcing sensitivities (dQ/dC) for a large selection of gases. For climate model calculations, however, only changes in CH4, N2O, CFC11, CFC12, HCFC22 and HFCl34a will be used--the other gases are included for GWP calculations. To change the methane lifetime, use the move-right arrow to move the edit bar to the right of the first decimal place, delete appropriately using the 'Backspace' key, and type in the new value. Clicking on Okay returns you to the GAS CYCLE MODEL PARAMETERS screen. You will see that the revised lifetime is now displayed below the 'Constant' button.

The next task is to alter the N2O and CFC11 lifetimes. To do this, move the mouse to the N2O and Halocarbons box and click on Edit (or type 'L'). This will cause the GAS LIFETIMES AND DQ/DC screen to be displayed again showing N2O details on the right. To edit details for any gas, click the mouse on the appropriate line on the left-hand display, and then edit the lifetime as described above for methane. To enter the change, click on the Okay button. This will return you to the GAS CYCLE MODEL PARAMETERS screen. The process of editing and returning to the GAS CYCLE screen must be repeated for each gas--it is only possible to edit a single gas at a time in the GAS LIFETIMES screen. It is, however, possible to edit both the Lifetime and dQ/dC at the same time for any particular gas.

Having completed the N2O and CFC11 lifetime changes and returned to the GAS CYCLE screen, the final task is to alter the aerosol forcing. Here, the user has only four choices, a best-guess ('medium') value, low and high values which are 0.5 and 1.5 times the medium value (following Wigley and Raper, 1992), and zero. Move the arrow to the SO2 forcing box and click on 'Low'.

2.4 Editing the climate and sea-level model parameters

The next step is to change the climate sensitivity, specified in MAGICC by the value of the equilibrium global-mean temperature change for a CO2 doubling ([[Delta]]T2x). To do this, return from the GAS CYCLE screen to the CLIMATE MODELLING screen by clicking on the Okay button. Next, move the mouse to and click on the User defined button in the Climate Model Parameters panel. This brings up the CLIMATE MODEL PARAMETERS screen, where the user can change the five main climate model parameters; viz. [[Delta]]T2x, the vertical diffusivity in the ocean (K), the mixed-layer depth (h), the upwelling rate (w) and the sinking-water to global-mean temperature change ratio [[Pi]]).

Editing climate model parameters is carried out as previously. The left and right arrow keys move the edit bar accordingly, 'Enter' moves the bar to the next box, and 'Backspace' deletes the figure to the left of the edit bar. New numbers appear to the left of the edit bar when typed on the keyboard. In the current example, only [[Delta]]T2x should be edited, from its default value of 2.5 deg. C to 4.0 deg. C. Now return to the CLIMATE MODELLING screen using the Okay button.

The next task is to edit the value of the Greenland sensitivity (i.e., the amount of ice melt per year per unit amount of warming). To do this, move the mouse arrow to the Sea Level Model Parameters panel and click on the User defined button (or type '6'). This will bring up the SEA LEVEL MODEL PARAMETERS screen, containing the parameters that determine the ice-melt (or, more strictly, mass balance change) contributions to sea level change. These contributions come from "small" glaciers (three parameters), Greenland (one parameter) and Antarctica (three parameters). Editing is carried out as for the climate model parameters: the Enter key moves the edit bar from item to item, the left and right arrows move the edit bar within an item, and Backspace deletes characters to the left of the edit bar. Move to the Greenland sensitivity box using Enter, key the right arrow four times to get to the right of 0.03 (the default value), and change the 3 to 5. (This screen contains convenient buttons to restore individual default values (A through G), or to select the full set of high or low values (X and Y)). Now click on Okay to return to the CLIMATE MODELLING screen.

2.5 Running MAGICC and viewing tabulated results

To run MAGICC, move the mouse arrow and click on the Run model button (or type 'R'). While running, the screen will display

			Running Climate model...
			(Policy case...)

			(Reference case...) 

On the left side of this screen (and of all the output table screens), there is a set of options that allows the user to. . .

• print the displayed Table
  
• view specific output tables for either the Policy or Reference scenarios
  
• move to the screen giving graphical display choices
  

• the gas-by-gas radiative forcing changes (together with global-mean temperature and sea level changes)
  
• global-mean temperature and sea level changes, with a breakdown of the sea level change components (together with total forcing and equilibrium temperature changes)
  
• gas concentration changes.
  

