1. Executive Summary

2. Introduction

2.1 The Arctic in Global Climate

2.2. Long-term Trends

2.3. Recent Changes

2.4. Community Reaction, Organizational Efforts to Date

3. Background: Changes in the Arctic

3.1. Atmospheric Changes

Atmospheric Pressure and Circulation

Surface Air Temperature and Cloudiness

Precipitation and Evaporation

3.2. Ocean and Sea Ice Changes

Ocean

Surface Currents and Ice Drift

Ice Extent and Thickness

3.3. Terrestrial Changes

Snow Cover

Permafrost

Glacier Mass Balance

River Runoff — Hydrology

3.4. Biological and Geo-chemical Changes

Marine Species Changes

Terrestrial Species Changes

Terrestrial Vegetation Changes

Carbon Dioxide and Methane Fluxes

3.5. Human Dimension

Local Effects

Sea Ice

Storm Patterns

Sea Level

Weather Changes and Contaminants

Snow

Rain

Water Temperatures — Food

Large Scale Effects

Contaminants

North Atlantic and Barents Sea Fisheries

North Pacific and Bering Sea Fisheries

Arctic Shipping

4. Hypotheses

4.1. Hypothesis 1. Onami and the Arctic Oscillation

4.2. Hypothesis 2. Relation to Climate Change

4.3. Hypothesis 3: Potential Feedbacks

4.4. Hypothesis 4: Impacts on the Ecosystem and Society

5. Objectives

5.1. Overall Objective

5.2. Special Aspects of Objectives by Component

5.2.1. Atmosphere

5.2.2. Ocean and Ice

5.2.3. Terrestrial

5.2.4. Biological System

5.2.5. Human Dimension

6. Strategy

6.1. Long-term Observations

6.1.1. Ice and Atmosphere Observations

In situ Measurements

Terrestrial

Marine

Data Assimilation Products

Remote Sensing

Traditional Knowledge

Air Chemistry and Contaminants

Climate Diagnostics

6.1.2. Ocean and Ice Observations

Eulerian Time Series — Moorings

Lagrangian Time Series — Drifting Stations

Hydrographic Surveys

Ice Observations from the Ocean

Acoustic Remote Sensing and New Technology

Coordinated Efforts

6.1.3. Terrestrial and Glaciological Observations

6.1.4. Biogeochemical Change Measurements

6.1.5. Human Dimension

6.1.6. Historical/paleo Information

6.2. Modeling Strategy

6.2.1. Defining and Quantifying Onami

6.2.2 Modeling the Connection Between AO and Onami

6.2.3. Modeling the Global Climate Connection of AO and Onami

6.2.4. Modeling for Critical Feedbacks

6.2.5. Is Onami Predictable?

6.3. Process Studies

6.4. Application

7. Organization and Relation to Other Programs

SEARCH as a Component of CLIVAR

IARPC and the Interagency Working Group for SEARCH

SEARCH Organizational Structure

Science Steering Committee

Project Office

8. References

It is alarmingly clear that a complex suite of significant, interrelated atmospheric, oceanic, and terrestrial changes has occurred in the Arctic in recent decades. This event is affecting virtually every part of the arctic environment and is now having both direct and indirect repercussions on human society. There is evidence that these changes are strongly connected with Arctic Oscillation (AO), which is a natural mode of atmospheric variation that is potentially active over a broad range of time scales including climatic time scales. We do not know if the recent complex of changes is part of a pattern of natural variability or the beginning of a long-term shift. However, there is theoretical evidence that the observed rising trend in the AO may be indicative of greenhouse warming, thus suggesting a human cause for the recent changes in the Arctic. There are other modes of human influence on the Arctic too, such as long-range transport of contaminants, and high rates of biomass removal from the marine environment. These intermingle and interact with the large-scale natural changes in complex ways. We do not know what feedback processes on climate or ecosystems may be involved in the recent changes, or what the long-term impacts may be. Because the observed changes have made it harder for those who live in the north to predict what the future may bring, we have given the name Onami (Inuit for "tomorrow") to the complex of intertwined, pan-arctic changes.

SEARCH has been conceived as a broad, interdisciplinary, multi-scale program with a core aim of understanding Onami. Part of gaining this understanding will be to determine the full scope of Onami. However, a working definition based on present knowledge is useful. For this we define Onami as the recent and ongoing, decadal (ex. 3-50 year), pan-arctic complex of intertwined changes in the arctic physical system. These changes include among other things a decline in sea level atmospheric pressure, increased surface air temperature, cyclonic ocean circulation, and decreased sea ice cover. The physical changes produce changes in the ecosystem and living resources and impact the human population. These biological and societal consequences may also be considered part of Onami. Although the dynamics are different, the situation is similar to the El Nino - Southern Oscillation phenomenon.

Activities undertaken as part of SEARCH will guided by a series of hypotheses. The first hypothesis is that:

Onami is related to or involves the Arctic Oscillation. A key objective of SEARCH will be to understand the interactions inherent in Onami in a rigorous quantitative way. Testing this hypothesis will tell us much about the interaction of the atmosphere, ocean, and land. It will allow us to tell how Onami is tied to the global atmosphere.

The second hypothesis is that:

Onami is a component of climate change. The AO is a fundamental mode of atmospheric variability and may be tied to climate change. Onami may be tied to climate change along with it or with the other large-scale patterns of atmospheric variability. The objective is to understand how Onami fits into the larger climate change picture.

The third hypothesis is related to the first two. It is that:

Yet unknown feedbacks amongst the ocean, land, and the atmosphere are critical to Onami. These feedbacks could determine whether the Onami, and the Arctic, play critical roles in climate change.

The final hypothesis is that:

Whether or not the recent Onami is tied to long-term climate change, the physical changes have effects on arctic ecosystems and societies that are mixed with the effects of other human activity. The imperative growing from this idea is that we must distinguish between the changes (physical, ecosystem, and societal) associated with the large-scale, physical Onami phenomenon and the changes due to other causes, including human activities. Ultimately we hope to predict not only the course of Onami, but also its impact on society.

SEARCH includes four major types of activities:

• long-term observations to track the environmental changes;
• modeling to test ideas about the coupling between the different components of Onami, and to predict its future course;
• process studies to test hypotheses about critical feedbacks; and
• application of what we learn to understand the ultimate impact of the physical changes on the ecosystems and societies, and to distinguish between climate-related Onami changes and changes due to other factors such as resource utilization, pollution, economic development and population growth.

The various components of the SEARCH program are at different stages of development. For example many of the long-term physical observations could justifiably be started immediately while much of the application component has yet to be determined.

This science plan is devoted to understanding changes in the arctic environment, especially the significant changes that have occurred roughly over the last decade. These recent changes stand out in a background of longer-term environmental trends and the idea that the Arctic may be a sensitive indicator of climate change. The recent changes have been identified as a complex of related atmospheric, oceanic, and terrestrial processes. Although the dynamics are different, the situation is similar to the El Nino - Southern Oscillation phenomenon. The physical changes produce changes in the ecosystem and living resources and impact the human population. They are making it harder for those who live in the north to predict what the future may bring. Because they are related to fundamental modes of change in the global atmosphere, they may comprise a template for future global climate change. For these reasons we have named the complex of recent changes Onami, Inuit for "tomorrow". Our goal is to understand Onami.

It has been argued that the Arctic may be a sensitive indicator of global change. Modeling studies such as Manabe et al. (1991), Manabe and Stouffer (1994), and Rind et al. (1995) have indicated this. Manabe et al. (1991) show that under a representative global warming scenario, temperature increases will be amplified in the Arctic, and the upper Arctic Ocean salinity will decrease due to enhanced precipitation at high latitudes. Rind et al. (1995) suggest that the pattern of temperature change in the Arctic may be a sensitive indicator of anthropogenic global warming.

The Arctic is a significant component of the global climate system. First, the Arctic Ocean's stratification and ice cover provide a control on the surface heat and mass budgets of the north polar region, and thereby on the global heat sink (ex. Manabe et al., 1991; Rind et al., 1995). For example, if the distribution of arctic sea ice were substantially different from the present, the altered surface fluxes would affect both the atmosphere and the ocean and would likely have significant consequences for regional and global climate. Second, the export of low-salinity waters, whether liquid or in the form of desalinated sea ice, has the potential to influence the overturning cell of the global ocean through control of convection in the subpolar gyres, which in turn feed the North Atlantic (Aagaard and Carmack, 1989). For example, recent suggestions that North Atlantic and Eurasian climate variability may be predictable on decadal time scales (Griffies and Bryan, 1997) rest in part on the variability of such upstream forcing in the Greenland Sea (Delworth et al., 1997). Third, arctic marine life is conditioned by sea ice, nutrient availability, and water density. Changes in these factors may impact marine ecosystems and biogeochemical cycling of essential nutrients and dissolved organic matter. Changes in the terrestrial hydrologic cycle may alter soil moisture, impacting plant communities and their grazers. Arctic soils serve as significant sources and sinks of global carbon dioxide and methane and appear to respond sensitively to altered soil moisture and temperature (Oechel et al., 1993, 1995; Oechel et al., 1998a; Zimov et al., 1993, 1996). Finally, the results of Thompson and Wallace (1998) and others show that the atmospheric circulation of the Northern Hemisphere has been changing as part of a pole-centered pattern, termed the Arctic Oscillation (AO). Recent modeling studies suggest the AO is a fundamental mode of atmospheric change and that the positive trend seen in recent decades may be symptomatic of anthropogenic climate change (Fyfe et al., 1999; Shindell et al., 1999)

There is evidence of multi-decadal and longer trends in several key arctic variables. There has been pronounced warming over northern Eurasian and North American land areas since the early 1970s, largest during winter and spring, partly compensated by cooling over northeastern North America (Chapman and Walsh, 1993). Temperatures have also increased over the Arctic Ocean in spring and summer (Martin et al., 1997; Martin and Munoz, 1997). These changes are broadly in agreement with those depicted in model anthropogenic change experiments. Reconstructions based on proxy sources indicate that late 20th century arctic temperatures are the highest of the past 400 years. Statistical analysis of this time series against records of known forcing mechanisms suggests that the recent warming has an anthropogenic component (Overpeck et al., 1997). Available observations point to long-term and recently augmented reductions in sea ice cover (Maslanik et al., 1996; Bjorgo et al., 1997; Cavalieri et al., 1997; Zakharov, 1997, Rothrock et al., 1999). Recent data suggest that past carbon accumulation has changed to a pattern of net loss, with growing season releases of up to 150 g m-2 y-1 (Marion and Oechel, 1993; Oechel et al., 1993, 1995; Zimov et al., 1993; 1996). The Arctic has been an overall significant sink for carbon over historic and recent geologic time scales, resulting in large stores of soil carbon of perhaps 300 gigatons (Miller et al., 1983). Present conditions appear to represent significant deviations from historic and Holocene carbon fluxes, and the potential for a positive feedback on global change through losses of CO₂ to the atmosphere of up to 0.7 Gt C y-1 (about 12% of the total emission from fossil fuel use) (Oechel and Vourlitis, 1994).

Most alarming are rapid changes that have occurred in the last decade. These will be described in more detail in sections to follow, but to understand the motivation for this science plan, it is useful to review a few key findings.

