This study provides further evidence of the impacts of tropical Pacific interannual [El Niño–Southern Oscillation (ENSO)] and Northern Pacific decadal–interdecadal [North Pacific index (NPI)] variability on the Pacific–North American (PNA) sector. Both the tropospheric circulation and the North American temperature suggest an enhanced PNA-like climate response and impacts on North America when ENSO and NPI variability are out of phase. In association with this variability, large stationary wave activity fluxes appear in the mid- to high latitudes originating from the North Pacific and flowing downstream toward North America. Atmospheric heating anomalies associated with ENSO variability are confined to the Tropics, and generally have the same sign throughout the troposphere with maximum anomalies at 400 hPa. The heating anomalies that correspond to the NPI variability exhibit a center over the midlatitude North Pacific in which the heating changes sign with height, along with tropical anomalies of comparable magnitudes. Atmospheric heating anomalies of the same sign appear in both the tropical Pacific and the North Pacific with the out-of-phase combination of ENSO and NPI. Both sources of variability provide energy transports toward North America and tend to favor the occurrence of stationary wave anomalies.
El Niño–Southern Oscillation (ENSO)-related sea surface temperature (SST) forcing has a direct effect on the large-scale atmospheric circulation through its impact on diabatic heating and subsequent upper-level divergence over the equatorial Pacific (e.g., Trenberth et al. 1998; Sardeshmukh and Hoskins 1988; DeWeaver and Nigam 2002). It has been suggested that ENSO SST forcing can selectively amplify natural forms of internal variability but cannot generate new structures (e.g., Palmer 1999, among others). Some studies indicate that the North American climate response to positive and negative interannual ENSO variability exhibits a phase shift in its teleconnection pattern, mainly due to the differences in tropical diabatic forcing (e.g., Hoerling et al. 1997; Wu et al. 2003); while others indicate the insensitivity of the extratropical response to the location of ENSO convection (e.g., Geisler et al. 1985; Held and Kang 1987).
It has also been proposed that North Pacific–North American climate variability is not strongly related to ENSO (Zhang et al. 1996). In particular, Straus and Shukla (2002) concluded that ENSO-related SSTs force a circulation pattern quite distinct from the internally generated Pacific–North American (PNA) mode of variability (Wallace and Gutzler 1981). Nigam (2003) examined the distribution of significant amplitude PNA events with respect to ENSO variability. He found the difference between the positive- and negative-phase PNA occurrences to be quite small and not statistically meaningful for most ENSO amplitudes, except weak ENSO conditions, during the 1958–93 winters. On the other hand, Gershunov and Barnett (1998) showed that the North Pacific oscillation [also known as the Pacific decadal oscillation (PDO); Mantua et al. 1997] exerts a modulating effect on ENSO teleconnections. Typical ENSO signals, such as those seen in sea level pressure over the North American sector and daily rainfall in the United States, tend to be stronger and more stable during preferred phases of the PDO. Barlow et al. (2001) reported that summertime U.S. hydroclimatic signals and long-term U.S. drought events are closely associated with both Pacific ENSO and decadal oscillations. DeWeaver and Nigam (2002) also showed that the ENSO responses in the PNA region might be modulated by a slowly varying decadal signal in different periods, so that there appears antisymmetry in the upper-level Pacific height responses to warm and cold ENSO events rather than a phase shift between El Niño and La Niña height patterns over the North Pacific. Recent studies further suggest that the variability of the Aleutian low pressure center and the associated SSTs in the North Pacific regulate the ENSO response over North America (Yang et al. 2002; Lau et al. 2004). The variability of the Aleutian low pressure center, which exhibits remarkable decadal variability (e.g., Deser et al. 2004), is represented by the North Pacific index (NPI; Trenberth and Hurrell 1994). The impact of this decadal Pacific mode of variability on North America is comparable in magnitude to that of ENSO (e.g., Hurrell 1996).
The Pacific interannual ENSO and decadal–interdecadal ENSO-like (or PDO) variability patterns have been compared and documented in Zhang et al. (1997). Similar spatial signatures were found in the associated atmospheric fields. Nevertheless, decadal–interdecadal SST anomalies in equatorial regions are smaller and generally broader in the meridional direction than their interannual counterparts, while the magnitude of equatorial and midlatitude SST anomalies is comparable. The corresponding sea level pressure signature is also stronger over the extratropical North Pacific, and its counterpart in the cold season 500-hPa height field more closely resembles the PNA pattern. Comprehensive reviews of mechanisms responsible for the interannual and decadal–interdacadal variability may be found in Neelin et al. (1998), Miller and Schneider (2000), and Sarachik and Vimont (2003). A recent study further suggests that the spatial structure of ENSO-like decadal variability is the result of variations in the interannual ENSO cycle (Vimont 2005).