The output tables screen also has a panel labeled "Regional Climate" which connects MAGICC to a regional climate change scenario generator called SCENGEN. SCENGEN is still at a development stage and will not be available until early 1995.

To return to the CLIMATE MODELLING screen, click on the Okay button.

2.6 Viewing the graphical output

Having viewed the tabulated (user model) results, clicking on the Graphs button will bring up the INPUT AND OUTPUT GRAPHS screen, where the user can choose to view the input emissions, or the output concentrations, temperature changes, sea level changes or radiative forcing changes. Clicking on Okay will return you to the previous screen.

2.6.1 Gas Emissions

Clicking on Gas Emissions (or typing 'E') will display a panel with eight buttons for CO2, CH4, N2O, CFC11, CFC12, HCFC22, HFC134a and SO2. These allow the user to view the emissions input for any of these gases for both the Policy (green line) and Reference (purple dots) scenarios. For example, clicking on CO2 displays both fossil emissions (fossil fuel plus cement production) and net land-use-change emissions (labeled here as deforestation). Note the large differences in fossil emissions between the two scenarios. Clicking on Okay returns you to the INPUT AND OUTPUT GRAPHS screen. As another example, clicking on HFC134a will display emissions for equivalent HFC134a, as explained elsewhere in this document.

In all of these displays, clicking on Print will print the displayed graph on an HP Paintjet printer. Other colour printer drivers will be added later. Clicking on Okay returns the user to the INPUT AND OUTPUT GRAPHS screen.

2.6.2 Gas Concentrations

Clicking on the Gas Concentrations button (or typing 'C') will display a control panel for the seven greenhouse gases.

CO2

One can see from the plot that the user model, which uses the IPCC90 value for Dn(1980s), produces concentrations noticeably less than for the current best-guess value of Dn(1980s): for IS92a, 679 ppmv at the end of 2100 compared with 714 ppmv; for IS92d, 523 ppmv at the end of 2100 compared with 547 ppmv, differences that amount to about 0.3 W/m² in terms of radiative forcing. Full details of the carbon cycle model results are given in the MAGOUT files, MAGOUTRD.DAT, *RU.DAT, *PD.DAT and *PU.DAT. These files also contain full details of all output information.

Clicking on Print will print the displayed graph on an HP Paintjet printer. Clicking on Okay returns the user to the INPUT AND OUTPUT GRAPHS screen.

CH4

Clicking on Print will print the displayed graph on an HP Paintjet printer. Clicking on Okay returns the user to the INPUT AND OUTPUT GRAPHS screen.

N2O and halocarbons¹

The Print and Okay buttons act as for CO2 and CH4.

2.6.3 Radiative Forcing

In terms of their climate-change significance, concentration results may be deceptive. The crucial parameter is the radiative forcing. To view these results, click on the Radiative forcing button. This will bring up a 6x4 control panel that allows the user to display forcing changes gas by gas (or totaled) for each of the two scenarios and for the default- and user-model cases. Additional buttons are provided to switch on or off all results, and to switch the key on or off. The Print button allows the user to print results on a HP Paintjet printer, while the Okay button returns the user to the previous (INPUT AND OUTPUT GRAPHS) screen.

It is interesting to compare the effect of the default- versus user-model forcing results for CO2, CH4², N2O and SO2 (i.e., sulphate aerosols) in the reference emissions scenario case (IS92a). This can be done by clicking on the buttons labeled I, J. K, M, R. S. T and V (or typing these letters). The full set pr 2100 forcing values (extracted from the MAGOUTR*.DAT files) is...

Click for data values.

It is clear from the plots, and the above Table, that CO2 and aerosols contribute the greatest uncertainty, with compensating effects in this case. If you now click on the Total buttons (N and W), you will see that the differences in total forcing resulting from the full set of user model-parameter changes is very small (see Table above)--additional compensations occur with CH4 and N2O.

Finally, clicking on the "switch all off" (X) button and then displaying the four Totals (using buttons G. F. N and W) will show the overall forcing differences between the two emissions scenarios for the default- and user-model cases. The differences in both cases are roughly 1.5 W/m² (i.e., IS92d represents a reduction in forcing of some 30% relative to IS92a). This is a large change, the result of even larger changes in the emissions of greenhouse gases for the Policy scenario relative to the Reference scenario (recall, for example, that the fossil CO2 emissions level in 2100 for IS92d was 10.4 GtC/yr, compared with 20.4 GtC/ yr for IS92a- see Tables 1 and 2). Most of the reduction in forcing for IS92d relative to IS92a is due to CO2, with the effects of emissions reductions in other gases largely offset by the reduction in SO2 emissions (which, globally, causes an increase in radiative forcing).

