Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 Here we report extreme near-surface permafrost warming resulting from a remarkable temperature anomaly during winter and spring 2005–2006 on Svalbard. We demonstrate that this atmospheric temperature anomaly fell well within the range of predicted warming scenarios for the late 21st Century. The mean December to May air temperature on Svalbard was as high as −4.8°C, some 8.2°C above the 1961–1990 average. The 2006 mean ground temperature at the permafrost table in a monitored borehole in bedrock was 1.8°C higher than the mean for the previous six years, and this corresponded to a 40% reduction in accumulated annual negative degree-days at that depth. The thermal response was detectable to a depth of at least 15 m. In future, a greater frequency of high-temperature anomalies such as this, superimposed on a warming trend, is likely to cause potentially hazardous irregular acceleration in near-surface permafrost thawing.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 In the Svalbard archipelago, located in the North Atlantic sector of the Arctic Ocean from 74° to 81°N and 10° to 35°E, permafrost is continuous in the non-glaciated areas [Liestøl, 1976], air temperatures are characterised by large inter-annual variability [Hanssen-Bauer and Førland, 1998], and are highly sensitive to the coupled sea ice-ocean-atmosphere system [Benestad et al., 2002]. Recently observed [Vinje, 2001; Stroeve et al., 2007] and predicted future [Serreze et al., 2007] shrinkage in Arctic sea-ice cover suggests that a rapid transition in atmospheric temperatures is in progress on Svalbard, associated with increasing frequency of high-temperature anomaly events [Christensen et al., 2007]. The response of Arctic permafrost to such events is not well documented, although Atkinson et al.  recently examined the impact of the extreme warm year in 1998 on the Canadian Arctic cryosphere.
 Here we report on near-surface permafrost temperature changes resulting from a remarkable temperature anomaly observed in the Svalbard region during winter and spring 2005–2006. We demonstrate that the temperature anomaly fell well within the range of predicted warming scenarios for the late 21st Century on Svalbard. It is concluded that future near-surface permafrost warming and degradation may be strongly influenced by the frequency and intensity of such high-temperature episodes.
2. Characteristics of the Arctic Temperature Anomaly
 The extreme temperature anomaly that affected the region during the winter and spring (December to May inclusive) 2005–2006 produced record breaking temperatures. The temperature anomaly for the Svalbard region was 4–9°C above the 1961–1990 average, with highest anomalies just east of Svalbard (Figure 1a).
 The homogenised mean air temperature for the period December 2005 to May 2006 inclusive, from the official weather station at Svalbard airport [Nordli and Kohler, 2004], was as high as −4.8°C, which is 8.2°C above the 1961–1990 average (Figure 1b). This is the warmest since records began in 1911, and 2.8°C higher than the previous extreme from 1954, amounting to an offset of 3.7 standard deviations from the mean (1961–1990). The most striking months were January and April, when mean air temperatures were respectively 12.6°C and 12.2°C above the 1961–1990 average. April 2006 was warmer than any previously recorded May, and January was warmer than any previously recorded April. The anomaly coincided with open water in most of the fjords through the whole winter and unusually early snow melt releasing extreme melt water discharges in the valleys.
 The extreme anomaly was associated with an unusual synoptic situation with long periods of strong mild southerlies that dominated the winter and spring seasons. In addition, sea temperatures were unusually high, with anomalous ocean heat transport observed to the west of Svalbard [Walczowski and Piechura, 2007], and periods of sustained along-shelf winds in western Svalbard generated upwelling and cross-shelf exchange causing extensive flooding of the costal waters with warm Atlantic Water from the West Spitsbergen Current [Cottier et al., 2007]. The extreme conditions interrupted the normal cycle of sea ice formation and contributed to an unusually large extent of marine open water around Svalbard during winter, spring and summer 2005–2006. This, in turn, contributed to the extreme temperatures observed on the islands, which are highly sensitive to the sea ice-edge location [Benestad et al., 2002]. An August 2006 Envisat MERIS satellite image (http://www.esa.int/esaEO/SEM7ZF8LURE_planet_0.html) highlights areas north of Svalbard where dramatic openings in the perennial sea ice pack and a very low sea ice concentration can be seen, extending all the way to the North Pole. The extent of perennial ice has declined rapidly in recent years [Stroeve et al., 2007], but the 2006 situation was unlike anything observed in previous low ice seasons.
3. Data and Methods
 The regional-scale temperature anomaly for December 2005 to May 2006 (Figure 1a) is based on NCEP Reanalysis data [Kalnay et al., 1996], provided by the NOAA/ESRL Physical Sciences Division, Boulder Colorado (http://www.cdc.noaa.gov/). The homogenised [Nordli and Kohler, 2004] Svalbard airport (78°15′N, 15°28′E, 28 m a.s.l.) monthly temperature series (1911–2006) is a composite of several shorter series of measurements carried out at a few nearby sites. All shorter series are adjusted to be valid for the current Svalbard Airport weather station (established in 1975).
