This Title All WIREs
How to cite this WIREs title:
WIREs Clim Change
Impact Factor: 4.571

Detection and attribution of climate change: a regional perspective

Full article on Wiley Online Library:   HTML PDF

Can't access this content? Tell your librarian.

The Intergovernmental Panel on Climate Change fourth assessment report, published in 2007 came to a more confident assessment of the causes of global temperature change than previous reports and concluded that ‘it is likely that there has been significant anthropogenic warming over the past 50 years averaged over each continent except Antarctica.’ Since then, warming over Antarctica has also been attributed to human influence, and further evidence has accumulated attributing a much wider range of climate changes to human activities. Such changes are broadly consistent with theoretical understanding, and climate model simulations, of how the planet is expected to respond. This paper reviews this evidence from a regional perspective to reflect a growing interest in understanding the regional effects of climate change, which can differ markedly across the globe. We set out the methodological basis for detection and attribution and discuss the spatial scales on which it is possible to make robust attribution statements. We review the evidence showing significant human‐induced changes in regional temperatures, and for the effects of external forcings on changes in the hydrological cycle, the cryosphere, circulation changes, oceanic changes, and changes in extremes. We then discuss future challenges for the science of attribution. To better assess the pace of change, and to understand more about the regional changes to which societies need to adapt, we will need to refine our understanding of the effects of external forcing and internal variability. Copyright © 2010 John Wiley & Sons, Inc.

This article is categorized under:

  • Paleoclimates and Current Trends > Detection and Attribution
Figure 1.

Observed global mean temperature changes from 1850 to 2008 (in red) from HadCRUT3v with uncertainties (yellow band as derived by Brohan et al.7 and expressed as anomalies relative to the mean temperature over the 1861–1899 period) overlain on a 1000 year segment of global mean temperatures from control simulations from the HadGEM1 model (black line).

[ Normal View | Magnified View ]
Figure 2.

Global mean surface temperature anomalies relative to the period 1901–1950, as observed (black line) and as obtained from climate model simulations with (a) both anthropogenic and natural forcings (red lines) and (b) natural forcings only (blue lines). Vertical gray lines indicate the timings of major volcanic eruptions. The thick red and blue curves show the multiensemble means and the thin lighter curves show individual simulations. (Reproduced from IPCC AR4 WGI report; Figure TS.23).

[ Normal View | Magnified View ]
Figure 3.

Estimated contribution from greenhouse gas (red), other anthropogenic (green) and natural (blue) components to observed global surface temperature changes. (a) 5–95% uncertainty limits on scaling factors based on an analysis over the 20th century, (b) the estimated contributions of forced changes to temperature changes over the 20th century expressed as the difference between 1990–1999 mean temperature and 1900–1909 mean temperature. The horizontal black line shows the observed temperature changes from HadCRUT2v.20 Five different analyses are shown using different models which are explained in more detail in the text. Adapted from Hegerl et al.2.

[ Normal View | Magnified View ]
Figure 4.

Regression analysis of 5 year, area mean temperature variations over Eastern North America during the 1900–1999 period. (a) Data from CRUTEM3 observations7 (black) and from simulations of two climate models including both natural and anthropogenic forcings (red and green). (b) 5–95% confidence intervals on scaling coefficients for the responses to various forcings in each climate model using the regression formula in Eq. (1). Simulations with historical greenhouse gas forcing only, historical natural forcings only, and both natural and anthropogenic forcings were input into the analysis. Scalings are shown for the responses to greenhouse gas forcing (red), sulfate aerosol and other anthropogenic forcing (green), and natural forcing (blue). (c) The resulting 5–95th percentile ranges on possible 5‐year average temperatures for the two climate models (red and green), compared to observations (black). (d) The warming between the 1900–1909 and 1990–1999 periods attributable to each of the external forcings. Diamonds show estimates from individual simulations (black). Lines show the estimated 5–95% confidence interval estimated using the linear regression analysis. Note that the larger warming of the PCM model over the UKMO‐HadCM3 model visible in (a) was adjusted through the different greenhouse gas forcing scalings in (b), producing better agreement between the models in (c) and (d).

[ Normal View | Magnified View ]
Figure 5.

