1. Introduction
Quite recently, several studies have reported recent abnormal climate changes in the northern extratropics. For example, Walsh et al. (1996) found a noteworthy strong polar vortex in the Arctic persistent since winter 1988/89. Perhaps associated with this change in the polar atmospheric circulation, reduction in summer sea ice in the offshore region of Siberia is observed (Maslanik et al. 1996). The decreasing of sea ice is also reported by Tachibana et al. (1996) in the southern Okhotsk Sea after winter 1989.
Watanabe and Nitta (1998, hereafter referred to as WN98) analyzed atmospheric circulation, sea surface temperatures (SSTs), and snow cover together, focusing on the decadal-scale changes that occurred in winter 1989. They have shown that the atmosphere experienced an abrupt shift in winter 1989 having characteristics as follows: dipole height anomalies between midlatitudes and the polar region with an equivalent barotropy; dominant appearance of the North Atlantic oscillation (NAO), Pacific–North American (PNA), and Eurasian (EU) teleconnection; weakening of the subtropical jet streams over the Pacific and Atlantic Oceans; and strong warming over Eurasia. The latter two features are consistent with the result by Ting et al. (1996), who focused on extreme phases of winter zonal flow. This mode of variability was previously detected by Kitoh (1991) in a 20-yr general circulation model (GCM) integration with climatological SSTs as well as with observed SSTs, although the temporal fluctuations for the two cases are not consistent with observational results by Ting et al. (1996) and WN98. He referred to the mode as an inherent variability in the atmosphere, which may be modulated by tropical and/or extratropical SST anomalies. On the other hand, results from the statistical analysis of WN98 suggest that the atmospheric shifts, including changes in surface air temperature, are coupled with large snow extent (denoted as SE) anomalies over eastern Eurasia in the preceding autumn. Although snow is an internal variable of the climate system, it was suggested that autumn SE anomalies force the atmosphere to change in the following winter. Hence, an examination of the physical impact of autumn snow anomalies on the atmosphere is attempted in this study by using an atmospheric GCM (AGCM).
In contrast to a number of diagnostic studies, such as Hahn and Shukla (1976), which focused on the relationship between Eurasian snow cover in spring and the succeeding summer monsoon, few observational studies were performed to investigate the role of an autumn Eurasian snow anomaly in the winter atmospheric circulation. At most, Foster et al. (1983) found Eurasian snow cover has a larger influence on in situ surface air temperatures in winter than the snow cover over North America. Similarly, sensitivity experiments with an AGCM were performed by several researchers (e.g., Barnett et al. 1989; Yasunari et al. 1991; Douville and Royer 1996) to investigate the responses of summer monsoons to anomalous Eurasian snow in spring. However, no GCM experiments have been performed to examine the sensitivity of winter extratropical atmospheric circulation to Eurasian snow anomalies in autumn. Recently, Walland and Simmonds (1997) performed a numerical experiment in perpetual January conditions to examine the impact of winter SE anomalies on the atmosphere. They have shown that snow anomalies in winter have a significant influence on the hemispheric atmosphere because of albedo feedback. It is expected that snow anomalies in autumn, when more insolation reaches the land surface than in winter, have an as strong as, or even stronger, impact on the atmosphere than winter snow anomalies.
SSTs are potentially a dominant cause of changes not only in the atmosphere but also in Eurasian snow due to their large thermal inertia. Therefore, we examine the relative effect of snow anomalies on the atmosphere to that of SST anomalies by taking the shift in winter 1989, for example, of an extreme phase in the atmosphere. In other words, this study is a sensitivity experiment, and does not directly deal with the decadal-scale persistence of atmospheric anomalies. However, the experiment may be a logical first step to investigate the decadal shifts.
This article is organized as follows. Observational data and observed anomalies in snow and SSTs, used for the validation and boundary input for the numerical experiments, are documented in section 2. In section 3, the model used and the design of numerical experiments are described. Section 4 presents results of the experiments, devoting attention to the responses of the surface climate and the atmospheric circulation, and the dynamics. Results are discussed in section 5, and summary and conclusions are given in section 6.
2. Observational background
a. Data
Northern Hemisphere snow cover data for 1972–91 are derived from the National Oceanic and Atmospheric Administration (NOAA) satellites and compiled at the National Environmental Satellite, Data and Information Service. The data have 89 × 89 grids equally spaced on a polar stereographic projection and give the information of presence (1) or absence (0) of snow weekly.
Global SST data used here consist of monthly mean SSTs on a 2° × 2° grid from January 1946 to February 1995. The data are compiled by the Japan Meteorological Agency (JMA).
The snow cover and SST data described above are the same as those analyzed in WN98 and are used here to show anomalies around winter 1989. The anomaly is calculated by subtracting the mean value averaged during the entire available period, and winter is defined as the average from December to February (DJF; autumn is also the 3-month average for September–November). In addition to this data, the following observations are used to validate the model climatology and anomaly fields.
Observed atmospheric states are shown by two data sources: monthly sea level pressure (SLP) compiled by the JMA and the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis dataset. The former is available on a 5° × 5° grid of the Northern Hemisphere from January 1951 to February 1995, whereas the latter was obtained from a special CD-ROM containing monthly mean variables in a global 2.5° × 2.5° grid on 17 standard pressure levels during 1982–94. Further details of the NCEP–NCAR reanalysis data are given by Kalnay et al. (1996).
We use monthly snow-depth climatology data compiled by the U.S. Air Force Environmental Technical Applications Center, which assembled all the available station data. The data cover the global domain with the spatial resolution of 1° × 1°. Although the data are quality controlled, the confidence of the value varies by the location (Foster and Davy 1988).
Surface albedo is obtained from the first World Climate Research Programme Surface radiation budget (SRB) data derived from the International Satellite Cloud Climatology Project (ISCCP) C1 data (Pinker and Laszlo 1992). The albedo is determined from the ratio of upward and downward shortwave radiations, so the value is undefined in higher latitudes where enough solar irradiance is not available. The data having a horizontal resolution of 2° lat × 2.5° long are available from March 1985 to December 1988.
b. Observed snow and SST anomalies in 1988/89
Autumn-mean snow cover anomalies in 1988 are displayed in Fig. 1a. As noted previously, a large decrease of snow cover is found over the eastern half of Eurasia at 40°–75°N, 70°–150°E. Snow cover reveals increasing anomalies over eastern Europe and central-eastern Canada. The SE anomalies over the Northern Hemisphere are equal to −291.15 × 104 km2 in Fig. 1a. It is evident that most of these large SE anomalies are attributed to the decrease of snow over the eastern Eurasian continent.
