Recent surface temperature trends in the interior of East Antarctica from borehole firn temperature measurements and geophysical inverse methods

Transcription

1 GEOPHYSICAL RESEARCH LETTERS, VOL. 38,, doi: /2011gl048086, 2011 Recent surface temperature trends in the interior of East Antarctica from borehole firn temperature measurements and geophysical inverse methods Atsuhiro Muto, 1,2,3 Ted A. Scambos, 1 Konrad Steffen, 2,4 Andrew G. Slater, 1 and Gary D. Clow 5 Received 10 May 2011; revised 7 July 2011; accepted 8 July 2011; published 12 August [1] We use measured firn temperatures down to depths of 80 to 90 m at four locations in the interior of Dronning Maud Land, East Antarctica to derive surface temperature histories spanning the past few decades using two different inverse methods. We find that the mean surface temperatures near the ice divide (the highest elevation ridge of East Antarctic Ice Sheet) have increased approximately 1 to 1.5 K within the past 50 years, although the onset and rate of this warming vary by site. Histories at two locations, NUS07 5 (78.65 S, E) and NUS07 7 (82.07 S, E), suggest that the majority of this warming took place in the past one or two decades. Slight cooling to no change was indicated at one location, NUS08 5 (82.63 S, E), off the divide near the Recovery Lakes region. In the most recent decade, inversion results indicate both cooler and warmer periods at different sites due to high interannual variability and relatively high resolution of the inverted surface temperature histories. The overall results of our analysis fit a pattern of recent climate trends emerging from several sources of the Antarctic temperature reconstructions: there is a contrast in surface temperature trends possibly related to altitude in this part of East Antarctica. Citation: Muto, A., T. A. Scambos, K. Steffen, A. G. Slater, and G. D. Clow (2011), Recent surface temperature trends in the interior of East Antarctica from borehole firn temperature measurements and geophysical inverse methods, Geophys. Res. Lett., 38,, doi: /2011gl National Snow and Ice Data Center, CIRES, University of Colorado at Boulder, Boulder, Colorado, USA. 2 Department of Geography, University of Colorado at Boulder, Boulder, Colorado, USA. 3 Now at Department of Geosciences, Pennsylvania State University, University Park, Pennsylvania, USA. 4 Cooperative Institute for Research in Environmental Sciences, University of Colorado at Boulder, Boulder, Colorado, USA. 5 U.S. Geological Survey, Lakewood, Colorado, USA. Copyright 2011 by the American Geophysical Union /11/2011GL Introduction [2] The recent climate trend of the interior of East Antarctica remains unclear relative to the rest of the globe because of a lack of climate station data. Sparse available records and spatial interpolation techniques have indicated a significant warming trend over the past 50 years for the Antarctic Peninsula and the region extending southwards to West Antarctica [e.g., Monaghan et al., 2008; Steig et al., 2009]. Trends in East Antarctica are still poorly constrained, although the most recent studies suggest marginally significant warming, but with various patterns across the region [e.g., Steig et al., 2009; O Donnell et al., 2011]. Despite differences in the details of the analysis techniques and consequent spatial patterns, observed strong warming in the Antarctic Peninsula region was found to extend further south to continental regions of West Antarctica. Our analysis gives new insight regarding East Antarctic climate based on in situ measurement of borehole temperatures. [3] Reconstruction of the past surface temperature history from sub surface temperatures, often called the borehole paleothermometry [e.g., MacAyeal et al., 1991], is a source of past climatic information independent of other methods such as isotopic analysis from firn/ice cores [e.g., Cuffey, 2007]. Borehole paleothermometry applied in West Antarctica [Barrett et al., 2009; Orsi and Severinghaus, 2010] have shown surface temperature trends of approximately 0.2 K/decade for the past 50 to 80 years, similar to estimates made by regression and/or re projection of station observations using spatial patterns of both reanalysis [Monaghan et al., 2008] and satellite derived [Steig et al., 2009] nearsurface air temperatures. However, trends for the East Antarctic interior remain less clear due to the coastal bias in the location of predictor data (temperatures recorded at stations) used for spatial interpolation approach and a lack of alternative data (e.g., borehole paleothermometry or oxygen isotope analysis in firn/ice cores) targeted at decadal to century time scales. [4] We present results of borehole paleothermometry analysis at four drill sites from the Norway U.S. IPY Scientific Traverse of East Antarctica (hereinafter called the Traverse) and infer the pattern of surface temperature trends of recent decades in the interior of Dronning Maud Land in East Antarctica. 