In the ocean, the influence of Atlantic Water is becoming more widespread and intense than previously found. Data collected during several cruises in 1993-95 (Carmack et al., 1995; McLaughlin et al., 1996; Carmack et al., 1997; Morison et al, 1998; Steele and Boyd, 1998) all indicate that the boundary between the eastern and western halocline types has advanced from over the Lomonosov Ridge to roughly parallel to the Alpha and Mendeleyev ridges (AMR). In terms of longitudinal coverage, this means the area occupied by the eastern water types is nearly 20% greater than previously observed. Results from these cruises all suggest a warming of the Atlantic water cores over the major ridge systems. Historical data of Gorshkov (1983) and Treshnikov (1977) and the Environmental Working Group (1997) give no indication of such warm cores and show a temperature over the Lomonosov Ridge nearly 1°C colder. The salinity and temperature changes appear to have begun in the late 1980s. Data from 1991 (Anderson et al., 1994; Rudels et al., 1994) show a slight warming near the Pole, and Quadfasel (1991) reports warmer than usual temperatures in the Atlantic Water inflow in 1990, but the differences from climatology seen during the subsequent cruises are much larger.

The observed shift in frontal positions is associated with a change in ice drift (Colony and Rigor, 1993, and Rigor and Colony, 1995) and atmospheric pressure patterns (Walsh et al., 1996). The ice drift and pressure fields for the 1990’s are shifted counterclockwise 40°–60° from the 1979-92 pattern, just as the upper ocean circulation pattern derived from the hydrographic data is shifted relative to climatology (Morison et al., 1998). This change is consistent with the findings of Walsh et al. (1996) that the annual mean sea level atmospheric pressure is decreasing and has been below the 1970-95 mean in every year since 1988. This change in atmospheric pressure is part of the recent large change in atmospheric circulation of the Northern Hemisphere documented by Thompson and Wallace (1998).

There have been changes in terrestrial variables as well. Changes in air temperature have been attended by reductions in spring snow cover since the mid-1980s (Robinson et al., 1993; 1995). Arctic glaciers have exhibited negative mass balances, paralleling a global tendency (Dyurgerov and Meier, 1997; Dowdeswell et al., 1997). Other studies point to increased plant growth (Mynemi et al., 1997), northward advances of the tree line (D'Arrigo et al., 1987; Nichols, 1988), increased fire frequency (Oechel and Vourlitis, 1996; Stocks, 1991) and thawing and warming of permafrost (Pavlov, 1994; Osterkamp and Romanovsky, 1996, 1999). Recent data suggest that past carbon accumulation has changed to a pattern of net loss (Marion and Oechel, 1993; Oechel et al., 1993; Zimov et al., 1993; 1996).

As the scientific community became aware of the magnitude of recent environmental changes, a number of us took a first step in exploring the scientific issues and opportunities by circulating an open "Dear Colleague" letter (April 1997) describing many of the observations outlined above. Ultimately 40 scientists from 25 institutions co-signed the letter. These included 30 scientists from 17 U.S. institutions and 10 scientists from 8 institutions in 6 other countries. The letter was also endorsed by the Arctic Systems Science - Ocean Atmosphere Ice Interaction (ARCSS-OAII) Steering Committee as consistent with the ARCS-OAII goals. Consequently the Arctic Systems Science (ARCSS) section of the NSF Office of Polar Programs agreed to sponsor a workshop to explore the extent of the Arctic change and to begin planning a program to study it.

The workshop was held November 10-12, 1997 in Seattle, Washington. It was open to al, and those with data or modeling results indicating changes in the physical characteristics of the Arctic over the last 10 years were strongly urged to attend. Invitations to the workshop were spread informally through the Internet and through the offices of the Arctic Research Consortium of the United States (ARCUS). A total of 74 scientists from many disciplines attended the meeting, 65 from the United States and 9 from other countries. The meeting had two dominant elements. The first was presentation of results related to the recent changes in the arctic environment. The second consisted of working group discussions of key questions and observables describing the changes. This evolved into a discussion of implications or overarching questions. Through the talks, and in preparing the workshop report, we learned of the temporal correlation between the shift in the AO index (Thompson and Wallace, 1998) and the temperature increase of the Arctic Ocean Atlantic water, the increase in the surface air temperature over the Russian Arctic, the Arctic Ocean circulation changes, and the freshening in the upper Beaufort Sea. The observations suggest the recent change in the Arctic is at least a decadal scale phenomenon with broad connections to changes at lower latitudes. The workshop participants agreed that this change must be tracked and analyzed. A workshop report (Morison et al, 1998) describes the results and conclusions. The report was approved and disseminated in the early Fall of 1998.

At the ARCS-OAII Science Steering Committee (SSC) meeting October 20, 1998, the ARCS-OAII SSC termed the new program the Study of Environmental Arctic Change (SEARCH) and formulated a broad organizational plan. SEARCH is envisioned to involve long time series measurements, modeling, and some process studies to track and understand the recent changes. The Committee also recognized that the scope of search might extend beyond the traditional bounds of OAII and even perhaps the polar research community. As the next step in the planning process, the ARCS-OAII SSC directed the formation of a Search Science Steering Committee (SSSC) to work with the scientific community to develop the SEARCH Science Plan for submission to the ARCSS OAII SSC and NSF-OPP.

With this mandate, a SEARCH SSC list was composed and submitted to the ARCSS-OAII SSC. It was reviewed, and revised to fit the broad scope of SEARCH. These members of the SEARCH SC met April 22-23, 1999 to write a preliminary outline for the Science Plan and formulate the agenda and invitation list for the Science Plan Workshop. The SEARCH SSC members in attendance were:

James Morison (Chair), Ocean (Sea Ice), University of Washington

David Battisti, Modeling (Atmosphere), University of Washington

Louis Codispoti (Geochemistry), Horn Point Laboratory, University of Maryland

Hajo Eicken, Sea Ice (land snow/ice), University of Alaska

James Overland, Atmosphere (Sea Ice & Remote Sensing), NOAA PMEL

Jonathan Overpeck, Paleoclimatology (land atmosphere), University of Arizona

Mark Serreze, Climatology, U. of Colorado

Edward Carmack, Douglas Martinson, Peter Schlosser, and Charles Vorosmarty could not attend. In response to the broadening interest in the program, the SSSC had appointed working-group leaders for the workshop to include biology (Jackie Grebmeier of the University of Tennessee) and human dimension issues (Jack Kruse of the University of Alaska and University of Massachusetts).

The SEARCH Science Plan Workshop was held June 30 - July 2, 1999 at the University of Washington with the support of NSF grant OPP-9978390. The invitees included experts from many fields: atmosphere, ocean, ice hydrology and frozen ground, paleoclimatology, glaciology, chemistry, biology, and the human dimension. A list of the invited attendees is provided in Appendix A. The meeting format was to form working groups to address the various sections of the outline. Broadly speaking, the main issues included assessment of the observed changes, the relation to climate, impacts, objectives, and approach.

Program scope was a prime concern of the Workshop. Because the increase in the Arctic Oscillation index appears to be so intertwined with the other environmental changes, there was some concern that SEARCH was being narrowed down to a study of the AO, with less consideration for the whole complex of atmosphere, ice, ocean and ecosystem changes. Most felt that a study of change in general would be much too broad. Several suggestions were made to limit SEARCH to particular time or length scales, but time and space scales of the observed changes are not fully known. In the end, the scope issue was resolved by agreeing that we would not limit SEARCH to a particular set of disciplines or scales; the discussions at the workshop suggest these boundaries are unknown. Instead, the scope of SEARCH will be defined by focusing on phenomena directly related to the complex of air-ice-ocean variations we have been observing and which appear connected to the Arctic Oscillation. As discussed at the outset, we christened this complex Onami. In many ways it is similar to the El Nino-Southern Oscillation (ENSO) phenomenon. Like ENSO, the AO-ocean-ice-land complex is a climate-driven phenomenon with important effects for the ecosystem and society.

It was agreed that the focus of SEARCH would be to understand Onami and its implications. The physical science effort will seek to identify and elucidate the feedbacks between land, air, ice, and ocean that drive Onami and couple it to the rest of the globe. The ultimate benefit will be the ability to predict the course of Onami and hopefully adapt to its consequences. The biological science effort will address associated ecosystem changes while the social science efforts will examine the human impact of Onami. In drafting the Science Plan, the SSC has used these as guiding foci. We think this will give SEARCH a strong, cohesive backbone from which the subjects may vary by discipline and scale as broadly as appropriate.

The plan that follows is broken into four main sections: 1) background describing the recent changes and long-term trends as we know them, 2) science and societal issues that make these critical, 3) the objectives of SEARCH, and 4) our recommended approach.

The behavior of the arctic atmosphere is changing (Serreze et al., 1999). There have been clear changes in the pressure field and circulation pattern, surface temperature, and cloudiness. Though the records are spotty there are some indications of changes in precipitation and evaporation.

Walsh et al. (1996) examined changes in sea level pressure (SLP) over the Arctic Ocean from 1979-1994. Their analysis shows reductions in SLP over the period 1987-1994, compared with the previous eight-year period, which are largest near the pole and statistically significant for autumn and winter. Serreze et al. (1997) present similar results. The yearly average pressure maps of the International Arctic Buoy Program (IABP) suggest that the shift in the atmospheric pressure pattern began around 1988–89. Before that time one of the dominant features of the arctic sea level pressure, the Beaufort high, was usually centered over 180° longitude, but after 1988 the annual average Beaufort high was weaker and usually confined to western longitudes. This change is consistent with the findings of Walsh et al. (1996) that the annual mean atmospheric surface pressure in the Polar Basin is decreasing and has been below the 1979-95 mean in every year since 1988. Serreze et al. (1999) also show significant increases in cyclone activity north of 65°N since at least 1958 for all seasons except autumn, and increased cyclone intensity for all seasons. There have been pronounced increases in cyclonic activity at higher latitudes during summer.

Proshutinsky and Johnson (1997) and Johnson et al. (1999) categorize the decadal variations in atmospheric pressure in terms of the response of sea level in a barotropic ocean model. They report that an anticyclonic mode, with sea level raised in the center of the Arctic Ocean and high atmospheric pressure over the basin, characterized 1945-52, 1957-62, 1971-1980, and 1984-88. The other periods, especially 1989-98, have been characterized by cyclonic mode with lower sea level in the center of the basin and lower atmospheric pressure in the basin. Their strongly cyclonic trend in the 1990s is in agreement with the findings of Walsh (1996).

Linkages have been found with changes at lower latitudes. Hurrell (1996) argues that almost half of the wintertime (December-March) temperature variance over the Northern Hemisphere (north of 20°N) since 1935 can be explained from the combined effects of circulation variability based on the North Atlantic Oscillation index (NAO) (31%) and the Southern Oscillation index (SO) (16%). The SO index is a common index for the state of ENSO. The positive phase of the NAO is associated with mutual strengthening of the Icelandic Low and Azores High. Under the positive mode, surface winds tend to be northerly over Greenland and eastern Canada, with associated negative temperature anomalies. Correspondingly, west to southwesterly winds tend to advect warm, moist air masses into northern Europe and Scandinavia. The NAO is best expressed during the cold season. While exhibiting considerable interannual variability, the NAO has been in a generally positive phase since about 1970 with several particularly large positive events since about 1980 (Hurrell, 1995; 1996). As will be discussed below, Deser and Blackman (1996), Maslanik et al. (1998), Swift et al. (1997), and Dickson et al. (2000) relate changes in the arctic ice cover and ocean properties to the NAO.

The SO represents the atmospheric component of the El-Nino Southern Oscillation (ENSO) phenomenon, with its index defined from the normalized sea level pressure difference between Tahiti and Darwin. ENSO effects on extratropical circulation, associated with a change towards a more negative SO index over the past two decades, account for part of the winter cooling over the Pacific Basin and warming over northern North America. With regard to the Pacific side of the Arctic Basin, Overland et al. (1997) report that the position of the tropospheric cold pool is approximately centered over the Canadian Arctic and the Beaufort Sea, displaced from the North Pole by orographic effects of the North American mountain ranges. The result is advection of atmospheric heat and moisture into the Greenland and Barents seas, and the eastern Arctic. The position of this arctic cold pool in turn affects the position of the arctic front and the atmospheric circulation in the western Pacific. New analyses reveal a polar pattern in these fluctuations of the cold pool and North Pacific circulation, and this polar pattern has undergone a marked shift since 1990. This seems to represent a new polar teleconnection pattern.

