SEARCH

SEARCH BY CITATION

Keywords:

  • Arctic haze;
  • trends;
  • interannual variation

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Trends and Seasonal Variations
  5. 3. Separation of Anthropogenic Emissions and Transports
  6. 4. Reconstructed Time Series via Transports and Emissions
  7. 5. Relative Contributions of Eurasia and North America
  8. 6. Conclusions
  9. Acknowledgments
  10. References

[1] Trends and interannual variations of the Canadian High Arctic aerosol record at Alert, Canada (82.5°N), from 1981 to 2007 are investigated and attributed to the influences of anthropogenic emissions and long-range transports. Sulfate and black carbon (BC) atmospheric mass concentrations declined from the mid 1980s to the late 1990s but have been relatively steady since. These tendencies are closely associated with those of the anthropogenic emissions of Eurasia (Europe and the Asian part of the former Soviet Union) and North America (United States and Canada). Interannual variations correlate with two indices derived from the 700 hPa geopotential heights. Variations in the emissions and the geopotential height indices can be used to reproduce up to 75% of the variations of the observed Arctic sulfate and BC mass concentrations. Over the 27 years of observational record, the relative contribution to sulfate and BC at Alert from Eurasia has decreased from more than 90% to about 75%. During the same time, the contributions from North American emissions has increased from less than 10% to about 25%. The increasing influence from North America was due to the faster reductions of sulfur and black carbon emissions in Eurasia during the period of these observations.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Trends and Seasonal Variations
  5. 3. Separation of Anthropogenic Emissions and Transports
  6. 4. Reconstructed Time Series via Transports and Emissions
  7. 5. Relative Contributions of Eurasia and North America
  8. 6. Conclusions
  9. Acknowledgments
  10. References

[2] Investigations into the spring Arctic haze [Greenaway, 1950; Mitchell, 1956] and subsequent “chemical fingerprinting” of its origin [Li and Barrie, 1993; Rahn et al., 1977; Rahn, 1989] revealed a significant contribution to the Arctic aerosol from anthropogenic emissions at lower latitudes. In particular, elevated levels of particles with sulfate and black carbon (BC) are regularly found in the Arctic air during the late winter and early spring. Sulfate particles scatter solar radiation and modify clouds and precipitation, which contribute to cooling of the atmosphere, whereas BC absorbs solar radiation and warms the atmosphere [Intergovernmental Panel on Climate Change, 2007]. When deposited on the snow surface in the Arctic, BC decreases the surface albedo, resulting in increased surface absorption of solar energy and hence faster melting [Clarke and Noone, 1985; Flanner et al., 2009; Hansen and Nazarenko, 2004]. For these reasons, the origin of changes in sulfate and BC in the Arctic atmospheric aerosols needs to be identified.

[3] The systematic monitoring of sulfate aerosol in the Arctic began in 1977 at Karasjok, Norway (69.47°N, 25.22°E), and in 1980 at Alert, Canada (82.46°N, 62.50°W). Since 1990, several other stations have been established to monitor sulfate, including Spitsbergen, Norway (78.90°N, 11.88°E), Barrow, Alaska (71.32°N, 156.60°W), and Janiskoski, Russia (68.93°N, 28.85°E). BC, as measured with aethelometers, was added to the long-term monitoring of the Arctic aerosols at Alert in 1990 [Sharma et al., 2002] and Barrow in 1988 [Bodhaine, 1995]. The observations from these sites indicate that the mass concentrations of sulfate and BC in the Arctic atmosphere decreased between the 1980s and early 1990s in concert with the reduction of sulfur emissions from the former Soviet Union and Europe during the 1990s [Quinn et al., 2007; Sirois and Barrie, 1999]. Sulfate has continued to decrease into the 21st century but at a slower rate. Analysis of the Arctic network data has also indicated the influence of climate variability on interannual changes on the levels of sulfate and BC in the Arctic during the winter and spring. Eckhardt et al. [2003] predicted that during positive phases of the North Atlantic Oscillation (NAO) surface concentrations of tracers in the Arctic winter were elevated by about 70% relative to negative phases of the NAO. Consistent with that prediction, Sharma et al. [2006] found that observed BC at Alert was 40% higher during the positive phase of the NAO. However, the detailed mechanisms of interannual variability remain unclear, and direct correlations of the BC with NAO are not high.