2. Data and Methodology [5] Details of our method for making firn temperature measurements and for solving the borehole paleothermometry inverse problem are presented by Muto [2010]. Therefore, we only give a concise presentation of the methodology in this paper. [6] We used firn temperature profile data obtained at four locations along the Traverse routes (Figure 1) using Automated Temperature Profiling Units (ATPUs). ATPUs consist of solar powered data loggers and ARGOS satellite data uplink systems with two thermometry strings placed in adjacent boreholes (one from near the surface to 3.5 m and the other from 3.5 m to m). The thermometry strings 1of6

2 Figure 1. (a) Map of the Traverse route in (blue) and (red), with locations of ATPUs (yellow stars). Background image is from MOA (MODIS Mosaic of Antarctica [Haran et al., 2006]). Contours are altitude above sea level at 200 m intervals, from the DEM of Bamber et al. [2009]. (b) Schematic diagram of the Automated Temperature Profiling Unit (ATPU). have a total of 16 platinum resistance thermometers (PRTs) distributed over 90 m depth (Figure 1). ATPUs were installed at four sites on the Traverse: NUS07 2, 5 and 7, and NUS08 5 (at NUS08 5, the PRT for 16 m depth was damaged). The overall uncertainty in PRT measurement was ±0.03 K. We collected multiple daily measurements (between 15 to 20 measurements per day) for more than one year at each site; January 2008 to April 2010 at NUS07 2, and 7, January 2008 to present at NUS07 5, and February 2009 to present (through early 2011) at NUS08 5. Data from the first month, which were contaminated with thermal disturbances from drilling, were discarded. Mean annual temperatures at 5 m and deeper were used for the inversion; hence years b.p. (before present) indicate before 2008 for NUS07 2, 5 and 7, and before 2009 for NUS08 5. [7] To solve the inverse problem, we require a forward model to represent the physics of heat transfer through firn and ice. We used a one dimensional heat diffusion advection equation in the ¼ @z ; where T is the temperature, t is time, z is depth (positive downwards), r is the density, c is the specific heat capacity, K is the thermal conductivity and w is the vertical firn or ice velocity. We used Dirichlet boundary conditions. The surface boundary condition is the time varying skin surface temperature that is the target of the inverse problem. The bottom boundary condition is the ice sheet basal temperature, assumed to be constant over time. In the forward model, we considered a range of basal temperatures between the pressure melting point of ice and 10 K lower. Although skin surface temperature and near surface air temperature ð1þ are not the same physical quantity, we will assume that they are closely related and do not diverge on annual or longerterm scales [e.g., Comiso, 2000]. [8] Input variables for the forward model are the thickness of firn ice column and its profile of r, c, K and w. Firn ice column thicknesses ranging from 2798 to 3530 m were obtained from the Community Ice Sheet Model at locations of NUS07 2, 5 and 7 (based on the DEM of [Bamber et al., 2009] and BEDMAP1 Plus ice sheet basal topography) or radar echo sounding at NUS08 5 [Langley et al., 2011]. We obtained the density profile based on the measured bulk density of firn cores and the parameterization of Severinghaus et al. [2010]. Thermal properties (c and K) were calculated in the same manner as those of Goujon et al. [2003] except we modified the parameterization of the thermal conductivity of firn (K f ). We allowed the parameters g and b in the equation f f i; K f ¼ K i i from Goujon et al. [2003] to be tunable parameters (subscripts i and f denote each variable for pure ice and firn, respectively). For each site, values for g and b were