2.6.4 Temperature and Sea Level Change

Return to the INPUT AND OUTPUT GRAPHS screen by clicking on Okay. Clicking on the Temperature and Sea Level Changes button will bring up the Temperature and Sea Level panel with the former button activated. First, click on the best guess buttons for the Policy and Reference scenarios. This will show the relatively small effect of the Policy on global-mean temperature (a reduction in the year 2100 from 2.3 deg. C to 1.8 deg. C). A decrease in the CO2 emissions change over 1990-2100 of 70% (20.4-6.1 GtC/yr to 10.46.1 GtC/yr) translates to a 30% reduction in radiative forcing, which translates to only a 20% reduction in global warming. Clicking now on the Sea level button will show that the averted sea level change amounts only to about 6 cm (i.e., 14%).

Next, the user may view the same results for the user-model case. First, click on the two user-model buttons to display the sea level results relative to the default model-parameter cases. Two results are immediately apparent: the user-model results are much higher; and the percentage reduction in sea level is similar in both cases. The user-model results are higher mainly because of the higher climate sensitivity used (4.0 deg. C compared with 2.5 deg. C)--we saw earlier that the user-model effects on radiative forcing were relatively small; indeed, negligible for IS92a. Clicking on Temperature shows the equivalent results for global-mean temperature.

Finally, an indication of uncertainty may be obtained by first clicking on the user-model buttons to remove these items from the temperature display, and then clicking on the range buttons. The uncertainties displayed by the dotted fields are those arising from uncertainties in climate sensitivity only. The central values are for [[Delta]]T2x = 2.5 deg. C, while the extremes correspond to [[Delta]]T2x = 1.5 deg. C and 4.5 deg. C. Uncertainties are much larger than the effect of the Policy. Clicking now on the Sea level button will display the same fields for sea level change. The uncertainty ranges are much larger than for temperature¹¹.

Note that these uncertainty ranges apply only to the individual simulations. They do not apply to the difference between the individual best guess results--indeed, it is likely that the uncertainty in the difference is very much smaller than the individual uncertainties and substantially less than the difference itself.

2.7 Greenland Ice-melt

One final model parameter change was made, namely an increase in the Greenland ice-melt sensitivity from 0.03 cm/yr/deg. C to 0.05 cm/yr/deg. C. Because of the simple structure of the Greenland ice-melt model (see Wigley and Raper, 1993), this must have a directly proportionate effect on the Greenland melt contribution to sea level change. It is not, however, possible to isolate this from MAGICC, nor to do so using the MAGOUT*.DAT files in DOS. Before showing how to get this specific information, consider the MAGOUT files. More specifically, consider the Reference scenario case and examine MAGOUTRD.DAT relative to MAGOUTRU.DAT.

MAGOUTRD.DAT and MAGOUTRU.DAT both contain climate and sea level output for three simulations, low, mid and high model parameter results for the former, and low, user and high model parameter results for the latter: see the Table below...

The specific effect of changing the Greenland sensitivity parameter cannot be isolated because it is (see simulation 2 column) concatenated with the user gas-cycle parameter values and the user choice of [[Delta]]T2x = 4.0 deg. C.

To extract the effect, new simulations must be carried out isolating this particular model parameter change. First, return to the CLIMATE MODELLING screen and click on Defaults for the Gas Cycle Parameters and the Climate Model Parameters. This will set all model parameters back to their default values^l, except for the Greenland sensitivity in the Sea Level Model Parameters set. You should examine this screen to check that this is the case. Now click on Run model.

After MAGICC has run, the INPUT AND OUTPUT GRAPHS screen can be selected by clicking on the Graphs button. Select the Temperature and Sea Level Changes screen, and then select the Sea level graphical output. The effect of the ice-melt sensitivity change can be viewed by comparing the best-guess and user-model outputs for either the Policy or Reference scenario--in both scenarios the difference is an increase of about 4-5 cm for the user-model parameter case in the year 2100. For a more precise value, return to DOS and compare MAGOUTRD.DAT with MAGOUTRU.DAT: see Table below...

