1. Introduction
a. Background
Perhaps the most well-known signal of climate variability is the El Niño–Southern Oscillation (ENSO), a pattern of warm and cold sea surface temperature (SST) anomalies in the central and eastern equatorial Pacific with coupled atmospheric changes. The atmospheric component is monitored by the normalized pressure difference between Tahiti (17.6°S, 149.6°W) and Darwin (12.4°S, 130.9°E) that is called the Southern Oscillation index (SOI). To quantify human impact on global climate change, ENSO must be understood in the context of anthropogenic signals (Houghton et al. 2001). This is a complex task, as ENSO affects temperature and precipitation on a global scale (Dai et al. 1997), and often does not follow recurring patterns (e.g., Houseago et al. 1998). Furthermore, ENSO occurs on a wide range of temporal scales from interannual (Cai and Watterson 2002) and decadal (Meehl et al. 1998) to millennial (Stott et al. 2002) and beyond.
In recent years, scientists have discovered surprising, and occasionally conflicting, climate changes taking place in the Antarctic. Jones (1995) finds an average warming at the surface over all of Antarctica (especially the Antarctic Peninsula) by analyzing station data from 1957 to 1994. Comiso (2000) notes a cooling trend between 1979 and 1998 from infrared Advanced Very High Resolution Radiometer (AVHRR) imagery and from station records, but the trends are not statistically significant for either dataset. Similarly, Doran et al. (2002) observe rapid surface warming on the Antarctic Peninsula and evidence of cooling elsewhere on the continent by analyzing station data from 1966 to 2000. Marshall et al. (2002) find a less pronounced warming trend in the free atmosphere above the Antarctic Peninsula; the annual mean tropospheric warming is not statistically significantly greater than the mean warming for the Southern Hemisphere. Others have documented declining sea ice extent in the mid and late twentieth century over the Antarctic as a whole (de la Mare 1997) and regionally in the Amundsen and Bellingshausen Seas (Jacobs and Comiso 1997). Ice sheet instability has become a topic of recent concern; Rignot and Jacobs (2002) present new evidence that rapid bottom melting is occurring near ice sheet grounding lines throughout Antarctica, especially in areas of West Antarctica bordering the Amundsen and Bellingshausen Seas. Perhaps the most dramatic recent change is the collapse of the Larsen B Ice Shelf on the Antarctic Peninsula, which had likely existed since the last glacial maximum (E. Domack et al. 2003, personal communication). ENSO may play an important role in these changes, and thus it is necessary to better understand ENSO forcing in the Antarctic, especially in light of the recent trend toward more frequent warm events; the mean of the SOI from 1982 to 1998 was −0.5 (Kwok and Comiso 2002). Trenberth and Hoar (1997) find that the trend toward more frequent El Niño episodes in recent decades is very unlikely to be accounted for solely by natural variability.
Vaughan et al. (2001) remind us that our current knowledge of the mechanisms and spatial distribution of climate change is limited, and we can only predict large-scale variations with some degree of confidence. It is important that we first expand our knowledge at the regional scale in order to truly understand global signals. With coarse resolution and simplified physics, the current generation of global climate models generally do not capture regional climate change with high skill (Connolley and O'Farrell 1998). Reanalysis products, which are intended to provide a coherent, uniform depiction of global climate back to the late 1950s, often exhibit large biases over Antarctica due to physics limitations, observational limitations, assimilation errors, and steep topography in the coastal regions (e.g., Bromwich et al. 2000; Hines et al. 2000; Marshall 2002; Trenberth and Stepaniak 2002). Global analyses and archived global numerical weather forecasts can provide good depictions of the atmosphere, and more recently at higher resolutions than are available in reanalyses, but are difficult to use for climate studies because frequent modifications to modeling systems can cause abrupt artificial changes in modeled fields (e.g., Kalnay et al. 1996). Thus, it is necessary to combine higher-resolution modeling that is consistent in space and time, and that can properly represent processes over and around polar ice sheets (e.g., ice crystals in clouds, sea ice thermodynamics, katabatic winds), in order to accurately reconstruct recent climate variability in the Antarctic.
Here, simulations from the Polar MM5 are used to examine interannual variability over the Antarctic for 3 years in the late 1990s (July 1996–June 1999). The Polar MM5 is a version of the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5; Grell et al. 1994), adapted by the Polar Meteorology Group at the Byrd Polar Research Center for high-latitude applications (e.g., Bromwich et al. 2001, Cassano et al. 2001). This period is chosen because it spans a strong El Niño (negative SOI) episode followed by a moderate La Niña (positive SOI). The model has demonstrated good skill in capturing atmospheric variability at diurnal, synoptic, seasonal, and annual time scales for extensive validations over Greenland and Antarctica (Cassano et al. 2001; Bromwich et al. 2001; Guo et al. 2003). This study gives, for the first time, a comprehensive depiction of interannual variability over Antarctica in the late 1990s.
b. ENSO and Antarctica in the late 1990s
Figure 1 shows the 5-month running mean of the SOI during the study period. A transition from a weak La Niña to a strong El Niño occurs in autumn (MAM) 1997. The El Niño episode lasts for about 1 yr, then shifts to a moderate La Niña by winter (JJA) 1998. The moderate La Niña conditions persist for nearly a year, and weaken toward a neutral state by late autumn 1999. For discussion, we divide the study period into three 1-yr increments, denoted by the vertical lines in Fig. 1. The first period (July 1996–June 1997) is distinguished by a weak La Niña changing to an El Niño by autumn 1997. The second period (July 1997–June 1998) is generally characterized by El Niño conditions. Likewise, the third period (July 1998–June 1999) is broadly characterized by La Niña conditions. Henceforth, “El Niño” refers to the second period and “La Niña” refers to the third period.
Figure 2 shows the progression of 500-hPa geopotential height anomalies simulated by the Polar MM5 over Antarctica during the study period. Most prominent in this sequence is the large blocking pattern that gradually migrates eastward from north of the Ross Sea in austral spring (SON) 1996 into the Bellingshausen Sea (locations given in Fig. 3) by SON 1997, as the El Niño matures (Fig. 1). This anomaly remains in the Bellingshausen Sea through MAM 1998 and then dramatically flips sign in JJA 1998. This occurs at the same time the SOI begins its abrupt transition from a strong El Niño to a moderate La Niña. The negative anomaly remains throughout the La Niña, and is still present at the end of the study period. This sequence of events, and primarily the large geopotential height anomalies in the eastern South Pacific sector of Antarctica, is relevant to subsequent discussion. The strongest ENSO variability in this study is found in 1) the Ross Sea–Ross Ice Shelf–Marie Byrd Land and 2) The eastern Antarctic Peninsula—Weddell Sea–Ronne/Filchner Ice Shelf regions, which lie within this circulation anomaly. Investigators have noted an ENSO-related dipole relationship between these regions in sea ice extent (SIE) and surface air temperature (SAT) (Yuan and Martinson 2000, 2001; Kwok and Comiso 2002). Kwok and Comiso (2002) extend the relationship to include sea ice concentration, sea/ice surface temperature, and meridional winds. This study examines atmospheric fields at the surface and in the troposphere in relation to the ENSO forcing.
During spring and summer, Renwick (1998) finds that the prominent blocking pattern seen here for the southeast Pacific during the 1997–98 El Niño is present in most of the other ENSO events in the 1980s and 1990s. Renwick observes that the 1982–83 warm event is an exception to this; blocking was slightly suppressed during this episode. Van Woert et al. (2003) note that the atmospheric forcing during the 1990s, including our period of interest, is anomalous with respect to observations during the 1980s. This implies that the results of this study do not necessarily represent a typical ENSO cycle; while an ENSO signal may be consistently strong in a given region, the conditions associated with individual events may vary significantly (e.g., Houseago et al. 1998).