sought that minimize the least squares misfit between the modeled and measured temperatures using the Simulated Annealing method of Sen and Stoffa [1995, chap. 4]. We ran the forward model with the top and bottom most observed temperatures as boundary conditions at daily time steps for one year and evaluated least squares misfit at each measurement depth. Vertical firn or ice velocity, assumed constant over time, was also calculated following Goujon et al. [2003], with the accumulation rate obtained from firn core chemistry (J. McConnell, personal communication, 2008) as the speed at the surface. For the speed at the base of the ice sheet, we used both melting and frozen conditions. To account for uncertainties in input variables mentioned above, we solved the inverse problem by varying firn ice column thickness and thermal conductivity by order ±10%, and accumulation rate by ±15%. We combined all solutions to derive uncertainty bounds of the inverted surface temperature histories. [9] To obtain the surface temperature history using observed sub surface temperatures, we used two markedly different methods: the linearized inversion (similar to Parker [1994]), and the Bayesian Reversible Jump Markov chain Monte Carlo (RJ MCMC [Green, 1995]). Both methods aim to extract the propagation of surface temperature anomalies in firn; hence, more recent anomalies will be better resolved while earlier fluctuations will be smoothed at depth as temperature disturbances dissipate further through the firn. [10] The linearized inversion follows the method of Parker [1994] with enhancements by one of the authors (G.D.C.). Here, a mildly non linear problem of heat transfer in firn ice column is linearized, justification of which is outlined by Muto [2010], and cast in a Fredholm integral equation of the first kind [e.g., Oldenburg, 1984; Aster et al., 2005, chap. 1]. The integral equation is numerically approximated and the surface temperature history is given as a linear combination of basis vectors that span the function space of the solution, using the numerically stable form of the spectral expansion method [Parker, 1994, ð2þ 2of6

3 Figure 2. Surface temperature histories from the linearized inversion (blue) and the RJ MCMC (red) and their associated error bounds in dashed lines for (a) NUS07 2, (b) NUS07 5, (c) NUS07 7 and (d) NUS08 5. Horizontal bars at the bottom of the figure are spreads for 5, 10, 20 and 50 years b.p which illustrate the temporal averaging window over which the temperatures are smoothed due to heat diffusion. chap. 3.05]. However, instead of truncating the basis vectors that lead to numerical instability, we assign appropriate weights to all basis vectors through regularization based on the model norm and the data misfit, which are the measures of the smoothness of the solution and ability of the solution to simulate observed data within the measurement uncertainty, respectively. We derived the error bound of the solution as a combination of the propagation of uncertainties in the temperature measurement and input parameters to the forward model. We also calculated the spread [see, e.g., Harris and Chapman, 1998] for selected times (5, 10, 20 and 50 years b.p.), represented by horizontal bars in Figure 2, following the procedure of Parker [1994, chap. 4.02], to assess the temporal resolution of the derived surface temperature histories (i.e., the bar represents the period over which the temperature is smoothed for a given point in time). [11] The RJ MCMC was first applied to terrestrial borehole paleothermometry by Hopcroft et al. [2007]. The inverse problem is cast in a Bayesian framework and the solution is given as the posterior probability distribution of the surface temperature history, conditioned on the data and prior information [e.g., Hopcroft et al., 2009]. The mean of the posterior distribution is given as the most likely model of the surface temperature history and the 95% confidence interval presented as its error bound. The sampling of the posterior distribution by RJ MCMC involves a two stage process of proposing a surface temperature history model probabilistically, forcing the forward model with this history and then accepting or rejecting the proposed model, similar to the well known Metropolis Hastings algorithm [Malinverno, 2002; Hopcroft et al., 2009]. With the RJ MCMC, the surface temperature history