Figure 1b shows global SST anomalies observed in winter 1989. Positive anomalies of about 1 K are found in the midlatitude Pacific and Atlantic Oceans. It is also known that strong La Niña has occurred during this period, so that large negative anomalies exist in the tropical central-eastern Pacific.
3. Numerical experiments
a. Model
The AGCM used in this study was jointly developed by the Center for Climate System Research (CCSR), University of Tokyo, and the National Institute for Environmental Studies (NIES), which is referred to as the CCSR–NIES AGCM. The CCSR–NIES AGCM is a global spectral model based on the primitive equations, and a finite-difference scheme is applied to the 20 vertical levels defined on σ surfaces. The model fields represented by spherical harmonics have a horizontal resolution of T21 (triangular truncation with wavenumber 21). In addition to the usual atmospheric variables, the mixing ratio of cloud liquid water and land surface variables such as soil moisture, soil temperature, and snow amount are also predicted in the model. The parameterization of cumulus convection employs a simplified Arakawa–Schubert scheme. The radiative transfer scheme was developed by Nakajima and Tanaka (1986) with the two-stream DOM/adding method. The land surface processes use a simple bucket model with a field capacity of 15 g cm−2. Further explanation of the model is given by Numaguti et al. (1998).
Morinaga and Igarashi (1991) showed from observational data that Eq. (3) is suitable as a first-order approximation. So the fractional snow cover at each grid box reaches 1 when the snow mass exceeds Wc, and is represented as
b. Validation
Before the description of numerical experiments, the model climatologies for the atmosphere and land surface conditions are compared with observations. The observational data and manner to obtain the climatological state in the model are, respectively, explained in section 2a and the next subsection.
Shown in Figs. 2a and 2b are winter (DJF) climatologies of 500-hPa height and 200-hPa zonal wind in observations and the model. The AGCM appears to reproduce the observed winter climatological fields in simulating the positions of the Pacific and Atlantic westerly jets associated with troughs/ridges. The magnitude of observed and modeled westerlies are similar to each other, and this suggests that the eddy feedback is acting in the model although the resolution of T21 is rather coarse, as presented in detail in section 4c. However, the difference map between two fields (Fig. 2c) shows that the model has serious deficiencies over the North Atlantic and Europe. The 500-hPa heights in the model are higher than observations over Greenland, whereas they are lower over the elongate regions of the midlatitude Atlantic and Europe. The maximum difference reaches above 150 m and these height differences result in the stronger westerlies over the Atlantic in the model. Smaller errors are found over the East Asian and Pacific regions but the sense is similar to errors over the Atlantic. Although the systematic biases in the model may affect the result of experiments, they do not alter the interpretation.
Mean states for the snow depth in February and surface albedo in October are compared between observations and the AGCM (Fig. 3). The northern snow accumulation is in general the maximum in February, whereas the albedo maps are displayed for October because observations north of 60° are not available after November. Although the resolution of T21 cannot resolve the observed detailed features, the snow depth in excess of 25 cm in the northern latitudes north of 60° in the model agrees with the observations (Figs. 3a and 3b). The difference is found with respect to regions of maximum snow depth in Eurasia; near 60°N, 90°E in observations and 60°N, 30°E in the model. The simple parameterization of the surface albedo in the model leads to somewhat different values from observations (Figs. 3c and 3d). Observed albedo is roughly 0.2 in regions where snow covers the ground, whereas the model albedo is slightly larger (about 0.25). In association with the differences in surface albedo, surface air temperatures in the model are cooler in Eurasia: winter surface temperatures in a region 35°–75°N, 70°–150°E are −16.06° and −20.69°C in the observations and model, respectively.
c. Design of the experiment
Six experiments are performed in this study. All the integrations of the model are started from 1 September and continued for half a year, since the study focuses on changes in the atmospheric circulation in winter 1989, but not decadal-scale persistence of anomalies. The initial conditions for all the atmospheric variables are adopted from the climatological (averaged for 1988–95) Global Analysis (GANAL) dataset produced by the JMA with the four-dimensional data assimilation. For each experiment, five integrations are started with slightly different initial conditions based on the lagged average forecast (LAF) method to assess the statistical significance of the atmospheric response. The LAF method employs the mean state from the GANAL during 1–5 September for the initial inputs for five members of integration. Observed climatological values are given for physical parameters such as the snow mass and soil wetness at the initial time.
To obtain the climatological atmosphere in the model, the AGCM is integrated as a control run (hereafter referred to as the CNTL run) with prescribed climatological values of SSTs obtained from the Atmospheric Model Intercomparison Project (AMIP) for the period 1979–88 (Gates 1992). Figures 2b, 3b, and 3d are derived from the ensemble means of the CNTL run. On the other hand, an experiment that imitates the deficient snow cover over eastern Eurasia in autumn 1988 (Fig. 1a) is referred to as the light snow run (the LSNW run) here. Unlike the snow experiments in spring described in the introduction, when the atmospheric response to anomalous snow in autumn is examined, introducing anomalous snow masses in the initial condition is not appropriate. Rather, the forcing by snow anomalies should be given during the accumulation of snow in autumn. Therefore, the following method is chosen for the experimental design.
Here δ is now set to 0.25 for the first three months (September–November) of the integration over the area from 35°N to 75°N, 70°E to 150°E as schematically presented in Fig. 4. This rectangular region roughly corresponds to areas having large negative anomalies of snow cover as shown in Fig. 1a. In a sensitivity experiment such that the perturbation given is defined by actual observations, the appropriateness of the experimental design, including the value of δ should be discussed enough.
