AR4 ensemble scenarios

National AR4-A1B Ensemble Seasonal and Annual Temperature and Precipitation Change with Model Standard Deviation

The CCDS has produced a summary of findings from the IPCC AR4 (2007) modelling assessment for Canada. More details resulting from this analysis are in production for publication. Twenty-four international modelling centres have contributed to the international dataset. The output used in this analysis is a mean ensemble from all available international modelling centres. Not all centres have produced runs for all emission scenarios. The models are outlined below:

The use of an ensemble approach (multi-model means/medians) has been demonstrated in recent scientific literature to likely provide the best projected climate change signal. Results between models can vary widely, and models each contain their own inherent biases. The use of a mean or median of many models reduces the uncertainty associated with any individual model. In effect, the individual model biases seem to 'cancel' out one another when considered as an ensemble. Compared against historical observed gridded data, ensemble results come closest to replicating historical climate. Although not a guarantee, an ensemble collection which can best represent historical climate, is more likely to represent future climate conditions.

Previously, due primarily to the difficulty in obtaining data, storing large data volumes and programming requirements, the ensemble approach has been overwhelming. Alternate methods included the use of a restricted set of model projections (two or three models), or the selection of scenarios from this small set which attempted to 'bound' projected climate variables (e.g., 'warm and dry, warm and wet, warmer and dry, warmer and wet'). Although a good approach, this technique still remains limited to a small sub-set of available models.

The current report provides an ensemble projection for all available GCM models outlined previously which have provided their data to the International Panel on Climate Change (IPCC) data warehouse or Program for Climate Model Diagnosis and Intercomparison (PCMDI). Even with these valuable resources, the retrieval of this data in its native format and subsequent conversion, formatting, interpretation, analysis and mapping is a large task not possible for many end-users and stakeholders (local government to the public). The CCDS has provided individual model results for some years now via the climate-scenarios.canada.ca website. This report provides stakeholders with the preliminary results of ensemble change fields for these models for the future from a common baseline period mean.

The mean monthly temperature and precipitation values are calculated for each model for the periods of 1971-2000 and 2020s, 2050s and 2080s. The mean grid-cell value is then used for further analysis.

Monthly GCM model output is interpolated to a common grid (National Centers for Environmental Prediction (NCEP) resolution) at 2.5 x 2.5 degrees across Canada. The NCEP grid resolution is shown in Figure 1.

Figure 1

Figure 1:

NCEP grid (click image to enlarge)

The results represent three periods of projected climate change in relation to the baseline of 1971-2000 for the "middle of the road" A1B emission scenario. The maps indicate the change in temperature in degrees Celsius, and for precipitation, the maps show the change in precipitation in percent from the baseline period.

In addition to a more rigorous estimate, the use of an ensemble versus a single or small combination of models can also provide valuable information of the spatial uncertainty of projections across Canada through the range of model results. Grid cells with greater model ranges indicate regions where different model formulations (sub-grid scale parameterization, atmospheric or oceanic circulations) differ, leading to an overall ensemble projection which has greater uncertainty. This is shown on the accompanying standard deviation maps for each variable. Areas of low standard deviation indicate model projections are more closely in agreement with each other (higher ensemble confidence), and vice-versa.