The most comprehensive picture of the change in the atmospheric pattern is presented by considering the leading empirical orthogonal function (EOF) of sea-level pressure variation for the Northern Hemisphere. Thompson and Wallace (1998), show that the variation of this EOF for winter, which they term the Arctic Oscillation (AO), is well correlated with other changes in atmospheric conditions. The time variation of the AO index resembles the more regional NAO index, and because the AO describes a hemispheric variation that includes the North Atlantic, the NAO can be considered as an expression of the AO (Deser, 1999). As shown in Figure 3.1a (Thompson and Wallace, 1998, Figure 1), the strong negative lobe of the AO spatial structure is nearly centered over the North Pole. It has strong positive lobes over the North Pacific and North Atlantic. As shown in Figure 3.1b, the time series of the wintertime AO index is similar to that of the wintertime NAO index. Although the AO was less energetic than the NAO before 1960, both indices have been rising since the mid-1960s, accompanied by an increase in Northern Hemisphere surface air temperature over the same period. Thompson and Wallace find that the AO index is more highly correlated with surface air temperatures over the Northern Hemisphere and the Eurasian continent in particular, than is the NAO. The positive mode of the AO is associated with reduced pressure over the Arctic Ocean. The particularly rapid increase in the AO index shown after the late 1980s (Figure 3.1b) thus agrees with the results of Walsh et al. (1996). Watanabe and Nitta (1997) also confirm the timing of the change. They compare annual 500-hPa height pattern changes to averages for the previous five and ten year periods, and conclude that the northern hemisphere 500-hPa field underwent significant change in 1989, with a strong height decrease centered over the Arctic Ocean.

One of the most remarkable aspects of the Thompson and Wallace (1998) study is the connection between the leading EOF of sea level pressure and that at 50-hPa. Figure 3.1c shows that the time series of the 50-hPa and surface coefficients are strongly correlated, indicaing that the change in the atmosphere extends from the stratosphere to the surface. The results of Thompson and Wallace (1998) suggest that the atmospheric change may be driven either by radiatively induced temperature changes in the stratosphere or by a barotropic response of the polar vortex to greenhouse warming in the troposphere.

The spatial distribution of surface temperature trends in the Arctic appear to be largely associated with AO. Generally the trends (Figure 3.1a taken from Thompson and Wallace, 1998) agree with those described by Chapman and Walsh (1993) and Martin et al. (1997). Figure 3.2 shows the spatial pattern of annual mean surface air temperatures trends for the Northern Hemisphere north of 40°N over the period 1966-1995 updated from Chapman and Walsh (1993) and based on data from Jones (1994). Temperatures have increased markedly over the Eurasian and northwest North American landmasses. Locally, trends exceed 0.5°C per decade. Over the ocean basins, temperature changes are generally smaller or negative. Pronounced cooling characterizes the western subpolar north Atlantic and extends into land areas over eastern Canada and southern Greenland. The annual results are primarily due to trends for winter and spring. Spatial trend patterns for summer and autumn are weaker, with autumn showing small negative trends over northern North America and Europe.

Interpretation of temperature trends changes substantially if decades prior to 1970 are included. Based on zonal means for the 55-85°N zonal band, annual mean temperatures fell during the period 1940-1970 (see Serreze et al., 2000). While recognizing sampling problems in the early part of the record, it appears that annual temperature from 1920-1940 rose even more markedly than during the post 1970s period. Recent warming, however, is not in doubt and appears to extend into the central Arctic Ocean. By combining all Russian North Pole (NP) drifting station records from 1961-1990, Martin et al. (1997) find statistically significant increases in air temperature during May and June of 0.89°C and 0.43°C per decade, as well as significant increases for summer as a whole. Further information is available from the gridded Polar Exchange at the Sea Surface (POLES) 2-meter air temperature data set for 1979-1995, which blends the NP data with International Arctic Buoy Program (IABP) drifting buoy and coastal station records. While these data should be viewed cautiously, due the short record length and the tendency for the buoys to overestimate summer temperatures because of radiational heating, results indicate that over the 17-year record, warming has occurred from late winter through early summer.

Rigor et al. (2000) have analyzed the IABP surface air temperatures for 1979-1997, and they finds a trend of +1°C per decade during winter in the eastern Arctic but a trend of -1°C per decade in the western Arctic. In spring there is a warming trend over the whole basin, but it is largest (up to 2°C per decade) in the eastern Arctic. The data show even larger trends over the Russian Arctic. Rigor et al. find the AO accounts for more than half the warming trends over Alaska, Eurasia, and the eastern Arctic.

Stone (1997) discusses temperature and cloud cover measurements at Barrow Alaska from 1965 to 1995. These indicate a 31-year warming trend in winter and spring and cooling in fall. Although Stone (1997) does not discuss short-term interannual variations, in the context of this review we notice that since 1989 there are enhanced positive trends in temperature and cloud cover for the months of November (counter to the long-term trend), January, February, and April. The strong positive correlation between temperature and cloud cover leads Stone (1997) to the conclusion that the warming is associated with changes in cloud distribution due to changes in atmospheric circulation.

The study of Overpeck et al. (1997) places the recent warming in the perspective of the record for the past several centuries. They attempt to explain variability in a 400 year Arctic temperature record reconstructed from proxy sources in terms of the relative roles of changes in trace gas loading, irradiance (solar radiation), aerosol loading from volcanic eruptions, and atmospheric circulation. They conclude that pronounced Arctic warming between 1820 and 1920 is primarily due to reduced forcing by volcanic aerosols and increasing irradiance. After 1920, both high insolation and low aerosol loading likely continued to influence arctic climate, but exponentially increasing trace gas concentrations probably played an increasingly dominant role. Their record indicates that arctic temperatures in the 20th century are the highest of the past 400 years.

Assessing changes in the atmospheric components of the northern high latitude hydrologic budget is difficult, even for "base" variables such as precipitation. The station network is fairly sparse and there are significant problems of undercatch of solid precipitation although some investigators (e.g., Groisman et al., 1991; Groisman and Easterling, 1994) have attempted to correct for gauge biases. Based on available data, annual precipitation for the period 1900-1994 has increased over both North America and Eurasia (Nicholls et al., 1996). For North America, positive trends in annual precipitation as well as snowfall are most apparent (up to a 20% increase) during the past 40 years over Canada north of 55°N (Groisman and Easterling, 1994). For the former Soviet Union, most of the increases occurred during the earlier part of the 20th century and are larger during winter than for summer, with a tendency for reduced precipitation in some areas since the middle of the century (Groisman et al., 1991). As summarized for zonal bands, annual precipitation for the period 1990-1995 has increased for the region 55°N-85°N, with the largest changes during autumn and winter (Serreze et al., 2000). The recent analysis of Dai et al. (1997) confirms these results.

From a hydrologic viewpoint, precipitation minus evaporation (P-E) is arguably more important than precipitation by itself. Two studies (Walsh et al., 1994; Serreze et al., 1995a) have utilized the network of northern high latitude rawinsonde stations to examine P-E averaged over the Arctic Basin north of 70°N via the "aerological approach." Based on data from the early 1970s through the early 1990s there are no obvious trends, with a mean annual value around 16-17 cm. Recent updates through the middle of 1996 also reveal no trends. Parallel efforts using analyzed wind and moisture fields from the NCEP/NCAR and European Center for Medium Range Weather Forecasts (ECMWF) reanalysis archives yield somewhat higher mean annual values (18-19 cm), but also no trends (Bromwich, personal. Communication, 1997).

The Arctic Ocean and sea ice have changed in concert with the change in atmospheric circulation (Serreze et al., 1999).

The results of several recent expeditions indicate that the presence of Atlantic derived water in the Arctic has increased. Data collected from the USS Pargo (Morison et al, 1998), and the Henry Larsen in 1993 (Carmack et al., 1995; McLaughlin et al., 1996), the Polar Sea and the Louis S. St. Laurent (Carmack et al., 1997) in 1994, and the USS Cavalla in 1995 (Steele and Boyd, 1998) all indicate that the boundary between the eastern (Atlantic) and western (Pacific) halocline types has moved. Earlier it was approximately aligned with the Lomonosov Ridge, but now lies roughly over the Alpha and Mendeleyev ridges. The area occupied by the eastern water types is therefore nearly 20% greater than previously observed. The greater Atlantic influence is also manifest in warm cores observed over the Lomonosov and Mendeleyev ridges in the USS Pargo and St. Laurent data, with temperatures over the Lomonosov Ridge greater than 1.5°C. Carmack et al. (1995) and McLaughlin et al. (1996) also observed an Atlantic layer temperature increase over the Mendeleyev Ridge. The earlier data of Gorshkov (1983) and Treshnikov (1977) give no indication of such warm cores over the Mendeleyev Ridge and show a temperature over the Lomonosov Ridge nearly 1°C lower. The recently prepared digital atlas of Russian hydrographic data (Environmental Working Group, 1997) confirms that no temperatures greater than 1° were observed during numerous investigations between 1950 and 1989.

Figure 3.3 (pdf) illustrates the differences between temperature and salinity measured in the fall of 1993 from the USS Pargo (Morison et al, 1998) and climatological temperature and salinity from the Joint U.S.-Russian Atlas of the Arctic Ocean: Winter Period (EWG, 1997; Gore and Belt, 1997). The EWG atlas (EWG, 1997) is a compilation of Russian and western wintertime hydrographic data taken from 1948 to 1987. It has been objectively gridded and separated into decadal and total statistics. The differences between 1993 data and the climatology are plotted as color contours along the USS Pargo track in the three-dimensional views of Figures 3.3a and 3.3b (pdf) (taken from Figure 1 of Morison et al., 1999). Figure 3.3a (pdf) shows that the salinity in the upper 250 m has increased dramatically in a wedge extending to a front roughly aligned with the Alpha and Mendeleyev ridges. Climatologies (Levitus, 1982, Gorshkov, 1983, and EWG, 1997) indicate this front was more nearly aligned with the Lomonosov Ridge in the past. The position of the front between the saltier surface waters of the eastern Arctic and the fresher western Arctic waters has advanced about 40° of longitude across the Makarov Basin. As a result, the presence of Atlantic-derived water in the basin has increased, and the surface salinity in the Makarov has increased 2.5 o/oo. The increase is likely a conservative estimate in the uppermost layers because the EWG data represent winter conditions, while the Pargo data is from the end of summer and early fall when the surface layers would normally be fresher. This is likely part of the reason for the negative salinity difference shown for the surface waters of the Canadian Basin. The salinity increase in the Makarov Basin is comparable to the instantaneous spatial variability over the whole upper Arctic Ocean. Comparison with statistics in the EWG atlas also indicates that the change is several times the typical interannual variability in the Makarov Basin. Furthermore, Steele and Boyd (1998) find that the Eurasian Basin winter mixed layer was saltier during the early 1990s than at any time in the 40-year span of the EWG atlas.

Figure 3.3b shows that temperature has also increased in the warm core of Atlantic water over the Lomonosov Ridge, with the maximum temperature over 1°C greater than at any time in the observed past. Furthermore, the Atlantic layer is shallower than in the past, so that the temperature is over 2° greater at 200 m. A less intense warm core appears over the Mendeleyev Ridge, and there is a general warming in the Makarov Basin centered near 200 m. The slight cooling centered at about 100 m in the Makarov Basin is associated with the influx of more saline water from the Eurasian Basin. These observations suggest that the whole Makarov Basin has taken on a more Atlantic character. Furthermore, the extent of the Bering Sea water temperature maximum has retreated behind the advancing Atlantic water front. Decadal statistics from the EWG atlas indicate that this change is greater than the normal variability. The cooling below 200 m in the Canada Basin is due to a 25-50 m downward displacement of the thermocline in the Pargo data relative to the EWG climatology.