Figure 5 illustrates the temporal transition of the snow depth averaged with area weighting over 35°–75°N, 70°–150°E in the CNTL run and LSNW run, respectively. The range of the natural snow variability is equivalent to one standard deviation of the snow depth obtained from the AMIP run by the CCSR–NIES AGCM (Shen et al. 1998). As expected, the snow depth in the LSNW run is nearly a quarter of that in the CNTL run during autumn. In winter when the boundary forcing is no longer given in the LSNW run, the snow depth in the LSNW run grows in parallel with that in the CNTL run, but having a depth around 10 cm less. It should be noted that the model snow excessively accumulates in winter compared to observations, although the observed and modeled snow depths are similar with each other in autumn. Table 1 shows a comparison of the SE anomalies in autumn between the observed values in 1988 and those in the LSNW run. The area for calculating the SE is the same as in Fig. 5. It is found that the SE anomalies averaged during autumn are similar between the observations and the model; however, the actual extent in the model accounts for 3.107 × 106 km2, which is roughly half of the observed SE (6.361 × 106 km2). This difference arises from the simple representation of the snow fraction, because such a large difference is not indicated in snow depth. Thus the simplified scheme for the snow accumulation in the AGCM leads to several discrepancies with observed snow depth and extent. Similarly, the forcing of δ = 0.25 appears to be somewhat large. However, Table 1 shows that the consequent surface albedo anomaly relative to the mean value in the model is rather small compared to the observed counterpart in autumn 1988. These results reveal that the design of the LSNW run is not absurd, although the anomaly may be slightly larger than observed situations in 1988.
Another experiment that evaluates the impact of SST anomalies observed in 1988/89 on the atmospheric circulation is performed as in the CNTL run, except that climatological SSTs are replaced with observed values in 1988/89 (Fig. 1b). This run, denoted as the SST run, employs observed SSTs over the entire globe (80°S–80°N). Three additional experiments, respectively, referred to as the LSgl, LStr, and LSex runs, are conducted with design similar to the LSNW run, but the climatological SSTs are replaced by observations for the SST run. It should be noted that the LSgl run is a fully combined experiment of the LSNW and SST runs, whereas the LStr run uses observed SSTs only in the tropical domain between 20°S and 20°N, and the LSex run adopts observed SSTs only in the northern extratropics north of 20°N. We intend that these three experiments may highlight two arguments: the relative importance of tropical and extratropical SST anomalies, and the combined effect of snow and SST anomalies.
In the following sections, results of those experiments are primarily represented by anomalies, namely, differences of ensemble averages of the five realizations of each experiment with those of the CNTL run. The significance of all the results are represented by using the t test as in Ferranti et al. (1994).
4. Results
a. Changes in surface climate
In AGCM experiments that gave anomalous snow in spring, two feedbacks have been taken into account (Yasunari et al. 1991): albedo and snow-hydrological feedbacks. It is speculated, however, that the former effect is predominant in the LSNW run, because less snowmelt will occur in the cold season. As described previously, surface albedos both in observations in autumn 1988 and the LSNW run show negative anomalies over the eastern Eurasian continent. Figure 6 displays observed and modeled albedo anomalies in November. It is evident that the observed anomalous albedos are dominant over eastern Eurasia (Fig. 6a) and they are consistent with a large decrease in SE over the region (Fig. 1a). The observed albedo anomaly averaged over 35°–75°N, 70°–150°E is −0.034, which is the largest in the period for which the SRB are available. The albedo anomalies over eastern Eurasia in the LSNW run directly reflect the artificial forcing to the snow accumulation, therefore, the significance of anomalies is meaningless.
Anomalous surface air temperatures obtained from the observations in winter 1989 and the model (anomalies in the LSNW run and SST run) are compared in Fig. 7. In the observed surface temperatures, the largest positive anomalies above 4 K are found over the eastern part of the Eurasian continent, which corresponds to areas having greatly decreased snow cover and surface albedo in the preceding autumn (Figs. 1a and 6a). Other warming areas are located over northwestern Canada and Europe, and cooling occurred over western Canada and Greenland. Notice that the pattern found in Fig. 7a is in good correspondence with the decadal-scale differences in surface air temperature (cf. Fig. 8 of WN98). Simulated anomalies in the LSNW run (Fig. 7b) reproduce the significant, strong warming over the Eurasian continent and Europe. Anomalies over North America are less resemblant to observed features—for example, over northwestern Canada. Temperature anomalies in the SST run are similar to those in the LSNW run, however, they are moderate and less significant over eastern Eurasia (Fig. 7c).
It is expected that the strong warming over Eurasia in the LSNW run is generated from the given snow anomalies in autumn (Fig. 8a). On the other hand, as mentioned in the introduction, SST anomalies may affect either surface air temperatures or snow amounts. Temperature anomalies in the SST run are apparently associated with snow anomalies in that run; however, no significant snow anomalies are found over the entire Northern Hemisphere in the SST run (Fig. 8b). Seasonal cycle of the snow depth in the SST run superimposed on Fig. 5 shows little change with the climatologies. These results are unexpected and suggest that, at least in the model, the Eurasian snow acts independently of SST anomalies. Actually, when the AMIP run is performed using the climatological SSTs, natural variability of the Eurasian snow has a magnitude nearly identical to that shown in Fig. 5. One possible explanation for those results is that the change in Eurasian snow is attributed to the internal processes of the atmosphere.
The surface warming over the Eurasian continent in the LSNW run (Fig. 7b) is partly explained by changes in local energy balance. In association with the reduced albedo, net shortwave radiation at the surface over eastern Eurasia reveals the anomaly of about −5 W m−2 during November and December. This anomaly account for 6.5% of the mean shortwave flux during the cold six months, although the anomalies of net shortwave fluxes are significant only in late November. The consequence of altered net radiation is a diabatic heating to the lower atmosphere. On the other hand, large-scale thermal advection also plays a role for generating the temperature anomalies. Winter SLP anomalies in the LSNW run and SST run are shown in Fig. 9 together with observed anomalies in winter 1989. The LSNW run produces significant low pressure anomalies in polar regions, although the magnitude is roughly a half of that observed. SLP anomalies in the midlatitudes are more tenuous and the anomalous Aleutian low presented in observations (Fig. 9a) is less clear in the LSNW run (Fig. 9b). Over the Eurasian continent, the anomalous northwesterly (southwesterly) wind concomitant with the SLP anomalies are found over the cooling (warming) region in Fig. 7b. This coincidence suggests that the large-scale advection influences temperature anomalies over Eurasia. In the SST run, winter SLP anomalies are similar to those in the LSNW run in higher latitudes, whereas they are larger and more significant in midlatitudes (Fig. 9c). In particular, conspicuous changes in the Aleutian low, which have a magnitude similar to observations, are found probably in response to the SST anomalies in the equatorial Pacific. Since surface energy budgets as well as the snow mass show little changes in the SST run, temperature anomalies over Eurasia appear to be dominated by the thermal advection.
b. Atmospheric responses
At the 500-hPa surface, observed anomalies during winter 1989 (Fig. 10a) have characteristics similar to the epoch differences on a decadal scale (cf. Fig. 1a of WN98): the dipole pattern between the midlatitudes and polar regions, a weakened Aleutian low, and the NAO and PNA–EU teleconnections. The zonally elongate features accompany large anomalies in zonal-mean zonal wind, and they are known as a kind of typical extreme phase in the extratropical atmospheric circulation (Ting et al. 1996).
