Frequently Asked Questions
1. How do I cite the Climate Futures method?
2. How do I tailor the projections for my project?
3. Why are there fewer models available in Pacific Climate Futures than in the IPCC reports?
4. Are the models used in Climate Futures the same as in the Climate Change in the Pacific reports?
1. How do I cite the Climate Futures method?
Conceptual Basis:
The conceptual basis for the climate futures approach, upon which Pacific Climate Futures is based, has been published in the international peer reviewed literature. The citation for this paper is:
Whetton P, Hennessy K, Clarke J, McInnes K, Kent D (2012) 'Use of Representative Climate Futures in impact and adaptation assessment.' Climatic Change. DOI: 10.1007/s10584-012-0471-z
This paper can be obtained directly from the publisher online at http://link.springer.com/article/10.1007%2Fs10584-012-0471-z
Application:
A detailed explanation of how to apply the approach, using the web-tool, in a hypothetical case study has been documented in a peer-reviewed conference paper. The case study is an Australian example, however the method is applicable to any region. The citation for this paper is:
Clarke JM, Whetton PH, Hennessy KJ (2011) 'Providing Application-specific Climate Projections Datasets: CSIRO’s Climate Futures Framework.' Peer-reviewed conference paper. In F Chan, D Marinova and RS Anderssen (eds.) MODSIM2011, 19th International Congress on Modelling and Simulation. Perth, Western Australia. December 2011 pp. 2683-2690. (Modelling and Simulation Society of Australia and New Zealand).
The full text of the paper is available for free download at http://www.mssanz.org.au/modsim2011/F5/clarke.pdf
2. How do I tailor the projections for my project?
Please note that to obtain project-specific, tailored projections from Pacific Climate Futures, you must be a registered user with access to Advanced Mode.
The basic steps in the process are set out below. For further detail see the following paper that details the process for a hypotheticl rainfall capture and storage project.
Clarke JM, Whetton PH, Hennessy KJ (2011) 'Providing Application-specific Climate Projections Datasets: CSIRO’s Climate Futures Framework.' Peer-reviewed conference paper. In F Chan, D Marinova and RS Anderssen (eds.) MODSIM2011, 19th International Congress on Modelling and Simulation. Perth, Western Australia. December 2011 pp. 2683-2690. (Modelling and Simulation Society of Australia and New Zealand).
[Free download available at http://www.mssanz.org.au/modsim2011/F5/clarke.pdf]
Refer to the User Guide for detailed instructions on use of the web-tool.
Outline of process:
1. Decide on the time periods and emissions scenarios of interest
2. Identify the climate variables required by the decision makers
Remember, if a variable of interest is not available in the web-tool, it is likely to be covered in the Climate Change in the Pacific report (available from http://www.pacificclimatechangescience.org).
3. Choose two of these variables to use as 'classifiers'
These will usually be surface temperature and rainfall or surface temperature and wind speed.
4. Define what constitutes the 'Best Case' and 'Worst Case' in terms of these two classifying variables.
You need to have some understanding of the probable impacts on your project of changes in these climate variables. For example, if you are investigating the impacts of climate change on swamp taro production, the hottest and driest climate future will probably represent the worst case.
5. Decide on the season to use as a classifier (it is helpful to explore the affect of season on the climate futures - if the range and likelihoods of the climate futures are very similar for each season, it will be acceptible to set the classifiers to 'Annual', if not you should treat each season separately)
6. Using the climate futures matrix display, identify the 'Best Case' and 'Worst Case' using the definitions from step 4 to guide the choice.Also identify the 'Maximum Consensus Case' if there is one. Document the short descriptions of each. For example:
Best Case: Warmer and Little Change in Rainfall
Worst Case: Hotter and Drier
Maximum consensus Case: Warmer and Drier
7. Use the Representative Model Wizard to identify one or more models to represent each case (usually use the default ranking method: ‘mean’). Document the model rankings.