For MAGOUTRD, the results are the same as previously, since both are default cases. For MAGOUTRU, the only difference is in the simulation 2 column, where Greenland melt increases from 7.0 to 11.6 cm. The proportional increase is 5/3, the ratio of the user sensitivity parameter choice to the default value.

3. ADDITIONAL OPERATIONAL ITEMS

The emissions library may be edited or added to by the user. To do this, return to DOS. All emissions scenario files are of the form *.GAS. To add a new scenario, the easiest method is to copy the existing file that is most similar to the new scenario to a new, appropriately-name file and then to edit this file using the editor of your choice. In doing so, note that the first two lines of the file give the file name (i.e., the * in *.GAS) and a plain language description of the file (up to 80 characters). These should be edited appropriately, since both bits of information appear in the library description displayed by MAGICC. Note also that MAGICC expects to read 11 lines of input data. The years specified are up to the user, but there must be 11 data lines. In running MAGICC, the active *.GAS file is copied to GASEM.$$$.

3.2 Gas lifetimes and radiative forcing sensitivities

These quantities are edited in the GAS LIFETIMES AND DQ/DC screen. For each gas, a default (or 'permanent') value is stored. It is possible for the user to change these values by returning to DOS and editing the file SUBST.DAT--this is not recommended for non-expert users. In addition to changing gas lifetimes, as was done in the example, the user may change the radiative forcing sensitivities (i.e., dQ/dC) for the halocarbons. To alter these values, select a gas and use the 'Enter' key to move the edit bar to the dQ/ dC box and edit as was described for lifetimes in Section 2.3. (In general, 'Enter' moves the edit bar between the Lifetime and dQ/dC boxes.) Note that, although editable values of dQ/dC are given for methane and nitrous oxide, editing these will have no effect on the model calculations--these gases have non-linear relationships between radiative forcing and concentration so that forcing changes, [[Delta]]Q, cannot be determined by a single dQ/dC value. The permanent values shown are those corresponding to 1990 concentration levels.

3.3 Gas cycle model options

There are a number of options available on the GAS CYCLE MODEL PARAMETERS screen that were not covered in the example. For CO2, the default option adds the CO2 produced by the oxidation of fossil-derived CH4 in the troposphere to the CO2 emissions before running the carbon cycle model. This is only a very small term; nevertheless, it is possible to set it to zero. For CO2 and CH4, there are buttons to add temperature feedback terms to the emissions. Although the FORTRAN code for MAGICC includes these feedbacks, they are not currently accessible to the user because of the large uncertainties in their quantification. Finally, in default mode the radiative effect of stratospheric ozone depletion arising from the emissions of chlorine and bromine-containing halocarbons is automatically included in calculating the total halocarbon radiative forcing. The method for doing this is the chlorine-loading method employed by Wigley and Raper (1992). This option may be turned off to give the direct radiative forcing only for halocarbons. For realistic emissions scenarios, the ozone depletion contribution to radiative forcing is positive (since the phase out of ozone-depleting substances leads to a recovery of stratospheric ozone levels) and small (since the total ozone depletion effect to date is small--only a few tenths W/m²).

3.4 Changing climate model parameters

In the example, only the climate sensitivity ([[Delta]]T2X) was altered for the User-model case. For the vertical diffusivity (K) and upwelling rate (w), these may be altered independently. However, since w/K determines the steady-state decay length for vertical temperature variations in the ocean model, they are more realistically constrained to vary together keeping w/K constant at about 4m/yr/cm²/sec. To facilitate keeping w/K constant, the 'restore' button for w on the CLIMATE MODEL PARAMETERS screen returns the value of w to 4K rather than to the default value of 4.0 m/yr that corresponds to K = 1 cm²/sec.

3.5 Stratospheric water vapour forcing

Apart from the five climate model parameters, there is one additional editable item on the CLIMATE MODEL PARAMETERS screen, namely the 'Stratospheric H2O forcing factor'. This factor controls the additional radiative forcing arising from CH4 emissions due to their oxidation to H2O in the stratosphere. Physically, this factor is the ratio of the stratospheric H2O forcing term to the methane forcing that occurs in the absence of N2O overlap (see Shine et al., 1990, for details). It gives a small but non-negligible contribution to the total radiative forcing. In 1990, IPCC judged this factor to be 0.3 (Shine et al., 1990), but the latest best estimate for it is 0.05 (Shine et al., 1994). This latter value is used as the default by MAGICC.

