Climate model projections

The recently completed Arctic Climate Impact Assessment (ACIA), an international activity with about 300 participating scientists, was aimed at evaluating and synthesizing Arctic climate variability and change (ACIA, 2005). One of the components of the ACIA was to provide scenarios of future Arctic climate. The ACIA made use of five different AOGCMs. These are: (1) CGCM2 (Canadian Climate Center, Canada); (2) CSM (National Center for Atmospheric Research, USA); (3) ECHAM4/OPYC3 (Max Plank Institute, Germany); (4) GFDL (Geophysical Fluid Dynamics Laboratory, USA) and; (5) HadCM3 (United Kingdom Meteorological Office, United Kingdom). The basic features of AOGCMs were outlined in Chapter 9. There are a total of 40IPCC SRES (Special Report on Emissions Scenarios) emission scenarios, built around four storylines that describe the evolution of the world in the twenty-first century (IPCC, 2001). Details of the emissions scenarios are provided by Nakicenovic et al. (2000). The results described here represent transient mean responses to the "middle of the road" SRES scenario B-2 (the primary scenario chosen by ACIA). A general overview of the IPCC 2001 Assessment and the SRES is provided in Barry and Chorley (2003, Chapter 13).

We first examine projected changes in SAT for the latter part of the twenty-first century (2060 through 2089) in comparison with the "present day" reference period 1960-90 for each model. Fields for winter, spring, summer and autumn are provided

Winter temperature

Winter temperature

Figure 11.21 Projected changes in surface air temperature (°C) for the period 2060-89 relative to the reference period 1960-90. These represent group averages from the five AOGCMs participating in the Arctic Climate Impact Assessment. Results are given for winter, spring, summer and autumn (courtesy of J. Walsh and W. Chapman, University of IllinoisU, rbana-Champaign, IL).

Spring temperature

Figure 11.21 Projected changes in surface air temperature (°C) for the period 2060-89 relative to the reference period 1960-90. These represent group averages from the five AOGCMs participating in the Arctic Climate Impact Assessment. Results are given for winter, spring, summer and autumn (courtesy of J. Walsh and W. Chapman, University of IllinoisU, rbana-Champaign, IL).

(Figure 11.21) based on the group average of the five ACIA models. The five model averages project the largest high-latitude warming in winter and autumn, with smaller warming in spring and summer. The winter and autumn fields show the changes to be larger over high as compared with middle latitudes. In turn, the projected changes are most pronounced over the Arctic Ocean (over a 6 °C change for most of the area, but larger nearer the Pole). Part of the high-latitude amplification relates to the strong

SutlllllCi Icmpetuluie Figure 11.21 (Cont.)

SutlllllCi Icmpetuluie Figure 11.21 (Cont.)

stability of the lower troposphere, which focuses the heating near the surface. It also manifests the retreat and thinning of sea ice. Melt starts earlier in spring and is amplified in summer due to greenhouse radiative forcing and attendant albedo feedbacks. While the melting process dampens summer temperature changes over the Arctic Ocean, autumn ice formation is delayed. The ice that forms is also thinner. This means strong heat fluxes to the atmosphere and higher autumn SATs. These effects continue through winter.

There is considerable scatter between the projections from the five models. The standard deviation between the winter projections of the five models is on the same order as the five-model means. Similar results emerge for the other seasons. There are a number of reasons for this. For each model, the effects of greenhouse gases will be superimposed upon low-frequency climate variability (e.g., NAO-like variability). For any given 30-year period, such as 2060-89, the climate of the different models may be in different phases of low-frequency variability. The resulting scatter between the models due to low-frequency variability should not be taken as evidence that any of the models are wrong. But the models also have some inherently different behaviors. Some exhibit strong low-frequency variability, while in others it is weaker. The state of the ocean circulation is important. Some of the models project significant reductions in the thermohaline circulation while others do not. Sea ice is not well handled in present AOGCMs and some of the models poorly simulate snow cover during the critical transition seasons. As a general statement, while feedbacks in the Arctic system associated with snow and sea ice act to amplify climate change, the same feedbacks amplify error and hence the scatter between different models.