Remember to take into account any additional information on model performance when making your selection. This information can be found in the model information pop-ups.
8. Repeat the process for each time-period/emissions scenario combination and document the results in a table. It will usually be desirable to use the same model/s to represent each ‘case’ at all time periods and emissions scenarios. This usually involves some compromises as the same model will not always be the most representative in all time/emisisons combinations.
9. Obtain the projections data:
a) by exporting the change values for the selected model from the Representative Model Wizard into Excel and adding observed data (the spreadsheet automatically applies the changes to the observed data to calculate plausible future values), or
b) by obtaining projections for the selected models from another source (these might be downscaled projections or data that you already have from existing reports)
10. Use these data in your impact assessment, ensuring that you clearly describe the results for Best Case, Worst Case and Maximum consensus Case throughout.
3. Why are there fewer models available in Pacific Climate Futures than in the IPCC Fourth and Fifth Assessment Reports?
Background
When interpreting projections arising from a given climate model, it is important to consider how skilful that model is in simulating the climate of the region of interest. With respect to each of the PCCSP/PACCSAP partner countries, detailed research has been undertaken to determine the skill of the models, by assessing their ability to simulate the present day climate (usually the last 20-30 years). The underlying assumption of this approach is that a model which adequately simulates the present climate will provide more reliable projections of the future.
The results of this research (Brown et al. 2011; Irving et al. 2011; Grose et al. 2014) and many other studies published in the literature (e.g. Gleckler et al. 2008) show that it is usually difficult to identify a single best climate model. For a given climate variable (e.g. surface air temperature, rainfall), climate feature (e.g. SPCZ, ITCZ) or driver of climate variability (e.g. ENSO), there is usually a number of models that are considered to perform well, depending on the precise statistic used to compare the model output with observations. A model that performs well against one measure, however may not perform well against another. In contrast, it is often possible to identify models that perform considerably worse than the remainder of the models. Consequently, seven models (six from CMIP3 and one from CMIP5) have been excluded for use in Pacific Climate Futures. A further eight have been ‘flagged’ as unsuitable in certain circumstances, as explained below. Pacific Climate Futures alerts the user to these models where appropriate.
Models that have been excluded from Pacific Climate Futures
From the CMIP3 models (those assessed in the IPCC’s Fourth Assessment Report in 2007) the INM-CM3.0, PCM and GISS-EH models perform particularly poorly with respect to their simulation of many aspects of the present day climate over the PCCSP region (Irving et al. 2011). The INGV-SXG model is associated with strong ‘drift’ (i.e. artificial trends in simulations of the future climate, which arise due to problems with the interaction between the model atmosphere and ocean; Brown et al. 2011; Irving et al. 2011). The GISS-AOM and GISS-ER models perform particularly poorly with respect to their simulation of the present day ENSO (van Oldenborgh et al. 2005; Brown et al. 2011; Irving et al. 2011). Since ENSO strongly influences year-to-year changes in the climate of the PCCSP region, all projections from these models should be interpreted with caution in all parts of the Pacific. Accordingly, these models are excluded from Pacific Climate Futures.
There were 30 CMIP5 models (those assessed for the 5th Assessment Report) that were available at the time of updating the climate projections for the Pacific. Of these, CSIRO-Mk3.6 was rejected for all projections for the PACCSAP region (Grose et al. 2014).
Models that are included in Pacific Climate Futures but should be treated with caution in certain circumstances
From the CMIP3 models: The MIROC3.2(medres) and MIROC3.2(hires) models perform particularly poorly with respect to their simulation of the present day SPCZ (Brown et al. 2011; Irving et al. 2011). Projections from these models for regions affected by the SPCZ should be interpreted with caution. Countries affected are: Solomon Islands, Vanuatu, Fiji, Tonga, Niue, Samoa, Cook Islands (North and South), Tuvalu, Kiribati (Phoenix, Gilbert and Line Islands) and Nauru.