is parameterized as a series of linear segments with nodes in the history and the number of nodes is treated as an unknown and the algorithm preferentially samples simpler models [Hopcroft et al., 2009], i.e., with fewer number of time nodes. This is an advantageous feature for borehole paleothermometry since we know a priori that the diffusive nature of the heat transfer in firn and ice removes high frequency temperature fluctuations, and we prefer to choose simple solutions to avoid resolving features that cannot be supported by the measured data. [12] We solved the inverse problem for 500 years, but surface temperature histories obtained from both inverse methods for 100 years before present (b.p.) and older had uncertainty bounds too large for meaningful assessments of temperature changes [Muto, 2010]. Therefore, only results for the most recent 100 years are presented here. 3. Results and Discussion [13] Inversion results for each site are shown in Figure 2. All temperature histories from the linearized inversion show temperature decreases within the most recent year. This results from our implementation of the method; basis functions from which the surface temperature history is derived represent the difference from the background longterm mean before the start of the inversion (500 years b.p.) and their values at the time of measurement equal the longterm mean [Muto, 2010]. In case of NUS07 7, however, the result from the RJ MCMC also shows a temperature drop within the most recent year. We consider this to be a true signal since the measured temperature at 5 m depth was significantly lower than at 10 m (Figure 3e). [14] At NUS08 5, both results indicate a pronounced temperature increase within the two most recent years; approximately 2.5 K and 2.6 K from the linearized inversion and the RJ MCMC, respectively. This most likely reflects a large temperature increase between the 10 and 5 m depths in 3of6

4 the measured temperature profile (Figure 3g). It may be that two anomalously warm years occurred consecutively. It is not surprising to see such an event in the inverted surface temperature histories since the signal from the most recent years should be relatively well resolved as can be seen from spreads at the bottom of the panels in Figure 2. [15] There are notable differences in the surface temperature histories from the two inverse methods at NUS07 5 and 7. Result from the linearized inversion for NUS07 5 (Figure 2b, blue line) shows a constant temperature increase up to 13 years b.p., then a large increase of approximately 1.4 K in the five most recent years. There is an event where the temperature decreases by about 0.2 K from 13 to 5 years b.p. although we can not determine if this is real since it occurs within the length of the spread calculated for 10 years b.p. Result from the RJ MCMC (Figure 2b, red line), on the other hand, indicates no marked change until 10 years b.p. then an increase of 1.4 K to present. At NUS07 7, the differences in surface temperature histories from the two methods are smaller. However, the linearized inversion shows an almost linear trend of 0.25 K/decade up to 5 years b.p., whereas the RJ MCMC shows no change until 25 years b.p., and then a non linear increase of approximately 1.3 K up to 1 year b.p. Therefore, when results from two methods are individually examined, different surface temperature trends can be inferred at NUS07 5 and 7. [16] To evaluate disparate seeming results at NUS07 5 and 7, we examined the misfits of data simulated by inverted surface temperature histories with the observed data (see Muto [2010] for details). As summarized in Table 1, surface temperature histories derived by both the linearized inversion and the RJ MCMC create data misfits lower than or equal to the expected value. Therefore, surface temperature histories from two inverse methods at all sites are both equally plausible solutions to the inverse problem. For NUS07 5 and 7 then, the onset and the duration of the warming at NUS07 5 and 7 are unclear. It is apparent that the warming has occurred within past several decades, whether it is confined to the most recent several years or it has been going on for a longer period of time. [17] Surface temperature histories at NUS07 2, 5 and 7, despite some discrepancies arising from the two inverse methods, indicate warming trends within the recent