The extensive data gathered during the Arctic Ocean Section (AOS) of 1994 (Carmack et al., 1997; Swift et al., 1997) give a measure of the timing, depth and breadth of the change in ocean structure, particularly the warming of the Atlantic water. The warming in the Atlantic layer represents more than a simple increase of the temperature maximum. Comparison of the 1994 data gathered over the Eurasian slope of the Lomonosov Ridge with data from the same area gathered during the Oden cruise in 1991 shows that the temperature maximum has become both warmer and shallower, and that the warming is also seen from the top of the thermocline to depths below 1500m. The temperature gradient in the thermocline is therefore also greater in the 1994 data.

One of the most remarkable aspects of the AOS observations is the geographic distribution of warm Atlantic water. Besides encountering the temperature maximum over the Lomonosov Ridge, temperature maxima near 1°C were observed at four places over the Chukchi boundary and Mendeleyev Ridge (Figure 3.3b). This region has been visited so rarely in the past that it is difficult quantify the warming, but it is likely at least 0.2°C. The position of the warm cores suggests that the Atlantic water moves with a barotropic flow following the isobaths along the slopes and ridges, and that the warm water is a tracer that is carried along by this flow. Swift et al. (1997) use the temperature as a tracer to infer the connection to the temperature of the Atlantic water inflow through Fram Strait. At various locations they estimate the time at which the Atlantic water entered Fram Strait, and by comparing the phase-shifted core temperatures with those actually measured near Fram Strait, they argue that the warming in the Arctic Basin is due to changes in the Atlantic water inflow temperature.

The observed salinity and temperature changes appear to have begun in the late 1980s. The differences from climatology as illustrated by Figure 3.3 are too large and spatially consistent to be attributed to instrument error or normal seasonal and interannual variability (Grotefendt et al., 1998). Data from the Oden cruise in 1991 (Anderson et al., 1994; Rudels et al., 1994) show a slight warming near the Pole, and Quadfasel (1991) reports warmer than usual temperatures in the Atlantic Water inflow in 1990, but the differences from climatology seen during the subsequent cruises are much larger. Comparison of the sigma-theta profiles from the 1991 Oden and 1994 AOS cruises (Carmack et al., 1997; Swift et al., 1997) indicates that the Atlantic water from 200 to 1500 meters was less dense in 1994 than in 1991, and the large differences between the 1991 and 1994 data suggest that we are seeing an event unique to the 1990s.

Even more recent observations reveal other aspects of change that have consequences for the thermodynamic balance of the Arctic Ocean. The shoaling of the Atlantic water discussed above suggests that the halocline, which isolates the surface from the warm Atlantic water, is growing thinner. Steele and Boyd (1998) show from observations during the 1995 cruise of the USS Cavalla that the cold halocline is indeed continuing to thin. They compare Arctic Ocean hydrographic data sets from the 1990s and the EWG atlas (EWG, 1997) and show that the Eurasian Basin cold halocline layer has retreated during the 1990s to cover significantly less area than in previous years. This agrees with a comparison of data from the 1991 Oden and the 1996 Polarstern cruises by Schauer and Björk (personal communication). Steele and Boyd (1998) find a retreat of the cold halocline from the Amundsen Basin into the Makarov Basin, and the latter is the only region with a true cold halocline layer found during the cruise of the USS Cavalla. Since the cold halocline layer insulates the surface and sea ice from the heat of the Atlantic water, the halocline changes could have profound effects on the surface energy balance and sea ice in the Arctic. The mid-Eurasian Basin winter mixed layer was also saltier in 1995 than ever recorded in the 40-year EWG (1997) climatology, continuing the Eurasian Basin trend seen in 1993 (Figure 3.3).

Although the results described so far have been for in the Eurasian and Makarov basins, there have been changes in the Canada Basin as well. McPhee et al. (1998) report that during the SHEBA (Surface Heat Budget of the Arctic) deployment phase in October of 1997, multiyear ice near the center of the Beaufort Gyre was anomalously thin. The upper ocean was also both warmer (relative to freezing) and substantially less saline in 1997 than in previous years. The total salinity anomaly in the upper 100 m of the water column, compared with conditions observed in the same region during the Arctic Ice Dynamics Joint Experiment (AIDJEX) in 1975, is equivalent to an excess of about 2.4 m of freshwater, and heat content (relative to freezing) has increased by 67 MJ m-2. During AIDJEX the change in salinity over the summer of 1975 implied melt equivalent to about 0.8 m of fresh water. Analogy with the seasonal progression observed during AIDJEX suggests that up to 2 m of freshwater input may have occurred during the 1997 summer, but from salinity changes alone we can not distinguish between changes in ice melt and runoff. The increased heat content, combined with the thin ice, does suggest that during the summer of 1997 the ice concentration was lower so as to allow more solar radiation to enter the upper ocean. McPhee et al. (1998) argue that these effects may be due to reduced ice convergence in the Beaufort Sea.

The changes in the Arctic Ocean appear to be related to changes in sea ice drift and atmospheric circulation. Morison et al. (1998) show that the observed shift in ocean frontal position causes a shift in the geostrophically balanced, along-front surface current. This has shifted (Morison et al, 1998, and Steele and Boyd, 1998) in association with the decadal trend in the sea level atmospheric pressure (Walsh et al., 1996). Figure 3.4 from Steele and Boyd (1998) shows the sea level pressure and ice drift fields of the IABP (Rigor, personal communication, 1997) averaged for 1979-87 (a) and for 1988-96 (b). The patterns of pressure and ice drift for 1988-96 are shifted counterclockwise about 35° from the 1979-87 pattern, similar to the change in frontal position inferred from Figure 3.3 (about 40°). The shift in ice drift and the pressure fields is consistent with a similar comparison by Morison et al. (1998). The yearly average pressure maps from the IABP indicate that the shift in sea level pressure pattern began around 1988–89. Before that time the Beaufort high was usually centered over 180° longitude, but after 1988 the Beaufort high was weaker and usually confined to western longitudes. This change is consistent with the findings of Walsh et al. (1996) and others discussed above that the annual mean sea level pressure in the Polar Basin has been below the 1979-95 mean in every year since 1988. The time of the atmospheric shift corresponds approximately to our estimate of when the ocean changes began. Morison et al. (1998) suggest that the atmosphere might in part drive the observed changes in ocean circulation by Ekman pumping, and that the effect of these circulation changes may reach deeper with time.

Maslanik et al. (1998) give consistent evidence of the change in ice drift as related to atmospheric circulation changes. Comparison of mean ice transport patterns for 1989-96 estimated from satellite microwave imagery show the contraction of the Beaufort Gyre and shift of the transpolar drift relative to conditions from 1979-88. They examine the relation of this change to the NAO index. Maslanik et al. (1998) indicate that the change in ice drift patterns has been in conjunction with the positive NAO index since 1989. However, in earlier years of positive NAO conditions this was not always the case, suggesting the low arctic sea level pressure of the 1990s (AO) is a critical ingredient in the circulation change. The ice transport through Fram Strait in the three years since 1989 with both the strongest positive NAO index and lowest arctic sea level pressure, are twice the transports in the three years with the most negative NAO index from 1978-96.

Using an ocean model run with atmospheric forcing over the last 50 years, Proshutinsky and Johnson (1997) report two decadally varying regimes corresponding to anticyclonic and cyclonic circulations of the arctic atmosphere and ocean. The anticyclonic and cyclonic circulations correspond to a "cold and dry" and a "warm and wet" atmosphere, and a "cold and salty" and a "warm and fresh" ocean, respectively. Shifts from one regime to another are forced by changes in location and intensity of the Icelandic low and Siberian high. Maslanik et al. (1998) indicate the ice transport patterns associated with positive and negative NAO resemble weak versions of the cyclonic and anticyclonic modes of ice drift modeled by Proshutinsky and Johnson (1997). Proshutinsky and Johnson (1997) report that wind-driven ice and water motion in the Arctic alternates between anticyclonic and cyclonic circulation states, with each regime persisting for 5-7 years (the period is 10-15 years). Their arguments suggest the recent change in the Arctic Ocean circulation is an extreme expression of the cyclonic pattern.

Based on analysis of the satellite passive microwave record from the Nimbus-7 Multi-channel Microwave Radiometer (SMMR) through 1987, Gloersen and Campbell (1991) demonstrated a small but significant downward trend in arctic sea ice extent. Chapman and Walsh (1993), using a longer record (1961-1990) based on weekly U.S. Navy/NOAA National Ice Center charts since 1973 and regional sea ice data sources for earlier years, confirm a downward trend. Johannessen et al. (1995) subsequently found that this downward trend has increased since about 1989. This view is reinforced by the more recent work Bjorgo et al., (1997), who address concerns over errors in sea ice retrievals from passive microwave data and problems in blending the earlier SMMR records (1978-1987) (Parkinson and Cavalieri, 1989) with the more recent time series (1987 onwards) from the Defense Meteorological Satellite Program Special Sensor Microwave/Imager (SSM/I). The most recent study using passive microwave data through 1996 (Cavalieri et al., 1997) shows arctic sea ice extent decreasing by 2.9 +/- 0.4% per decade. Also based on the passive microwave time series, Smith (1998) shows that these ice reductions have been accompanied by a general increase in the length of the ice melt season.

As seen in the time series of Northern Hemisphere ice extent (Figure 3.5), ice extent exhibits large variability, superimposed on an overall downward trend. Further inspection of time series shows that the annual trend is strongly driven by the trends in late summer and early autumn. Extreme minima, unprecedented within the passive microwave record, are found during 1990 and 1995. These reflect primarily reduced ice cover over the Laptev and East Siberian seas where (based on data through 1995) ice extent has decreased fairly steadily since about 1990 (Maslanik et al., 1996). Recent data show record low ice extents in the Beaufort Sea in summer 1998, consistent with reports from the manned camp of the Surface Heat Budget of the Arctic (SHEBA) experiment. However, in terms of total ice extent, this anomaly is partly offset by more extensive ice on the Eurasian side of the Arctic, hence contrary to the general pattern seen in the 1990s (Maslanik et al., 1999)

With regard to longer-term (century-scale) changes, Zakharov (1997) has shown a substantial decrease of sea ice coverage in the eastern North Atlantic during the twentieth century. This trend is also apparent in the charts of the Danish Meteorological Institute (Walsh et al., 1998b), although the data used in these syntheses are primarily for the spring-summer portion of the year. Vinje and Colony (1998) extend the time series back several centuries in the vicinity of the Norwegian Sea. Large decreases of sea ice extent since the 1890s are apparent. The Koch Index of sea ice near Iceland also indicates that the twentieth century has been relatively ice-free near Iceland in comparison with the previous century.

Recent thickness results of Rothrock et al. (1999) are the most compelling evidence for change. The upper Arctic Ocean is an ice bath. Thus, its thermodynamic state is not determined by temperature; this is constrained to be near the freezing point. The thermodynamic state is determined by ice mass; take away heat and ice forms, add heat and ice melts. Rothrock et al. (1999) compare ice thickness measured by U.S. Navy submarines over the last 20 years and find an average 43% reduction in thickness for the central Arctic Ocean. This suggests a substantial change in the thermal state of the Arctic Ocean with important implications for the ice cover and the ice-albedo feedback.

The patterns of change on land have been highly variable, but generally there has been a decrease in snow cover, increase in permafrost temperatures in some areas, changes in coastlines, and a long term trend toward reduced river runoff of some major rivers.