In section 2, we describe the Polar MM5, observational data, and processing methods used. In section 3, we evaluate the quality of the Polar MM5 simulations by comparing model output to automatic weather station (AWS) data, manned observations, radiosonde data, and satellite radiances. Results are presented in section 4 and discussed in section 5. A summary is given in section 6.
2. Data and methods
a. The Polar MM5
The Polar MM5 version 1 (henceforth, PMM5) is based on the nonhydrostatic version 2 of MM5 (Dudhia 1993; Grell et al. 1994). A description of the polar modifications available in PMM5 is given by Bromwich et al. (2001) and Cassano et al. (2001). The model domain (Fig. 3) is a polar stereographic projection consisting of 120 grid points in orthogonal directions centered at the South Pole, with a horizontal resolution of 60 km. The top pressure is set at 10 hPa with a rigid-lid upper boundary condition. A total of 32 sigma levels are used in the vertical; 7 are located within the lowest 400 m of the atmosphere. The lowest sigma level is at a nominal height of 12 m above ground level (AGL). This relatively high resolution near the surface is required to accurately represent the evolution of the shallow katabatic layer over the Antarctic ice sheet (Bromwich et al. 2001). The model topography over the Antarctic continent is interpolated from a modern 5-km resolution digital elevation model for Antarctica (Liu et al. 1999), the so-called RAMP DEM. The regions spanned by the Ronne/Filchner Ice Shelf and Ross Ice Shelf are manually identified from climatic maps. The surface height and land use type for both ice shelves are set to 50 m and permanent ice, respectively.
The 2.5° horizontal resolution European Centre for Medium-Range Weather Forecasts (ECMWF) Tropical Ocean–Global Atmosphere (TOGA) surface and upper-air operational analyses are used to provide the initial and boundary conditions for the model atmosphere. These data are interpolated to the PMM5 model grid using the standard preprocessing programs provided by NCAR for use with the MM5 modeling system. In addition, the 1.125° ECMWF/TOGA global surface analyses are used to specify the initial surface temperature, SST, and subsurface temperature. The daily polar gridded sea ice concentration data with 25-km horizontal resolution obtained from the National Snow and Ice Data Center are used to identify the sea ice surface type and its fractional coverage at each model grid. The PMM5 produces short duration (36 h) simulations of the atmospheric state over Antarctica from July 1996 through June 1999. The model is initialized with the 0000 UTC ECMWF/TOGA analyses for each preceding day, with the 12–36-h PMM5 forecasts used for data analyses (the first 12 h are discarded to account for spinup).
b. Observational data
Table 1 summarizes the observational datasets used in this study. Definitions of acronyms that have not previously been defined may also be found in the table. Additional details concerning these data can be found at the web addresses provided, and some specific points are discussed here.
AWS and manned station data are used to examine the quality of PMM5 simulations at the surface and in the free atmosphere. Model grid points nearest the location of each station are selected for comparison. Locations of the stations are shown in Fig. 3. Modeled surface pressure is adjusted to the reported AWS elevation using the hypsometric equation. Modeled near-surface temperatures and winds are adjusted from the lowest model sigma level (∼14 m) to the nominal AWS instrument heights of 2 m (temperature) and 3 m (winds) using Monin–Obukhov similarity theory (Stull 1988).
Monthly means for the statistical comparison of PMM5 versus AWS and radiosonde observations are taken on a point-by-point basis to account for missing observations; they are compiled from 3- and 6-hourly surface data and 12-hourly radiosonde data. Where seasonal or annual means are taken from monthly means or where the means of several sites are averaged together, these are done using weighted averaging to account for the number of observations taken in a given month, or for a given station. In this manner, the statistics reflect the model performance at synoptic rather than monthly time scales. The statistics compiled are bias (model minus observed), root-mean-square error (rmse), and correlation coefficient. In addition to the statistical comparisons, seasonal anomalies of monthly mean manned and AWS observations, where complete records are available, are plotted on several of the subsequent PMM5 plots for comparison to model data. All seasonal and annual means used in the figures are calculated by linearly averaging monthly means.
Monthly mean surface (skin) temperatures inferred from AVHRR are used to evaluate the spatial distribution of the PMM5 surface (2 m) temperatures over Antarctica. An extensive description and validation of this dataset is given in Comiso (2000). The author compared the satellite-derived surface temperatures with those from 21 surface stations for 1979–98 and found the data to be in good agreement, having a correlation coefficient of 0.98 and a standard deviation of about 3°C. Because infrared observations of the surface are only available during cloud-free periods, the monthly means are cooler than the true monthly mean by ∼0.5 K.
Monthly cloud frequency fields inferred from AVHRR (Key 2002, and references therein) are compared to those from PMM5. Due to the newness of the dataset, relatively little validation has been done. Note that satellite-inferred cloud fraction fields (which would allow direct comparison with model output and manned station observations) are not standard products; hence, cloud frequency (monthly means of cloud mask at 0200 LST) is used. The PMM5 cloud frequency fields are computed from once-daily cloud mask data (0300 UTC). The PMM5 cloud mask is considered present if cloud fraction at a grid point is greater than five-tenths.
3. Evaluation of Polar MM5 simulation quality
Table 2 summarizes the bias, rmse, and correlation coefficient for modeled versus observed surface pressure, temperature, zonal wind speed, and meridional wind speed for winter (JJA) and summer (DJF). The statistics are divided into two categories: above and below 500-m elevation. For those interested, an expanded version of this table showing the statistics for each station is given in the appendix (Table A1); the locations of the stations are shown in Fig. 3. As mentioned above, the JJA and DJF statistics are compiled for 3- and 6-hourly data to demonstrate that the model skill is good over short time scales, and that taking monthly means does not significantly “average out” model deficiencies. However, because this text primarily employs monthly means, the last column presents the correlation of the monthly mean anomalies over the annual cycle, intended to demonstrate the model skill and ability to capture variability over monthly time scales.
High correlations and low biases are characteristic of the surface pressure, indicating that the model is capturing the surface pressure with high skill in both seasons. Pressure biases (large for some individual sites not shown) may in part be influenced by inaccurate elevation measurements for these sites, which are often taken using aircraft altimeters (G. Weidner, AMRC, 2002, personal communication). A negative bias is observed for surface temperature for the winter months and is greatest at high elevations (continental interior). This has been noted in other evaluations of PMM5 and is probably caused by deficient simulated wintertime cloud cover (Guo et al. 2003). Generally small biases are apparent in the wind fields, although at low elevations in winter a negative zonal wind bias of about −3 m s−1 is compensated by a positive meridional wind bias of about +3 m s−1; this may be related to the generally steep, complex topography in the coastal regions interacting with the persistent wintertime katabatic wind regime. The rmses for all variables are highest in winter, which is likely due to greater atmospheric variability in these months (Das et al. 2002). Examination of the correlations of monthly mean anomalies over the annual cycle (last column) indicates that the model is capturing surface pressure (synoptic variability) on monthly time scales with high skill (correlations are nearly unity for high and low elevations). At monthly time scales, correlations between modeled and observed temperatures and winds are at or above 0.73 for all but the low-elevation meridional winds, which are likely degraded by the model coastal topography at 60-km resolution.
Table 3 is similar to Table 2, but for 500-hPa geopotential height, temperature, dewpoint, and wind speed for the mean of seven sites. An expanded version of this table showing the statistics for each station is given in the appendix (Table A2); the locations of the stations are shown in Fig. 3. As with surface pressure, PMM5 shows generally high skill in simulating geopotential height at synoptic timescales, and correlations are nearly unity for monthly time scales (final column). A negative height bias is present during winter but absent during summer. This is in part due to the large wintertime surface cold bias noted in Table 2. The model shows good skill in capturing the variability of the 500-hPa temperature, especially when correlated over the monthly mean anomalies (0.89). The model generally has a positive (slightly moist) dewpoint bias at 500 hPa, especially in summer, but large rmses indicate that these fields are highly variable, which may be partly due to the small amount of moisture in the air at this level of the atmosphere. A slight negative bias is present in the modeled 500-hPa wind during summer. Interestingly the skill of the PMM5 in simulating the atmosphere at 500-hPa is lowest over Amundsen–Scott, McMurdo, and Dumont D'Urville (appendix, Table A2). The lower model skill in this region appears to be a result of the model not fully capturing the correct magnitude and placement of synoptic forcing; if synoptic gradients are not properly depicted, the model will oversimulate or undersimulate the winds, which in turn affects the advection of moisture, temperature, and other variables.