several decades. A clearly different pattern is seen at NUS08 5, where slight cooling to no change is observed except the most recent several years in results from both inversion methods. NUS07 2, 5 and 7, sites along the Traverse in the season are 1000 to 1200 m higher in altitude than NUS08 5 (Figure 1). Hence, a broad pattern of the recent surface temperature trends in East Antarctica, at least Figure 3. (a, c, e, and g) Observed data (black) and those simulated by surface temperature histories inverted by linearized inversion method (blue) and RJ MCMC (red). Error bars on black line indicate the uncertainty of temperature measurements with ATPUs (0.03 K). (b, d, f, and h) Data misfit of simulated data from the linearized inversion (blue) and the RJ MCMC, and the uncertainty of measurements (black dashed lines). Table 1. The Misfit of Data Simulated by Surface Temperature Histories, Derived by the Linearized Inversion and the RJ MCMC, With the Observed Data, and the Expected Misfit for Each Site a Site Linearized RJ MCMC Expected NUS07 2 (76.06 S, E) NUS07 5 (78.65 S, E) NUS07 7 (82.07 S, E) NUS08 5 (82.63 S, E) a Expected misfit for NUS08 5 is lower than other sites since there is one less measurement point. 4of6

5 for the Dronning Maud Land, emerges: a warming trend near the ice divide and cooling to no change off the divide. [18] As previously mentioned, there are now several examples of Antarctic wide near surface air temperature reconstruction for the past 50 years using spatial interpolation techniques. The annual trend for East Antarctica as a whole from four sources of reconstructions vary between 0.05 ± 0.12 to 0.14 ± 0.16 K/decade, mostly statistically not significant [Schneider et al., 2011]. However, maps of various reconstructions, summarized by D. P. Schneider (Antarctic trends compared, 2011, available at cgd.ucar.edu/ dschneid/david_p._schneider/antarctic. html) (the figure is made from published data sources although the webpage itself is not peer reviewed), indicate non uniform near surface air temperature trends. Dronning Maud Land is shown to be warming uniformly at a rate of 0.05 to 0.1 K/decade and 0.1 to 0.2 K/decade by Steig et al. [2009] using trended and detrended AVHRR data, respectively. Monaghan et al. [2008] reconstruction on the other hand, indicates warming trend of 0.2 to 0.3 K/decade over an area close to NUS07 2 and 5, surrounded by areas of weaker trend (0.05 to 0.1 K/decade) at lower altitudes. It also shows a small area of cooling trend of 0.2 to 0.1 K/decade at and near Halley in Coats Land that transitions to warming several hundred kilometers inland. A similar pattern is shared by GISTEMP of Hansen et al. [1999, 2010] although area of cooling between 0 and 45 W longitudes extends further inland to South Pole. In the reconstruction by the regularized least squares (RLS) method of O Donnell et al. [2011], a region in the vicinity of a line connecting Halley and South Pole, including the Recovery Subglacial Lakes area [Bell et al., 2007] where NUS08 5 is located, is also indicated to be cooling. These different patterns arise as a result of differences in details of each respective technique and there are active discussions on this topic. However, it appears as though each of these reconstructions represent some aspect of the spatial pattern we infer from our borehole paleothermometry results. [19] An interesting pattern of temperature trends are seen in the mid troposphere which may be related to the pattern at the surface. Turner et al. [2006] found a warming trend of 0.5 to 0.7 K/decade in winter (June August) at the 500 hpa level between 1971 and 2003, based on radiosonde profiles. Although this strong, statistically significant warming trend is in winter, 7 out of 8 stations in East Antarctica showed warming on an annual basis, 3 of which (South Pole, Syowa and Casey) were statistically significant (>95% confidence) trends between 0.2 and 0.4 K/decade. Altitudes of NUS07 2, 5 and 7 are higher than 3500 m above sea level where the mean annual surface air pressures are around 600 hpa [see, e.g., Schwerdtfeger, 1970; Yamanouchi et al., 2007], hence these three sites are closer to the midtropospheric level than NUS08 5. Moreover, the existence of the persistent katabatic surface airflow from the interior towards the periphery of the Antarctic Ice Sheet implies that subsidence must exist over the interior [King and Turner, 1997, chap. 4.3; Parish and Bromwich, 2007]. Therefore, we expect stronger coupling of the surface and the 500 hpa level at the higher elevation near divide sites that are close to the origin of the katabatic airflow [see Parish and Bromwich, 2007, Figure 2] than the lower elevation site (NUS08 5), which may explain the similar temperature trends. 