The LSNW run generally succeeds in generating consistent features of the 500-hPa heights (as shown in Fig. 10b), such as dipole anomalies. However, there are several disagreements with the observed height anomalies that cannot be overlooked. In particular, the amplitude of the height response is somewhat weaker than observed anomalies. The positive anomalies over the northeast Pacific shown in the observed field (Fig. 10a) are not well simulated in the model response (Fig. 10b). In addition, modeled height anomalies over the North Atlantic are slightly distorted and shifted eastward. The discrepancy over the Atlantic sector may be due to difference mean state in observations and the model (Figs. 2a and 2b). This possibility is examined in the dynamical analysis in section 4c. It is noteworthy that Fig. 10b shows large areas having significant anomalies at the 5% level in the Northern Hemisphere. In addition to the spatial resemblance, the height response in the LSNW run represents a strong equivalent barotropic structure (figures are not shown), which is consistent with the observed atmospheric changes in 1989. These results support the hypothesis presented by WN98: the importance of the large decrease in the Eurasian snow found in autumn 1988 as an amplifier for the atmospheric changes in the subsequent winter. Since this study focuses on linkages during the extreme phase in a single year, the results of these experiments do not give any insight into the mechanism of continuous anomalies in the atmosphere after 1989.
The differences between the height anomalies in observations and the LSNW run over the North Pacific are probably due to the effect of equatorial SST anomalies, because remarkable La Niña conditions that may weaken the Aleutian low were observed in 1988/89. The anomalous response of 500-hPa heights in the SST run shown in Fig. 10c reveals positive height anomalies exceed 60 m over the North Pacific, and amplified negative anomalies north of them. These anomalies presumably result from large negative SST anomalies in the tropical Pacific (Fig. 1b), which correspond to a heat sink. It should be noted that (as is seen in Fig. 10c) height anomalies in the SST run resemble those in the LSNW run (Fig. 10b), but the former have a larger amplitude than the latter. The spatial correlation coefficient between Figs. 10b and 10c is +0.887, and the spatial standard deviation of height anomalies in the SST run is roughly 40% larger than that in the LSNW run. Over the Atlantic, height anomalies shown in Fig. 9 are 10–20 m larger than those in the LSNW run, although the pattern shows the eastward shift of anomalies as in the LSNW run compared to observations. These results suggest that the influence of tropical and extratropical SST anomalies during autumn 1988 and winter 1989 have an effect greater than that of the Eurasian snow anomaly on the atmospheric changes in winter 1989.
Since little changes in the Eurasian snow are found in the SST run as noted earlier, the height response to the combined forcing of LSNW plus SST (LSgl run) is examined with respect to a linked effect of snow and SST anomalies. Furthermore, responses to the snow anomalies incorporated with SST anomalies in the Tropics (LStr run) and extratropics (LSex run) are compared. First, in order to examine the amplification of these height anomalies, results of three additional experiments are summarized in Table 2 by a comparison of similarity and amplitude of their atmospheric responses with those in the LSNW and SST runs over the entire Northern Hemisphere. As expected from the similarity of anomalies in the LSNW and SST runs, anomaly patterns in 500-hPa heights in all three experiments show strong positive correlations with those in the LSNW run and SST run. The ratios of the spatial standard deviation (SD) of the height anomalies in the LSgl and LStr runs indicate that the amplitude of response in these runs is about 40% larger than that in the LSNW run, and nearly identical to that in the SST run. These ratios manifest that the height response in the SST run is mainly dominated by tropical SST anomalies. The SD ratio for the LSex run is nearly 1.0 to the LSNW run and roughly 0.7 to the SST run. Figure 11 shows anomaly patterns in 500-hPa height fields as in Figs. 10b and 10c, except for the LSgl, LStr, and LSex experiments. A common structure as in the LSNW and SST runs is found in all the patterns: the north–south seesaw between polar regions and midlatitudes. The height anomalies in the full-combined LSgl experiment quite resemble the observed anomalies (Fig. 10a) and reveal the significant amplification over the North Atlantic relative to the LSNW and SST runs. When the negative SST anomalies over the tropical Pacific are given as boundary conditions (i.e., LStr run), positive 500-hPa height responses over the North Pacific and negative anomalies north of them are apparently strengthened in comparison with the LSNW run (Fig. 10b). As in the SST run, negative SST anomalies in the tropical Pacific appear to produce those large anomalies, which are reminiscent of the reversed western Pacific pattern (Wallace and Gutzler 1981). A relative increase of the hemispheric amplitude of the height response in the LStr run compared to the LSNW run (as found in Table 2) is mainly caused by these changes. It is interesting that the anomalies over the North Atlantic are significantly suppressed compared to those in the LSNW run in addition to the SST run. In contrast, 500-hPa height anomalies in the LSex run (Fig. 11c) represent a different modification than those in the LSNW run. The positive height anomalies over the central North Pacific and negative anomalies north of them are considerably weak in comparison with those in the LStr run. On the other hand, the positive height response over the North Atlantic between 30° and 60°N is slightly intensified. In Fig. 11c, significant dipole anomalies are also found over the eastern North Pacific around 130°W, but these anomalies are inconspicuous in the SST run. The result that the amplitude of height response in the LSex run is nearly identical to that in the LSNW run (Table 2) implies that the impact of midlatitude SST anomalies is tenuous in the model.
It should be noted that the amplitude of anomalies in the LSgl run is not a linear sum of each response to snow and SST forcings, rather it is smaller. This suggests that the atmospheric response in the LSgl run is produced and/or maintained not only directly by the combined forcing of the Eurasian snow and SST anomalies but also by nonlinearities in the atmosphere. As in the dynamical analysis of the height response presented in the following subsection, the amplitude of response to the boundary forcing in the model atmosphere may be partly determined by nonlinearities within the atmosphere.
c. Dynamical analysis of the height responses
The height anomalies produced by the LSNW run shown in Fig. 10b are regarded as a steady response of the atmosphere to the Eurasian snow anomaly. However, in order to understand how the Eurasian snow anomaly generates the atmospheric response not only over the Eurasian continent but also over the opposite hemisphere, that is, the North Atlantic and Europe, further analysis of the dynamics is necessary. The analysis should include evaluation of the net forcing of the time-mean field by transient eddies. Thus, the following analysis technique is used to diagnose the process of the response, such as the contribution of changes in the eddy activities, and the formation of the time-mean field.
The terms on the lhs of Eq. (9) are interpreted as the net forcing by stationary nonlinearities (SN), high- and low-frequency eddies (HF and LF), interaction within eddies (EI), anomaly interactions with climatological stationary waves (SW), interactions between anomalies and climatological zonal symmetries (ZA), and all the remaining components such as diabatic effects (R2). The derivation of Eq. (9), mathematical representation for each forcing, and numerical procedure to obtain the anomalous tendencies in Eq. (9) are described in the appendix. The following results are represented by the corresponding height tendencies.
Before evaluating the anomalous geopotential height tendencies, (∂Ẑ/∂t), the geopotential height tendencies due to HF obtained from the CNTL run are computed to compare the eddy forcings in the AGCM with those in the observational results (as shown in Fig. 12). The result shows that features of the climatological height tendencies in the model are, in general, consistent with those observed (cf. Fig. 3 of LH84). However, model-derived tendencies associated with eddy heat fluxes (∂Z/∂t)H are somewhat magnified relative to observations in the lower troposphere. This discrepancy may be attributed to (∂Z/∂t)H being sensitive to small-scale noise due to the interpolation to pressure surfaces.
It turns out that the anomalous geopotential height tendencies, (∂Ẑ/∂t), have more complicated spatial distributions than the climatological tendencies. Therefore, the results are summarized with respect to their spatial standard deviations, which represent the amplitude of tendencies, and spatial correlation coefficients of (∂Ẑ/∂t) with Ẑ at the same level, which is a measure of whether the corresponding forcings act to reinforce the anomalous time-averaged height field Ẑ. It is noted, for example, that Ẑ at 500 hPa is equivalent to Fig. 10b. The anomalous height tendencies are calculated not only for the LSNW run but also for the SST run to examine whether height responses can be explained in a common dynamics.
Figure 13 displays the summary of the tendencies due to each total forcing (i.e., vorticity plus heat fluxes) at 300 and 900 hPa for the LSNW and SST runs. When height tendencies are calculated in the conventional way as in Eq. (5), those associated with vorticity forcings prevail in the upper troposphere, whereas tendencies due to thermal forcings predominate in the lower troposphere. A similar relationship is found in tendencies for the anomalous time-averaged fields. Hence, total anomalous tendencies, (∂Ẑ/∂t)V+H, in the upper (lower) troposphere appear to reveal characteristics similar to the tendencies associated with net vorticity (thermal) forcings. In the upper troposphere, results obtained from two experiments are similar to each other (Fig. 13a). It is found that tendencies associated with a forcing by interactions between height anomalies and climatological stationary waves,
On the other hand, tendencies in the lower troposphere at 900 hPa (Fig. 13b) reveal somewhat different dynamical balances from the upper troposphere. In the LSNW run, forcings that help height anomalies grow are three components of LF, interactions of anomalies with zonal-mean states (ZA), and residual (R2) terms, whereas the net LF forcing tends to weaken the Ẑ field in the SST run. Anomalous tendencies due to stationary nonlinearities (SN) and eddy interactions (EI) show negative and positive correlations with Ẑ in both runs, although the latter amplitudes are negligibly small. Positive correlations between
Net changes in high-frequency eddies, which act to amplify height anomalies in the upper troposphere (Fig. 13a), are closely related to changes in storm tracks. Anomalies in the root-mean-squares of transient eddies in geopotential height at 300 hPa evidently indicate a significant weakening (∼8 m) or northward displacement of the storm tracks over the North Pacific and North Atlantic (as shown in Figs. 14a and 14c). These suppressed or displaced storm tracks can cause weaker-than-normal vorticity flux convergence due to transient eddies north of the strong westerlies, and lead to relatively weak negative height tendencies around 40°N, where negative tendencies exist in the climatology (Fig. 12).
The tendency due to anomaly interactions with climatological zonal asymmetries (SW), which is also important to the growth of anomalies, is interpreted by drawing contours of climatological stationary waves (denoted as
These results of the internal dynamics in the upper troposphere are reminiscent of results of the linear budget analysis by Branstator (1992), who investigated the relative importance of internal processes in maintaining low-frequency variations. He obtained the conclusion that the interactions of low-frequency anomalies themselves with climatological stationary waves are crucial, whereas transient eddies within the low-frequency anomalies are of secondary importance in the maintenance of low-frequency patterns. The former corresponds to SW, and the latter should be replaced by HF here. This consistency warrants attention. When the finding by Branstator (1992), that anomaly interaction (SW) is the major source of low-frequency fluctuations, is taken into consideration in Fig. 13, the following interpretation is possible: if Ẑ is further amplified, low-frequency anomalies gain energy and act to reduce the zonal asymmetries and to suppress the amplified time-mean anomalies. If these processes are found in the real atmosphere, the magnitude of extreme height anomalies such as in winter 1989 is partly self-determined in the upper troposphere.
5. Discussion
Model results presented in the previous section suggest the importance of the Eurasian snow anomaly in autumn 1988 as well as the global SST anomalies in the atmospheric changes in the following winter. The numerical experiments reinforce the diagnostic inference in WN98; however, we need to take two matters into consideration.
First, Randall et al. (1994), who examined snow feedbacks in 14 GCMs as part of AMIP, reported that the impacts of snow anomalies on the atmosphere largely depend on the GCMs used, varying from strong positive feedback to even weak negative feedback. The variable strength of snow feedback is attributed not only to different snow responses to the boundary forcing (i.e., SSTs) but also to the complexity in indirect effects, such as changes in water vapor, surface radiation, and cloudiness. Thus, there may remain some ambiguities in the results of the LSNW run, although the results of the LSNW run appear to show positive snow feedback in the CCSR–NIES AGCM.
Second, a more fundamental matter is the necessity to understand what contributed to the changes in Eurasian snow in autumn. It may be possible that snow anomalies and/or ground wetness in the eastern Eurasian continent are associated with the tropical SSTs (Kitoh 1996). We calculated temporal correlations between the Eurasian SE in autumn and global SSTs in the preceding summer using the observations (not shown). Large negative correlations significant at the 1% level are evident in two regions: the western Pacific and tropical Atlantic. Kitoh (1991) has indicated that SST anomalies in the western tropical Pacific largely affect the location and/or strength of the subtropical jet over East Asia, which can alter the transient eddy activity over the region. On the other hand, a thermal forcing in the equatorial Atlantic effectively influences the atmospheric state over the North Atlantic (Simmons et al. 1983), where disturbances which cause snowfall in downstream regions (i.e., Eurasia) can be produced. To confirm this possibility, it will be necessary to investigate changes in observed eddy activities, and the water vapor transport over Eurasia and associated teleconnections over the Atlantic, by using a reliable daily dataset.
Dynamical analyses for the LSNW run and SST run suggest that dipole height anomalies as in Figs. 10b and 10c easily grow in the atmosphere although the anomalies are forced responses. This inference is in agreement with results by Kharin (1995), who examined a dominant mode of variability in 19-yr integrations of an AGCM with observed SST anomalies. He pointed out that the dominant canonical height pattern, which resembles the dipole mentioned in this study, is similar to a normal mode of a linear barotropic system. As shown by Branstator (1985) in his eigenanalysis of a linear system, on the other hand, the steady solution is represented by the projection of forcing function onto normal mode vectors. If the responses in the LSNW run and SST run exactly coincide with a normal mode, the amplitudes of height anomalies in the experiment of the combined forcing (LSgl run) ought to be equal to the sum of each response because of the orthogonality of normal mode vectors. Thus, a smaller amplitude of height anomalies in the LSgl run than the sum of two responses (LSNW plus SST) indicates that the forcing vectors by snow and SST anomalies are toward different directions in a phase space, although their projections onto normal modes are similar to each other.
The issue of the role of extratropical SST anomalies is quite ambiguous, although a number of model studies have been carried out (Palmer and Sun 1985; Pitcher et al. 1988; Lau and Nath 1990; Ferranti et al. 1994; Graham et al. 1994). First, Palmer and Sun (1985) found in their GCM that midlatitude SST anomalies do influence the atmospheric circulation. They proposed a hypothesis that SST anomalies in the western portion of the extratropical oceans cause not only local diabatic heating anomalies but also changes in the momentum forcing by baroclinic eddies, which can drive the atmospheric response over the downstream of SST anomalies. Their results are verified by several studies (Pitcher et al. 1988; Ferranti et al. 1994); however, some studies failed to reproduce the impact of extratropical SST anomalies on the atmosphere (Lau and Nath 1990; Kushnir and Held 1996). Ferranti et al. (1994), based on Palmer and Sun’s hypothesis, speculated that experiments having SST anomalies in the central or eastern part of extratropical oceans or having a low horizontal resolution in the GCM will underestimate eddy momentum fluxes and are not able to reproduce the significant atmospheric response. Thus the low resolution of T21 used here may be partly responsible for the anomalies in the LSex run, which suggest the vague influence of midlatitude SST anomalies. Moreover, the experimental design is not matching to investigate an impact of extratropical SST anomalies [e.g., observed anomalies in 1988/89 show small magnitude (<1 K) in the midlatitudes as in Fig. 1b], because this study mainly focuses the impact of snow anomalies compared to SST anomalies. We are starting an attempt to investigate the extratropical atmosphere–ocean interaction using the CCSR–NIES AGCM with a denser resolution, so the results of the LStr and LSex runs will be further tested.
6. Summary and conclusions
On the basis of the observational result shown by WN98 that the Eurasian snow anomalies in autumn may have an important role in the atmospheric shifts in the following winter, numerical experiments were performed to examine the relative impact of autumn snow anomaly versus SST anomalies on the atmospheric circulation, taking the anomalous situation in 1988/89 into account.
The AGCM, which employed artificial negative snow anomalies over eastern Eurasia, was integrated during the six cold months (September–February) with climatological SSTs as the LSNW run. The LSNW run reproduces strong surface warming over Europe and central-eastern Eurasia, and significant dipole anomalies in 500-hPa height, both of which resemble the observed anomaly patterns in winter 1989. However, height anomalies observed over the North Pacific (weakening of the Aleutian low) are not simulated by the LSNW run. The SST run, which adopted global SSTs observed in 1988/89, also produced the dipole height anomalies and furthermore the weakened Aleutian low. It was ascertained that the anomalies over the North Pacific in the SST run arose from the La Niña condition of SSTs in the equatorial Pacific in winter 1989. An influence of extratropical SSTs is tenuous relative to the tropical SST and Eurasian snow anomalies, because midlatitude SST anomalies given to the SST run are small (<1 K).
The magnitude of the height response in the LSNW run is roughly 60% of that in the SST run. This suggests that the extreme snow anomalies over eastern Eurasia have a substantial impact on the atmospheric circulation, with magnitudes two-thirds of that by large SST anomalies such as in La Niña. It should be noted that the SST run did not change the Eurasian snow and failed to simulate the strong surface warming over eastern Eurasia found in observations. This implies that the surface warming over eastern Eurasia is mainly induced from the anomalous snow through albedo feedback, while the cause of changes in Eurasian snow is not clear and should be further investigated.
When the AGCM is integrated with the combined boundary forcing of Eurasian snow anomalies and observed SSTs, the height anomalies reveal a pattern similar to that in the previous two experiments. The amplitude of height responses to the combined forcing is, however, smaller than the sum of responses in the LSNW and SST runs. Dynamical analyses of the results obtained from the LSNW run and SST run show that height anomalies in the AGCM are dependent not only upon the diabatic heating associated with given boundary forcings but also upon nonlinearities in the atmosphere, such as the anomalous forcing by changes in transient eddies and interactions of anomalies themselves with climatological zonal asymmetries. These internal dynamics can determine in part the amplitude of the response.
The mechanism for the maintenance of the atmospheric anomalies on the decadal scale was not addressed in this study. It is expected that the snow anomalies have little contribution to the multiyear persistence of the atmospheric anomalies because the snow is considered to have no memory durable over several years. Observational results suggest that the SST anomalies in the extratropical oceans may be important for preserving the atmospheric anomaly patterns over several years. This possibility should be examined in the future.