Distributions of near‐surface temperature trends during 1950–1997 in different regions constrained by the global analysis in a climate forced with both anthropogenic and natural forcings (red lines) and with natural forcings only (green lines). The y axis gives the normalized likelihood. The observed trend in each region is marked on each panel as a black line. The regions are South Australia (SAU), North Australia (NAU), Central America (CAM), western North America (WNA), central North America (CAN), eastern North America (ENA), Mediterranean Basin (MED), northern Europe (NEU), South Asia (SAS), Atlantic (ATL), and Pacific (PAC). Their geographical extents are defined in Christidis et al. (2009).

[ Normal View | Magnified View ]
Figure 6.

Observed (top row) and simulated (bottom row) trends in specific humidity over the period 1973–1999 in grams per kilogram per decade. Observed specific humidity trends (a) and the sum of trends simulated in response to anthropogenic and natural forcings (d) are compared with trends calculated from observed (b) and simulated (e) temperature changes under the assumption of constant relative humidity; the residual actual trend minus temperature induced trend is shown in (c) and (f). Adapted from Willett et al.47.

[ Normal View | Magnified View ]
Figure 7.

Observed (solid black) and simulated zonal‐mean land precipitation trends for 1950‐1999. Black dotted lines indicate the multimodel mean from all available models (including both anthropogenic and natural forcings (denoted ALL) in (a), anthropogenic forcings only (ANT) in (b), and natural forcings only (NAT) as represented by ALL‐ANT in (c), and black dashed‐dotted lines from the subset of four models that simulated the response to each of the forcing scenarios (ALL4, ANT4, and NAT4). The model‐simulated range of trends is shaded. Black dashed lines indicate ensemble means of ALL and ANT simulations that have been scaled (SALL and SANT) to best fit the observations based on a one‐signal analysis. Colored lines indicate individual model mean trends.

[ Normal View | Magnified View ]
Figure 8.

Seasonal evolution of observed and simulated Arctic sea ice extent (ASIE) over 1953–2006. ASIE anomalies relative to the respective 1953–1982 means from observations (OBS) and 20C3M model simulations with anthropogenic only (ANT) and natural plus anthropogenic (ALL) forcings are plotted. Adapted from Min et al.68.

[ Normal View | Magnified View ]
Figure 9.

Trends in seasonal mean sea level pressure in seven 25°‐latitude bands between 87.5°S and 87.5°N calculated from 5‐year means over the period December 1949–November 2009. Solid lines show observed trends from HadSLP2, and dashed lines show ensemble mean trends in the ALL simulations of HadGEM1. Gray bands represent the approximate 5–95% range of simulated SLP trends in a control simulation of HadGEM1. MAM, JJA and SON trends are offset from DJF trends by 2, 4 and 6 hPa, respectively. Vertical dotted lines indicate zero trend for each season. Adapted from Gillett and Stott.75.

[ Normal View | Magnified View ]
Figure 10.

Time series of global ocean temperature above the 14°C isotherm relative to 1950–1999 average for (a) global ocean, (b) Atlantic Ocean, (c) Pacific Ocean, and (d) Indian Ocean. Shown are the observations (black); the ensemble average of four HadCM3 simulations including both anthropogenic and natural forcings (red) and ensemble standard deviation (orange shading); and the ensemble average of four HadCM3 simulations including only natural forcings (blue) and ensemble standard deviation (light blue shading). The model data have been regridded and subsampled to match the observational coverage. The vertical lines show the approximate timing of the major volcanic eruptions. Adapted from Palmer et al.90.

[ Normal View | Magnified View ]
Figure 11.

Change in probability of mean European summer temperatures exceeding the 1.6 K threshold showing histograms of frequency of such events under late 20th century conditions in the absence of anthropogenic climate change (green line) and with anthropogenic climate change (red line). From Stott et al.105 The distributions represent the uncertainty in this calculation's estimate of the frequencies of such events in the two cases.

[ Normal View | Magnified View ]

Related Articles

China National Climate Change Program
Climate Change Master Class

Browse by Topic

Paleoclimates and Current Trends > Detection and Attribution

Access to this WIREs title is by subscription only.

Recommend to Your
Librarian Now!

The latest WIREs articles in your inbox

Sign Up for Article Alerts