[4] This paper presents the observed sulfate aerosol from 1981 to 2007 and BC aerosol from 1989 to 2007 in the Canadian High Arctic station Alert and separates the contribution and interannual variability from anthropogenic emissions and transport factors. Furthermore, the changes in the relative contributions to the Alert aerosols from Eurasian and North American sources are investigated from the data analysis.

2. Trends and Seasonal Variations

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Trends and Seasonal Variations
  5. 3. Separation of Anthropogenic Emissions and Transports
  6. 4. Reconstructed Time Series via Transports and Emissions
  7. 5. Relative Contributions of Eurasia and North America
  8. 6. Conclusions
  9. Acknowledgments
  10. References

[5] The declining trend observed in sulfate at Alert from 1991to 1995 [Sirois and Barrie, 1999] continued until about 2002 and since then has leveled off (Figure 1). The smoothed trend line is a fit of all months of data and represents the general trends of sulfate (Figure 1a) and BC (Figure 1b) aerosols. Application of the Mann-Kendall test and the nonparametric Sen method to the monthly mean sulfate shows decreasing trends for January–April at significance levels α < 0.001. Assuming a linear trend, the Sen slope or rate of concentration change for sulfate for these 4 months ranges from −38 to −65 ng m−3 yr−1, which translates into about 2.4%–2.7% per year. BC also shows a decreasing trend from January to April from 1990 to 2007, with α < 0.01 for January and February, α < 0.001 for March, and α < 0.05 for April. The Sen slope for BC ranges from −4.2 to −8.7 ng m−3 yr−1, or about 2.3%–3.2% per year for January–April. There was a slight increase of BC since around 1997, but it leveled off and declined around 2002. This is somewhat consistent with the increase in BC from Eurasian and North American regions for the same period of time (Figure 6b). However, a general declining trend since 1980 is still dominant.

image

Figure 1. Trends and seasonal variations of sulfate and black carbon (BC) aerosols at Alert, Canada. Weekly integrated data of sulfate by an ion chromatography from a high volume sampler [Sirois and Barrie, 1999] and weekly averaged BC data presented here were measured by aethalometer and adjusted to elemental carbon concentrations derived from a thermal method [Sharma et al., 2004].

Download figure to PowerPoint

[6] As with most anthropogenic aerosol constituents measured in the Arctic boundary layer, sulfate and BC have similar seasonal variations that peak during the late winter to early spring and are minimized in the summer. The elevated sulfate and BC during the winter and spring are a result of the intensification of meridional transport from the midlatitudes to the Arctic and the stable atmosphere and reduced precipitation during these seasons [Barrie et al., 1981; Heintzenberg and Larssen, 1983; Iversen and Joranger, 1985; Shaw, 1981, 1995]. The sulfate and BC remain in the Arctic for 1–2 weeks [Stohl, 2006] to 15–30 days [Shaw, 1981, 1995].

[7] On the basis of published data, Stern [2005] provided global sulfur emissions from 1850 to 2000 for each country, which was extended to 2007 using the European Monitoring and Evaluation Programme emission database. Coefficients of correlation (r2) between monthly averaged sulfate for January–April from 1981 to 2007 at Alert and the sulfur emissions are 0.73 for Eurasia (Europe and the Asian part of the former Soviet Union) and 0.74 for North America (United States and Canada), indicating that the emissions from these two regions are significant factors for the trends in sulfate at Alert. In order to separate the influences of the emissions and transport, which could be mostly regarded as human and natural factors, the time series for January were used to explore the influences of long-range transport on the interannual variations of Alert sulfate. In contrast to the directly emitted BC, sulfate is largely secondary, that is, produced from the atmospheric oxidation of SO2. During the polar winter, a less photochemically driven aerosol is expected [Sirois and Barrie, 1999], which enable us to link more explicitly the impact of long-range transport and emissions to the ambient levels at Alert and minimizes the influences due to the changes in photochemistry productions. This is not quite so after polar sunrise in the Arctic, when SO2 oxidation can contribute to sulfate.