Snow cover over the Northern Hemisphere has historically been quite variable but has been significantly below average in recent years, especially during spring. Weekly National Oceanic and Atmospheric Administration (NOAA) Northern Hemisphere charts of snow covered area (SCA), derived primarily from analysis of visible-band satellite imagery, are available since 1972 (Robinson et al., 1993; 1995). NOAA data analyzed through August 1998, presented as monthly anomalies and twelve month running means of Northern Hemisphere SCA (Figure 3.6), show generally (but by no means always) above-average coverage from the beginning of the record through the mid 1980s. Within this period, snow cover was particularly extensive in the 1970s and mid 1980s. By comparison, the late 1980s through August 1998 has been a period of generally subnormal SCA. This pattern is seen over both North America and Eurasia. The difference in annual means between 1987-present and the preceding period is statistically significant and the largest changes have occurred during spring and summer. Overall, Northern Hemisphere annual SCA has declined by about 10% since 1972 (Groisman et al., 1994b).

There is evidence that for Canada (Brown and Goodison, 1996), there has been a general decrease in snow depth since 1946, especially during spring. Winter snow depths have declined over European Russia since the turn of the century (Meshcherskaya et al., 1995; Fallot et al., 1997). However, reconstructions for Canada suggest that while there has been a general decrease in spring SCA since 1915, winter SCA has increased. Winter snow depths over parts of Russia also appear to have increased in recent decades. The common thread between studies that have examined seasonality is an overall reduction in spring snow cover.

Permafrost studies have shown strong warming and thawing trends in many areas of the Arctic (Lachenbruch and Marshall, 1986; Burn, 1992; Pavlov, 1994; Serreze et al., 1999; Osterkamp and Romanovsky, 1999). The United States Geological Survey has measured permafrost temperatures from deep drill holes in northern Alaska since the late 1940s. Based on data through the mid-1980s (Lachenbruch et al., 1982; Lachenbruch and Marshall, 1986), permafrost in this region generally warmed. Typical changes were 2-4°C although some holes showed little or no change or a cooling. The recent part of the records points to cooling in the early 1980s. Northern Alaskan data (1983 to 1993) reveal a cyclic variation in permafrost temperatures superimposed on the century- long warming and with similar amplitude (Osterkamp et. al., 1994; Osterkamp and Romanovsky, 1996). Permafrost cooled initially until the mid-1980s, warmed until the early 1990s and then cooled until 1993, followed again by warming. This pattern is supported by other investigations for Alaska (Nelson et al., 1993; Lachenbruch, 1994).

Osterkamp and Romanovsky (1999) report that warming and thawing have been occurring in the areas of Alaska partially covered by frozen ground, the discontinuous permafrost region. Their model results suggest that permafrost in this area warmed in the late 1960s and 1970s. Permafrost temperatures varied little from the late 70s to the late 1980s and then warmed again through at least 1996. The warming in recent years is about 0.5°C at 20 m depth and the thaw rates are about 0.1 m yr-1. Osterkamp and Romanovsky (1996) find that in the continuous permafrost on the North Slope of Alaska the permafrost was cooling prior to the late 1980s. Since that time, however, and coincident with the changes in the Arctic Ocean, the permafrost has been warming and thawing. Pavlov (1994) indicates that the near-surface temperatures of permafrost in northern Russia have increased 0.6-0.7°C over the 1970s and 1980s. An opposite trend is reported by Wang and Allard (1995) for northern Quebec. They observe decreasing permafrost temperatures that they relate to observed lower air temperatures in that region. The permafrost warming trend therefore is broadly consistent with the spatial variation of surface air temperature trends (Rigor, 1999)

Glacier mass balance is also highly variable. The earliest records start in 1940s. The mass balance is unknown for Greenland, but there has been a generally negative cumulative balance for small glaciers over the Arctic as a whole. Based on a comprehensive data set including the Arctic islands, Antarctica, Greenland, the mountainous areas of Siberia, central Asia, and the Caucasus , the area-weighted global mass balance of small glaciers evaluated for the period 1961-1990 is -130+/- 33 mm, or 0.25+/ -0.10 mm per unit area in sea level equivalent (Dyurgerov and Meier, 1997). This represents approximately 16% of the average rate of sea level rise in the past 100 years. Area-weighted balances have been positive only for the European sector. Over the period 1961-1990, the contribution of sea level rise to the melt of small glaciers is estimated at about 7.36 mm, and of this, the Arctic Islands contribute 1.36 mm (about 18% of the total). Alaska makes a smaller contribution of 0.54 mm (7%). The largest contribution has been from Asia of 3.34 mm (45%). In the Arctic, negative annual balances have been particularly persistent for small glaciers on Svalbard. Balances were the most negative over the period of record for the Arctic Islands in 1991 and 1993. It is stressed that conditions for individual glaciers vary. Dowdeswell et al. (1997) examined 40 arctic ice caps and glaciers with records extending back to the 1940s. They find that, while most arctic glaciers have experienced predominantly negative balances over the past few decades, some, such as in the montane parts of Scandinavia and Iceland have been positive due to increased winter precipitation.

Arctic river runoff shows a substantial seasonal variability. For dominant Russian rivers, 65-95% of the total typically occurs in late spring and summer (Pavlov, personal communication, 1999). The inter-annual variations are typically 20-30% peak to peak. Most of these rivers, and certainly the largest, Ob, Yenisey, and Lena, show a trend of increasing runoff from as far back as the 40s up to 1990. Trends in the last decade are unknown. However, Johnson et al. (1999) compare their two regime, cyclonic and anticyclonic, sea level height signal (Proshutinsky and Johnson, 1997) with the Ob, Yenisey, and Mackenzie River discharges. They find a correlation indicating the runoff from these rivers is relatively high during the cyclonic regime. The other rivers emptying into the Arctic Ocean show no correlation with these circulation regimes.

Changes in the physical environment are causing changes in the ecosystems of the North. Some of these changes are associated with biogeochemical conditions and some are related to the change in gross physical properties (e.g. ice cover).

Among the unique features of the Arctic Ocean is the distinct difference in nutrient concentration between the Atlantic and Pacific waters entering the basin. The Pacific waters are markedly higher in silicate and phosphate and modestly higher in nitrate (e.g. Codispoti and Richards, 1968; Salmon and McRoy, 1994; Anderson, 1995). Carbonate system components in the rivers draining into the Arctic also vary considerably from river to river (e.g. Anderson, 1995). Thus, it must be expected that changes in the distribution of water masses are reflected in changes in nutrient and carbonate system distributions.

Changes that impact the stratification of arctic waters will also influence nutrient distributions and can thus impact biological productivity. For example, the remarkable freshening of the surface layer noted during the SHEBA experiment was associated with low nutrient concentrations in the surface waters, suggesting that the increased stratification reduced vertical transport of nutrients into the photic zone. Any reduction of nutrient input to the northern shelves will likely limit primary production, which on these shallow shelves supports water and benthic faunal populations that are tightly linked to marine mammals and birds that are consumed by native populations.

Nutrients may also be limited at the Bering Sea source. Since the Bering Strait inflow represents a major source of nutrients to the Arctic Ocean, and since the highest nutrient values in this inflow tend to be associated with the highest salinities, the reduction in transport and decreased salinity of the Bering Strait inflow in the 1990's (Roach et al. 1995; Aagaard and Weingartner unpubl. data) may represent a significant reduction of the nutrient transport into the Arctic. Since this system is shallow and water column processes are directly coupled to the underlying benthos (Grebmeier and Barry, 1991), there could well be a rapid cascading result from lower water column production to higher trophics in this region (Aagaard et al., 1999-stategic plan, pg.37-47).

For example, recent studies in high benthic biomass regions in the northern Bering Sea indicate population declines during the 1990's that are coincident with reduced transport through Bering Strait (Grebmeier and Cooper, 1995; unpubl. data; Schell, 2000). These studies indicate a reduced carbon deposition south of St. Lawrence Island and an increase in the silt and clay content of underlying sediments, indicative of reduced transport conditions (Grebmeier, unpubl. data). These on-going benthic studies also indicate that the dominant bivalve populations in the region are declining in size as well changing in dominance, which may influence the observed decline in populations of the threatened diving Spectacled Eider Seaduck in the region (Lovvorn et al., 2000).

The extent, thickness and duration of the ice cover in the Arctic can have a major impact of on biological as well as physical systems. The recent overall decline in ice cover can influence primary production, algae species type, as well as the amount of organic carbon available to faunal communities in the water and sediments. For example, sea ice studies on the SHEBA ice camp indicate that the sea ice underwent a change in algal species composition from decades before, with the species observed in 1997/98 characterized by more brackish and freshwater forms (Melnikov et al., 1997). This may be linked to the increased diversion of Mackenzie River runoff into the central and western Beaufort Sea reported on by Macdonald et al. (1999). Finally, both scientific and indigenous people observations indicate a reduced ice extent in the 1990s (except 1999). For example, early ice breakup was observed from 1995-1998. In the Bering and Beaufort Seas in 1998 many ringed seal pups were abandoned and underweight walrus were observed in the Bering Strait area. Recent observations (albeit limited) suggest that walrus populations may be in decline (Brendon Kelly, pers. comm.).

There are possibly effects of change as far south as the Bering Sea. The ecosystem there has changed dramatically in the last decade, contemporaneous with the some of the most pronounced physical changes in the Arctic Basin. According to Brodeur et al. (1999) the biomass of large jellyfish has soared in the 1990s. Saar (2000) reviews the causes and impact of massive blooms of the small phytoplankton Emiliana huxleyi in 1997. This has boosted the population of copepods at the expense of euphausids, and thus resulted in massive die-off of the short-tailed shearwater that feeds on euphausids. The growth in Emiliana huxleyi has been attributed to unusually warm and sunny weather conditions in the spring and summer of 1997, corresponding to pronounced ice melt in the Beaufort Sea (McPhee et al., 1998).

Recent changes have been observed in fisheries, including sightings of Pacific salmon species entering rivers in the eastern Arctic and more salmon being caught off Barrow, Alaska. These shifts in the geographic location of various species are also seen on the other side of the Arctic Ocean in the Barents and White Seas. Studies there indicate that the normal fisheries in the regions that characteristically are located further south near the ice edge, were in recent years found further north with the retreating ice edge. Thus, the fisheries migrated north in conjunction with reduced ice extent.

Tynan and DeMaster (1997) have hypothesized that the decreases in ice extent and warming trends may have a profound effect on marine mammals. Bowhead whales were reported feeding very close to shore in 1997, a light-ice record-year when the ice edge receded over 200 km from shore in the Alaskan Beaufort Sea (Treacy, 1998). This observation reinforces the results of habitat selection analyses, which showed that from 1982-91 bowhead whales selected ice-free inner shelf habitat in light ice conditions, but remained offshore in slope habitat when in heavy ice cover (i.e., >70% surface cover). Conversely, habitat selection analyses for white whales over this same period (1982-91) suggest that white whales are more strongly affiliated with ice and remain in continental slope habitat no matter the ice cover or the degree of in-flow (transport) at the Bering Strait (Moore, 1997; Moore and DeMaster, in review). In the 1980s, gray whales were concentrated north of St. Lawrence Island (Moore et al. 1986; Moore and DeMaster 1997; Highsmith and Coyle, 1992), but recently are more spread out. In addition, benthic sampling in recent years suggests a decline in the ampeliscid amphipod populations in this region that support gray whale populations, thus declining food supply may also be limiting these whales (Grebmeier unpubl. data). The number of stranded gray whales reported to the National Marine Fisheries Service (NMFS) in 1999 far exceeded numbers for the preceding five years (Moore et al., in prep - Mortality Event Report). While the cause for this "event" remains unclear, it is hypothesized that the Eastern Pacific stock of gray whales may be at or near carrying capacity, especially if their primary foraging areas are not as productive as in the past (Rugh et al., 1999).