4. SCIENTIFIC DETAILS

Most of the models used in MAGICC are described in the published literature. In these cases, the user is referred below to the appropriate papers. Where the methods used are not published, a more comprehensive explanation is given.

4.2 Carbon Dioxide

The carbon cycle model employed is that of Wigley (1993). This model comprises an ocean section in which the atmosphere-to-ocean flux changes are represented as a convolution integral, and a 4-box terrestrial biosphere that is one of the hierarchy of similar models described by Harvey (1989). The ocean model is based on the 3D ocean GCM carbon cycle model of Maier-Reimer and Hasselmann (1987), adjusted to allow the 1980s-mean ocean flux to be varied. It uses a particularly efficient numerical algorithm to evaluate the convolution integral (see Wigley, 1991a). Although the ocean flux can be varied in general, the option to change this flux is not accessible through MAGICC. Instead a fixed value of 2.0 GtC/yr for the 1980s-mean flux is used, corresponding to the current best-guess value (see Schimel et al., 1994).

The model carbon budget is given by

2.123 d[[Delta]]C/dt = Efossil + Dn - Socean - Sfert 	(1)

Since three components of the budget (fossil emissions, atmospheric buildup and ocean flux) are fixed, the only balance involved is that between net "deforestation" (i.e., net land-use-change emissions) and CO2 fertilization (Sfert). Dn(1980s) is uncertain, with an a priori estimated range of 1.1 + 1.1 GtC/yr (Schimel et al., 1994). The uncertainty range used in MAGICC, however, is less than this because the balancing term, Sfert, is also constrained. On physiological grounds, the fertilization parameter (r¹) that determines Sfert most probably lies in the range 1.05 to 1.30. The implied range for Dn(1980s) is then 1.1+ 0.7 GtC/yr, as used in MAGICC.

The budget balance issue is far more complex than portrayed here, because there are other feedbacks that are not included in equ. (1). As explained by Wigley (1993), the best way to view this is to assume that the Dn term represents that part of the "observed" Dn that is balanced by CO2 fertilization. In this way, specifying a particular Dn value (as may be done by the user in MAGICC) is more correctly viewed as selecting a particular value for the fertilization coefficient. The way the carbon cycle model in MAGICC is set up, the difference between the selected Dn and the true "observed" Dn is then effectively balanced by a non-CO2-fertilization terrestrial biosphere sink that reduces to zero magnitude over 1990-2000. This is a conservative assumption, likely to lead to a slight overestimate of projected CO2 concentrations.

For most emissions scenarios, the MAGICC carbon cycle model behaves very similarly to other models. For low emissions scenarios, MAGICC tends to produce concentrations in the low end of the inter-model range. Further details of the performance of MAGICC relative to other models is given in Enting et al. (1995).

4.3 Methane

The methane model used is that of Osborn and Wigley (1994). This is a mass balance model of the form

d[[Delta]]C/dt = E/B - C /T - C/Tsoil 		(2)

where B (= 2.75 TgCH4/ppbv) is a concentration-to-mass conversion factor, T is the atmospheric lifetime and C/Tsoil is a soil sink term (Tsoil = 150 yr). The atmospheric lifetime is assumed to be a function of OH concentration. OH concentration in turn is assumed to depend on CH4 concentration and the concentrations of CO, NOx and VOCs (with the latter three assumed proportional to their respective emissions because of their short lifetimes). The empirical relationship determining changes in T as a function of changes in these quantities has been determined by fitting the model to other, more complex, atmospheric chemistry models. A best-guess set of model parameter values has been chosen based on the average of the various models, adjusted to account for more recent information. Inter-model differences allow an estimate of the range of uncertainty. For a full description, see Osborn and Wigley (1994).

4.4 Nitrous Oxide

The N2O model is a simple mass balance model of the form

d[[Delta]]C/dt = E/[[Beta]] - C /T 		(3)

where [[Beta]] (= 4.81 TgN/ppbv) is a concentration-to-mass conversion factor. The lifetime T is assumed to be constant.