The MIROC3.2(hires) model performs particularly poorly with respect to its simulation of the present day ITCZ in the west Pacific (Irving et al. 2011). Projections from this model for countries affected by the ITCZ should be interpreted with caution. Affected countries are: Palau, FSM (East and West), Marshall Islands (North and South).
The IPSL-CM4 and MIROC3.2(medres) models fail to simulate the reversal of wind direction that defines the western Pacific monsoon. Affected countries are: East Timor, Marshall Islands (South), Federated States of Micronesia (East and West), Palau, Papua New Guinea, Solomon Islands, Tuvalu.
From the CMIP5 models, MIROC-ESM and MIROC-ESM-CHEM were shown to perform very poorly in simulating the SPCZ (Grose et al. 2014). Affected countries are: Cook Islands (North and South), Fiji, Kiribati (Phoenix, Gilbert and Line Islands), Nauru, Niue, Papua New Guinea, Samoa, Solomon Islands, Tonga, Tuvalu, and Vanuatu.
References
Brown J, Colman R, Katzfey J, Irving D, Sen Gupta A, Abbs D, Perkins S, Arthur C, Brown J, Chand S, Chattopadhyay M, Church J, Delage F, Durack PJ, Kokic P, McGregor J, Moise A, Muir L, Murphy B, Nguyen K, Platten S, Power S, Smith I, Summons N, Tory K, Wijffels S and Zhang X (2011) Climate model reliability. In 'Climate Change in the Pacific: Scientific Assessment and New Research Volume 1: Regional Overview (Australian Bureau of Meteorology and CSIRO).' (Australian Bureau of Meteorology and Commonwealth Scientific and Industrial Research Organisation (CSIRO))
Gleckler PJ, Taylor K and Doutriaux C (2008) Performance metrics for climate models. Journal of Geophysical Research 113(D6), D06104.
Grose MR, Brown JN, Narsey S, Brown JR, Murphy BF, Langlais C, Sen Gupta A, Moise AF and Irving DB (2014) Assessment of the CMIP5 global climate model simulations of the western tropical Pacific climate system and comparison to CMIP3. International Journal of Climatology.
Irving DB, Perkins SE, Brown JR, Sen Gupta A, Moise AF, Murphy BF, Muir LC, Colman RA, Power SB, Delage FP and Brown JN (2011) Evaluating global climate models for climate change projections in the Pacific island region. Climate Research C1028.
van Oldenborgh GJ, Philip SY and Collins M (2005) El Niño in a changing climate: a multi-model study. Ocean Science 1, 81-95.
4. Are the models used in Climate Futures the same as in the Climate Change in the Pacific reports?
Not necessarily.
Because Climate Futures is an online resource, we are able to add new model data as they become available. In the case of the CMIP5 data (i.e. the models used for the IPCC's Fifth Assessment Report), 26 models were used for the updated reports. Since then, additional models have become available and these have been added to Climate Futures.
In the case of the CMIP3 data, however, the same models are available in Climate Futures as were used in the original reports. This means the results from Climate Futures are consistent with the original (2011) reports.
If you need the results from Climate Futures to match those in the updated (2014) reports, you should use the "Models Selection" panel in the Advanced interface to de-select the additional models.
The models used in the updated (2014) reports were:
ACCESS1-0
ACCESS1-3
BCC-CSM1-1
CanCM4
CanESM2
CCSM4
CNRM-CM5
FGOALS-g2
FGOALS-s2
GFDL-CM3
GFDL-ESM2G
GFDL-ESM2M
GISS-E2-H
GISS-E2-R
HadCM3
HadGEM2-CC
HadGEM2-ES
inmcm4
IPSL-CM5A-LR
IPSL-CM5A-MR
MIROC4h
MIROC5
MPI-ESM-LR
MRI-CGCM3
NorESM1-M
NorESM1-ME