Diagnostic Analysis of GCM Simulations Driven by SRES A2 and B2 Emissions Scenarios
Xianfu Lu
Tyndall Centre for Climate Change Research
School of Environmental Sciences, University of East Anglia

Introduction
To conduct climate change impact, vulnerability and adaptation (I, V&A) assessments, integrated scenarios of socio-economic and environmental features are required. As a result from the initial project development and two AIACC training workshops, most regional study teams have indicated that SRES driving GCM (General Circulation Model) experiments will be used to derive climate scenarios for these projects. Due to the limitation of both the computing and analytical resources, most project teams may have to make choices from a large number of available GCM experiments.

As an attempt to assist the AIACC study partners in selecting the most "appropriate" GCM experiments to develop climate scenarios for their projects, we performed a preliminary diagnostic analysis of climate simulations from 9 different GCMs driven by SRES A2 and B2 emissions scenarios, for 13 regions covered by 11 AIACC projects.

Methodology and Data
To evaluate the performance of different GCMs for different regions over the globe, we looked at two sets of quantities: different model simulations of quantities (e.g. mean temperature, precipitation,…) versus global 0.5 degree observed climatology for the baseline period (i.e. 1961~90); simulations of changes in seasonal climate quantities by different GCMs for the period of 2071~2100. In order to do so, datasets were extracted and plotted as maps for each region/area.

By comparing model simulated baseline period climatology against observations, one could make a judgment on which simulations perform better over the others for a specific region/area, in term of reproducing observed average climate features. Meanwhile, by examining the range of changes in primary climate features simulated by various GCMs, one could then decide, depending on how much of the uncertainty span should be explored in the I, V &A assessment, which model simulation(s) to use to derive regional climate scenarios.

CRU05 0.5 degree seasonal mean climatology was used and compared against GCM simulations. This data set is described in New, M., M. Hulme and P. Jones, 1999. Simulations of "present-day" (1961~90) climatology and climate futures towards the end of this century (2071~2100) from 9 state-of-the-art GCMs driven by SRES emissions scenarios were analysed. These model simulations were reviewed by the IPCC Third Assessment Report (TAR).

Two variables (mean temperature and precipitation), two seasons [JJA(June-July-August) and DJF (December-January-February)], and two SRES emissions scenarios (A2 and B2) are included in this analysis.

Results
Results of the analysis are presented in three sets of maps:

  • baseline (1961~90) period observed and simulated seasonal climatology (in maps);
  • simulations of seasonal anomalies from 9 GCMs (in maps);
  • scatter plots for each region/area illustrating the range of anomalies simulated by different GCMs1

Follow the links below to download the results corresponding to your region/area. (The following are zipped files that contain several images)

Additional Papers and Resources

1 The aggregation of anomalies over each region was achieved by simply averaging the numbers for GCM grid boxes falling inside each region. Scatter plots are missing for a few regions. This is due to the fact that some models produce 0 mm of seasonal precipitation over the baseline period. Therefore, the percentage change of precipitation in the 2080s was not attainable.