4.5 Halocarbons

The halocarbons considered in MAGICC are divided into four groups based partly on similarities in their chemical composition, and partly on similarities in their projected future emissions under the scenarios developed by IPCC in 1992 (Leggett et al., 1992). The reason for this is to simplify the emissions input process, recognizing that the bulk of the radiative forcing effects of halocarbons arises from a small number of gases. Thus, instead of specifying emissions for each individual gas, the user need only specify the emissions for the dominant or 'key' member of each group. Because of within-group similarities, total radiative forcing for the group is well estimated simply by scaling up the forcing for the key member. A judicious choice of group membership allows this scaling-up procedure to work well for both the direct forcing and for the forcing when stratospheric ozone depletion is accounted for. The groups are:

• CFC11 group: Comprises CFC11, CFC113, CC14, CHCl3, methyl chloroform (CH3CCl3), Ha1211 (CF2ClBr) and Hal301 (CF3Br). All of these gases are strong ozone depleters (having three or more chlorine atoms, or bromine in their structure).
• CFC12 group: Comprises CFC12, CFC13, CFC114 and CFC115. These gases are moderate ozone depleters (one or two chlorine atoms) and have long lifetimes.
• HCFC22 group: Comprises HCFC22, CH2Cl2, HCFC123, HCFC124, HCFC141b and HCFC142b. These are all hydrogenated species with one or two chlorine atoms and relatively short lifetimes.
• HFC134a group: Comprises HFC134a, HFC152a, HFC125, CFC14 (CF4) and CFC116 (C2F6). The primary similarity here is that these are all non-ozone depleting gases. Because of differences in their lifetimes and in their future emissions projections, they are considered in equivalent HFC134a terms by combining their emissions with GWP weights. The weights to use are based on averages of 20-year and 100-year time horizon GWPs from the 1994 report (Albritton et al., 1994), viz.

  E(equiv - HFC134a) = E(HFC134a) + 3.04 E(CFC14) + 6.05 E(CFC116)
  + 1.96 E(HFC125) + 0.12 E(HFC152a)

  Note that this has a non-zero value in 1990, even though emissions of HFC134a were zero at this time.

The full set of gases here covers all those considered by IPCC in 1992 except for HFC225 (whose future forcing role is minimal), and includes additionally CFC13, CHCl3, CH2Cl2, CFC14 and CFC116.

The scaling method used in MAGICC required a number of steps in its development. First, radiative forcing changes had to be calculated for each gas to get the true group totals. Up to 1990, this was done using observed concentrations. Beyond 1990, projected concentrations were deduced from IPCC92 emissions scenarios using standard first-order mass balance equations of the form

dC/dt = E(t)/[[Beta]] - C /T

where [[Beta]] converts concentration to mass and T is the lifetime. Radiative forcings were calculated using [[Delta]]Q = [[Alpha]] [[Delta]]C. T and [[Alpha]] values were as given by IPCC94 (Prather et al., 1994; Shine et al., 1994). Scaling factors were then estimated by comparing group forcing totals with key member forcings. This was done for a range of emissions scenarios, and "best-estimate" scaling factors were chosen to minimize the mean errors in the scenarios. Next, the same procedure was repeated for radiative forcings reduced by stratospheric ozone depletion effects.

To quantify the ozone depletion effect, the chlorine loading method of Wigley and Raper (1992; see also Wigley, 1994) was used. For bromine-containing compounds, bromine loading was scaled by 40 to account for its greater ozone-depletion potential. (Wigley and Raper used a factor of 5, which is too low. However, this has a negligible effect on the results.) The total forcing based on this method is given by

[[Delta]]QTOTAL = [[Delta]]QDIRECT - [[Delta]]QOZONE

where

[[Delta]]QDIRECT = [[Alpha]] [[Delta]]C

[[Delta]]QOZONE = [[radical]] NCl [[Delta]]C

Here [[Delta]]C is the concentration change, [[Alpha]] = dQ/dC is the radiative forcing sensitivity for the gas, NCl is the number of chlorine atoms in the molecule (replaced by 40 NBr for halons) and [[radical]] is a constant calculated by comparing the results of the method with the detailed calculations of Ramaswamy et al. (1992)--which gives [[radical]]= 0.0762. Note that ozone depletion in the stratosphere not only changes the radiative balance, but may also affect tropospheric chemistry in a way that amplifies its direct radiative effect (Bekki et al., 1994). This indirect effect is probably somewhat smaller than the direct effect. It is not included in MAGICC calculations.