 Two instrumented permafrost boreholes, 102 m and 15 m deep, were established at Janssonhaugen, western Svalbard (78°10′N, 16°28′E, 270 m a.s.l.), in May 1998 [Sollid et al., 2000; Harris et al., 2001b] and the latter provides data for the present analysis. Borehole temperatures were measured with negative-temperature-coefficient (NTC) thermistors; Yellow Spring Instruments YSI 44006 with a absolute accuracy estimated as ±0.05°C and the relative accuracy ±0.02°C [Vonder Mühll and Holub, 1992]. The measurement interval of the thermistors is every six hours. The boreholes penetrate Cretaceous sandstone bedrock with low ice content overlain by a thin (0.2–0.5 m) weathering layer containing no organic material, the ground surface has no vegetation, and during winter snow cover is thin or completely absent due to deflation. Thus, a high correlation is observed between air temperature and ground surface temperature and the climate signal that penetrates the ground shows little disturbance by near-surface latent heat effects [Isaksen et al., 2000]. The chosen period for the ground surface mean value in Figure 2a is 1 December 2005 to 30 November 2006. To be representative and detect the full effect of the 2005–2006 air temperature anomaly that penetrated into the permafrost, the period for the mean annual ground temperature values in the series is adjusted successively with depth for the phase lag of the annual wave with respect to the surface. At, for instance, 2.0 m depth, the annual wave is delayed by 45 days and at 3.0 m the lag is 71 days [Isaksen et al., 2000].
 The data representing the climate change simulations were downscaled time series of monthly mean temperature. The 20th century runs were obtained from 24 different simulations from the WCRP CMIP3 Multi-Model ensemble [Benestad, 2005], and driven with historical emissions and natural forcing. The 2071–2100 simulations were based on 15 different simulations following the SRES A1b, from WCRP CMIP3 [Benestad, 2005]. In both cases, large-scale temperatures were taken as predictors for the downscaling [Benestad, 2004], and the European Centre for Medium-range Weather forecasts (ECMWF) reanalysis [Simmons and Gibson, 2000] (ERA40) was used to calibrate the downscaling models. Daily observations for the period 1975–2006 were taken from the Norwegian Meteorological Institute data archive. The daily values were supplemented with homogenised monthly data [Nordli and Kohler, 2004] for the period 1911–1974, interpolated to points on a daily basis by the means of cubic splines.
 The near-surface thermal response at Janssonhaugen to the extreme anomaly is clearly evident from observed mean ground temperatures for the one year period 1 December 2005 to 30 November 2006 (Figure 2a). The mean ground surface and permafrost table (2.0 m) temperatures were respectively 2.3°C and 1.8°C above the 1999–2005 average and 1.5°C and 1.1°C higher than the previously recorded maximum values. The 2006 minimum ground temperature profile for 2–15 m depth (Figure 2b) showed a positive displacement ranging from 1.6°C at 2 m to 0.3°C at 15 m depth, respectively 0.8°C and 0.2°C higher than the previously recorded maximum values. The extreme ground temperature, coupled with summer 2006 air temperatures ∼2°C above the 1961–1990 normal, also affected the thaw depth of the active layer. The start of thawing was the earliest in the 8-year record and active layer thickness was 1.80 m, exceeding the mean of the previously recorded years by 0.18 m (∼11% increase). The response of permafrost to a change in ground surface temperature is strongly modulated by active-layer thermal processes, especially phase change of water, so that the temperature response in dry bedrock reported here is much greater than in, for instance, ice-rich unconsolidated materials in which latent heat exchanges are more dominant [Smith and Riseborough, 1996; Sazonova et al., 2004]. However, at a monitored site in Endelan, approximately 12 km from Janssonhaugen, where permafrost is developed in ice-rich fine-grained soil [Harris et al., 2007], thaw penetration totalled just over 1.0 m in 2006 compared with just 0.9 m in 2005 (C. Harris, manuscript in preparation, 2007).