Figure 4 compares the average surface temperature over Antarctica from July 1996 to June 1999 for PMM5 (Fig. 4a) to that inferred from satellite infrared radiances (Fig. 4b). The cold bias in PMM5 is evident, especially over the continental interior. For example, the 220-K isotherm covers a larger area in the PMM5 temperature plot. Many features are resolved: the tongue of cool air extending over the South Pole into West Antarctica, the temperature distribution over the Amery Ice Shelf, and the placement of the 270-K isotherm surrounding the continent, which closely follows the mean annual sea ice edge (Gloersen et al. 1992). The PMM5 temperature minimum is within ∼0.6 K of the satellite-inferred minimum over East Antarctica. In general, the spatial distribution of the PMM5 surface temperature closely follows that of the satellite-derived temperature both qualitatively and quantitatively.
Figure 5 compares percent change in cloud frequency between the La Niña (DJF 1998–99) and El Niño (DJF 1997–98) summers for the PMM5 with those inferred from AVHRR satellite radiances (Key 2002). The percent change is taken from the summer that has the lower cloud frequency to the summer with the higher cloud frequency; this method of normalization gives a better quantitative estimate of change than normalizing the data to values between zero and one, and allows for regions that have less cloud cover climatologically (i.e., the interior) to be accounted for. This period is chosen because it reflects high interannual variability in the PMM5 output. Similar features are resolved in both; the largest anomalies appear to reflect a wavenumber 2 pattern encircling the continent. During the El Niño episode, more clouds are present over the Ross Ice Shelf and West Antarctica, the South Pole, and East Antarctica from ∼30° to 90°E. Fewer clouds are present over the Ronne/Filchner Ice Shelf, the Antarctic Peninsula, and over East Antarctica from ∼90° to 140°E. In addition, fewer clouds are generally present over the entire Southern Ocean surrounding Antarctica, although this is difficult to discern in Fig. 5a due to the shading scale. The PMM5 anomalies, both positive and negative, are more amplified, especially over land and permanent ice; this may be partially due to the difficulty of distinguishing clouds from the cold Antarctic surface in the satellite-inferred data (i.e., it nearly always appears to be overcast over these areas) (e.g., Key 2002, and references therein).
In conclusion, the PMM5 appears to be accurately capturing the monthly variability from July 1996 to June 1999. The pressure and geopotential fields are especially well simulated, indicating that the synoptic variability is well represented. The best performance is in the free atmosphere where the direct effects of terrain are minimal.
4. Results
a. Spatial anomalies
Figure 6 shows the July 1997–June 1998 (El Niño) minus July 1998–June 1999 (La Niña) surface pressure anomaly over the PMM5 domain. Station observations are plotted for comparison to modeled fields, and are generally in good agreement. For example, at Rothera, an observed surface pressure anomaly of +8.8 hPa corresponds well with a modeled anomaly of about +9 hPa. The negative anomaly over the Ross Ice Shelf is present in the observations, but appears to be overestimated by the PMM5 by about −2 hPa. This is most likely due to a low-level warm bias over the Ross Ice Shelf during the coldest months, which is discussed in detail below. Note the similarity to the blocking pattern in the Bellingshausen Sea shown for the 500-hPa geopotential height anomaly in Fig. 2 (the four frames spanning JJA 1997 through MAM 1998), indicating a consistent synoptic situation from the surface through the troposphere; the anomaly at 300 hPa is also similar (not shown). This blocking feature implies enhanced onshore flow over West Antarctica from approximately 90° to 150°W and offshore flow over the Antarctic Peninsula and Weddell Sea regions during the El Niño episode. The most prominent low pressure anomalies are over the Ross Sea/Ice Shelf and to the northeast of the Weddell Sea, centered at ∼15°W, implying enhanced cyclone activity or cyclone intensity in these areas during the El Niño episode.
Figure 7 shows the MAM 1997 (El Niño) minus MAM 1999 (La Niña) surface temperature anomaly. The autumn months are chosen because they reflect the strongest interannual variability in surface temperature throughout the study period. Note that MAM 1998 is not used because it occurs during the abrupt transition of the SOI from negative to positive (Fig. 1), and thus does not show significant variability. The most prominent feature is the warmer temperatures over the Ross Ice Shelf and Marie Byrd Land during the El Niño episode, due to enhanced onshore flow during March and May 1997 (this is not obvious in Fig. 2 for MAM 1997 because of an offset in the flow pattern during April 1997). Over the Ronne/Filchner Ice Shelf and in the Weddell Sea, cooler temperatures are observed for the same period. Large warm anomalies are also present near Novolazarevskaja (∼15°E) and between Mirnyj and Casey (∼100°E), and a large cool anomaly is to the east of Dumont D'Urville (∼150°E). These are supported by the circulation anomalies in Fig. 2. The observations in Fig. 7 indicate that PMM5 is capturing the spatial distribution and sign of the anomalies with good skill over most of the model domain. Over the Weddell Sea and Antarctic Peninsula, the magnitude of the observed versus modeled anomalies are in good agreement. The PMM5 appears to be overestimating the magnitude of the negative anomaly over the interior of East Antarctica from ∼10°–70°E. The magnitude of the positive anomaly over the interior of East Antarctica from ∼70°–140°E is also somewhat overestimated, and this “tongue” of warm air should perhaps extend further eastward and inland according to the observations at Dumont D'Urville, Dome C, and LGB35. The most marked differences between observed and modeled values are in the large positive anomaly over the Ross Ice Shelf. The observations indicate that this anomaly definitely exists, but perhaps is about +5 K at its maximum, compared to about +11 K in the model. This discrepancy is examined in detail below.
Figure 8 shows the ECMWF/TOGA (see Table 1 for description of this dataset) average temperature anomalies over the Ross Ice Shelf and Ronne/Filchner Ice Shelf at 2 m, 850 hPa, 700 hPa, and 500 hPa. The initial and boundary conditions for the PMM5 are derived from ECMWF/TOGA. It is immediately apparent that the coldest months (February/March through August/September) have the largest temperature anomalies at the surface and that an abrupt switch from positive to negative occurs sometime in early 1998. This is due to a change in the treatment of the ice shelves in the ECMWF/TOGA analyses. Prior to 1 April 1998, the ice shelves were treated as permanent 2-m-thick sea ice. As discussed in Renfrew et al. (2002) this is much different from the reality of ice shelves, which are several hundred meters thick. They noted that for the same ocean–atmosphere temperature difference, 1-m-thick sea ice conducts 100 times more heat than 100-m-thick sea ice. Over the annual cycle, they found that ECMWF/TOGA was about 10.2 K warmer than AWS observations over the Ronne/Filchner Ice Shelf and the largest biases occurred at cooler temperatures (i.e., late autumn, winter, early spring). On 1 April 1998, this issue was resolved in the ECMWF/TOGA analysis, which the authors find considerably reduced the bias. The influence of this change on the anomalies is not detectable in the summer months due to a smaller temperature gradient between the air above and the (supposed) water below the 2-m-thick ice. Also, there is decreased sea ice extent and likely more warm air advection onto the ice shelves in summer. The temperature anomalies at 850, 700, and 500 hPa (Fig. 8) do not appear to be affected by the treatment of the ice shelves in any season, indicating that any artificial biases are confined to the inversion layer in the cold months. As a result, the authors do not feel the large-scale forcing in the PMM5 (in which the ice shelves are treated as permanent ice) has been significantly altered by this bias (or its correction) in the ECMWF/TOGA initial conditions. This was confirmed by examining similar time series to those in Fig. 8, but for surface pressure and geopotential height at 850, 700, and 500 hPa (not shown). The anomalies indicate that there is an artificial negative contribution to the pressure bias over the ice shelves in the cold months due to their treatment as 2-m-thick sea ice in ECMWF/TOGA, which makes the PMM5 surface pressure anomalies over the Ross Ice Shelf and Ronne/Filchner Ice Shelf (Fig. 6) too low by about 2 hPa; however, the influence is not seen in the geopotential height at 850 hPa and above.
