U.S. Dept. of Commerce / NOAA / OAR / PMEL / Publications

Recent Temperature Changes in the Western Arctic during Spring

James E. Overland1, Muyin Wang2, and Nicholas A. Bond2

1Pacific Marine Environmental Laboratory, National Oceanic and Atmospheric Administration, Seattle, Washington, 98115
2Joint Institute for the Study of the Atmosphere and Ocean, University of Washington, Seattle, Washington, 98195

Journal of Climate, 15(13), 1702–1716, (2002).
Copyright ©2002 by American Meteorological Society. Further electronic distribution is not allowed.

5. Low-level heat budget for warm and cold years

In this section we investigate the processes important to the springtime heat budget for the lower troposphere (850 hPa) over northern Alaska and the southern Beaufort Sea. As noted in the discussion of Fig. 2, April had six warm years in the 1990s: 1990, 1991, 1993, 1995, 1997, and 1998. Heat budgets are evaluated for the four years (1990, 1993, 1995, and 1997) that had cold stratospheric temperature anomalies in March. For comparison we investigate four years from the 1980s with a relatively cold lower troposphere in April: 1980, 1982, 1984, and 1987. Of these years, 1980 and 1984 had warm anomalies in the Arctic stratosphere in March.

The heat budget equation in pressure coordinates can be written as

Equation 1

where T is temperature, V is the horizontal wind vector, omega is the vertical velocity in pressure coordinates, and CP is the specific heat capacity. The diabatic heating rate Q will be treated as residual and is primarily related to longwave radiative processes, but does include the effects of shortwave radiation and sensible and latent heating. The static stability parameter, Gamma, is defined as

Equation 2

where R is the gas constant, and P is pressure. The temperature tendencies are calculated from daily values from the NCEP–NCAR reanalysis averaged over the five closest grid points to Barrow, Alaska. We chose 850 hPa as a representative level for the lower troposphere. Results were qualitatively similar at other levels between 925 and 500 hPa. The results at 925 hPa are compromised, to a certain extent, because improper specification of the static stability at the 925-hPa level produces errors in the vertical advection term.

Because we are interested in how the cumulative effects of the processes impacting the heat budget contribute to the monthly temperature anomalies, we have chosen to present time integrals of the terms in (1). Thus, a particular warming or cooling event is manifested as a change in slope of the traces on the plots. Units are in K, which provide a quantitative comparison of the influence of each term on the temperature record.

The March–April time series of 850-hPa temperature, and the time integral of the horizontal advection, vertical advection, and residual terms of (1) during 1990, 1993, 1995, and 1997 for Barrow, are shown in Fig. 7. The time series of temperature show that the springtime warming can be more-or-less slow and steady for periods as long as a month, but is also punctuated by positive and negative temperature changes as large as 20 K over a few days. In large part, these fluctuations are attributable to day-to-day changes in horizontal heat advection. This term is more episodic than the other terms in most years, with 1993 representing an exception. For example, in 1990 there was one event, from 19 to 23 March, that produced roughly two-thirds of the total warming of 30 K due to horizontal advection. In 1995, there were two periods of enhanced warm advection, and two periods of substantial cold advection, each lasting roughly 4–7 days. Alternating cold and warm advection, with a near-neutral baseline, characterized the horizontal advection term in 1997. The magnitude of the horizontal advection term integrated over the 2-month period ranged from 30 K in 1990 and 1997 to 80 K in 1995. Strong warm advection events were often, but by no means universally, accompanied by upward motion and hence the compensating effects of the vertical advection term on the heat budget. Except for these occasional events, there was usually subsidence, and hence a rather steady contribution from the vertical advection term toward the springtime warming. The range in magnitude of this term over the 2-month period is from 40 K in 1990 to 80 K in 1993. The residual term in each year showed more-or-less steady cooling, with occasional brief periods of slight warming. The rate of the cooling due to the residual (and mostly diabatic) term varied from year to year, ranging from about –120 K in 1993 and 1995 to roughly –60 K in 1990 and 1997.

Figure 7

FIG. 7. The heating terms at 850 hPa based on Eq. (1). (top) The daily temperature from 1 Mar to 30 Apr. (bottom) The time accumulated horizontal heat flux divergence (solid line with *), the time accumulated vertical heat flux divergence (solid line with square), and the residual term (solid line with triangle) that include diabatic heating due to radiation, sensible, and latent heat. From left to right, the years are 1990, 1993, 1995, and 1997.