To carry out the full set of halocarbon concentration and radiative forcing calculations, estimates had to be made for a number of quantities not covered in standard data sets; specifically, current concentrations and emissions, and future emissions for CFC13, CHCl3, CH2Cl2, CFC13 and CFC116. These estimates were based on information in the literature and personal communications with Stuart Penkett (UEA) and Archie McCulloch (ICI). The data used are given below (concentrations in pptv, emissions in Gg/yr, lifetimes in years, and dQ/dC in Wm^-2/ppbv).

Click for data.

The scaling factors used in MAGICC 1.2 are ....

Click for data.

These have yet to be fully updated to accord with IPCC94 lifetimes and radiative forcing sensitivities, but any future changes will be small.

The output of MAGICC gives actual concentrations for CFC11, CFC12 and HCFC22. For HFC134a, the output is an equivalent concentration, defined as the concentration of HFC134a required to give the total radiative forcing for the group. Since these gases have no ozone depletion effects, there is no ambiguity in this definition (as there would be for the other key gases). Mathematically, the relationship is

[[Delta]] [equiv- C(HFC134a)] = [[Delta]] [Q(HFC134a group) / [[Alpha]] (HFC134a)

Because CFC14 and CFC116 have non-zero concentrations in 1990, the equivalent HFC134a concentration is also non-zero in 1990.

As a final point, it should be noted that the effect of editing lifetimes or radiative forcing sensitivities for the key halocarbons is automatically passed on to other gases in each group (because of the scaling method used). For CFC11, CFC12 and HCFC22, the correct concentration changes will, of course, be calculated and shown. For HFC134a, the interpretation of results is more difficult since only equivalent HFC134a details are used. Equiv-HFC134a results are, nevertheless, a useful qualitative guide to the specific results for HFC134a.

4.6 Sulphate Aerosols

MAGICC 1.2 uses the aerosol radiative forcing formulation and quantities described in Wigley (199lb) and Wigley and Raper (1992). Only SO2 emissions from fossil sources are considered (biomass emissions will be included in a later version of MAGICC--their effect is relatively small). Forcing is divided into a direct (clear-sky) term that is taken to be linear in emissions, and an indirect (cloud albedo) term that is logarithmic in emissions. The 1990 emissions is assumed to be 75 TgS/yr¹ and the corresponding global-mean forcing is 0.75W/m², split 80:20 between the direct and indirect terms. Because of the functional forms for the dependence of forcing on emissions, the relative importance of the cloud albedo term is greater before 1990 and less after 1990. Because most fossil SO2 emissions are from land areas in the Northern Hemisphere, forcing is split 9:1 between the Northern and Southern Hemispheres, and it is assumed that direct forcing is over land only. Since cloud albedo changes are expected mainly over ocean areas, indirect forcing is over ocean only. The results are insensitive to these assumptions.

For uncertainty ranges, MAGICC follows Wigley and Raper (1992) in assuming that the uncertainty is +/-50% about the best guess value. The lower bound agrees with a priori estimates, but the upper bound is considerably less than a priori estimates (e.g. as given by Shine et al., 1994). The relatively low upper bound used here is based on a subjective interpretation of observed temperature changes; it agrees well with an independent estimate from Hansen et al. (1993). If the chosen upper bound applied, then this would give an aerosol forcing of about -2W/m² in the Northern Hemisphere and -0.25W/m² in the Southern Hemisphere; compared with a total greenhouse-gas forcings of around 2-2.5W/m² in each hemisphere. The net forcing would then be near zero in the Northern Hemisphere and +2W/m² in the Southern Hemisphere, a differential that should be observable in the hemispheric-mean temperature records, but isn't (see Wigley, 1989). Furthermore, if the total aerosol forcing were above the chosen upper bound value, then the observed global-mean warming of 0.5 deg. C over the past century can only be explained by assuming an unrealistically large value for the climate sensitivity (Wigley, 1989).

It is, of course, possible to accept higher aerosol forcing values if one assumes that the above-dalcribed discrepancies are explainable through natural climate variability. Our understanding of the magnitude of this natural variability is limited, and we have no firm, quantitative knowledge of the partitioning of observed changes between natural and anthropogenic causal factors. Occam's razor, however, demands that we give the natural variability explanation a relatively low probability (since it requires invoking this as an additional explanatory factor). We therefore retain a +50% factor to define the upper aerosol forcing "bound", admitting a relatively small probability that it is higher.