Similar to Fig. 7, MAM 1997 minus MAM 1999 temperature anomalies were plotted for ECMWF/TOGA and for two AVHRR-inferred temperature datasets from Comiso (2000) and Key (2002) (not shown). In the ECMWF/TOGA plot, the corresponding temperature anomaly over the Ross Ice Shelf has a maximum magnitude of +22 K. Considering the PMM5 data is based on a series of short forecasts, which are reinitialized every 24 h by ECMWF/TOGA, the PMM5 partially, but not completely, adjusts for the artificial temperature biases induced by the ECMWF/TOGA initial conditions, resulting in a much smaller maximum of 11 K over the Ross Ice Shelf for the same period. Along with the observations, the satellite-inferred data confirm that the warm bias over the Ross Ice Shelf does exist. However, these data imply that its maximum magnitude is most likely to be 5–8 K. Likewise, they indicate the cold bias in the Weddell Sea probably extends inland over the Ronne/Filchner Ice Shelf. The maximum magnitude of this bias as depicted in PMM5 is probably close to the “truth” (about −5 K near Halley).
Figures 9 and 10 show the percent change in PMM5 cloud fraction and precipitation, respectively; between the La Niña (DJF 1998/99) and El Niño (DJF 1997/98) summers (from the summer that has the lower cloud fraction or precipitation to the season with the higher cloud fraction or precipitation); this is the same method of normalization described for Fig. 5. The summer months are chosen because they display strong interannual variability over the dipole region. This appears to be due to the magnitude and location of the blocking feature in the Bellingshausen Sea, which is oriented most favorably for anomalous flow into (out of) the Marie Byrd Land (Weddell Sea) region during DJF 1997/98 (DJF 1998/99), and vice versa (Fig. 2). Favorable blocking is also present during spring in these two years, but less variability in clouds and precipitation is observed (not shown). It is speculated that this is due to a greater extent of sea ice during spring, which may dampen the signal. Additionally, summer was chosen because it is barely affected by the low-level warm bias resulting from the treatment of the ice shelves as 2-m-thick sea ice in the ECMWF/TOGA initial conditions (discussed above). Over the Ross Ice Shelf and Marie Byrd Land cloud fraction is more than 200% greater in the El Niño summer than the La Niña summer. Conversely, over the Ronne/Filchner Ice Shelf and Coats Land cloud fraction is 100%–150% less in the El Niño summer. Other notable anomalies in El Niño cloud fraction occur over East Antarctica from ∼30° to 90°E (positive), over East Antarctica from ∼90° to 150°E (negative), and over the South Pole (positive). Only a few observations are available in areas of high variability. At Vostok, the observed change in clouds is −69% compared to about −85% in the simulations; at McMurdo, 21% compared to ∼25% in the simulations; and at the South Pole, 11% compared to ∼45% in the simulations. The cloudiness trends simulated by PMM5 generally follow those that are observed, consistent with the comparison to satellite-inferred cloud frequency shown in Fig. 5. In the areas of greatest variability, the magnitude of change appears to be overestimated, although the degree is uncertain due to the lack of observations.
Enhanced summer precipitation occurs over the Ross Sea/Ross Ice Shelf and Marie Byrd Land extending inland to the South Pole during the El Niño episode (Fig. 10), in excess of 400% more than the La Niña summer over much of the area. Decreased precipitation occurs over the Ronne/Filchner Ice Shelf and Coats Land, in excess of 400% less than the La Niña summer in some areas. Similarly, Guo (2003) found that the 12-month centered running means of precipitation between these two regions is significantly anticorrelated (r ≈ 0.45) over the 1980s and 1990s using a variety of datasets. Other large minima are located along the East Antarctic coastline near the Amery Ice Shelf (∼70°E) and Dumont D'Urville (∼120°–150°E). Few observations are currently available for comparison to the PMM5 simulations; it is anticipated that accumulation data from the International Trans Antarctic Scientific Expedition will be available in the future. However, comparison of Figs. 9 and 10 indicates these precipitation trends correspond closely to the cloud trends in the areas of highest variability, which are shown to be in broad agreement with observed and satellite-inferred clouds.
Figure 11 shows the July 1997–June 1998 (El Niño) minus July 1998–June 1999 (La Niña) surface meridional wind anomaly. This period is chosen because it reflects the strong interannual wind variability that is present throughout all seasons of the ENSO cycle and because it corresponds to the circulation anomalies for the same period shown in Fig. 6. Greater onshore flow occurs during the El Niño episode from the eastern Ross Sea east to about 80°W and greater offshore flow occurs from about 80°W east across the Weddell Sea, extending northward into the South Atlantic. This is consistent with the large blocking pattern in the Bellingshausen Sea in Fig. 6. Greater offshore flow occurs in the western Ross Sea/Ross Ice Shelf against the Transantarctic Mountains due the low pressure anomaly in the central Ross Sea/Ross Ice Shelf implying greater cyclone activity that would induce more frequent barrier flow along the mountains (O'Connor et al. 1994). The variability is not as pronounced in East Antarctica, but does follow a marked circumpolar wavelike spatial pattern of alternating onshore/offshore anomalies extending nearly to the South Pole. Fewer observations are available in these plots because of the difficulty in obtaining 24 complete months of wind observations from AWSs (most of the observations shown are from manned stations). In the area of high variability between ∼20° and 70°W, the observed and modeled meridional winds agree well qualitatively, and to some degree quantitatively, at Rothera (1.8 m s−1 observed compared to ∼1.3 m s−1 simulated), Bellingshausen (0.7 m s−1 compared to ∼2.5 m s−1), and Halley (1 m s−1 compared to 0.9 m s−1). Over the rest of Antarctica, the observations are generally in agreement with the simulations as well. An exception to this is at Dumont D'Urville (−0.5 m s−1 observed compared to ∼+0.3 m s−1 modeled) and Swithinbank (+0.3 m s−1 compared to about −0.4 m s−1). However, at Marilyn, which lies between these two stations, the observed and modeled winds are in better agreement (+0.8 m s−1 compared to ∼1.2 m s−1).
b. Temporal anomalies
Figure 12 shows the July 1996–June 1999 seasonal anomalies of several variables averaged over the Marie Byrd Land (MBL: 130°–160°W, 75°–85°S) and Weddell Sea-Ronne/Filcher Ice Shelf (WRF: 40°–70°W, 70°–80°S) sectors (Fig. 3). These are chosen because they are regions of generally high interannual variability in the model output, as demonstrated in Figs. 6–7 and 9–11. Moreover, they have nearly opposite responses to ENSO forcing throughout the study period. As mentioned above, this phenomenon is associated with the “Antarctic Dipole” (Yuan and Martinson 2000, 2001). The seasonal surface temperature anomalies (Fig. 12a) indicate warmer (cooler) than normal temperatures in the MBL (WRF) sector during the El Niño episode, and the reverse during La Niña. The largest variability occurs in the spring and autumn months in both regions. The meridional winds (Fig. 12b) are primarily onshore (offshore) in the MBL (WRF) sector during El Niño and opposite in the La Niña year; this is consistent throughout the troposphere, for example, at 500 hPa (Fig. 12c). The largest seasonal variability occurs in springtime in the WRF sector and is considerable in all El Niño and La Niña seasons throughout the study period for both sectors. Cloud fraction (Fig. 12d) and P minus E (Fig. 12e) are closely related. Increased (decreased) cloudiness and precipitation occur in the MBL (WRF) sector during the El Niño year. The opposite occurs in the La Niña year. The largest anomalies occur in spring and summer. Overall, Fig. 12 demonstrates that strong ENSO variability, of opposite sign, is present in the MBL and WRF regions.
One factor in choosing the spatial plots in section 4a was to demonstrate the most significant anomalies between El Niño and La Niña seasons. The shaded areas in Fig. 12 correspond to months for which the figures in section 4a are plotted. In general, the seasons chosen reflect the highest interseasonal (or interannual) variability around the continent, and this is reflected in the MBL and WRF regions. Other factors, such as the low-level temperature bias over the ice shelves in winter motivated the decision to use the autumn months for surface temperature and summer months for the cloud fraction and precipitation fields (even though strong ENSO variability is indicated year-round in Figs. 12d and 12e for the latter fields).
5. Discussion
a. Blocking in the southeast Pacific
1) Mechanism
Figures 6–7 and 9–11 indicate the strongest interannual variability in Antarctica during the study period occurs in the MBL and WRF regions, and Fig. 12 clearly shows the dipole relationship between the two regions. The Antarctic dipole (ADP) is an interannual variance structure in SIE and SAT organized as a quasi-stationary wave and characterized by an out-of-phase relationship between the ice and temperature anomalies in the central/eastern Pacific and Atlantic sectors of the Antarctic (Yuan and Martinson 2001). They show that sea level pressure (SLP) is more closely related to ENSO forcing than SAT (the leading EOF mode eigenvector of SLP anomaly associated with ENSO explains 28% of the variance, compared to 18% for SAT). This SLP variance is located in the Amundsen and Bellingshausen Seas (between the dipole regions), in the same area where the large circulation anomaly that is generally present in contemporary ENSO warm events is located. The feature may be part of a wave train of circulation anomalies extending from the subtropical and extratropical Pacific that generally resembles the Pacific–South American (PSA) pattern (e.g., Ghil and Mo 1991) and appears to be forced by ENSO. Several authors have noted PSA or “PSA-like” wave trains extending to high southern latitudes and detected at all levels of the troposphere (e.g., Karoly 1989; Houseago et al. 1998; Renwick 1998 and references therein; Renwick and Revell 1999; Harangozo 2000; Cai and Watterson 2002). This mechanism merits further discussion.
Figure 13a shows the 500-hPa geopotential height anomalies for all DJF El Niños minus all DJF La Niñas for the period 1979–99, from the National Centers for Environmental Prediction (NCEP)-2 dataset (see Table 1). Similarly, Fig. 13b shows the anomalies for the DJF 1997/98 El Niño minus the DJF 1998/99 La Niña (note that other seasons are similar to DJF). El Niño and La Niña seasons were defined following Kwok and Comiso (2002). El Niño summers occurred when the DJF mean of the 3-month running mean of the SOI < −1; La Niña summers were defined similarly for SOI > 0. Using this convention, five El Niño summers (DJF 1982/83, 1986/87, 1991/92, 1992/93, 1997/98) and seven La Niña summers (DJF 1981/82, 1983/84, 1984/85, 1988/89, 1995/96, 1996/97, 1998/99) were defined. Inspection of both plots reveals prominent blocking in the southeast Pacific Ocean (Amundsen/Bellingshausen Seas) that appears to be part of a wave pattern originating to the east of Australia and propagating southeastward.1 It is also evident that the signal is amplified for the late 1990s ENSO cycle; for example, the center of the positive anomaly in the southeast Pacific is about 150 gpm in the late 1990s, versus the mean value of about 90 gpm for 1979–99. Considering that the origin of this teleconnection appears to be related to tropical forcing, it is instructive to examine the ENSO modulation of tropical convection.
Figure 14 is temporally the same as Fig. 13, but for top-of-atmosphere outgoing longwave radiation (OLR; see Table 1 for details of dataset) between 40°N and 40°S; in the Tropics, this is a proxy of convection (e.g., Chelliah and Arkin 1992). Noting that negative OLR anomalies indicate enhanced tropical convection, the typical ENSO modulation of convection is seen, with enhanced convection in the central equatorial Pacific during warm events, and suppressed convection in the Indo–Pacific. Comparing Figs. 14a and 14b, it is noteworthy that convection in the central Pacific is stronger and covers a broader east–west region in the late 1990s compared to the climatology. Even more prominent is the suppressed convection in the Indo–Pacific, which is much stronger in the late-1990s ENSO forcing than for the mean of 1979–99. This suggests that the amplified wave train (and subsequent blocking in the southeast Pacific) observed in the late 1990s (Fig. 13b) is due to enhanced, ENSO-modulated tropical convection (suppression) in the central (Indo–) Pacific.
2) Placement
Comparison of Figs. 13a and 13b also indicates an eastward shift of the blocking pattern in the southeast Pacific from about 130°W for the mean of the 1979–99 events to about 90°W in the late 1990s. Likewise, inspection of Figs. 14a and 14b indicates that the ENSO-modulated anomalies of tropical convection in the equatorial Pacific are also shifted eastward from about 170°W for the mean of the 1979–99 events to about 155°W in the late 1990s. This suggests that the eastward shift in the blocking pattern in the late 1990s is in part due to an eastward shift in ENSO-modulated tropical convection; the position of this anomaly is examined in greater detail here.
Chen et al. (1996), in a study of the 1986–89 ENSO cycle, show that the southern branch of the New Zealand split jet (the polar front jet; PFJ) is modulated by eddy momentum fluxes from both high and low latitudes, and that the PFJ is weakened (enhanced) during warm (cold) ENSO events. They note that the fluctuation of the PFJ exerts an influence on the position of the Amundsen Sea low. The position of the Amundsen Sea low is closely related to the anomaly discussed here, which Connolley (1997) termed the “West Antarctic Pole of Variability” (PV), expressed as the standard deviation of the mean long-term SLP field located in the Amundsen Sea and extending into the Bellingshausen and Ross Seas. Marshall and King (1998) show that the variability of the PV is related to shifts in the phase of the New Zealand split jet system, analogous to the results of Chen et al. (1996). It is inferred from these findings, then, that the position of the PV is influenced from both low and high latitude forcing mechanisms that are manifested as fluctuations in the PFJ. With respect to high-latitude forcing, Lachlan-Cope et al. (2001) conduct atmosphere-only GCM experiments in which axisymmetric sea ice, sea surface temperatures, and orography are imposed at high southern latitudes and find that the nonaxisymmetric nature of the Antarctic orography is necessary to maintain the PV (this is consistent with the earlier findings of Connolley 1997).
Connolley (1997) suggests the PV is centered at 65°–70°S, ∼120°W, about the same position as for the “normal” ENSO position in Fig. 13a. In this study, an eastward shift in the PV to about 90°W may reflect a stronger ENSO component than for the climatological mean (i.e., Figs. 13 and 14). It is interesting that this is adjacent to the ridge in the Antarctic topography that extends from the South Pole to the coast along the 90°W meridian (Fig. 3). The spatial variation of the meteorological fields presented (i.e., Figs. 7, 9, and 10) is roughly constrained by this ridge. Collectively, these observations raise the question as to whether the low-latitude (ENSO) and high-latitude (topographic) forcing mechanisms that appear to modulate the location of the PV at 90°W are acting independently.
b. ENSO variability in precipitation over West Antarctica
Cullather et al. (1996) observe an abrupt change in the correlation of the SOI with West Antarctic precipitation [the moisture flux convergence (MFC) averaged from 75° to 90°S, 120°–W-180° using ECMWF/TOGA operational analyses] from positive in the late 1980s to negative after 1990 (through 1994). Genthon and Krinner (1998), using ECMWF 15-yr reanalysis data (ERA-15), do not find the abrupt switch in the SOI–MFC correlation after 1990 that Cullather et al. (1996) obtain using ECMWF/TOGA operational analyses. In response, Bromwich et al. (2000) extend the ECMWF/TOGA record to 1999 and compare the differences between this dataset and the ERA-15 precipitation. They find that the ERA-15 signal is damped due to errors in the elevation of Vostok used to assimilate surface pressure into ERA-15, which causes large geopotential height biases. They conclude that the ECMWF/TOGA operational analyses more accurately depict the interannual variability in precipitation before and after 1990. More recently, Guo et al. (2004) examine several precipitation datasets over the last 2 decades, including NCEP-2, (1979–2000; Table 1), the ECMWF/TOGA operational analysis (1991–2000), ERA-15 (1979–93), and modeled precipitation from a dynamical retrieval method applied to ECMWF/TOGA and ERA-15 (Bromwich et al. 2004). The weight of the available evidence supports the findings of Cullather et al. (1996) and Bromwich et al. (2000), namely that a shift from a positive correlation in the mid-to-late 1980s to a strong negative correlation in the 1990s is present between West Antarctic precipitation and the SOI.
The use of multiple data sources by Guo et al. (2004) helps to reduce the impacts of errors in the individual datasets; however, uncertainties are present in all of the precipitation syntheses due to the absence of more comprehensive observations and ice core records. The PMM5 simulations in this study span a comparatively short time; however, they closely follow the precipitation trends in the ECMWF operational analyses (Fig. 15). It may be speculated that this is a residual effect of deriving the initial fields for PMM5 from ECMWF/TOGA. However, it is shown in section 4 that the PMM5 shows considerable adjustment from the initial ECMWF/TOGA temperature fields, which are biased over the ice shelves during winter. In addition, the ECMWF/TOGA temperature bias is confined to the low levels, indicating that the predicted PMM5 precipitation is not significantly influenced by the ECMWF/TOGA initial temperature fields (note that there is no initial precipitation field). Taking into account that the PMM5 precipitation compares very similarly to the ECMWF/TOGA precipitation over West Antarctica (Fig. 15) and that comparisons to observations and satellite data suggest the PMM5 is capturing the interannual variability associated with many precipitation-related fields, it is inferred that the record of ECMWF P-E over West Antarctica, at least in the late 1990s, is realistic.
c. Why is the temperature signal over the ice shelves so large?
Marked temperature anomalies of opposite sign are present over the MBL and WRF regions between autumn El Niño and La Niña years (Fig. 7), and this is present throughout the ENSO cycle (Fig. 12a). While the anomalies are amplified in the PMM5 (mainly in winter) due to the ECMWF/TOGA initial conditions (discussed above), the observations and satellite-inferred data indicate that these do exist and are the most prominent anomalies in the Antarctic. Sharp gradients exist along the ice shelf–land transition over both ice shelves; over the Ross Ice Shelf, a steep gradient is present along the ice shelf–ocean edge as well. While the shelves serve as the major outlets for atmospheric mass from the cold, dry interior of Antarctica (e.g., Parish and Bromwich 1987), their relatively low elevation with respect to the continental interior makes them more sensitive to maritime forcing as well (e.g., Bromwich et al. 1994). This implies that a high degree of interannual variability can be expected over the ice shelves and is most likely dependent on whether the large-scale, low-level synoptic forcing is primarily onshore or offshore. The geographic position of the shelves, both located in/near the MBL and WRF dipole regions, is favorable for this variability in the meridional flow. The 500-hPa height anomalies (Fig. 2), which are similar to the low-level circulation anomalies (not shown), suggest primarily onshore (offshore) forcing component during the period of warm (cool) anomalies over both ice shelves. Over the MBL region the surface manifestation of the circumpolar vortex, whose climatological center is over the northeast Ross Sea, may additionally enhance the onshore flow (Bromwich et al. 2000). This may in part be why the largest variability in temperature, cloud fraction, and precipitation is observed in the MBL region during this episode.
6. Summary
This study employs the Polar MM5 to examine the ENSO modulation of Antarctic climate for July 1996–June 1999, which is particularly strong compared to the period from 1979–99. This appears to be largely due to an eastward shift and enhancement of convection in the tropical Pacific Ocean. The results, which demonstrate large variability associated with ENSO, emphasize the importance of identifying the magnitude and role of ENSO in Antarctic climate variability, especially in light of recent and dramatic changes over the continent during a trend toward more frequent warm events. The strongest ENSO forcing is observed over the Ross Ice Shelf–Marie Byrd Land and over the Weddell Sea–Ronne/Filchner Ice Shelf. In addition to having the largest climate variability associated with ENSO, these two regions exhibit dipolar anomalies throughout the study period, which supports and extends the results of previous investigators (Yuan and Martinson 2000, 2001; Kwok and Comiso 2002). This dipolar relationship is observed in surface temperature, meridional winds, cloud fraction, and precipitation. The ENSO-related variability appears to be primarily controlled by the large-scale circulation anomalies surrounding the continent, which are consistent throughout the troposphere. When comparing the El Niño/La Niña phases of this late-1990s ENSO cycle, the circulation anomalies are nearly mirror images over the entire Antarctic, indicating their significant modulation by ENSO. Large temperature anomalies, especially in autumn, are prominent over the major ice shelves; this is most likely due to their relatively low elevation with respect to the continental interior making them more sensitive to shifts in synoptic forcing offshore of Antarctica.
The PMM5 precipitation simulations are found to be in good agreement with those of Cullather et al. (1996) and Bromwich et al. (2000), who observe an abrupt change in the correlation of the SOI with West Antarctic precipitation (using ECMWF/TOGA) from positive in the late 1980s to negative in the 1990s. Comparisons to observations and satellite data suggest the PMM5 is capturing the interannual variability in many precipitation-related fields; from this it is inferred that the precipitation record of ECMWF/TOGA over West Antarctica, at least in the late 1990s, is realistic.
It is stressed that this is a case study of the ENSO modulation in the late 1990s, which appears to be especially strong. While recent ENSO warm/cold events may have had similar signals over Antarctica, long-term records indicate significant variability in the climatic response to ENSO forcing. The PMM5 captures ENSO-related variability reliably in this study, and in the future it may be practical to apply PMM5 to the 1980s and 1990s (or longer) to examine the reversal in correlation of West Antarctic precipitation and the SOI. This would provide a high-resolution, continuous record from a model adapted for Antarctic conditions. It is envisioned that the model could be initialized using the new ECMWF 40-yr reanalysis (ERA-40), which is expected to benefit from the lessons learned from deficiencies in current reanalyses.
Acknowledgments
This research was supported by NSF Grant OPP-9725730, UCAR Subcontract SO1-22961, and NASA Grant NAG5-9518. Data were obtained from the Antarctic Meteorological Research Center and the Cooperative Institute for Meteorological Satellite Studies at the University of Wisconsin—Madison, the Arctic and Antarctic Research Center at Scripps Institution of Oceanography, the British Antarctic Survey, NASA Goddard Space Flight Center, the National Center for Atmospheric Research, the European Centre for Medium-Range Weather Forecasts, and the National Weather Service's Climate Prediction Center. Thanks go to Sheng-Hung Wang for processing the data for Fig. 8.
REFERENCES
Bromwich, D. H., Y. Du, and T. R. Parish, 1994: Numerical simulation of winter katabatic winds from West Antarctica crossing Siple Coast and the Ross Ice Shelf. Mon. Wea. Rev., 122 , 1417–1435.
Bromwich, D. H., A. N. Rogers, P. Kallberg, R. I. Cullather, J. W. C. White, and K. J. Kreutz, 2000: ECMWF analyses and reanalyses depiction of ENSO signal in Antarctic precipitation. J. Climate, 13 , 1406–1420.