We now examine the heat budgets at 850 hPa for the cold years in the 1980s (Fig. 8). The horizontal advection term in the 1980s, as in the 1990s, included substantial week-to-week variability. But the cold years in the 1980s also include periods of longer and stronger cooling due to horizontal advection. While net warming was found for this term in the 1990s, the March–April time integrals in the 1980s ranged from a cooling of –80 K in 1980 to near-zero values in 1982 and 1984. The vertical advection term in the 1980s was similar in its character in the 1990s in that it contributed toward warming at a fairly steady pace. The magnitudes of this term were larger in the 1980s (ranging from 60 K in 1982 to 140 K in 1980), because the subsidence tended to be stronger in the 1980s than the 1990s. Finally, the residual term contributed toward roughly three-quarters of the cooling in the 1980s (–100 K in 1984 to –50 K in 1980 and 1987) as it did in the 1990s. We note little temperature change from March to April in these years.

Figure 8

FIG. 8. As in Fig. 7, but for the "cold" years in the 1980s. From left to right, the years are 1980, 1982, 1984, and 1987.

The heat budget results for the Barrow region for the selected years in the 1980s and 1990s are summarized in Table 1. The amount of warming over the March– April period in the selected years of the 1990s was double that during the selected years in the 1980s. But this warming represents a relatively small difference between large terms. We are more struck by differences in the nature of the heat balance between the contributions in the 1980s versus the 1990s. In particular, the horizontal advection term acted to cool in the 1980s and to warm in the 1990s; this change was largely compensated for by changes in the vertical advection term and to a lesser extent, the residual term. These results can be compared with those from the heat budget carried out by Tanaka and Milkovich (1990) for northern Alaska in the winter of 1988/89. They also found that prominent subseasonal changes in low-level temperature were associated with variations in horizontal advection, but that this contribution to the heat budget was largely compensated for by variations in radiative cooling, rather than due to variations in vertical motion.

Table 1

TABLE 1. Means and standard errors of the means for the heat budgets at 850 hPa for four notable years in the 1980s and 1990s.

The fundamental difference between cold and warm years relates to the sense, number, and intensity of episodic horizontal temperature advection events. The preponderance of cold advection events during the selected years in the 1980s versus warm advection events in the 1990s in the vicinity of Barrow can be attributed to differences in the mean circulation during the two periods. As shown by the 925-hPa geopotential height anomaly maps for the two groups of years during March–April (Fig. 9), the 1980s included anomalously high pressure centered over the Barents Sea extending across the eastern Arctic, and hence also anomalous northeasterly low-level flow, while the 1990s included relatively low pressures over the central Arctic and high pressures south of Bering Strait and hence a greater tendency for southwesterly low-level flow. From our inspection of individual horizontal advection events, it appears that strong warm advection in the spring near Barrow is most frequently associated with westerly low-level winds, while strong cold advection is typically accompanied by northerly winds.

Figure 9

FIG. 9. Geopotential height anomalies at 925 hPa averaged over Mar and Apr for the four cold years in the 1980s: (a) 1980, 1982, 1984, and 1987; and the four warm years in the 1990s: (b) 1990, 1993, 1995, and 1997. The data are from the NCEP-NCAR reanalysis. The "B" indicates the location of Barrow, Alaska.

These overall changes in the low-level circulation in the western Arctic from the 1980s to the 1990s can be considered a near-surface signature of the AO. Previous work has revealed that geopotential height anomalies with the AO are closely coupled in an equivalent barotropic sense between the lower troposphere and stratosphere (Thompson and Wallace 1998, Black 2002). To extend these studies, we examine the degree of this coupling in spring on subseasonal timescales and for the subregion represented by the western Arctic. As a measure of this coupling, we evaluated spatial correlations of monthly 200- and 850-hPa height anomalies in the sector north of 70°N and between 140°E and 90°W during the 16 selected months for which heat budgets were estimated. The results are strong positive correlations for 14 of the 16 months, with values typically greater than 0.90. This leads to our conceptual model for the tropospheric temperature changes in the western Arctic from the 1980s to the 1990s. These changes occurred because the polar vortex persisted into March in the 1990s, with coincident low-level circulation anomalies. These low-level flow anomalies brought about more episodes of warming due to horizontal advection; the accumulated contribution of these warming events tended to cause lower-tropospheric temperature anomalies to peak in April. It bears noting that this effect of the AO on the climate of the western Arctic exists because the Beaufort Sea represents the location of a node in the EOF pattern for sea level pressure (SLP; Thompson and Wallace 1998); this node represents a region of variability in the SLP gradient, and hence lower-tropospheric winds. It is therefore no surprise that the AO substantially influences the climate of the western Arctic.

Return to previous section or go to next section

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage