for example, where pattern scaling will break down. As the climate warms, temperature changes will be large as the ice edge moves across a particular location, but then return to small values with additional warming, because the ice edge is now further poleward.

Given other uncertainties, we find that pattern scaling is justified for current attempts to link stabilization targets and impacts, keeping in mind limitations due to the evolution of the pattern of warming on long, stabilization time scales and the limitations in regions of sharp gradients.

On the basis of CMIP3 simulations, Chapters 10 and 11 of IPCC AR4, WG1 analyzed geographical patterns of warming and measures of their variability across models and across scenarios. The executive summary of Chapter 10 reports that “[g]eographical patterns of projected SAT warming show greatest temperature increases over land (roughly twice the global average temperature increase) and at high northern latitudes, and less warming over the southern oceans and North Atlantic, consistent with observations during the latter part of the 20th century …”. Figure 10.8 of the report depicts the patterns of annual average warming across three scenarios (A2, A1B and B1) and three time periods (2011-2030, 2046-2065, and 2080-2099) over which change is computed. Figure 10.9 shows seasonal patterns for DJF and JJA under A1B. Chapter 10 also reports that the spatial correlation of fields of temperature change is as high as 0.994 in the model ensemble mean when considering late 21st century changes between A2 and A1B. A table in the same section (Table 10.5) quantifies the strict agreement between the A1B field, as a standard, and the other scenario patterns using a measure proposed by Watterson (1996) with unity meaning identical fields and zero meaning no similarity. Values of this measure are consistently above 0.8 and increase as the projection time increases (later in the 21st century fields agree better than earlier in the century), with values of 0.9 or larger for the late 21st century. The same table also shows that the agreement deteriorates if considering commitment scenarios. The results are documented as applying to seasonal warming patterns besides annual averages.

On the basis of the previous discussion and results, we compute patterns of standardized warming from the available CMIP3 SRES scenario simulation and produce maps of the ensemble average warming (these—in their non-normalized version—are available from AR4 WG1 Figures 10.8 and 10.9, and individual models’ maps are available in supplementary material in Chapter 10 (http://ipcc-wg1.ucar.edu/wg1/Report/suppl/Ch10/Ch10_indiv-maps.html, and Chapter 11 (http://www.ipcc.ch/pdf/assessment-report/ar4/wg1/ar4-wg1-chapter11-supp-material.pdf) along with measures of the regionally differentiated variability of this pattern across models and scenarios (IPCC, 2007a).