Such scatter is by no means limited to the ACIA models. Holland and Bitz (2003) summarized annual temperature changes projected by 15 different AOGCMs in terms of 20-year averages centered over the year of doubled carbon dioxide, which was assumed to increase by 1% per year. All of the modes project Arctic amplification. However, the ratio of the high-latitude to low-latitude warming varies from two to four.

Figure 11.22 provides seasonal summaries of the model group average changes in precipitation. While there is a great deal of spatial variability for each season, the overall picture is one of higher precipitation. The general increase in high-latitude precipitation is understood in that, with a warmer climate, the atmosphere will hold more water vapor. Some of the largest changes occur outside of the Arctic. Again, the standard deviation of the five-model group average is quite large. Figure 11.23 provides a summary of precipitation, evaporation and P — ET for the Arctic Ocean and the major terrestrial Arctic drainages, in this case for 2070-90 relative to the 1960-1990 reference period. On the basis of the five-model averages, precipitation is projected to increase in all regions. While changes in evaporation are less coherent, net precipitation (hence runoff to the Arctic Ocean) is also projected to increase. On the basis of the HadCM3 model, Wu et al. (2005) argue that greenhouse-forced increases in precipitation have already occurred and can help explain the increased river Siberian discharge documented by Peterson et al. (2002).

As a final example, we examine projected changes in Northern Hemisphere sea ice extent for March and September. These are shown as time series of ice extent (at least 15% ice concentration) for the period 2000 through 2100 (Figure 11.24). Two sets of time series are presented. The top panels give the raw model time series, which start from the reference climatology for each model. The reference climatology for each

Winter precipitation

Spring precipitation

Figure 11.22 Projected changes in precipitation (millimeters per day) for the period 2060-89 relative to the reference period 1960-90. These represent group averages from the five AOGCMs participating in the Arctic Climate Impact Assessment. Results are given for winter, spring, summer and autumn (courtesy of J. Walsh and W. Chapman, University of IllinoisU, rbana-Champaign, IL).

model is different, with some having too much ice compared with observations and some too little. The bottom panels give time series in which the raw model values are adjusted based on the bias of the model reference climatology compared with observations. For example, if the model reference climatology is 20% too high, for

Sum met piccipii.d'ion Figure 11.22 (Cont.)

Sum met piccipii.d'ion Figure 11.22 (Cont.)

Autumn precipitation

each year the modeled ice extent is reduced by 20%. This provides for a more direct comparison between the models.

March ice extent shows only small changes. This indicates that even in the warming climate, it still remains cold enough in winter for thin, firstyear ice to form. However, thin, firstyear ice will be quick to melt the following spring and summer. Hence the

1 1 1

| Precipitation |

1 Î I '

ï î

-

-

-

-

¡¡111!"

Arctic Ocean Ob Yenisey Lefia

Pechora Mackenzie

Arctic Ocean Ob Yenisey Lefia

Pechora Mackenzie

Evaporation 1

1

-

-

[

1

Ï

-

1

J

L t -

-

-

-

] !

-

Pechora Mackenzie

Figure 11.23 Projected percentage changes for the Arctic Ocean and the major Arctic draining watersheds of annual precipitation, evaporation and net precipitation (P-ET) for the period 2070-90 relative to the reference period 1960-90. These represent group averages of the five AOGCMs participating in the Arctic Climate Impact Assessment (courtesy of J. Walsh and W. Chapman, University of Illinois, Urbana-Champaign, IL).

Ardic Ocean Ob

Pechora Mackenzie

i i

1

1 Ip-E'TI

-

r

i

-

i

-

-

L

-

-

-

-

Pechora Mackenzie

Yenisey

Pechora Mackenzie projected changes in September ice extent are stronger, with one of the models showing a complete loss of September ice by about 2070. By contrast, and pointing to the differences in model behavior, the CSM shows very little September ice loss.

0 0

Post a comment