Acknowledgments
The authors are grateful to Drs. M. Kimoto, H. Nakamura, and anonymous reviewers for their comments on an earlier version of the manuscript. We thank Dr. A. Numaguti for developing a substantial part of the CCSR–NIES AGCM. Suggestions on the numerical experiments by Dr. X. Shen were also helpful. This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, and Culture of Japan.
REFERENCES
Barnett, T. P., L. Dümenil, U. Schlese, E. Roeckner, and M. Latif, 1989: The effect of Eurasian snow cover on regional and global climate variations. J. Atmos. Sci.,46, 661–685.
Branstator, G., 1985: Analysis of general circulation model sea-surface temperature anomaly simulations using a linear model. Part II: Eigenanalysis. J. Atmos. Sci.,42, 2242–2254.
——, 1992: The maintenance of low-frequency atmospheric anomalies. J. Atmos. Sci.,49, 1924–1945.
Douville, H., and J. F. Royer, 1996: Sensitivity of the Asian summer monsoon to an anomalous Eurasian snow cover within the Météo-France GCM. Climate Dyn.,12, 449–466.
Ferranti, L., F. Molteni, and T. N. Palmer, 1994: Impact of localized tropical and extratropical SST anomalies in ensembles of seasonal GCM integrations. Quart. J. Roy. Meteor. Soc.,120, 1613–1645.
Foster, D. J., Jr., and R. D. Davy, 1988: Global snow depth climatology. USAFETAC/TN-88/006, Scott Air Force Base, IL, 48 pp.
Foster, J., M. Owe, and A. Rango, 1983: Snow cover and temperature relationships in North America and Eurasia. J. Climate Appl. Meteor.,22, 460–469.
——, and Coauthors, 1996: Snow cover and snow mass intercomparisons of general circulation models and remotely sensed datasets. J. Climate,9, 409–426.
Gates, W. L., 1992: AMIP: The Atmospheric Model Intercomparison Project. Bull. Amer. Meteor. Soc.,73, 1962–1970.
Graham, N. E., T. P. Barnett, R. Wilde, M. Ponater, and S. Schubert, 1994: On the roles of tropical and midlatitude SSTs in forcing interannual to interdecadal variability in the winter Northern Hemisphere circulation. J. Climate,7, 1416–1441.
Hahn, D. G., and J. Shukla, 1976: An apparent relationship between Eurasian snow cover and Indian monsoon rainfall. J. Atmos. Sci.,33, 2461–2462.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteor. Soc.,77, 437–471.
Kharin, V. V., 1995: The relationship between sea surface temperature anomalies and atmospheric circulation in GCM experiments. Climate Dyn.,11, 359–375.
Kitoh, A., 1991: Interannual variations in the atmospheric GCM forced by the 1970–1989 SST. Part II: Low-frequency variability of the wintertime Northern Hemisphere extratropics. J. Meteor. Soc. Japan,69, 271–291.
——, 1996: Comparison between the influences of land processes and SST to the atmospheric variability (in Japanese). Grosswetter,34, 20–30.
Kushnir, Y., and J. M. Wallace, 1989: Low-frequency variability in the Northern Hemisphere winter: Geographical distribution, structure, and time-scale dependence. J. Atmos. Sci.,46, 3122–3142.
——, and I. M. Held, 1996: Equilibrium atmospheric response to North Atlantic SST anomalies. J. Climate,9, 1208–1220.
Lau, N. C., and E. O. Holopainen, 1984: Transient eddy forcing of the time-mean flow as identified by geopotential tendencies. J. Atmos. Sci.,41, 313–328.
——, and M. J. Nath, 1990: A general circulation model study of the atmospheric response to extratropical SST anomalies observed in 1950–79. J. Climate,3, 965–989.
——, and ——, 1991: Variability of the baroclinic and barotropic transient eddy forcing associated with monthly changes in the midlatitude storm tracks. J. Atmos. Sci.,48, 2589–2613.
Maslanik, J. A., M. C. Serreze, and R. G. Barry, 1996: Recent decreases in Arctic summer ice cover and linkages to atmospheric circulation anomalies. Geophys. Res. Lett.,23, 1677–1680.
Morinaga, Y., and H. Igarashi, 1991: Relationships between snow cover extent and snow mass derived from NIMBUS-7 SMMR data over the Eurasian continent. Climatol. Note,40, 101–110.
Nakajima, T., and M. Tanaka, 1986: Matrix formulation for the transfer of solar radiation in a plane-parallel scattering atmosphere. J. Quant. Spectrosc. Radiat. Transfer,35, 13–21.
Numaguti, A., T. Nakajima, M. Takahashi, and A. Sumi, 1998: A description of CCSR/NIES AGCM. J. Meteor. Soc. Japan, in press.
Palmer, T. N., and Z. Sun, 1985: A modeling and observational study of the relationship between sea surface temperatures in the north-west Atlantic and the atmospheric general circulation. Quart. J. Roy. Meteor. Soc.,111, 947–975.
Pinker, R. T., and I. Laszlo, 1992: Modeling surface solar irradiance for satellite applications on a global scale. J. Appl. Meteor.,31, 194–211.
Pitcher, E. J., M. L. Blackmon, G. T. Bates, and S. Muñoz, 1988: The effect of North Pacific sea surface temperature anomalies on the January climate of a general circulation model. J. Atmos. Sci.,45, 173–188.
Randall, D. A., and Coauthors, 1994: Analysis of snow feedbacks in 14 general circulation models. J. Geophys. Res.,99, 20 757–20 771.
Shen, X., M. Kimoto, and A. Sumi, 1998: Role of land surface processes associated with interannual variability of broad-scale Asian summer monsoon as simulated by the CCSR/NIES AGCM. J. Meteor. Soc. Japan,76, 217–236.
Simmons, A. J., J. M. Wallace, and G. Branstator, 1983: Barotropic wave propagation and instability, and atmospheric teleconnection patterns. J. Atmos. Sci.,40, 1363–1392.
Tachibana, Y., M. Honda, and K. Takeuchi, 1996: The abrupt decrease of the sea ice over the southern part of the sea of Okhotsk in 1989 and its relation to the recent weakening of the Aleutian low. J. Meteor. Soc. Japan,74, 579–584.
Ting, M., M. P. Hoerling, T. Xu, and A. Kumar, 1996: Northern Hemisphere teleconnection patterns during extreme phases of the zonal-mean circulation. J. Climate,9, 2614–2633.