Bromwich, D. H., J. J. Cassano, T. Klein, G. Heinemann, K. M. Hines, K. Steffen, and J. E. Box, 2001: Mesoscale modeling of katabatic winds over Greenland with the Polar MM5. Mon. Wea. Rev., 129 , 2290–2309.
Bromwich, D. H., Z. Guo, L. Bai, and Q-S. Chen, 2004: Modeled Antarctic precipitation. Part I: Spatial and temporal variability. J. Climate, in press.
Cai, W., and I. G. Watterson, 2002: Modes of interannual variability of the Southern Hemisphere circulation simulated by the CSIRO climate model. J. Climate, 15 , 1159–1174.
Cassano, J. J., J. E. Box, D. H. Bromwich, L. Li, and K. Steffen, 2001: Verification of Polar MM5 simulations of Greenland's atmospheric circulation. J. Geophys. Res., 106 , 13867–13890.
Chelliah, M., and P. A. Arkin, 1992: Large-scale interannual variability of outgoing longwave radiation anomalies over the global Tropics. J. Climate, 5 , 371–389.
Chen, B., S. R. Smith, and D. H. Bromwich, 1996: Evolution of the tropospheric split jet over the South Pacific Ocean during the 1986–89 ENSO cycle. Mon. Wea. Rev., 124 , 1711–1731.
Comiso, J. C., 2000: Variability and trends in Antarctic surface temperatures from in situ and satellite infrared measurements. J. Climate, 13 , 1674–1696.
Connolley, W. M., 1997: Variability in annual mean circulation in southern high latitudes. Climate Dyn., 13 , 745–756.
Connolley, W. M., and S. P. O'Farrell, 1998: Comparison of warming trends over the last century from three coupled models around Antarctica. Ann. Glaciol., 27 , 565–570.
Cullather, R. I., D. H. Bromwich, and M. L. Van Woert, 1996: Interannual variations in Antarctic precipitation related to El Niño–Southern Oscillation. J. Geophys. Res., 101 , 19109–19118.
Dai, A., I. Y. Fung, and A. D. Del Genio, 1997: Surface observed global land precipitation variations during 1900–1988. J. Climate, 10 , 2943–2962.
Das, S. B., R. B. Alley, D. B. Reusch, and C. A. Shuman, 2002: Temperature variability at Siple Dome, West Antarctica, derived from ECMWF re-analyses, SSM/I and SSMR brightness temperatures and AWS records. Ann. Glaciol., 34 , 106–112.
de la Mare, W. K., 1997: Abrupt mid-twentieth-century decline in Antarctic sea ice extent from whaling records. Nature, 389 , 57–60.
Doran, P. T., and Coauthors, 2002: Antarctic climate cooling and terrestrial ecosystem response. Nature, 415 , 517–520.
Dudhia, J., 1993: A nonhydrostatic version of the Penn State–NCAR mesoscale model: Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev., 121 , 1493–1513.
Genthon, C., and G. Krinner, 1998: Convergence and disposal of energy and moisture on the Antarctic polar cap from ECMWF reanalyses and forecasts. J. Climate, 11 , 1703–1716.
Ghil, M., and K. Mo, 1991: Intraseasonal oscillations in the global atmosphere. Part II: Southern Hemisphere. J. Atmos. Sci., 48 , 780–790.
Gloersen, P., W. J. Campbell, D. J. Cavalieri, J. C. Comiso, C. L. Parkinson, and H. J. Zwally, 1992: Arctic and Antarctic sea ice, 1978–1987: Satellite passive-microwave observations and analysis. National Aeronautics and Space Administration NASA SP-511, Washington, DC, 290 pp.
Grell, G. L., J. Dudhia, and D. R. Stauffer, 1994: A description of the fifth-generation Penn State/NCAR mesoscale model (MM5). NCAR Tech. Note NCAR/TN-398 + STR, 117 pp.
Guo, Z., 2003: Spatial and temporal variability of modern Antarctic precipitation. Ph.D. dissertation, The Ohio State University, 150 pp. [Available from Byrd Polar Research Center, 1090 Carmack Rd., Columbus, OH 43210.].
Guo, Z., D. H. Bromwich, and J. J. Cassano, 2003: Evaluation of Polar MM5 simulations of Antarctic atmospheric circulation. Mon. Wea. Rev., 131 , 384–411.
Guo, Z., D. H. Bromwich, and K. M. Hines, 2004: Modeled Antarctic precipitation. Part II: ENSO modulation over West Antarctica. J. Climate, in press.
Harangozo, S. A., 2000: A search for ENSO teleconnections in the West Antarctic Peninsula climate in austral winter. Int. J. Climatol., 20 , 663–679.
Hines, K. M., D. H. Bromwich, and G. J. Marshall, 2000: Artificial surface pressure trends in the NCEP/NCAR reanalysis over the Southern Ocean and Antarctica. J. Climate, 13 , 3940–3952.
Houghton, J. T., Y. Ding, D. J. Griggs, M. Noguer, P. J. van der Linden, X. Dai, K. Maskell, and C. A. Johnson, Eds.,. 2001: Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, 881 pp.
Houseago, R., G. R. McGregor, J. C. King, and S. A. Harangozo, 1998: Climate anomaly wave-train patterns linking southern low and high latitudes during the South Pacific warm and cold events. Int. J. Cliamtol., 18 , 1181–1193.
Jacobs, S. S., and J. C. Comiso, 1997: A climate anomaly in the Amundsen and Bellingshausen Seas. J. Climate, 10 , 697–711.
Jones, P. D., 1995: Recent variations in mean temperature and the diurnal temperature range in the Antarctic. Geophys. Res. Lett., 22 , 1345–1348.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437–471.
Karoly, D. J., 1989: Southern Hemisphere circulation features associated with El Niño–Southern Oscillation events. J. Climate, 2 , 1239–1252.
Key, J., 2002: The Cloud and Surface Parameter Retrieval (CASPR) system for Polar AVHRR data user's guide. Space Science and Engineering Center, University of Wisconsin, Madison, WI, 62 pp.
Kwok, R., and J. C. Comiso, 2002: Southern Ocean climate and sea ice anomalies associated with the Southern Oscillation. J. Climate, 15 , 487–501.
Lachlan-Cope, T. A., W. M. Connolley, and J. Turner, 2001: The role of the non-axisymmetric Antarctic orography in forcing the observed pattern of variability of the Antarctic climate. Geophys. Res. Lett., 28 , 4111–4114.
Liu, H., K. C. Jezek, and B. Li, 1999: Development of an Antarctic digital elevation model by integrating cartographic and remotely sensed data: A geographic information system based approach. J. Geophys. Res., 104 , 23199–23213.
Marshall, G. J., 2002: Trends in Antarctic geopotential height and temperature: A comparison between radiosonde and NCEP–NCAR reanalysis data. J. Climate, 15 , 659–674.
Marshall, G. J., and J. C. King, 1998: Southern Hemisphere circulation anomalies associated with extreme Antarctic Peninsula winter temperatures. Geophys. Res. Lett., 25 , 2437–2440.
Marshall, G. J., V. Lagun, and T. A. Lachlan-Cope, 2002: Changes in Antarctic Peninsula tropospheric temperatures from 1956–1999: A synthesis of observations and reanalysis data. Int. J. Climatol., 22 , 291–310.
Meehl, G. A., J. M. Arblaster, and W. G. Strand, 1998: Global scale decadal climate variability. Geophys. Res. Lett., 25 , 3983–3986.
O'Connor, W. P., D. H. Bromwich, and J. F. Carrasco, 1994: Cyclonically forced barrier winds along the Transantarctic Mountains near Ross Island. Mon. Wea. Rev., 122 , 137–150.
Parish, T. R., and D. H. Bromwich, 1987: The surface windfield over the Antarctic ice sheets. Nature, 328 , 51–54.