3. Separation of Anthropogenic Emissions and Transports

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Trends and Seasonal Variations
  5. 3. Separation of Anthropogenic Emissions and Transports
  6. 4. Reconstructed Time Series via Transports and Emissions
  7. 5. Relative Contributions of Eurasia and North America
  8. 6. Conclusions
  9. Acknowledgments
  10. References

[8] A time series Y(t) of an observed species can be expressed as consisting of two quite different unobservable parts, a so-called trend T(t) and a stochastic component X(t):

  • equation image

The recent advent of wavelets as a tool for time series analysis has sparked interest in handling trends via this approach [Capilla, 2008; Kumar and Founfoula-Georgiou, 1997]. A wavelet analysis is a transformation of Y(t) into two types of coefficients: wavelet coefficients and scaling coefficients. Together these coefficients are fully equivalent to the original time series because they can be used to reconstruct Y(t). Wavelet coefficients are related to changes of averages over specific scales, whereas scaling coefficients can be associated with averages on a specified scale. Since the scale associated with the scaling coefficients is usually fairly large, the information that these coefficients capture agrees with the notion of trend. The general idea behind trend analysis with wavelets is to associate the scaling coefficients with the trend T(t) and the wavelet coefficients (particularly those at the smallest scales) with the stochastic component X(t). Details of the mathematical derivation for the scaling and wavelet coefficients are beyond the scope of this paper but can be found in a number of references [e.g., Daubechies, 1992].

[9] Figure 2 shows the time series of original January sulfate and its decomposed components based on a three-level Daubechies 6 wavelet with a low-frequency (LF) component (i.e., T(t)) and three high-frequency (HF) components (i.e., X(t)). The analysis was done using the commercial product MATLAB with Wavelet Toolbox. In mathematics, the sum of three HF components and an LF component is equal to the original data. The LF component, which describes longer time trends, matches the declining trend of the sulfur emissions from 1980 to 2000 [Stern, 2005] for Eurasia (r2 = 0.98) and North America (r2 = 0.94) (Figure 2) better than the original time series. The LF component could also be affected by long-term trends in transport. However, a recent study [Osborn, 2006] found no significant trends in NAO from 1981 to 2005, which exerts a significant control on pollutant transport to the Arctic. It is evident that the observed declining trends of sulfate are primarily associated with the trends in emissions in Eurasia and North America and reflect the anthropogenic influences on the Arctic aerosols.

image

Figure 2. Time series of January sulfate at Alert and its wavelet-decomposed components superimposed by the annual Eurasian and North American sulfur emissions. Data for 1981–2000 are from Stern [2005], and data for 2001–2007 are from the European Monitoring and Evaluation Programme emission database (dotted lines). HF, high frequency; LF, low frequency.

Download figure to PowerPoint

[10] Considerable interannual variability is shown by the HF components that represent shorter time scales. These three HF wavelet components were found to correlate with the 700 hPa geopotential heights. This was done by correlating the two-dimensional geopotential height in the domain with the sum of the three HF components at Alert for multiple years. As the atmospheric motion or circulation results from the geopotential height patterns following the geotrophic wind theory, air circulation in the lower troposphere is meteorologically featured by the 700 hPa geopotential height distribution. Since the emission and transport factors influencing the Arctic are separated and the LF component is highly correlated with the emissions, three HF components should be closely related to the transport, which is governed by the atmospheric motion or circulation. As will be discussed later, other minor factors, such as photochemistry activity and precipitation, may also contribute to the HF or LF components, but the transport is the dominant factor in controlling the HF components.

[11] Figure 3 shows the correlation between the 700 hPa geopotential height and the sum of the three HF components for the mean January sulfate at Alert from 1981 to 2007. The correlation coefficients are indicated by the orange and purple (dark and light for b/w) shading, and superimposed are the averaged streamlines at 700 hPa denoting the climatological patterns of the westerly wave, Arctic anticyclones, and cyclones in winter. The two broad arrows indicate the predominant pathways for air masses originating from Europe and North America to reach Alert. Four high-correlation centers are evident, one on each side of the pathways. Two positively correlated centers, EA+ (118.75°E, 61.25°N) and NA+ (93.75°W, 41.25°N), with correlation coefficients r of 0.54 and 0.55, are located near the westerly stream ridges over the Eurasian and North American continents. Two negatively correlated centers, EA− (1.25°E, 81.25°N) and NA− (73.75°W, 61.25°N), with correlation coefficients of −0.35 and −0.44, correspond to the two cyclonic circulations in the Arctic. Since the geostrophic wind is determined by the horizontal pressure gradients, which govern the transport fluxes of pollutants to the Arctic, the pressure gradients between the EA+ (NA+) and EA− (NA−) centers (Figure 3) can serve as a gauge for the major transport paths of pollutants from Eurasia and North America to Alert, respectively, which regulate the predominant pathways as indicated by the two broad arrows. A third highly positive center over the Atlantic coupled with the NA− center coincides with the NAO-positive phase, but its correlation with observed interannual variability is much weaker than that of the other two positive centers. This also implies that the NAO exhibits a smaller influence on the interannual variability of transport to the Arctic.