4. Conclusion [20] Based on our results of borehole paleothermometry, an emerging picture of the surface temperature trends for the last 30 to 50 years in interior Dronning Maud Land of East Antarctica is, warming near the ice divide and cooling to no change off the divide. This indicates that the recent climatic changes in the Antarctic interior may be more complex than originally assumed [e.g., Thompson and Solomon, 2002]. Significant warming trends have been observed in the midtroposphere although its connection to the inferred pattern at the surface is speculative at this point. Hence, the exact mechanism of coupling between the mid troposphere and the near surface, and the vertical atmospheric circulations, need to be investigated in the future. In order to determine more precise magnitude and duration of the surface temperature trends, future efforts of borehole paleothermometry should attempt to reduce measurement uncertainties and extend measurements as deep as possible since these two parameters are most influential in determining the temporal extent and resolution of the surface climate signal one obtains [e.g., Clow, 1992; Hartmann and Rath, 2005]. [21] Acknowledgments. We would like to thank the members of the Traverse and personnel from the Norwegian Polar Institute, Raytheon Polar Services Company, the 109th Air National Guard and the Antarctic Logistics Center International for their help in the field. We are also grateful to Frank Urban and Mark Ohms of the USGS for their help in calibrating temperature sensors. Accumulation rate and density data were kindly provided by Joe McConnell of the DRI, University of Nevada. We thank two anonymous reviewers whose comments greatly improved the text. This work was financially supported by the U.S. National Science Foundation grant OPP to the University of Colorado. [22] The Editor thanks two anonymous reviewers for their assistance in evaluating this paper. References Aster, R., B. Borchers, and C. Thurbur (2005), Parameter Estimation and Inverse Problems, Academic Press, New York. Bamber, J. L., J. L. Gomez Dans, and J. A. Griggs (2009), A new 1 km digital elevation model of the Antarctic derived from combined satellite radar and laser data Part 1: Data and methods, Cryosphere, 3, , doi: /tc Barrett, B. E., K. W. Nicholls, T. Murray, A. M. Smith, and D. G. Vaughan (2009), Rapid recent warming on Rutford Ice Stream, West Antarctica, from borehole thermometry, Geophys. Res. Lett., 36, L02708, doi: /2008gl Bell, R. E., M. Studinger, C. A. Shuman, M. A. Fahnestock, and I. Joughin (2007), Large subglacial lakes in East Antarctica at the onset of fastflowing ice streams, Nature, 445, , doi: /nature Clow, G. D. (1992), The extent of temporal smearing in surface temperature histories derived from borehole temperature measurements, Palaeogeogr. Palaeoclimatol. Palaeoecol., 98, 81 86, doi: / (92) C. Comiso, J. C. (2000), Variability and trends in Antarctic surface temperatures from in situ and satellite infrared measurements, J. Clim., 13, , doi: / (2000)013<1674:vatias>2.0.co;2. Cuffey, K. M. (2007), Ice core methods: Borehole temperature records, in Encyclopedia of Quaternary Science, edited by S. A. Elias, pp , Elsevier, Amsterdam, doi: /b /00332-x. Goujon, C., J. M. Barnola, and C. Ritz (2003), Modeling the densification of polar firn including heat diffusion: Application to close off characteristics and gas isotopic fractionation for Antarctica and Greenland sites, J. Geophys. Res., 108(D24), 4792, doi: /2002jd Green, P. J. (1995), Reversible jump Markov chain Monte Carlo computation and Bayesian model determination, Biometrika, 82, , doi: /biomet/ of6