In addition to potential biological changes associated with changing ice conditions, an increase in wave action and erosion on shorelines of the Bering to Beaufort seas may both directly impact native village sites as well as provide a source of older "peat" carbon to the marine system. However, most of this old carbon is not utilizable by fauna and may actually dilute the utilizable carbon descending from phytoplankton production in the overlying waters, thus limiting benthic populations.

Mynemi et al. (1997) present evidence that photosynthetic activity of terrestrial vegetation in northern high latitudes increased from 1981 through 1991 suggestive of an increase in plant growth and a lengthening of the active growing season. The largest increases in photosynthetic activity (10-12%) are found between 45-70°N, which they argue is consistent with marked springtime warmings. Results are based on two independent records of the normalized difference vegetation index (NDVI) derived from NOAA Advanced Very High Resolution Radiometer (AVHRR) satellite records. Further analyses show continuation of the increasing NDVI on the North Slope of Alaska into 1997 (Hope et al., unpublished).

Results appear consistent with an increased amplitude in the seasonal cycle of atmospheric carbon dioxide of over 20% since the 1970s at Point Barrow, Alaska and an advance of up to seven days in the timing of CO2 draw down in spring and early summer (Keeling et al., 1996). Fung (1997), in arguing in general for the veracity of these results, points out that over the same period, CO2 has increased by only 4% (from 340-355 ppmv) and could not have enhanced photosynthesis at the NVDI rate. In addition to the possibility that temperature increases may have stimulated photosynthesis directly or indirectly by accelerating snowmelt and increasing the length of the growing season, Fung (1997) also argues those higher temperatures may have mobilized nutrients previously frozen in the soil. The NDVI record is obviously too short to make firm conclusions. In this regard, Jones and Briffa (1995) find that over the former Soviet Union, there have been no coherent changes in the duration, start or end of the growing season for the period 1950-1989 (and since the 1880s for selected stations with long temperature records). This is consistent with observations that observed temperature increases have been strongest for the winter season.

While interpretation of the NDVI time series is open to question, observations do point to a northward movement of the arctic tree line in recent decades (D'Arrigo et al., 1987; Nichols, 1998). The 1980s and 1990s have also seen an increased abundance of shrubs in northern Alaska (Chapin et al., 1995). To put these observations in perspective, it should be understood from paleological studies and timberline changes, that fairly dramatic vegetation changes have occurred over decades to centuries in the past.

There have also been recent increases in fire frequency in Alaska between 1955 and 1992 (Oechel and Vourlitis, 1996) and in other circumpolar zones that have experienced regional warmings (Stocks, 1991). Whether the recent changes are climate induced or a result of reckless human behavior is open to question.

The Arctic has been an overall significant sink for carbon over historic and recent geologic time scales, resulting in large stores of soil carbon of perhaps 300 gigatons (Miller et al., 1983). Carbon dating (¹⁴C) of peat accumulation indicates carbon uptake by arctic terrestrial ecosystems on the North Slope of Alaska through the Holocene (Marion and Oechel, 1993). Studies conducted under the International Biological Program (IBP) in the 1970s showed uptake rates of 30-100 g m-2 per year (Chapin et al., 1980; Miller et al., 1983). However, recent data suggest that past carbon accumulation has changed to a pattern of net loss, with growing season releases of up to 150 g m-2 y-1 (Marion and Oechel, 1993; Oechel et al., 1993; Zimov et al., 1993; 1996). These changes represent significant deviations from historic and Holocene carbon fluxes. They show the potential for a positive feedback on global change through losses of CO2 to the atmosphere of up to 0.7 Gt C y-1 (about 12% of the total emission from fossil fuel use) (Oechel and Vourlitis, 1994).

To investigate this apparent change, net CO2 fluxes measured an IGB site 2 in Barrow Alaska in 1971 (Coyne and Kelley, 1975) were reassessed in 1992. Both data sets comprise chamber and aerodynamic measurements. Data were also collected at surrounding sites at Barrow in the 1960s and first half of the 1990s (Oechel et al., 1995). The new measurements show that by 1992, the net CO2 sinks of -25 g m-2 y-1 had become small sources of about 1 g m-2 y-1. Even in years when the tundra appears to be a sink for CO2 during the growing season, it is a source when the full year is analyzed (Oechel et al., 1998a). These results are representative for wet sedge tundra, a common coastal vegetation type in arctic regions. Wet sedge tundra accounts for about 18% of the circumpolar tundra (Oechel and Billings, 1992).

On a larger scale, the Kuparuk Basin (North Slope of Alaska) now appears as a net source of CO2. This region is comprised mainly of acidic and non-acidic tussock tundra and wet sedge tundra (Walker et al., 1998). Approximately 20% of the growing season loss is from carbon transported to lakes and streams in groundwater and then released from water sources to the atmosphere (Kling et al., 1991). Studies from Europe, Russia and Canada also show a preponderance of arctic sites now losing carbon dioxide to the atmosphere (Zimov et al. 1993, 1996; Zamolodchikov and Karelin, unpublished). However, there are arctic sites that are neutral or a sink for CO2 (Sogaard et al., submitted).

Existing evidence suggests that the change in carbon flux to a small atmospheric source is due to the effects of recent warming and resultant change in P-E on soil moisture content and soil water table and not to the direct effects of increasing temperature on ecosystem respiration. Drying has been shown to cause increased carbon loss in the Arctic under experimental conditions (Oechel et al., 1998b) and drying has been observed in Barrow and the surrounding area (Oechel et al., 1995). Warming, where soil moisture is unchanged, would not be expected to cause a decrease in net ecosystem carbon sequestration (Shaver et al., 1992; Oechel and Vourlitis, 1994; Oechel et al., 1998b). Whether the results for Barrow can be extended to elsewhere in the Arctic remains to be established.

Thermokarst, which is expected to increase in response to observed warming of permafrost, could increase methane fluxes by increasing the area of wetlands and ponds. High-latitude wetlands currently account for 5-10% of global fluxes of methane (Reeburgh and Whalen, 1992). In addition, Siberian thermokarst lakes, which emit most of their methane in winter, could contribute to the recent increase in seasonal amplitude and winter concentration of atmospheric methane observed at high latitudes (Zimov et al., 1997). Methane release from thermokarst lakes is fueled primarily by Pleistocene carbon of terrestrial origin. However, the time series of methane release are too short to detect trends (Whalen and Reeburgh, 1992).

There is a strong human dimension to the environmental changes of recent years. These have local effects on the residents of the Arctic directly because many of them live so close to the environment. Moreover, the changes seem to be having farther reaching affects that touch society in sub-arctic and even temperate regions through fisheries and conceivably transportation.

The ecosystem changes discussed in Section 3.4 may affect the residents of the Arctic that may subsist wholly or in part on naïve species. Indeed, the hunters and fishers of the north have made many of the observations of ecosystem change described in 3.4. They have recounted recent declines in abundance of a variety of fish species as well as marine mammals and sea birds. They have reported changes in the terrestrial environment - such things as drying of lakes, wetlands, drying of summer vegetation, and the thawing of discontinuous permafrost.

Sea Ice -------------------------------------

"There have been a lot of changes in the sea ice currents and the weather. Solid ice has disappeared and there are no longer huge icebergs during fall and winter. The ice now comes later and goes out earlier and it is getting thinner. The current is stronger. It is windier on the island. We had a bad hunting season with lots of high winds. Some years ago there was a massive amount of dead murres that floated on the water. I think they caught the warm currents from Japan. Our elders tell us that our earth is getting old and needs to be replaced by a new one."

Jerry Wongittilin Sr. (Wongittilin, 2000), Savoonga, St. Lawrence Island

"A lot of the elders don’t read but they know what is coming ahead. If we don’t get any ice up here until late March or something the north wind always takes that ice out and then it is open all winter. I couldn’t go anywhere last winter. I went out there and got a few seals and took them home but after that I never went out again. It was too tough on me."

Robert Tocktoo (Tocktoo, 2000), Brevig Mission

"20 years ago we used to have ice all the way from Shaktoolik to Unalakleet - now there's no ice about 1.5 -2 miles out. The ice that's there is thinner. The thickness changed in the last 20 years."

William Takak (Takak, 2000), Shaktoolik

"We have the same problem with ice; it's not freezing up as thick. When we go hunt seal by snowmachine - used to be late October now it's getting to be December."

Enoch Scheidt (Scheidt, 2000), Kotzebue

Storm Patterns -------------------------------------

"Especially this time of the year we usually get clams when the northwest and northeast winds would come. But right after the northwest winds come, we are getting a south wind that messes up the cycle. At this time of the year, usually in the fall time, we have clams and starfish, shrimp, crabs, ducks, flounders wash up. The winds keep changing. We have all kinds of things wash up and we harvest them. This year all we got was a few starfish and little bits of wood. The right kind of wind would come and before the clams would wash up, the wind would switch to south. There are a lot less clams than when I was growing up. It depends on the weather too." Ellen Richard (Richard, 2000), Wales

"Last year we didn't get any silvers, very few salmon. I was thinking about the El Nino. Maybe the weather changes - before we used to get storms that lasted 3 days. Now it lasts a day and goes on. El Nino may be affecting the warming trend."

Peter Buck (Buck, 2000), White Mountain

"Last spring we only got 6 walrus because of the weather and ice moving out too quick. I talked to elders about the weather. A long time ago it used to be real nice for weeks and even sometimes for months. Now we only have a day or two of good weather. And a lot of times it is real windy now. They don’t know what is causing that either. And the hunters that I talked with about the ice conditions say it is getting a lot thinner. It is going out too quick. Maybe it is because of the weather. Maybe it is because of that global warming."

Herman Toolie (Toolie, 2000), Savoonga, St Lawrence Island

Sea Level -------------------------------------

"The sea level is rising and that's why at Sishmaref we've had to move 9 houses and 6 more are scheduled to be moved. The storms undercut right underneath the houses. I work for the National Weather Service. Every ten years we update our average temperature - daily and monthly normal temperature. And during the last 20 years the average temperature has risen - a difference of even 1 degree makes a tremendous difference. We think it's melting the ice and raising the sea level."

Delano Barr (Barr, 2000), Shishmaref

Weather Changes and Contaminants

"Is there going to be some conclusion at the end of three years about the effect of climate change on the observations people are making? When the winds and weather patterns are different, it will bring a lot of change. It could change how contaminants travel. It might not be just the temperature, but it may also be the sand flowing on the ice and melting it faster."

Charlie Johnson, (Johnson, 2000), Nome

Snow -------------------------------------

"It used to snow lots and there used to be not very much wind. I remember when I was a kid when there use to be lots of snow on the river. Now days there is not very much snow on the river because of the wind. Mostly last winter there was hardly any snow. I work in airport and there was no snow. And then it started raining this summer...lots! I think that some of the berries don’t grow when there is not much snow in the winter. That is what some people say. And it must have rained too much this summer and maybe that is why there are not any blackberries."

Alfred Adams (Adams, 2000), Koyuk

"Old people talk about die off winters - when the snow gets so deep the deer go onto the beach and can't get back up. We haven't had one in a long time."

Larry Willard (Willard, 2000), Ketchikan

Rain -------------------------------------

"Our seaweed is a good indication of changes in the climate. It wasn’t this last season but the one before. It turned brown and like a lot of it was destroyed and we never even had a chance to use or dry it. When it was picked and it was pretty black. Then somehow it turned brown. Too much rain is what we all thought. Something touched it, once it was picked and taken from the environment. Very discouraging, all that time and energy."

Linda Carroll (Carroll, 2000), Sitka

Water Temperatures - Food -------------------------------------

"There's a real hesitancy with clams. We used to eat them raw as my parents were cleaning them out - the buttons. I won't let my kids eat them anymore. Now you don't know. We have a generation that's scared of eating their native foods. It's from so much poison in the red tides."