Wallace, J. M., and D. S. Gutzler, 1981: Teleconnections in the geopotential height field during the Northern Hemisphere winter. Mon. Wea. Rev.,109, 784–812.
Walland, D. J., and I. Simmonds, 1997: Modelled atmospheric response to changes in Northern Hemisphere snow cover. Climate Dyn.,13, 25–34.
Walsh, J., W. Chapman, and T. Shy, 1996: Recent decrease of sea level pressure in the central Arctic. J. Climate,9, 480–486.
Watanabe, M., and T. Nitta, 1998: Decadal changes in the atmospheric circulation and associated surface climate variations in the Northern Hemisphere winter. J. Climate, in press.
Yasunari, T., A. Kitoh, and T. Tokioka, 1991: Local and remote responses to excessive snow mass over Eurasia appearing in the northern spring and summer climate—A study with the MRI GCM. J. Meteor. Soc. Japan,69, 473–487.
APPENDIX
Anomalous Geopotential Tendency
Such a separation for the time-mean field like Eq. (A4) is basically the same as in Branstator (1992), although he attempted to assess linear budgets. According to his definition, terms C to E in Eq. (A4) are called stationary nonlinearities (SN), anomaly interactions with climatological stationary waves (SW), and interactions between anomalies and climatological zonal symmetries (ZA), respectively. The separation for the geopotential tendency associated with the heat flux,
Since we apply the tendency method to investigate the dynamical balance of the atmospheric response revealed in the LSNW run, daily products from the AGCM are used for the analysis. To distinguish LF components, which involve fluctuations with period between 8 days and a season, a four-pole Butterworth filter was used as in Kushnir and Wallace (1989). The HF components are obtained by subtracting LF quantities from the unfiltered data.
(a) Snow cover anomalies observed in autumn 1988. Units are percent fractional coverage, and the snow extent anomaly over the hemisphere is depicted at the top of figure. (b) SST anomalies observed in winter 1989. Contour interval is 0.5 K and negative contours are dashed. Shaded regions represent areas having positive anomalies more than 0.5 K. In (a), a fan-shaped rectangle indicates the region that the snow forcing is given in the model.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
(a) Observed 500-hPa height (contour) and 200-hPa zonal wind (shaded) in winter averaged over 13 yr. Contour interval is 60 m and regions of westerly more than 40 m s−1 (60 m s−1) are lightly shaded (dark shaded). (b) As in (a) except for ensemble averages obtained from the control (CNTL) run calculated by the CCSR–NIES AGCM. (c) Differences between (a) and (b). Contour interval is 20 m; light-shaded (dark-shading) denotes wind difference of more than +5 m s−1 (less than −5 m s−1).
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
(a) Observed snow depth climatology and (b) snow depth in the CNTL run in February. Contours are 2, 25, 50, and 75 cm and areas having snow above 5 cm (40 cm) are lightly shaded (dark shaded). (c) and (d) As in (a) and (b) except for surface albedo in October. Contour interval is 0.05 and areas having snow above 2 cm are shown by a light shading.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
Design of the experiments for the CNTL run and light snow (LSNW) run.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
Snow depth in the CNTL (thick solid), LSNW (thick dashed), and SST (thin dashed) runs averaged over the area of 35°–75°N, 70°–150°E, where δ = 0.25 in the LSNW run. Units are kg m−2. Shaded area denotes the range of the natural variability in the AGCM derived from the 10-yr AMIP run. Temporal transition of the observed snow depth climatology is together displayed by the thin solid line.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
(a) Observed surface albedo anomalies in November 1988. Shading is every 0.05 and contours are only shown for −0.03 and −0.06. (b) As in (a) except for the albedo anomalies in the LSNW run.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
(a) Surface air temperature anomalies observed in winter 1989. Contour interval is 1 K and negative contours are dashed. (b) and (c) As in (a) except for winter temperature anomalies in the LSNW run and SST run, respectively. Areas having anomalies significant at the 5% level by the t test are shaded.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
(a) As in Fig. 7a except for the anomalous snow mass in the LSNW run. The contour interval is 5 kg m−2. (b) As in (a) except for the SST run.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
As in Fig. 7 except for SLP anomalies in winter. Contour intervals are 2 hPa and negative contours are dashed.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
As in Fig. 7 except for 500-hPa height anomalies in winter. The contour interval is 20 m.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
As in Fig. 10b except for the height anomalies in the (a) LSgl, (b) LStr, and (c) LSex runs.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
(a) Geopotential height tendencies due to vorticity fluxes of the high-frequency (<8 days) eddies at 300 hPa in the CNTL run. Contour interval is 2 m day−1 and negative contours are dashed. (b) As in (a) except for tendencies due to eddy heat fluxes at 300 hPa. (c) As in (a) except for tendencies at 900 hPa. (d) As in (a) except for tendencies due to eddy heat fluxes at 900 hPa.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
Spatial standard deviations (bars) and spatial correlations with Ẑ (lines) of anomalous geopotential height tendencies due to vorticity and heat fluxes, (∂Ẑ/∂t)V+H, for each forcing presented in Eq. (9) at (a) 300 hPa and (b) 900 hPa. Results for the LSNW run (SST run) are presented by cross-hatched bars (open bars) and the solid line (dashed line). The notation of each forcing term can be found in the text. The term “ALL” is equivalent to the forcing −R2 and obtained from the sum of all the other tendencies.
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
(a) Differences in root-mean-squares of high-frequency eddy components of the winter geopotential heights at 300 hPa in the LSNW run with those in the CNTL run. Contour interval is 2 m and negative contours are dashed. Shading denotes areas having significant differences at the 5% level, whereas the thick solid lines indicate contours of 60 m for rms of transient eddy heights in the CNTL run. (b) Zonal asymmetries:
Citation: Journal of Climate 11, 11; 10.1175/1520-0442(1998)011<2837:RIOSAS>2.0.CO;2
Surface albedo, SE, and their anomalies averaged over the area 35°–75°N, 70°–150°E in autumn. Values are derived from observations in 1988 and the LSNW run. Units for the SE are 106 km2.
Hemispheric spatial correlation coefficient and ratios of the spatial standard deviations of the 500-hPa height anomalies in the LSgl, LStr, and LSex experiments with those in the LSNW and SST runs.