image

Figure 3. Correlation between HF components 1–3 for the mean January sulfate measured at Alert and January 700 hPa geopotential heights of National Centers for Environmental Prediction data from 1981 to 2007 (filled contours; solid line denotes positive and dotted line denotes negative correlations). The streamlines of January mean wind at 700 hPa represent the mean circulation patterns. The broad arrows indicate the prevailing transport pathways from Eurasia and North America. The four high-correlation centers are marked EA+, EA−, NA+, and NA−. The location of the monitoring site, Alert, is at A. The correlation coefficients of 0.27, 0.34, 0.40, and 0.51 are significant at 80%, 90%, 95%, and 99% confidence levels, t-test, respectively.

Download figure to PowerPoint

[12] The index ΔZEAZNA) is defined as the difference of 700 hPa geopotential heights between the fixed EA+ (NA+) and EA− (NA−) centers (Figure 3), to represent the seasonal and interannual variations of transport of sulfate to Alert. The January mean wind streamlines at 700 hPa and the anomalies of the wind velocities relative to the 27 year mean for the five highest and lowest years of ΔZEA and ΔZNA are shown in Figure 4. For the lower-index years, both Eurasia and North America have negative anomalies of meridional circulation over the Eurasian or North American regions in mid and high latitudes with weak north-south wind flows inhibiting the transport of pollutants into the Arctic and vice versa. The wind streamlines averaged over the high- and low-index years also reflect the circulation patterns, with strong and weak air flows from the Eurasian and North American source regions to the Arctic, respectively; this leads to the interannual variability of observed aerosols at Alert.

image

Figure 4. Circulation patterns over the Arctic in January as represented by the wind streamlines at 700 hPa averaged for low- and high-index ΔZEAZNA) years (below each plot) for (a) Eurasia and (b) North America. The anomalies of wind velocities relative to the 27 year mean for low and high transport index are represented by the filled contours (solid line denotes positive and dotted line denotes negative anomalies). The broad arrows indicate the prevailing pathways of pollution transport from Eurasia and North America.

Download figure to PowerPoint

4. Reconstructed Time Series via Transports and Emissions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Trends and Seasonal Variations
  5. 3. Separation of Anthropogenic Emissions and Transports
  6. 4. Reconstructed Time Series via Transports and Emissions
  7. 5. Relative Contributions of Eurasia and North America
  8. 6. Conclusions
  9. Acknowledgments
  10. References

[13] Ruling out a significant contribution from Southeast Asia to the aerosol at Alert [Stohl, 2006], the focus is placed on emissions and transports from North America and Eurasia. An empirical reconstruction of the time series of the surface concentration (SC) of sulfate or BC is made using the transport indices and emissions from Eurasia (EEA) and North America (ENA) based on the following linear equation:

  • equation image

EEA and ENA represent the annual emissions from Eurasia and North America, respectively. Since the ΔZEA and ΔZNA reflect no information on the relative contributions of transport from Eurasia and North America to the Alert observations, two weighing coefficients aEA and aNA were introduced, which can be derived if the ΔZEA, ΔZNA, EEA, ENA, and relative contributions from Eurasia and North America are known for a specific year. According to a recent study on the pollution transport to the Arctic using the averaged results of 17 models [Shindell et al., 2008], the relative contributions from Eurasia and North America to the Arctic in 2001 are estimated to be 73% and 13% for sulfate, respectively, and 72% and 10% for BC, respectively. It should be noted that none of the 17 models captured the seasonal variations of the Arctic aerosols (both sulfate and BC) successfully when compared to observations. No Arctic haze was simulated for BC, and some models even had an unrealistic summer peak of BC at Barrow. Results for sulfate were more scattered that those for BC, with no obvious high winter or early spring haze simulated. Nevertheless, the intermodel differences should be systematic, so that the relative importance of different regions is robust. In a recent study, a model with very good performance in both the Arctic and source regions gave similar contributions from Eurasia and North America [Huang, 2010]. Actual contribution to the Alert station may fluctuate from these values but should be within the same ranges. With these numbers, aEA and aNA are estimated for sulfate and BC. Together with ΔZEA and ΔZNA, the transport functions iEA and iNA were obtained, which reflect the observed surface concentrations contributed from unit Eurasian and North American emissions. For sulfate, the transport functions iEA and iNA averaged for January–April from 1981–2007 are 0.57 and 0.48 ng m−3 Gg−1 S km2, respectively. There is no apparent long-term trend in either of these functions, but there is significant interannual variability.

[14] The time series of the observed sulfate concentrations and those reconstructed using equation (2) are shown in Figure 5 for each of January, February, March, and April. Over the 27 year record, our empirical model is able to explain 51%–75% of the variation in the observed sulfate for these 4 months. A t-test of the correlation surpassed 99.95% confidence in each of the four time series. It is hypothesized that since the two transport functions iEA and iNA represent only the effects of transport, the correlations are stronger for January and February relative to March and April due to effects of increased photochemistry or precipitation removal during and after polar sunrise.

image

Figure 5. Comparisons of the original sulfate time series at Alert with the reconstructed time series of January–April from 1981 to 2007. The insets show the correlation between these two time series.

Download figure to PowerPoint

[15] The same analysis applied to the BC observations from Alert for 1990–2005 with annual emissions of BC aerosols explains 36%–70% of the monthly variance (January–April) at a 95% confidence level based on a t-test (Figure 6). For this study, previous BC emissions [Sharma et al., 2004] were updated to 2005 by using the methodology of Cooke et al. [1999]. However, compared with the reconstructed sulfate time series, the BC reconstruction was not as good as for sulfate. The different source regions and seasonal variations between sulfate and BC emissions could cause those results. It is also noted that a better correlation during March and April was achieved for BC than during January and February, which is opposite to the sulfate reconstructions. The increased photochemistry activity for sulfate production in March and April may have been the reason for the difference. Biomass burning may also influence the BC in the Arctic. During the winter and spring seasons, most of the biomass burning is located south of the temperate regions, which are outside the Arctic air mass. Therefore, only certain episodic biomass-burning events may occur in spring within the Arctic air mass [Warneke et al., 2009] and may have limited impact on the surface BC at Alert.

image

Figure 6. Same as for Figure 5, but for BC, for observed and reconstructed time series. The annual mean emissions of BC from 1990 to 2005 plotted in Figure 6b were updated for this study.

Download figure to PowerPoint

[16] It should be noted that the reconstruction analysis only includes two major contributions: transport and emissions. Other factors, such as changes in the chemical transportation and meteorology parameters, that is, clouds and precipitation, could also have contributed to the trends and interannual variations of observed aerosol concentrations. Furthermore, like other climate indices such as the NAO, the Pacific/North American pattern (PNA), and the European (EU) pattern, the EA and NA indices are based on correlations of winter means in January from sulfate aerosol observations. The differences between overestimates and underestimates are associated with seasonal evolution. Similarly, NAO, PNA, and EU describe the atmospheric circulation better in winter than in the other seasons.

5. Relative Contributions of Eurasia and North America

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Trends and Seasonal Variations
  5. 3. Separation of Anthropogenic Emissions and Transports
  6. 4. Reconstructed Time Series via Transports and Emissions
  7. 5. Relative Contributions of Eurasia and North America
  8. 6. Conclusions
  9. Acknowledgments
  10. References