Elaine Abraham (Abraham, 2000), Yakutat

"I fish. There are some things in talking with different fishermen, when the salmon season was opened, they couldn't find them. They'd show up and then they'd disappear for a while. In talking with one fisherman, he looked on some chart and he saw the water temperatures. He went where the water was colder and then that's where he found the fish. In Chatham, this year for some reason the fish weren't there like they've been in the past. Fish & Game is trying to figure it out. Why didn't the fish come in? It was taking a lot of baited hooks to get the quota. Temperature of the water, or maybe the bait hadn't moved in. Bait may have gone to the predators."

Dan Moreno (Moreno, 2000), Sitka

"Our river - we've noticed that it doesn't freeze across in the last 10 years. The temperatures are warmer. The lakes are drying up. The water is low in June, affecting the fish run - over the last two years. Sockeyes are much smaller and so are hatchery fish. When I was growing up, our fish racks were full by June - 8 bales of fish. Now we only have a bale by June."

Gloria Stickwan (Stickwan, 2000), Copper Center.

"My people hunted beaver in Hay Slough for over 100 years, and in one house we had 32 beaver. Because a lot of our lakes don't freeze as deep. We are having more warmer winters than usual on a consecutive basis. What's happening is that because the winters are warmer, the lakes don't freeze all the way down and more of the young beaver survive. We now have more beaver than ever in this slough because of warm winters that give the beaver the most favorable conditions to survive. The beaver then proceed to dam and tier off the sloughs so resident species of fish, which again provides the Indians with a very viable source of food, cannot reach their spawning ground to provide the next generation of food for the Indians of the Interior."

Paul Erhart (Erhart, 2000), Fairbanks

"There is a video that I have that shows a big green thing where the mouth of the Yukon goes. It is just like a big thing. The Japanese noticed that about 15 years ago. They said it was going to be big and that it would be detrimental to the fish. They figured that out 15 years ago. And the state never did anything. They just left the processor go and the fisheries go. And now look at where those fisheries are today?"

Gerald Nicholia (Nicholia, 2000), Tanana

Thinking in terms of larger scales, there is growing concern that the Arctic is a final destination for airborne contamination from the rest of the Northern Hemisphere. As evidenced by the comments above, this is a major concern of the indigenous population. Further, the recent changes in the arctic environment seem to have a connection with changes in the fisheries of the North Atlantic, Bering Sea Barents Sea and the Yukon River. These have resulted in regional economic change and a re-distribution of income in many areas.

Contaminants -------------------------------------

It is a mistake to think of the Arctic Ocean as being almost pristine. One reason why this is so is because of the atmospheric transport of semi-volatile organic pollutants (e.g. DDT, PCBs, etc.) that enter the atmosphere in lower latitude regions and condense out in the Arctic. Due to this mechanism, we find concentrations of pollutants such as PCB's in arctic and antarctic fauna. In addition, there is local atmospheric pollution, and the largest arctic rivers drain some heavily industrialized zones, including portions of the Former Soviet Union that were used heavily for the production and processing of radionuclides. Finally, there has been direct dumping of pollutants into the Arctic Ocean (Edson et al., 1997). It is difficult to predict what the future holds for transports of pollutants into the Arctic, except to say that the transports are likely to change, and that for some pollutants such as organochlorines and mercury there is legitimate concern (Macdonald and Bewers, 1996). There is also concern that, just as the rising AO enhanced the northward heat flux, it may be increase in the northward flux of contaminants. These concerns give rise to a lack of confidence in the safety of native foods.

North Atlantic and Barents Sea Fisheries -------------------------------------

Over the past decade, large-scale ecological changes have impacted fisheries-dependent societies around the world. Fishing pressure has been one driver for these changes, but often the changes have coincided with climatic variations as well. Economically critical groundfish populations, for example, exhibited steep declines or collapses off Norway, the Faeroe Islands, Iceland, West Greenland, Newfoundland and New England during the late 1980s or early 1990s (Hamilton et al. 1998; see Figure 3.7a). The collapse of Newfoundland’s Northern Cod fishery in 1991—92 occurred in conjunction with unusual ice conditions and a broadening of the cold intermediate layer of the Labrador Current, during a Northwest Atlantic cooling phase of the NAO. Norway’s cod fishery was partially recovering from its own crises (1989) during the same years, assisted by a Northeast Atlantic warming phase (Figure 3.7b). West Greenland’s cod fishery first developed as the warm Irminger Current extended northwards around 1920, but later declined and eventually collapsed (1992) as fishing increased and waters cooled (Vilhjálmsson 1997; see Figure 3.7c).

Climate and circulation variations directly affect commercial fish populations (particularly their reproduction, larvae, and food webs) through parameters such as water temperature, salinity anomalies, vertical mixing, and currents (Jakobsson 1992; Jakobsson et al. 1994; Klyashtorin 1998; Laevastu 1993). Moreover, fisheries themselves can increase the vulnerability of target populations to climatic change by altering age structure (e.g., removing most of the robust and high-fertility older individuals) and densities among predatory fish populations, and reducing populations of food fish (Marteinsdottir and Thorarinsson 1998). Human adaptive efforts, in response to these ecological changes, include technological intensification, shifts to alternative species, economic diversification, government subsidy, and out-migration. Fisheries-dependent communities throughout the northern Atlantic have experienced population losses during the past decade (Hamilton and Otterstad 1998; Hamilton and Haedrich 1999).

North Pacific and Bering Sea Fisheries -------------------------------------

In the North Pacific, a physical regime shift took place with an intensification of the Aleutian low pressure system in the mid-1970s. Among the many changes associated with that shift were increased Alaskan salmon catches and a change from shrimp to groundfish dominance in the Gulf of Alaska (Hare and Francis 1995, Botsford et al. 1997). Similarities have been observed between the effects on fisheries of ecological changes in the Bering Sea, and those in the Newfoundland and Barents Sea, and Newfoundland coast. The groundfish stocks associated with these areas have historically contributed to relatively stable fisheries over fairly long periods of time until recently. Surprisingly, the cod and pollock fisheries seem to be drawing down mature age classes at rates that exceed recruitment in most years. Periodically, however, a really good year provides exceptional juvenile survival, which builds the fishable stocks back up several years later as the young fish mature. (Figure 3.8 for the Barents Sea cod stock). We do not know the ecosystem changes that are causing this. Both the Barents Sea fishery and the Canadian Atlantic fisheries saw a rapid increase in the industrial fishery in the 1950s and 60s, combined with boundary disputes that frustrated fishery managers. It seems that when the natural fluctuations in productivity of the marine ecosystem are great, "normal fishing pressure" can be enough to deplete stocks beyond recovery in just a few years if oceanographic changes cause the good years to become less frequent. These fisheries, then -- which by the way are among the world's largest -- may be extremely vulnerable to climate change.

An example of the potential interaction of climate and fisheries management is the recent collapse of some western Alaska salmon stocks and the curtailment of groundfish operations in the Bering Sea due to declines in the western population of the Steller sea lion and northern fur seal. These are important current management issues. The basic science problem with resource management is that fisheries agencies with responsibility over stocks important for human harvest are driven toward solving narrow short-term problems. There is a large research effort on North Pacific fishery stock assessment (Quinn reference). For productive fisheries management, we need to understand how the whole system works, from climate influences to ocean circulation to ecosystem productivity to specific species important to humans.

Arctic Shipping -------------------------------------

Clearly, the changes in ice conditions and weather in recent years have had an impact on local transportation. This can be seen in the remarks of the Native people above. The changes may be more far reaching for their effect on the northern sea route. The Northern Sea Route has been a primary concern of Russian polar scientists for many years. Much of their research was done with the aim of improving predictions of shipping conditions along their Arctic Ocean coast. Now other nations, notably Japan and Russia, are examining the new potential of the Northern Sea Route for trade. If the Arctic change affects navigability of the northern sea route (Brigham, 1998; Brigham et al., 1999), this may change shipping patterns between Asia and northern Europe and the world economic significance of the Arctic Ocean.

We have identified a complex of related atmospheric, oceanic, and terrestrial changes that have dominated the Arctic in the last 2 decades. Because they have made it harder for those who live in the north to predict what the future may bring, we have named the complex of recent changes Onami, Inuit for "tomorrow". Onami is characterized among other things by:

Learning the full scope of Onami will be an ongoing part of SEARCH. However, a working definition based on present knowledge is useful. For this we define Onami as the recent and ongoing, decadal (ex. 3-50 year), pan-arctic complex of intertwined changes in the arctic physical system. These changes include those above. The physical changes produce changes in the ecosystem and living resources and impact the human population. These biological and societal consequences may also be considered part of Onami.

We have developed several key working hypotheses to help guide SEARCH. Our first hypothesis is that:

A second hypothesis is that:

A third hypothesis is related to the first two. It is that:

Our final hypothesis is that:

A reiteration of some background provides justification for this hypothesis. There is a large body of literature addressing climate and environmental variability associated with the NAO. As examples of recent work, Swift et al. (1997) and Dickson et al. (2000) suggest that the warming of the Atlantic water in the Arctic Ocean is linked to the NAO, for which the winter index increased since 1970 to the highest-ever values in the early 1990s. Dickson et al. (2000) argue that the southerly airflow that accompanies the positive index has resulted in warming of the two streams of Atlantic water entering the Arctic Ocean across the Barents Sea shelf and along the continental slope west of Spitsbergen. It also has resulted in increased precipitation in the Norwegian Sea and decreases in the salinity of the Atlantic water inflow. Swift et al. (1997) similarly relate changes in the Atlantic water temperatures to changes in the Fram Strait inflow, and they show that these temperature changes relate to with fluctuations in the NAO. Swift et al. (1997) also suggest that the ultimate cause of the warming is reduced winter cooling of the Atlantic water in transit through the Norwegian Sea. This agrees with arguments by Dickson et al. (2000).

While the exact relationship between the NAO and the AO is still being debated (e.g. Deser, 2000), the view taken by SEARCH is that NAO is a major component of the more fundamental AO. By virtue of this reasoning and the hemispheric scope of the AO, Hypothesis 1 considers the recent changes in the arctic environment in the context of the Arctic Oscillation.

The basis of this working hypothesis is that as the AO index rises, the strength of the polar vortex increases, and the surface pressure in the Arctic Basin decreases, weakening the Beaufort high (Walsh et al., 1996). This applies positive vorticity to the sea ice and the ocean circulation (Proshutinsky and Johnson, 1996), resulting in reduced convergence in the Beaufort Gyre. This in turn results in more open water, greater radiative heat input, and increased summer melt (McPhee et al., 1998). The change in circulation may also account for the decreased ice cover on the Siberian shelves (Maslanik et al., 1996). Steele and Boyd (1998) argue that the change in circulation re-routes Siberian river runoff and is thereby responsible for thinning the cold halocline layer. The shift of Siberian runoff to the east may also be in part responsible for the freshening of the upper layers of the Beaufort Sea (McPhee et al, 1998; Macdonald, 1999). The increased cyclonic vorticity added to the Arctic Ocean may also act to draw surface water from the lower salinity, western region of the basin and increase the amount of fresh surface water flowing out through Fram Strait. This could increase stratification in the Greenland Sea and contribute to the weakened deep convection there in recent years (Aagaard et al., 1991; Schlosser et al., 1991).

The AO pattern (Thompson and Wallace, 1998) indicates a northward component in the average winds across the Atlantic sector, carrying warm air over the Greenland-Norwegian Sea, Scandinavia, and northern Russia. It also should advect moisture northward to these regions, producing greater cloudiness and enhanced downward longwave radiation (Stone, 1997). An increase in warm air and moisture advection results in an increase in surface air temperature in the Greenland-Norwegian Sea and northern Russia (Thompson and Wallace, 1998; Rigor et al., 2000). The warming over the Norwegian Sea reduces the heat loss from the Atlantic Water before it enters the Arctic Ocean, leading to the warmer Atlantic layer observed by Carmack et al. (1997), Morison et al. (1998), Steele and Boyd (1998), and Swift et al. (1997). Following earlier arguments, the change in atmospheric circulation is also represented in a rising NAO index and the observed changes in Fram Strait inflow temperatures as described by Dickson et al. (2000) and Swift et al. (1997).