 The effect of the extreme event on 2006 ground temperatures is clearly illustrated by the cumulative negative ground temperature at the permafrost table (∼2 m) (Figure 3a), which shows a 600°C-day (∼40%) reduction compared with the average of the previous six years. To put the severity of the anomaly into perspective, an empirical-statistical downscaled ensemble [Benestad, 2005] was used. It is based on the multi-model World Climate Research Programme (WCRP) Coupled Model Intercomparison Project (CMIP3) of the most recent Special Report Emission Scenario (SRES) A1b (in which atmospheric CO2 reaches 720 parts per million by 2100) produced for the Intergovernmental Panel on Climate Change (IPCC) Assessment Report 4 (AR4). The annual cumulative air temperature curve for 2006 at Svalbard airport is comparable with equivalent curves based on empirical-statistical downscaling for the period 2071–2100 (Figure 3b), where the year 2006 lies well within the range of the predicted scenarios. The plot of cumulative air temperature at Svalbard airport against cumulative permafrost table temperature at Janssonhaugen, compared with normal (1961–1990) (Figure 3c), emphasises a high correlation between air and ground temperatures, and shows 2006 to have been a marked outlier, conditions falling within the modelled scenario, just outside the 25–75% inter-quartile range for 2071–2100. Clearly, the temperature in 2006 falls well within the range of predicted scenarios for 2071–2100, although seasonal warming in 2006 commenced more than one month earlier than in most previous years and in most simulations (Figure 3b). During summer, the observed accumulated temperature is in the upper range of the model simulations largely because temperatures in the first three months of the year were higher than in most previous years and higher than most projected values for the late 21st century.
5. Discussion and Conclusions
 In permafrost higher frequency atmospheric signals are filtered [Lachenbruch and Marshall, 1986] and monitoring to a depth of 100 m provides direct evidence of longer-term (decadal to multi-decadal) trends. Ground surface temperature reconstruction based on a heat conduction inversion model using data on the near-surface geothermal profile at Janssonhaugen indicated warming of the permafrost surface of 1–2°C over the last 6–8 decades [Isaksen et al., 2000]. The observed trend since 1999 is for accelerated warming, the calculated rate at the top of the permafrost being in the order of 0.6–0.7°C/decade [Isaksen et al., 2007]. Thus the 2005–2006 extreme event was superimposed on a significant warming trend.
 The downscaled temperature scenarios for Svalbard used here indicate annual mean warming during the 21st century of only 0.3–0.4°C/decade. Empirical-statistical downscaling for Arctic locations tend to underestimate warming due to a sparse observational network and lower quality in the gridded re-analyses used as predictors in model calibration. The large variability in scenario data (Figures 3b and 3c) is strongly related to coupling of ice-ocean-atmosphere systems within modelled simulations. The scatter among models reflects many factors, including the initial (late-20th century) simulated ice state, aspects of the modelled ocean circulation, simulated cloud conditions, and natural variability in the modelled system. These tie in strongly to the strength and characteristics of the positive ice-albedo feedback mechanism [Serreze et al., 2007]. While projected changes in winter sea ice extent are moderate, late-summer sea ice is projected to disappear almost completely toward the end of the 21st century in the IPCC AR4-A1b scenario [Serreze et al., 2007]. The reduction is accelerated by a number of positive feedbacks in the climate system, recognised as the “polar amplification”.
 Very little work has been done in analysing future changes in extreme events in the Arctic [Christensen et al., 2007]. However, simulations from the IPCC multi-model data sets indicate that the increase in mean temperature will be combined with an increase in the frequency of very warm winters and summers [Christensen et al., 2007]. The region of Svalbard is highly sensitive to the coupled sea ice-ocean-atmosphere system [Benestad et al., 2002] and recently observed [Vinje, 2001; Stroeve et al., 2007] and predicted future [Serreze et al., 2007] shrinkage in Arctic sea-ice cover suggests that a rapid transition in atmospheric temperatures is in progress on Svalbard associated with increasing frequency of high-temperature anomaly events [Christensen et al., 2007]. In the continuous permafrost of Svalbard, the observed and predicted general rise in ground temperatures may not pose an immediate threat to natural and human systems, although in the long-term there might be severe impacts on the stability of infrastructure. However, if the frequency of high-temperature anomalies increases, then the process of near-surface permafrost warming will be irregular rather than gradual and punctuated by rapid warming events such as that described here. Resulting increases in active-layer thickness may in certain areas be associated with unprecedented thaw settlement as ice-rich soils in the upper permafrost layer melt [Nelson et al., 2001], and in consequence, a marked increase in slope instability [Harris et al., 2001a].
 This work was supported by the Norwegian Meteorological Institute. The borehole on Janssonhaugen was mainly funded by the European Community through the 4th Framework Permafrost and Climate in Europe (PACE) Project (Contract ENV4-CT97-0492, period 1997–2001). Support was also received from the University of Oslo and the University Courses in Svalbard. The PACE project was co-ordinated by C. Harris and the leader of the Norwegian group was J. L. Sollid. At the end of the PACE project in 2001 the Norwegian Meteorological Institute has been the main financial supporter for the Janssonhaugen permafrost station. Two anonymous reviewers gave important improvements to the manuscript. The contribution of all persons and institutions mentioned is gratefully acknowledged by the authors.