6 Hansen, J., R. Ruedy, J. Glascoe, and M. Sato (1999), GISS analysis of surface temperature change, J. Geophys. Res., 104(D24), 30,997 31,022, doi: /1999jd Hansen, J., R. Ruedy, M. Sato, and K. Lo (2010), Global surface temperature change, Rev. Geophys., 48, RG4004, doi: /2010rg Haran, T., J. Bohlander, T. Scambos, T. Painter, and M. Fahnestock compilers (2006), MODIS mosaic of Antarctica (MOA) image map, digital media, Natl. Snow and Ice Data Cent., Boulder, Colo. Harris, R., and D. Chapman (1998), Geothermics and climate change: 1. Analysis of borehole temperatures with emphasis on resolving power, J. Geophys. Res., 103(B4), , doi: /97jb Hartmann, A., and V. Rath (2005), Uncertainties and shortcomings of ground surface temperature histories derived from inversion of temperature logs, J. Geophys. Eng., 2(4), , doi: / /2/4/ S02. Hopcroft, P. O., K. Gallagher, and C. C. Pain (2007), Inference of past climate from borehole temperature data using Bayesian Reversible Jump Markov chain Monte Carlo, Geophys. J. Int., 171, , doi: /j x x. Hopcroft, P. O., K. Gallagher, and C. C. Pain (2009), A Bayesian partitioning modeling approach to resolve spatial variability in climate records from borehole temperature inversion, Geophys. J. Int., 178, , doi: /j x x. King, J. C., and J. Turner (1997), Antarctic Meteorology and Climatology, Cambridge Univ. Press, Cambridge, U. K., doi: / CBO Langley, K., J. Kohler, K. Matsuoka, A. Sinisalo, T. Scambos, T. Neumann, A. Muto, J. G. Winther, and M. Albert (2011), Recovery Lakes, East Antarctica: Radar assessment of sub glacial water extent, Geophys. Res. Lett., 38, L05501, doi: /2010gl MacAyeal, D. R., J. Firestone, and E. Waddington (1991), Paleothermometry by control methods, J. Glaciol., 37, Malinverno, A. (2002), Parsimonious Bayesian Markov chain Monte Carlo inversion in a nonlinear geophysical problem, Geophys. J. Int., 151, , doi: /j x x. Monaghan, A. J., D. H. Bromwich, W. Chapman, and J. C. Comiso (2008), Recent variability and trends of Antarctic near surface temperature, J. Geophys. Res., 113, D04105, doi: /2007jd Muto, A. (2010), Multi decadal surface temperature trends in East Antarctica inferred from borehole firn temperature measurements and geophysical inverse methods, Ph.D. dissertation, Univ. of Colo. at Boulder, Boulder. O Donnell, R., N. Lewis, S. McIntyre, and J. Condon (2011), Improved methods for PCA based reconstructions: Case study using the Steig et al. (2009) Antarctic temperature reconstructions, J. Clim., 24, , doi: /2010jcli Oldenburg, D. W. (1984), An introduction to linear inverse theory, IEEE Trans. Geosci. Remote Sens., GE 22, Orsi, A. J., and J. P. Severinghaus (2010), Evidence of recent warming in polar latitudes from borehole temperature, Abstract C24 04 presented at 2010 Fall Meeting, AGU, San Francisco, Calif., Dec. Parish, T. R., and D. H. Bromwich (2007), Reexamination of the nearsurface airflow over the Antarctic continent and implications on atmospheric circulations at high southern latitudes, Mon. Weather Rev., 135, , doi: /mwr Parker, R. L. (1994), Geophysical Inverse Theory, Princeton Univ. Press, Princeton, N. J. Schneider, D. P., C. Desser, and Y. Okumura (2011), An assessment and interpretation of the observed warming of West Antarctica in the austral spring, Clim. Dyn., doi: /s x. Schwerdtfeger, W. (1970), Theory and observations of the wind in the friction layer over the Antarctic plateau, Ant. J. U.S., 5, Sen, M. K., and P. L. Stoffa (1995), Global Optimization Methods in Geophysical Inversion, Elsevier, Amsterdam. Severinghaus, J. P., et al. (2010), Deep air convection in the firn at a zeroaccumulation site, central Antarctica, Earth Planet. Sci. Lett., 293, , doi: /j.epsl Steig, E. J., D. P. Schneider, S. D. Rutherford, M. E. Mann, J. C. Comiso, and D. T. Shindell (2009), Warming of the Antarctic ice sheet surface since the 1957 International Geophysical Year, Nature, 457, , doi: /nature Thompson, D. W., and S. Solomon (2002), Interpretation of recent Southern Hemisphere climate change, Science, 296, , doi: / science Turner, J., T. A. Lachlan Cope, S. Colwell, G. J. Marshall, and W. M. Connolley (2006), Significant warming of the Antarctic winter troposphere, Science, 311, , doi: /science G. D. Clow, U.S. Geological Survey, Box 25046, Mail Stop 980, Lakewood, CO 80225, USA. A. Muto, Department of Geosciences, Pennsylvania State University, 408 Deike Bldg., University Park, PA 16802, USA. (aum34@psu.edu) T. A. Scambos and A. G. Slater, National Snow and Ice Data Center, CIRES, University of Colorado at Boulder, 449 UCB, Boulder, CO 80309, USA. K. Steffen, Cooperative Institute for Research in Environmental Sciences, University of Colorado at Boulder, 216 UCB, Boulder, CO 80309, USA. 6of6