The advection of warm air into the Russian Arctic and the corresponding positive temperature trend there favors warming and thawing of permafrost (Pavlov, 1994). Rigor et al. (2000) show some of this warming spilling eastward from Russia to Alaska, where permafrost warming is also observed by Osterkamp and Romanovsky (1994, 1999). The strengthened AO pattern adds a northerly component to the airflow over eastern Canada, accounting for the surface cooling there (Rigor, 1999) and the decrease in permafrost temperatures (Wang and Allard, 1995).

Although efforts to isolate the effects of AO on the rest of the arctic system are few as yet, several modeling studies have shown the response of the Arctic Ocean to recent changes in atmospheric forcing. Using an ocean model run with atmospheric forcing over the last 50 years, Proshutinsky and Johnson (1997) report two decadally varying regimes corresponding to anticyclonic and cyclonic circulations of the arctic atmosphere and ocean. The regimes are defined by the sea surface response of a barotropic ocean model, but the average atmospheric circulation for each regime is also examined. They argue that the anticyclonic and cyclonic circulations correspond to a "cold and dry" and a "warm and wet" atmosphere, and a "cold and salty" and a "warm and fresh" ocean, respectively. Shifts from one regime to another are forced by changes in location and intensity of the Icelandic low and the Siberian high. Maslanik et al. (1998) indicate the ice transport patterns associated with positive and negative NAO resemble weak versions of the cyclonic and anticyclonic modes of ice drift modeled by Proshutinsky and Johnson (1997).

Proshutinsky and Johnson (1997) find decadal oscillations between their regimes even in the 1950s-80s when the AO variability was relatively small. This may be in part because their domain is confined to the Arctic Basin while the AO is a larger scale phenomenon. Also, they consider the total average field rather than the most significant EOF of Thompson and Wallace (1998). However, the strong cyclonic regime in the 90s found by Proshutinsky and Johnson (1997) corresponds to the high AO (cyclonic) index for this period. This suggests the recent change in the Arctic Ocean and terrestrial regions may be thought of at least in part as a response to an extreme cyclonic regional-scale pattern of Proshutinsky and Johnson (1998), brought on by a hemispheric-scale change of the AO. Using a coupled ice-ocean model forced by different years (1987, 1992) of the Proshutinsky and Johnson two regime signal, Polyakov et al (1999) predict how the anticyclonic and cyclonic regimes affect the Arctic Ocean. Their results indicate that under cyclonic forcing, precipitation increases over the ocean and decreases over land. In agreement with our working hypothesis, ice divergence under the cyclonic regime results in thinner ice and a fresher surface layer. Also, the cyclonic regime substantially reduces deep convection in the Greenland Sea by exporting more fresh water from the Arctic Ocean.

Numerical simulations by Zhang et al. (1998) suggest a strengthened inflow of Atlantic waters through the Barents Sea in recent years. This ice-ocean simulation of the past 18 years is driven by observed daily varying winds and air temperatures. It shows significant warming and salinization beginning in 1989, due mainly to a marked increase in the inflow of Atlantic water across the Barents Sea shelf. The result is a warming of the Atlantic layer within the Arctic Ocean, a weakening of the halocline in the eastern Arctic, and a decrease in sea ice volume and extent.

Zhang and Hunke (1999) report on simulations with the Parallel Ocean Program (POP) model of Los Alamos National Laboratory, which has been adapted to the Arctic Ocean. Simulated surface distributions of tracers under conditions representative of the first half of the 1990s show the Beaufort Gyre significantly decreased in extent. The simulation also shows the central Arctic having a cyclonic circulation similar to that discussed by Proshutinsky and Johnson (1996, 1997). The Transpolar Drift is absent over the Lomonosov Ridge, having been shifted over the Mendeleyev and Alpha ridges. Maslowski et al. (2000) describe similar results from their coupled Arctic Ocean model and a high-resolution global ocean model. They trace both the freshwater and the Atlantic Water using ‘dyes’ in the model runs. Their modeled response to 1979-93 winds is similar to the observed ocean changes. Taken together, these models suggest that in the 1990s most of the fresh water on the Russian shelves, instead of moving off-shelf and across the basin, drifts eastward and exits through the Canadian Archipelago.

In conclusion, there is ample reason to suggest the cyclonic spin up of the atmosphere associated with a positive AO may drive many of the changes we associate with Onami. Testing this hypothesis helps us define Onami and the area of our concerns. It will lead to a quantitative understanding of the physical mechanisms controlling Onami.

There is growing observational and theoretical evidence that the Arctic Oscillation is both a natural pattern of variability of the atmosphere and distinctive component of climate change. The AO appears over a broad range of frequencies and has effects well beyond the Arctic. In observations Thompson and Wallace (1999a) find that the annualized AO follows the same decadal pattern as the wintertime index of Thompson and Wallace (1998). Moreover, they find that a complementary annular mode in the Southern Hemisphere, the Antarctic Oscillation (AAO), has displayed the same pattern as the AO. Thompson and Wallace (1999b) find that the AO ranges across a wide range of frequencies and that daily weather as far south as the northwestern United States can be correlated with the daily AO index.

Numerous simulations with different models reveal that the AO (and the AAO) are important natural modes of variability of the global atmospheric circulation. Shindell et al. (1999a, 1999b) find a strong AO signal in simulations with models based on the Goddard Institute of Space Studies (GISS) atmospheric General Circulation Model (GCM). Fyfe et al. (1999) find a strong AO signal in results from the Canadian Center for Climate Modeling and Analysis (CCCMA) coupled GCM. Yamazaki and Shinya (1999) have found the AO to be a dominant mode of variability in simulations with the Center for Climate Systems Research/National Institute for Environmental Studies (CCSR/NIES) atmospheric GCM. Hall and Visbeck (1999) find clear evidence of strong annular modes (such as the AO and AAO) in simulations with the Geophysical Fluid Dynamics Laboratory (GFDL) R15 coupled ocean-atmosphere model.

Because the AO is a natural mode of oscillation, it is reasonable to think it may be an important part of a climate change pattern. A priori, the rising trend in the AO appears consistent with greenhouse warming in that it involves heating of the lower atmosphere and cooling of the upper atmosphere. In fact, substantial components of greenhouse warming scenarios from state-of-the-art climate models such as those of Fyfe, et al. (1999) and Shindell et al. (1999a) conform roughly to the form and spatial pattern of observed changes associated with the recent increase in the AO index. However, the recent changes in the AO are not only larger than previously observed in this century, but they are larger and have come sooner than predicted by these 100-year simulations of greenhouse warming. Fyfe, et al. (1999) find the Northern Hemisphere long-term climate change includes a 28% contribution from AO, 35% is from the 3rd EOF, and very small contributions from the other EOFs. Consistent with observations, the simulations, indicate that AO and AAO also oscillate at high frequencies (Fyfe et al., 1999; Shindell et al, 1999). Using the GISS Middle Atmosphere Model, Shindell et al. (1999b) find that the simulated response to increased greenhouse gases leads to an increasing trend in the AO index comparable to that observed over the last 30 years. Increased stratospheric aerosols due to volcanic eruptions have a similar effect, but the surface climate response to solar cycles is more complicated. This suggests the rising AO trend is a possible fingerprint of anthropogenic change. Similarly, Robock et al.. (1999) find in observations and in simulations with the Max Planck Institute ECHAM 4 and GFDL SKYHI GCMs that the AO responds positively to volcanic eruptions in the tropics. The tropical lower stratosphere heats up due to the volcanic aerosols. This increases the pole-to-equator temperature gradient and thus strengthens the polar vortex. The response propagates into the troposphere and due to the standing wave pattern of the AO, advectively warms much of the Northern Hemisphere land mass.

Many questions remain about how AO is driven. For example the role of the stratosphere is unclear. Fyfe, et al. (1999) produce an AO response in a model without an active stratosphere. This suggests the AO can be driven from the surface, for example by CO₂ induced warming. In contrast, the model results of Shindell et al. (1999a, 1999b), Yamazaki et al. (1999), Robock et al. (1999), and Christiansen (1999), and the observational results of Baldwin and Dunkerton (1999), Baldwin et al. (1999), and Dunkerton et al. (1999) argue that the stratosphere is critical in changing the AO. Generally speaking, their results suggest that warming of the low-latitude lower-stratosphere leads to strengthening of the polar vortex and propagation of the response downward and poleward.

Assuming the Onami is related to the AO (first hypothesis), the simulations suggest that the Onami is the type of change we might expect under greenhouse warming. Even if we find Onami is not tied to the AO, it must surely be related to some larger scale atmospheric process that includes the basic AO characteristics of the last decade. In either event some of the changes in the ocean and on land are consistent with certain aspects of greenhouse warming simulations. For example the increase in precipitation, increase in river runoff, thinning of sea ice and the decrease in surface salinity of the western Arctic are trends roughly consistent with the simulations of greenhouse warming by Manabe et al. (1991, 1992).

Testing the hypothesis that Onami is related to climate change is critical to SEARCH. Testing this must rely heavily on modeling studies. As noted, existing studies suggest the AO is an important natural mode of variability and a component of modeled climate change scenarios. To verify this we must determine if Onami has happened before and provide a sound physical basis for how it may develop in the future. We will also need to explore the relationships between Onami and processes outside the Arctic.

This hypothesis leads to determination of whether Onami is simply driven by a large-scale atmospheric circulation perturbation, or includes feedbacks that are now or will be important in maintaining change. Some relevant feedbacks have received considerable attention already, but must be put in the context of Onami to gauge their importance. A leading example is ice-albedo feedback. This is the process whereby a reduction in snow cover and sea ice extent decreases albedo. This allows more radiation to be absorbed resulting in higher temperatures and promoting additional melt of snow cover and sea ice. It is being investigated extensively at local scales (scale of a typical climate simulation model) as part of the Surface Heat Balance of the Arctic (SHEBA) experiment (Perovich et al., 1999). The ice-albedo feedback has been investigated implicitly in general large-scale simulations (Covey, et al., 1991; Curry et al., 1995). Specific to SEARCH and ice-albedo feedback, Drobot and Andersen (1999) find that the onset of summer snow melt is earlier during periods of positive AO. Bamzai (1999) indicates the recent positive trend in wintertime AO index partially accounts for the recent negative trend in Northern Hemisphere snow cover. These results indicate AO is associated at least in part with decreased albedo.

Cloud-radiation feedback is another process which may enhance heating in the Arctic. There are a number of mechanisms. One is that increased heating at the surface causes more evaporation from open water. This enhances cloud cover, increasing the longwave radiation flux to the surface, and further increasing surface heating. Curry et al.(1996) have suggested that the cloud radiation feedback is positive for the Arctic in that clouds lead to surface warming, counter to the cooling effect of clouds globally. This process at the local scale has also been a focus of SHEBA. At basin and larger scales, it is reasonable to speculate that if the AO circulation brings more moisture into the Arctic, this could result in more cloudiness and higher surface temperatures.

Saenko and Holloway (1999) find that the response of a coupled sea/ice/snow model (based on the GFDL MOM model) to atmospheric forcing with extremes in AO produces enhanced freshwater storage during positive extremes of the AO. This is consistent with the ice/ocean modeling results of Polyakov et al. (1999). A more vigorous water cycle associated with Onami, resulting in enhanced precipitation and runoff, could be part of at least one feedback whereby the increased freshwater may increase st