[17] The reconstructed time series of sulfate and BC from the derived transport functions and emissions from Eurasia and North America were used to estimate the relative contributions of Eurasia (iEAEEA/SCAlert) and North America (iNAENA/SCAlert) to the sulfate and BC at Alert over the last 30 years (Figure 7) based on those of Shindell et al. [2008] for year 2001. For both aerosol constituents, the Eurasian contributions have declined over nearly 30 years from close to 90% to about 75%, while the North American contributions have increased from about 10% to over 25%. The high Eurasian contributions in the 1970s and 1980s were predicted to be at about 90% in this analysis and were confirmed by other studies, with values of 94% reported for the late 1970s [Barrie et al., 1989] and about 90% for late 1980s [Christensen, 1997]. The decreasing influence from Eurasia is a result of two combining factors: transport functions and emissions. Because of the circulation patterns in the high latitudes, Alert is more susceptible to the Eurasian emissions [Barrie et al., 1989; Christensen, 1997]. A higher transport function from Eurasia implies a larger contribution to Alert aerosols from a unit Eurasian emission. Conversely, any reduction in a unit emission from Eurasia will translate into a larger decrease of aerosol concentration at Alert than from North America. Furthermore, the ratio of sulfur emissions from Eurasia to those from North America decreased from 5.4 in 1980 to 3.2 in 2000 [Stern, 2005]. The BC emissions ratio decreased from 7.7 in 1990 to 3.8 in 2005 [Sharma et al., 2004]. The faster reduction in Eurasian emissions and higher transport function has led to an increasing relative influence of North American emissions to Alert. The lead isotope ratio (Pb 206/207) observed at Alert suggests the same trend as derived here. It has slightly increased from the value of 1.160 ± 0.010 in 1983–1984 [Sturges and Barrie, 1989] to 1.165 ± 0.013 averaged from 1998 to 2003, reflecting the greater influence of North American emissions at Alert, as the North American lead is higher in isotopic ratio than Eurasian emissions.

image

Figure 7. Estimated relative contributions of Eurasian and North American emissions to the Alert (a) sulfate aerosol from 1981 to 2007 and (b) black carbon from January–April of 1990–2006. The results from this analysis are shown by the thick lines for the average contributions for January–April; the shaded areas show the min–max ranges. The black vertical line at 2001 indicates the values from the Shindell et al. [2008] model for deriving the transport functions. The line at 1980 indicates the values deduced by the trajectory transport model analysis of Barrie et al. [1989] .

Download figure to PowerPoint

6. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Trends and Seasonal Variations
  5. 3. Separation of Anthropogenic Emissions and Transports
  6. 4. Reconstructed Time Series via Transports and Emissions
  7. 5. Relative Contributions of Eurasia and North America
  8. 6. Conclusions
  9. Acknowledgments
  10. References

[18] The declining trends of sulfate and BC aerosols were found to be closely associated with emission reductions in both Eurasia and North America, with reductions in emissions from Eurasia occurring at a faster rate over the period of interest. Through the analysis, factors controlling the interannual variation of Alert aerosols were derived, and the relative contributions from Eurasia and North America were assessed. Two pairs of centers of 700 hPa geopotential heights were found to partially control the interannual variations of the aerosols arriving at Alert. The relative contributions of emissions from Eurasia and North America to the Arctic atmospheric pollutions have changed. For sulfate and BC aerosols, the Eurasian contribution declined from 90% in 1980s to about 75% in late 2000, while the North American contribution increased from 10% to about 20% in the same period. The research finding is relevant to the development of climate and air quality mitigation strategies and to the evaluation of their impact and effectiveness. The methodology developed for this study can be used to investigate other atmospheric components observed in the Arctic. The current analysis indicates that the most effective way to reduce surface aerosol concentrations at Alert is through a reduction of anthropogenic emissions in Eurasia and North America and that a unit decrease in Eurasia is about 20% (∼0.57/0.48) more effective than one in North America.

[19] A recent study concluded that that decreasing concentrations of sulfate aerosols and increasing concentrations of black carbon have substantially contributed to rapid Arctic warming during the last three decades [Shindell and Faluvegi, 2009] from modeling studies. Since both sulfate and BC are decreasing, the separation of their climate impact is complicated, as they have opposite effects. Further study is needed to find the cause and consequence of these two coincidental declines in the Arctic.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Trends and Seasonal Variations
  5. 3. Separation of Anthropogenic Emissions and Transports
  6. 4. Reconstructed Time Series via Transports and Emissions
  7. 5. Relative Contributions of Eurasia and North America
  8. 6. Conclusions
  9. Acknowledgments
  10. References