Anomalous circulation or stationary wave anomalies contribute to the climate response to SST forcing over North America. Physically, stationary wave anomalies may be produced either by internally generated dynamical perturbations of the atmosphere or by external forcing such as orography, land–sea thermal contrast, and SST anomalies. Substantial variability can arise in midlatitudes without external forcing as a result of internal nonlinear atmospheric dynamics. For instance, the midlatitude jet stream and storm tracks vary considerably from year to year in models in the absence of external forcing (e.g., Zwiers 1987; Hartmann 1995). Nevertheless, external forcing (e.g., from SST anomalies) may tend to favor the occurrence of certain types of variations. Investigations have suggested that SST anomalies in the tropical Pacific produce stronger and more reproducible atmospheric responses than those in the extratropics (e.g., Trenberth et al. 1998, and references therein). In the Tropics, the local atmospheric response to tropical heating is well explained by linear equatorial dynamics (Gill 1980). In the midlatitudes, three dynamical mechanisms appear to be important in the response: horizontal energy propagation resulting from the dispersion of planetary waves (e.g., Hoskins and Karoly 1981), feedbacks between transient eddies and low-frequency anomalies (e.g., Branstator 1995), and interactions between the disturbances and the climatological mean state (e.g., Trenberth et al. 1998 for a review).
In this study, we further detail the atmospheric response to Pacific variability by demonstrating the enhanced climate response over North America as a result of the out-of-phase effects of interannual ENSO and decadal–interdecadal North Pacific variability. We then analyze the anomalous stationary wave activity and focus on associated changes in diabatic heating and atmospheric energy transports to help physically understand changes in the stationary wave associated with the combination of these two modes of Pacific variability. We will show that atmospheric heating during this out-of-phase state appears in both the tropical Pacific and the North Pacific and gives rise to enhanced energy transport toward North America, and we suggest that this leads to a robust PNA-like stationary wave structure.
2. Data and diagnostic methods
a. ENSO and NPI indices
We characterize interannual ENSO variability and decadal–interdecadal North Pacific variability by means of the Niño-3.4 and NPI indices. The Niño-3.4 index is defined as the SST anomaly relative to 1950–79 in the region 5°S–5°N, 120°–170°W (Trenberth 1997), while NPI is given by the area-weighted sea level pressure over the region 30°–65°N, 160°E–140°W (Trenberth and Hurrell 1994). These indices were obtained from http://www.cgd.ucar.edu/cas/catalog/climind. An extended winter mean (November–March) Niño-3.4 time series is further constructed and values beyond thresholds of ±0.5°C are used to define ENSO events. Accordingly, we identify 31 El Niño and 26 La Niña events from 1900 to 2004 (Table 1, excluding the transition winters 1924–25 and 1976–77; discussed further below) and use them to construct the anomalous ENSO winters. The selected events are similar to those identified in previous studies (e.g., Kiladis and von Loon 1988; Trenberth 1997).
We use the NPI to identify decadal–interdecdal North Pacific variability rather than the North Pacific decadal oscillation index (Mantua et al. 1997) derived directly from SST anomalies in order to avoid the possible effects of sparse data coverage early in the PDO record (Fig. 4 of Deser et al. 2004). In particular, we are concerned that the PDO index, which is derived from SST observations, may contain spurious signals due to poor coverage during the period 1901–10. Substitution of the PDO index by the NPI is justified by the strong lag-0 correlation between the two (Deser et al. 2004) and the fact that they both exhibit similar climate transitions between predominantly positive (1900–24 and 1947–76) and negative phases (1925–46 and 1977–2003). It is these climate transition points that are of primary concern in this study. Here 1946–47 is taken as the boundary between complete cycles of the interdecadal variability (also see Fig. 1 of Deser et al. 2004). Composites around different phases of decadal variations (e.g., Fig. 5) are taken as differences between 1977–2003 and 1947–76 averages. The structure of ENSO teleconnections during different decadal periods are taken as the average of all El Niño (or La Niña) events during negative (1900–24 and 1947–76) or positive (1925–46 and 1977–2003) decadal NPI periods (Table 1).
b. Diagnostic analysis data
An updated monthly Northern Hemisphere sea level pressure (SLP) dataset (Trenberth and Paolino 1980) and observed monthly North American surface temperature are used to determine the SLP and surface temperature responses to the Niño-3.4 and NPI variations. As above, the data are averaged in each extended winter. Results for the winters 1900–01 to 2003–04 are analyzed for SLP over the region 20°–70°N, 180°–360°W on a 5° latitude–longitude grid. We also use a high-resolution North American surface temperature dataset, on a 0.5° latitude–longitude grid for the winters 1900–01 to 2000–01 that was obtained from the Climatic Research Unit (CRU) at the University of East Anglia (Jones and Moberg 2003; also see http://www.cru.uea.ac.uk/cru/data).
In addition, we use the National Centers for Environmental Prediction–National Centers for Atmospheric Research (NCEP–NCAR) reanalyses (Kistler et al. 2001) for January 1949 to December 2003. The variables considered include geopotential height, temperature and horizontal and vertical velocities in the troposphere, precipitation, net shortwave and longwave radiation at the top of the atmosphere (TOA), net shortwave and longwave radiation at the surface, and surface latent and sensible heat fluxes. Both 6-hourly (0000, 0600, 1200, and 1800 UTC) and monthly mean isobaric fields were extracted on a 2.5° latitude–longitude global grid at 17 pressure levels (10, 20, 30, 50, 70, 100, 150, 200, 250, 300, 400, 500, 600, 700, 850, 925, and 1000 hPa). The monthly European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40; Simmons and Gibson 2000; Uppala et al. 2005) from September 1957 to August 2002, on a 2.5° latitude–longitude global grid, was used to confirm our diagnostic results, especially those concerning the vertically integrated total diabatic heating and atmospheric energy transports. In all cases, anomalies are calculated relative to the 1961–90 30-yr climatology. There are 16 El Niño and 17 La Niña events during the NCEP–NCAR reanalysis period, and 14 El Niño and 13 La Niña events during the ERA-40 reanalysis period (Table 1). Those are the winters used in the corresponding reanalysis composites.
c. Dynamic and thermodynamic diagnostics
Responses of the tropospheric circulation and North American surface temperature to the Pacific interannual and decadal–interdecadal variability are constructed and analyzed with respect to the Niño-3.4 and NPI indices. The Plumb flux diagnostic (Plumb 1985) of quasi-stationary wave activity is used as an aid in locating source and sink regions of anomalous wave activity and in identifying wave propagation characteristics (Black and Dole 1993). As mentioned above, stationary wave anomalies may be produced either by internally generated dynamical perturbations of the atmosphere or by external forcing. Hence, we also study the atmospheric heating anomalies related directly to such perturbations and/or forcing by examining both the three-dimensional (3D) net diabatic heating and the vertically integrated heating and energy transports. We diagnose the 3D diabatic heating anomaly of the atmosphere as a residual from the time-averaged thermodynamic equation (e.g., Hoskins et al. 1989; Nigam 1994). The vertically integrated total diabatic heating and atmospheric energy transports (e.g., Boer 1986; Trenberth and Solomon 1994) are calculated and analyzed by using energy balance variables at the TOA and at the surface from the NCEP–NCAR and ERA-40 reanalyses.
3. Responses of tropospheric circulation and North American temperature
a. SLP and geopotential heights
SLP anomalies for El Niño and La Niña winters during the opposite NPI phases are displayed in Fig. 1. The anomalous signatures are generally stronger for the out-of-phase combinations of Pacific ENSO and NPI variability (El Niño/−NPI and La Niña/+NPI) than those for the in-phase combinations (El Niño/+NPI and La Niña/−NPI). The strongest anomalies occur over the North Pacific, with downstream anomalies of opposite sign over North America and anomalies of the same sign over the subtropical–midlatitude Atlantic. This indicates a train of atmospheric waves, indicated by a sequence of high and low pressure centers that extends from the Tropics toward the North Pacific, North America, and then the Atlantic, and suggests a PNA-like impact of the two modes of Pacific variability on the North American climate. In contrast, SLP anomalies tend to be weak when the ENSO and NPI indices are in phase. Similar results are obtained for both interdecadal cycles, denoted as C1 and C2 cycles in Fig. 1, before and after year 1946. These results are also in accord with the SLP composites of Gershunov and Barnett (1998) for the period 1933–93.
Significant PNA-like responses to the combined effects of Pacific interannual and decadal variability are more pronounced in geopotential height. Figure 2 displays the Φ500 composites obtained from NCEP–NCAR reanalysis for the winters 1949/50 to 2002/03. A PNA-like pattern is clearly evident when the ENSO and NPI indices are out of phase, while there are zonally homogeneous anomalies extending from the North Pacific to the North Atlantic when the two indices are in phase. The centers of action between the composites differ significantly at the 5% level (von Storch and Zwiers 1999) and appear throughout the troposphere (see Fig. 4 later).
b. Surface temperature over North America
The combined effects of ENSO and NPI variability are also clear for surface temperatures over North America. Figure 3 shows surface temperature anomalies over North America, derived from the CRU dataset and constructed for El Niño and La Niña events during the opposite NPI phases for the winters 1946/47 to 2000/01. Results from 1900 to 1946 are similar to those from the later period and are not shown.
As with the SLP anomalies, surface temperature anomalies over North America have greater amplitude when the ENSO and NPI indices are out of phase compared to their in-phase combinations. Most of Canada and the northern United States are characterized by pronounced anomalous warming, with a band of intense anomalies greater than 1.0 K extending northwestward from the Canadian Midwest to Alaska, along with smaller negative temperature anomalies over the central-southern areas of the United States and Mexico when positive ENSO events superimpose on the negative NPI phase. These warming anomalies are consistent with the anomalous circulation indicated by the depressed SLPs and reduced heights over the North Pacific (Figs. 1 and 2). The opposite happens when La Niña occurs with positive NPI, in which case there are generally modest temperature anomalies of less than 1.0 K over most of Canada and the United States. Exceptions occur in the La Niña composite during 1977–2001, where there are positive anomalies of about 1.0–2.0 K over the Canadian Midwest. This is closely associated with the abnormally long warm spell over North America during the 1998–2000 La Niña event (e.g., Hoerling et al. 2001). In fact, the temperature anomalies over the Canadian Midwest reduce by about 0.5–1.0 K over most of Canada and the northern United States when this La Niña event is excluded from the composite (not shown). Numerical studies indicate that the protracted warming during 1998–2000 is unrelated to equatorial eastern Pacific SSTs, but is forced by the global SSTs with abnormally higher SSTs in the Indo-Pacific warm pool possibly contributing to the abnormal warmth (Hoerling et al. 2001; Kumar et al. 2001).
Thus, the responses of both tropospheric circulation and North American surface temperature to combined interannual ENSO and decadal–interdecadal North Pacific variability show an enhanced PNA-like climate anomaly and intense impacts on North America. These results also suggest that the combined impacts of the two modes of variability may have important implications for climate prediction over North America (e.g., Mantua et al. 1997; Trenberth et al. 1998; Gershunov and Barnett 1998; Derome et al. 2001), provided that one can skillfully predict these phenomena.
4. Stationary wave activity
Following Plumb (1985), the horizontal components of the stationary wave activity flux Fs can be written as
where p is the pressure, p0 = 1000 hPa, (λ, ϕ) are the longitude and latitude, (u, υ) are the zonal and meridional geostrophic wind components, Φ is the geopotential, Ω is the earth’s rotation rate, and a is the earth’s radius. The overbar indicates the time mean and the asterisk indicates the departure of the variable from its zonal mean. As suggested by Karoly et al. (1989), the wave activity flux is calculated using geostrophic winds from the height anomalies instead of total winds, consistent with its quasigeostrophic derivative in the wave activity flux. We calculated the wave activity flux from the 55-yr monthly NCEP–NCAR dataset. Results are displayed poleward of 15°N and 15°S.
Figure 4 displays geopotential height anomalies at 200 hPa (Φ200) and corresponding stationary wave activity flux Fs components constructed for El Niño and La Niña events during opposite NPI phases. The wave activity flux anomalies were calculated based on the anomalous Φ200 fields during the identified winters and then constructed for each composite. As for Φ500 (Fig. 2), the Φ200 anomalies also feature a PNA-like wave train. The anomalous height patterns differ markedly when ENSO variability is in phase with the NPI, where the anomalous height generally extends east from the North Pacific to the midlatitude North Atlantic. The main features in the Tropics resemble the theoretical representation of circulation response to heating variations (Gill 1980). In the immediate vicinity of the forcing over the tropical central-eastern Pacific, the response is characterized by an anticyclonic vorticity pair that is nearly symmetric about the equator.
When interannual ENSO and decadal–interdecadal North Pacific variability modes are out of phase, larger wave activity fluxes originate from the North Pacific and flow downstream toward North America, indicating that the major source of wave activity may lie over the North Pacific Ocean. This source is located close to the midlatitude jet and storm track (Peixoto and Oort 1992) suggesting that its forcing mechanism may be associated with local processes such as instabilities of the jet or with transient eddies (e.g., Simmons et al. 1983; Lau and Holopainen 1984; Hoskins and Ambrizzi 1993; Trenberth and Hurrell 1994; Branstator 1995; Chen and Van den Dool 1997), consistent with instability theory for the PNA pattern (e.g., Karoly et al. 1989; Trenberth et al. 1998). There is also some evidence of propagation of Fs from the Tropics into midlatitudes around 150°–160°W and equatorward propagation over the Atlantic Ocean with the out-of-phase ENSO and NPI variability. On the contrary, the wave activity fluxes are considerably weaker over the PNA sector when the two modes of variability are in phase. We have also analyzed the separate effects of ENSO and NPI variability on 200-hPa height and wave activity and its propagation. There is more evidence of propagation of Fs from the Tropics to midlatitudes for ENSO variability than for the NPI variability (not shown), resembling the results from a comparison of the ENSO and PNA cases in Karoly et al. (1989).
Black and Dole (1993) and Lyon and Dole (1995) have suggested a number of possible mechanisms for generating stationary wave anomalies. These include anomalous diabatic heating, topographic forcing, or transient eddy forcing on different spatial scales and interactions between stationary waves and background mean flow. The PNA-like wave train and propagation of wave activity out of the Tropics is clearly seen in a simple model response to tropical thermal forcing (Karoly et al. 1989). However, the observed location of the main stationary wave activity source for the PNA-like pattern is inconsistent with tropical forcing of the PNA pattern and Rossby wave–like propagation from low latitudes (Fig. 4). On the other hand, as noted by Sardeshmukh and Hoskins (1985, 1988), it is conceivable that alteration of the local Hadley circulation by the tropical forcing may have effectively moved the stationary wave source from the Tropics to the subtropics. In fact, while examining the causes of the 1988 drought over North America, Trenberth et al. (1988) suggested that the anomalous flow pattern was primarily forced by diabatic heating related to warmer than normal SSTs in the subtropical Pacific. The above results, therefore, cannot clearly demonstrate where the potential forcing is located.
5. Three-dimensional diabatic heating
To help identify the source regions and forcing of the anomalous stationary wave activity, the three-dimensional net diabatic heating Q is calculated and analyzed. Recent studies have shown that diabatic heating contributes substantially to divergence and large-scale atmospheric circulation anomalies in the Tropics and midlatitudes during ENSO events (e.g., Nigam et al. 2000; DeWeaver and Nigam 2002). We thus ask here how the heating relates to the anomalous stationary waves and what is the 3D heating structure in association with the combined effects of ENSO and NPI variability.
where T is the temperature, V (u, υ) are the horizontal velocities, ω is the vertical velocity, θ is the potential temperature, and κ = R/Cp = 0.286. The double overbar designates the monthly average and the prime indicates the departure of the 6-h analysis from the monthly average. The robustness of our results is established by comparing with results from previous studies. The seasonal and annual mean climatology of atmospheric diabatic heating calculated for the 55-yr NCEP–NCAR reanalysis resembles that from the ECMWF initialized analyses for March 1979–February 1989 (Hoskins et al. 1989) and from the NCEP–NCAR reanalyses for a 15-yr period from 1980 to 1994 (Yanai and Tomita 1998). Diabatic heating associated with ENSO also compares well with that reported in Nigam et al. (2000) from both the ECMWF and the NCEP–NCAR reanalyses during the overlapping 1979–93 period.
a. Spatial patterns
Figure 5 displays the diabatic heating anomalies at 500 hPa (Q500) associated with the two modes of variability. The ENSO diabatic heating is marked by positive anomalies over the tropical central-eastern Pacific, accompanied by off-equatorial negative anomalies in the far western North Pacific and South Pacific convergence zone (SPCZ). The heating anomalies are asymmetrically distributed about the equator, with heating extending somewhat farther into the Southern Hemisphere (Fig. 5a). These features are essentially in agreement with the ENSO heating anomalies at 400 hPa in Nigam et al. (2000) and with typical ENSO precipitation anomalies (e.g., Wallace et al. 1998). In addition, there are weaker heating anomalies elsewhere over the midlatitude Pacific, the Indian Ocean, and the North Atlantic. Over the extratropical North Pacific, positive anomalies are observed along the west coast of the United States and Canada and around Japan. The Q500 anomalies tend to have zonal and meridional heating contrasts over the Indian Ocean and the North Atlantic, respectively.
The heating anomalies associated with the Pacific decadal–interdecadal variability differ considerably and are generally modest (about 3–4 times weaker) over the tropical Pacific compared to ENSO heating anomalies. The Q500 differences between the negative and positive NPI phases (Fig. 5b) feature similar positive anomalies over the tropical central Pacific and east of the date line in the midlatitude North Pacific, with negative anomalies in the subtropical North Pacific. The anomalies over the Indian and Atlantic Oceans are somewhat less well organized, especially over the Atlantic. The heating anomaly differences between the ENSO and NPI variability are statistically significant at the 5% level over the equatorial Pacific and the midlatitude North Pacific (not shown). Figure 5c further shows Q500 anomalies for the composite of El Niño events during the negative NPI phase. As would be expected, the heating anomalies here are dominated by the ENSO diabatic heating, but are also accompanied by a clear heating over the midlatitude northeast Pacific, reflecting the combined effects from both ENSO and NPI variability.
Figure 6 displays the heating anomalies in the lower troposphere at 850 hPa (Q850). The ENSO heating anomalies are generally weaker at this level, especially in the tropical Pacific, compared to those in the midtroposphere. Positive heating anomalies now elongate zonally over the North Pacific and their amplitudes are more comparable with those in the Tropics. The heating anomalies over the midlatitude North Pacific are rather striking and even stronger than its tropical counterpart for the NPI variability (Fig. 6b). Consequently, the heating anomalies associated with the combined effects of the two modes of variability feature positive anomalies with analogous amplitudes both in the tropical central-eastern Pacific and in the North Pacific (Fig. 6c).
b. Equatorial and midlatitude cross sections
The heating distributions in the middle and lower troposphere differ markedly. In general, in the Tropics, the thermodynamic balance is between diabatic heating and adiabatic cooling. A tropical heat source away from the surface is hence balanced by upward motion. The midlatitude situation, however, is in complete contrast (e.g., Hoskins and Karoly 1981). In midlatitudes, heating at any level is balanced by horizontal advection of temperature. The sensitivity of low-level temperature to the vertical distribution of midlatitude heating sources is consistent with the thermal wind relation.
Figure 7 compares vertical cross sections of diabatic heating anomalies in the troposphere over the tropical Pacific (10°S–10°N, 120°E–90°W) and over the midlatitude Pacific (30°–50°N, 150°E–120°W), together with SST anomalies in the Pacific basin, for the composite of El Niño events during the negative NPI phase. The vertical structure of heating along the equator is dominated by ENSO-related diabatic heating, while the midlatitude structure is dominated by NPI-related heating (not shown). The equatorial cross section (Fig. 7a) is characterized by deep warming anomalies to the east of 170°E and cooling anomalies to the west in association with the El Niño–like SST warming (Fig. 7c). The anomalies generally have the same sign throughout the troposphere, with maximum heating anomalies at ∼400 hPa. This heating structure also suggests an anomalous Walker circulation that is thermally direct with enhanced convection and the release of latent heat providing a source of energy for rising motion in the central Pacific. In contrast, the midlatitude heating anomalies (Fig. 7b) are dominated by positive anomalies in the lower-middle troposphere. To the west of 150°–160°W, in association with the cold SST anomalies, there tend to be positive heating anomalies centered around 850–950 hPa in the lower troposphere (below ∼600 hPa). These warming anomalies are overlaid by negative anomalies with higher values above about 500 hPa. This may also suggest an enhancement of low cloud in this sector. To the east of 150°W, there is positive heating in the middle troposphere and negative heating in the upper and lower troposphere, overlying the warm SST anomalies along the west coast of the northern United States and Canada in this region (Fig. 7c).
Recently, White and Chen (2002) analyzed the tropospheric response to SST anomalies in the Antarctic Circumpolar Wave and found that warm SST anomalies coincide with anomalous mid- to upper-level diabatic heating and low-level cooling throughout the analyzed period from 1983 to 1992. The heating structure over the midlatitude Pacific (Fig. 7b) bears some resemblance to that reported in White and Chen (2002), although the thermodynamic mechanisms (such as the anomalous vorticity budget) have not been diagnosed further in our analysis. Numerical studies have also suggested that atmospheric responses to midlatitude SST anomalies can be complex. For instance, Peng and Whitaker (1999) and Peng et al. (1995, 1997) examined the extratropical tropospheric response to SST anomalies and found that warm SST anomalies produce different responses in different months, depending on the influences that different basic states have on the transient eddy interaction with the weak SST-induced circulation.
6. Vertically integrated total diabatic heating and atmospheric energy transport
As noted in Nigam et al. (2000) and documented in White and Saha (1996), the diabatic heating generated during a 6-h model forecast starting from each time step’s reanalysis circulation is also available and partitioned into six components: large-scale condensation, deep and shallow convective, longwave and shortwave, and vertical diffusion heating rates. In this section, we examine the vertically integrated heating and atmospheric energy transports based directly on the radiation at both TOA and surface, and based on the surface heat fluxes from the reanalysis datasets. This also serves to validate the 3D diabatic heating results discussed above.
respectively, where RT is the downward radiation at TOA, FS is the downward surface flux (including radiative and turbulent heat fluxes), L is the latent heat due to evaporation, P is the precipitation rate, and E is the surface evapotranspiration. In a steady state, the vertically integrated total atmospheric energy transport Ha and the heating terms Q̃1 and Q̃2 can be related as
where the divergence describes the transport link from source to sink (e.g., Trenberth et al. 2001; Yu and Boer 2002). In addition, the associated potential function (χ) in the atmosphere can be obtained from
We may infer the divergent component of atmospheric energy transport vectors either by calculating directly the vertically integrated and time-averaged total atmospheric energy (formally, the moist static energy since kinetic energy is small and can be neglected) or by obtaining it indirectly from the heating terms Q̃1 and Q̃2. We have adopted the latter approach here as in Yu and Boer (2002). The vertically integrated heating and energy transports are computed based on both the NCEP–NCAR and ERA-40 monthly datasets.
The vertically integrated atmospheric heating anomalies related to ENSO and NPI variability are then analyzed by diagnosing the anomalies of vertically integrated heating (Q̃1) and energy transports (Ha). Trenberth et al. (2001) made comparisons and evaluations of the radiation and heat fluxes based on NCEP–NCAR and ECMWF reanalyses and pointed out that nontrivial systematic differences exist between the two. Nevertheless, the climatological patterns of Q̃1 and Ha from NCEP–NCAR and ERA-40 compare reasonably well (not shown). The main differences occur in the Tropics, particularly in the intertropical convergence zone, and are dominated by precipitation differences (Uppala et al. 2005). The results reported in the following differ only modestly between the two datasets because systematic errors and biases in the means are partially removed by the compositing process (WCRP 2000).
Figure 8 displays the spatial distributions of heating Q̃1 anomalies and the anomalous atmospheric energy transports for the two modes of Pacific variability and for the composite of El Niño events during the negative NPI phase. The anomalous divergent fluxes and associated potential functions, reflecting the atmospheric energy transports, are a consequence of all the processes taking place with the variability. The geographical distributions of heating Q̃1 associated with ENSO and NPI variability are markedly different. The ENSO-related heating anomalies are dominated by tropical centers with the strongest heating in the equatorial central-eastern Pacific surrounded by cooling in the western Pacific and vice versa. In contrast, for the NPI variability, there is a center of anomalous heating over the North Pacific together with anomalies of comparable magnitudes in the tropical Pacific. These results are consistent with the 3D diabatic heating anomalies analyzed above (Figs. 5 and 6). Consequently, for the entire atmospheric column, energy transports that flow from the Tropics to North America are associated primarily with ENSO-related tropical heating anomalies. In contrast, energy transports from the North Pacific toward North America are associated with NPI-related heating anomalies over the North Pacific. The energy fluxes associated with ENSO variability also display a large-scale divergence of atmospheric energy from the central Pacific, which is characteristic of El Niño events such as those of 1982/83 and 1986/87 (Boer 1989; Sun and Trenberth 1998). The heating anomaly and energy transport differences between ENSO and NPI variability are statistically significant at the 5% level over the tropical and North Pacific regions (not shown). There are also energy sources/sinks and associated energy transports over the Indian Ocean, the western Pacific, and the Sahara that appear to be NPI related. This may suggest that there are potential impacts of Pacific variability on East Asia and Africa, which remain to be investigated.
Thus heating anomalies of the same sign appear in both the tropical and North Pacific (Fig. 8, lower panels) when interannual ENSO variability superimposes on an out-of-phase change of NPI variability. Two potential function (χ) centers are evident in the tropical central-eastern Pacific and in the midlatitude North Pacific. Both vertically integrated total diabatic heating and atmospheric energy anomalies produce energy transports propagating toward North America by means of anomalous atmospheric circulation patterns (Figs. 1, 2 and 4) reflecting changes in the position and intensity of the Aleutian low pressure center and the subtropical jet streams (e.g., Wallace and Gutzler 1981; Trenberth and Hurrell 1994; Renwick and Wallace 1996; Shabbar et al. 1997; Higgins et al. 2002; Yang et al. 2002), and may hence strengthen the PNA-like stationary wave structure. It is worth noting, as pointed out in Trenberth and Hurrell (1994) and indicated from this analysis, that changes in diabatic heating in the Northern Hemisphere can change the planetary waves and poleward heat fluxes. Nonetheless, detailed dynamical processes, in particular changes in storm tracks and the eddy–mean flow interaction, in association with the combined Pacific variability remain to be explored.
7. Summary and discussion
In this study, we analyze the atmospheric response to tropical Pacific interannual ENSO and North Pacific decadal–interdecadal NPI variability, with a focus on the Pacific–North American sector. We then diagnose the Plumb quasi-stationary wave activity flux and three-dimensional atmospheric diabatic heating anomalies in association with the Pacific variability to help understand the stationary wave anomalies.
With the out-of-phase combination of NPI and ENSO variability, the responses of both tropospheric circulation and North American surface temperature show an enhanced PNA-like climate anomaly and intense impacts over North America. Large stationary wave activity fluxes appear primarily in the mid- to high latitudes originating from the North Pacific and flowing downstream toward North America, indicating that the major source of wave activity may lie over the North Pacific Ocean. Atmospheric heating anomalies associated with ENSO variability are confined primarily to the Tropics; while for the NPI variability, the heating anomalies are characterized by a center over the North Pacific along with tropical anomalies of comparable magnitudes. The ENSO diabatic heating is characterized in the vertical by deep warming anomalies to the east of 170°E and cooling anomalies to the west along the equator, a generally equivalent barotropic structure with maximum heating anomalies at 400 hPa. In contrast, the NPI-related heating features a generally baroclinic structure in the midlatitude North Pacific, and is dominated by anomalies in the lower-middle troposphere. Heating in the North Pacific region and over the tropical Pacific changes in opposite directions with in-phase NPI and ENSO changes. Consequently, the out-of-phase combination of NPI and ENSO would provide anomalous atmospheric energy transports toward North America from both the North Pacific and the tropical Pacific, which tends to favor the occurrence of stationary wave anomalies and would lead to a robust PNA-like wave anomaly structure.
As indicated in several numerical studies, the protracted warming during 1998–2000 is unrelated to equatorial eastern Pacific SSTs, but is forced by the global SSTs (Hoerling et al. 2001; Kumar et al. 2001). We have checked the above results by excluding this La Niña event. The surface temperature, SLP, and height anomalies over Canada and the United States are generally weaker when this event is excluded in the in-phase combination of NPI and ENSO composite from 1977 to 2003.
The impact of the low-frequency NPI variability on North America may be underestimated in this analysis because our results are based on composites calculated for the entire phase period of the NPI instead of using extreme NPI years as for ENSO. Such a compositing approach was necessary because there were only a small number of NPI cycles available for study. The results may be confirmed by considering other observational variables and by analyzing long climate model integrations (Yu and Zwiers 2007). The introduction of satellite data in 1979 may have had an effect on some of our diagnostics. In addition, dynamical processes, such as variations of momentum fluxes by high-frequency eddies and in particular the variation of storm tracks, in association with the combined effects of the Pacific variability remain to be examined. This may help clarify the process of atmospheric energy transport in conjunction with the planetary wave anomaly. The impacts of Pacific variability beyond the PNA region also merit further consideration.
We thank S. Lambert and A. Niitsoo for help with data issues and K. Szeto, Q. Teng, and E. Watson for helpful comments on this study. In addition, we thank three anonymous reviewers and the editor (S. Nigam) for their constructive suggestions and advice, which considerably improved the original manuscript.
Corresponding author address: Bin Yu, Climate Data and Analysis Section, Climate Research Division, Environment Canada, 4905 Dufferin Street, Toronto, ON M3H 5T4, Canada. Email: Bin.Yu@ec.gc.ca