Annex B: Simulation of Regional Climate Change with Global Coupled Climate Models and Regional Modeling Techniques

In IPCC (1990) and IPCC (1992), very low confidence was placed on the climate change scenarios produced by general circulation model (GCM) equilibrium experiments on the sub-continental, or regional, scale (order of 105-107 km²). This was mainly attributed to coarse model resolution, limitations in model physics representations, errors in model simulation of present-day regional climate features, and wide inter-model range of simulated regional change scenarios. Since then, transient runs with Atmosphere-Ocean GCMs (AOGCMs) have become available that allow a similar regional analysis. In addition, different regionalization techniques have been developed and tested in recent years to improve the simulation of regional climate change. This section examines regional change scenarios produced by new coupled GCM runs. Following the 1990 and 1992 reports, emphasis is placed on the simulation of seasonally averaged surface air temperature and precipitation, although the importance of higher order statistics and other surface climate variables for impact assessment is recognized (Kittel et al., 1995; Mearns et al., 1995a,b).

B.1. Regional Simulations by GCMs

In IPCC (1990), five regions were identified for analysis of regional climate change simulation: Central North America (CNA; 35-50°N, 85-105°W), South East Asia (SEA; 5-30°N, 70-105°E), Sahel (Africa) (SAH; 10-20°N, 20°W-40°E), Southern Europe (SEU; 35-50°N, 10 W- 45°E), and Australia (AUS; 12-45°S, 110-155°E). Output from different coupled model runs with dynamical oceans for these regions was analyzed by Cubash et al. (1994a), Whetton et al. (1996), and Kittel et al. (1997), while analysis over the Australian region from equilibrium simulations with mixed-layer ocean models was performed by Whetton et al. (1994). Results over two additional regions were analyzed by Raisanen (1995) for Northern Europe (NEU; land areas north of 50°N and west of 60°E) and Li et al. (1994) for East Asia (EAS; 15-60°N, 70-140°E). To summarize the findings of these works, Figure B-1 shows differences between region-average values at the time of CO₂ doubling and for the control run, and differences between control run averages and observations (hereafter referred to as bias), for winter and summer surface air temperature and precipitation. Note that these models contain increases of CO₂ only. Experiments including increased CO₂ and the effects of sulfate aerosols will be discussed later.

Figure B-1: Difference between averages at time of CO2 doubling and control run averages (CO2-Control) and difference between control run averages and observed averages (Control-Observed) as simulated by nine AOGCM runs over seven regions. Units are °C for temperature and percentage of control run, or observed, averages for precipitation. In (f) and (h), values in excess of 200% have been reported at the top end of the vertical scale. In (e), values in excess of 60% have been reported at the top end of the vertical scale. Winter averages are for December-January-February (DJF) for Northern Hemisphere regions, and June-July-August (JJA) for Australia; summer averages are for JJA for the Northern Hemisphere, and DJF for Australia. See Kittel et al. (1997) for further details.

The biases are presented as a reference for the interpretation of the scenarios, because it can be generally expected that the better the match between control run and observed climate (i.e., the lower the biases), the higher the confidence in simulated change scenarios. The model runs are labeled d, g, m, x, p, q, r, s, and t as shown in the legend of the figure. Letter designations here refer to model descriptions in Table B-1. Note also in Table B-1 that the models employ different spatial resolutions and flux adjustments.

(continues on next page...)

IPCC Homepage