4.7 The Climate Model

The climate model is a standard upwelling-diffusion, energy balance model of the form originally developed by Hoffert et al. (1980) and used by many authors (including IPCC90; Bretherton et al., 1990). The specific model is described in various papers by Wigley and Raper (1987, 1991, 1992). It differs slightly from some other models in that land and ocean "boxes" are identified in each hemisphere, partly in order to allow regionally-differentiated aerosol forcing. The climate model also calculates the oceanic thermal expansion component of global-mean sea level rise, using a non-linear temperature- and pressure-dependent expansion coefficient. Further details are given in Wigley and Raper (1993).

4.8 Ice-melt Models

The ice-melt models for "small" glaciers, and the Greenland and Antarctic ice sheets, are minor modifications of those developed by Wigley and Raper for use by IPCC in 1990 (Warrick and Oerlemans, 1990)--see Wigley and Raper (1993) for full details.

5. ACKNOWLEDGMENTS

*mailing address
fax: (303) 497-2699 October 1994
email: wigley@ncar.ucar.edu

Footnotes

¹ Default, best guess and mid are used interchangeably.

² This is not included in the present version, but will be included in later versions.

³ The effects of emissions from biomass burning will be included in a later version of MAGICC.

⁴These four halocarbons are used as proxies for a much larger set of gases, which are divided into four groups with the named gases as key members. For HFC134a, the emissions input required is equivalent HFC134a, the GWP-weighted sum of emissions for CFC14 (CF4), CFC116 (C2F6), HFC125, HFC143a and HFC152a. The calculated radiative forcing for equiv-HFC134a closely approximates the total forcing for this group of gases. For the other three halocarbons, individual gas concentrations are carried through the calculations, converted to radiative forcing, and then scaled up to account for other gases in the groups. While approximate, the errors involved in this scaling up process are very small (of order 0.01-0.02 W/m²) for all realistic emissions scenarios.

⁵ All, that is, except alternate (non-ozone-depleting) halocarbons, as characterized here by HFC134a. For CFC substitutes, IS92a and IS92d differ in the split between HCFC22 and HFC134a: IS92d assumes a phase out of HCFC22 and a greater usage of HFC134a than IS92a. This is more in accord with the Copenhagen amendment to the Montreal Protocol, which was agreed to after the IS92 scenarios were derived.

⁶ Hitting the 'Esc' key at any time will return the user to the previous menu. For most screens 'Enter' has the same effect.

⁷ An unused version of MAGICC will have no emissions scenarios installed. It is necessary to install a policy and an emissions scenario in order to run MAGICC.

⁸ This is the fractional increase in global Net Primary Productivity for a CO2 doubling from 340 to 680 ppmv.

⁹ There is a 'bug' in the present version of MAGICC for the display of these gases. Only the user-model results are shown, and these are incorrectly labeled as default ('best guess') model results.

¹⁰ CH4 forcing includes the contribution from stratospheric water vapour changes.

¹¹ For sea level, upper and lower bounds are obtained by using the same three values of [[Delta]]T2x (with other climate model parameters fixed at their best guess values) to calculate global-mean temperature and oceanic thermal expansion, and using high and low ice-melt model parameters coupled with the high and low temperatures to obtain the ice-melt contributions to sea-level change.

¹² Note that the user-selected model parameters will not be changed. Clicking on the User defined button will reset the previously chosen values.

¹³ This is the fractional increase in global Net Primary Productivity for a CO2 doubling from 340 to 680 ppmv.

¹⁴ This is 2 TgS/yr above the IPCC92 value used by Leggett et al. (1992). A 2 TgS/yr 'correction' has been applied to all post-1990 IS92 scenario values to account for this difference (e.g. as in Tables 2 and 3).

Acknowledgement

Data Errors, Corrections and Disclaimer
CIESIN follows procedures designed to ensure that data disseminated in CIESIN's Host are of reasonable quality. If, despite these procedures, users encounter apparent misstatements in CIESIN's Host, they should contact CIESIN Customer Services at 517/797-2614 or via Internet e-mail at CIESIN.Info@ciesin.org. CIESIN will notify the original data provider of the apparent errors or misstatements, and will attempt to correct any errors or misstatements. Neither CIESIN nor NASA verifies or guarantees the accuracy, reliability, or completeness of the data provided.

Text Browser Utilities:
[CIESIN Home Page, CIESIN Gateway, Compass, User Services, Register, FAQs, Full-Text Search]