Renfrew, I. A., J. C. King, and T. Markus, 2002: Coastal polynyas in the southern Weddell Sea: Variability in the surface energy budget. J. Geophys. Res.,107, 3063, doi:10.1029/2000JC000720.
Renwick, J. A., 1998: ENSO-related variability in the frequency of South Pacific blocking. Mon. Wea. Rev., 126 , 3117–3123.
Renwick, J. A., and M. J. Revell, 1999: Blocking over the South Pacific and Rossby wave propagation. Mon. Wea. Rev., 127 , 2233–2247.
Rignot, E., and S. S. Jacobs, 2002: Rapid bottom melting widespread near Antarctic ice sheet grounding lines. Science, 296 , 2020–2023.
Stott, L., C. Poulsen, S. Lund, and R. Thunell, 2002: Super ENSO and global climate oscillations at millennial time scales. Science, 297 , 222–226.
Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic Press, 666 pp.
Trenberth, K. E., and T. J. Hoar, 1997: El Niño and climate change. Geophys. Res. Lett., 24 , 3057–3060.
Trenberth, K. E., and D. P. Stepaniak, 2002: A pathological problem with NCEP reanalyses in the stratosphere. J. Climate, 15 , 690–695.
Van Woert, M. L., E. S. Johnson, L. Langone, D. L. Worthen, A. J. Monaghan, D. H. Bromwich, R. Meloni, and R. B. Dunbar, 2003: The Ross Sea circulation during the 1990s. Biogeochemical Cycles in the Ross Sea, G. Ditullio and R. Dunbar, Eds., Antarctic Research Series, Amer. Geophys. Union, in press.
Vaughan, D. G., G. J. Marshall, W. M. Connolley, J. C. King, and R. Mulvaney, 2001: Devil in the detail. Science, 293 , 1777–1779.
Yuan, X., and D. G. Martinson, 2000: Antarctic sea ice extent variability and its global connectivity. J. Climate, 13 , 1697–1717.
Yuan, X., and D. G. Martinson, 2001: The Antarctic dipole and its predictability. Geophys. Res. Lett., 28 , 3609–3612.
Five-month running mean of SOI (hPa) during Jul 1996–Jun 1999. Vertical lines divide the study period into three 1-yr increments
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Five-month running mean of SOI (hPa) during Jul 1996–Jun 1999. Vertical lines divide the study period into three 1-yr increments
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Five-month running mean of SOI (hPa) during Jul 1996–Jun 1999. Vertical lines divide the study period into three 1-yr increments
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Sequence of seasonal anomalies from 3-yr seasonal means (Jul 1996–Jun 1999) for 500-hPa geopotential height. (Contour interval: 10 gpm)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Sequence of seasonal anomalies from 3-yr seasonal means (Jul 1996–Jun 1999) for 500-hPa geopotential height. (Contour interval: 10 gpm)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Sequence of seasonal anomalies from 3-yr seasonal means (Jul 1996–Jun 1999) for 500-hPa geopotential height. (Contour interval: 10 gpm)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Map of PMM5 model domain showing model topography at 500-m intervals. Names and locations of AWS (stars), manned stations (circles), and sectors over which time series are extracted are shown
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Map of PMM5 model domain showing model topography at 500-m intervals. Names and locations of AWS (stars), manned stations (circles), and sectors over which time series are extracted are shown
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Map of PMM5 model domain showing model topography at 500-m intervals. Names and locations of AWS (stars), manned stations (circles), and sectors over which time series are extracted are shown
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Three-year mean (Jun 1996–Jul 1999) of surface temperature for (a) PMM5 and (b) inferred from infrared satellite radiances (Comiso 2000). (Contour interval = 10 K)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Three-year mean (Jun 1996–Jul 1999) of surface temperature for (a) PMM5 and (b) inferred from infrared satellite radiances (Comiso 2000). (Contour interval = 10 K)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Three-year mean (Jun 1996–Jul 1999) of surface temperature for (a) PMM5 and (b) inferred from infrared satellite radiances (Comiso 2000). (Contour interval = 10 K)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Percent change in cloud frequency between La Niña (DJF 1998/99) and El Niño (DJF 1997/98) as described in the text for (a) PMM5 and (b) inferred from satellite radiances (Key 2002). Positive (negative) areas indicate greater (less) cloud frequency for the El Niño summer. Positive (negative) values greater than 25% (less than −25%) are dark (light) gray. (Contour interval: 50% with additional contours at ±25%)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Percent change in cloud frequency between La Niña (DJF 1998/99) and El Niño (DJF 1997/98) as described in the text for (a) PMM5 and (b) inferred from satellite radiances (Key 2002). Positive (negative) areas indicate greater (less) cloud frequency for the El Niño summer. Positive (negative) values greater than 25% (less than −25%) are dark (light) gray. (Contour interval: 50% with additional contours at ±25%)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Percent change in cloud frequency between La Niña (DJF 1998/99) and El Niño (DJF 1997/98) as described in the text for (a) PMM5 and (b) inferred from satellite radiances (Key 2002). Positive (negative) areas indicate greater (less) cloud frequency for the El Niño summer. Positive (negative) values greater than 25% (less than −25%) are dark (light) gray. (Contour interval: 50% with additional contours at ±25%)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
July 1997–Jun 1998 minus Jul 1998–Jun 1999 PMM5 surface pressure anomalies. Observed anomalies are given in the white boxes next to stations. (Contour interval: 1 hPa; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
July 1997–Jun 1998 minus Jul 1998–Jun 1999 PMM5 surface pressure anomalies. Observed anomalies are given in the white boxes next to stations. (Contour interval: 1 hPa; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
July 1997–Jun 1998 minus Jul 1998–Jun 1999 PMM5 surface pressure anomalies. Observed anomalies are given in the white boxes next to stations. (Contour interval: 1 hPa; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
MAM 1997 minus MAM 1999 PMM5 surface temperature anomalies. Observed anomalies are given in the white boxes next to stations. (Contour interval: 1 K; areas with change > |4 K| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
MAM 1997 minus MAM 1999 PMM5 surface temperature anomalies. Observed anomalies are given in the white boxes next to stations. (Contour interval: 1 K; areas with change > |4 K| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
MAM 1997 minus MAM 1999 PMM5 surface temperature anomalies. Observed anomalies are given in the white boxes next to stations. (Contour interval: 1 K; areas with change > |4 K| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
July 1996–Jun 1999 time series of temperature anomalies (K) for ECMWF/TOGA data averaged for 10 points over the (a) Ross Ice Shelf (80°–82.5°S; 170°–180°W) and (b) Ronne/Filchner Ice Shelf (77.5°–80°S; 60°–70°W). Data are given for 2 m (T2m), 850 hPa (T850), 700 hPa (T700), and 500 hPa (T500)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
July 1996–Jun 1999 time series of temperature anomalies (K) for ECMWF/TOGA data averaged for 10 points over the (a) Ross Ice Shelf (80°–82.5°S; 170°–180°W) and (b) Ronne/Filchner Ice Shelf (77.5°–80°S; 60°–70°W). Data are given for 2 m (T2m), 850 hPa (T850), 700 hPa (T700), and 500 hPa (T500)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
July 1996–Jun 1999 time series of temperature anomalies (K) for ECMWF/TOGA data averaged for 10 points over the (a) Ross Ice Shelf (80°–82.5°S; 170°–180°W) and (b) Ronne/Filchner Ice Shelf (77.5°–80°S; 60°–70°W). Data are given for 2 m (T2m), 850 hPa (T850), 700 hPa (T700), and 500 hPa (T500)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Percent change in PMM5 cloud fraction between La Niña (DJF 1998/99) and El Niño (DJF 1997/98), as described in the text. Observed changes are given in the white boxes next to stations. Positive (negative) changes indicate greater (less) cloud fraction for the El Niño summer. (Contour interval: 25%; areas with change > |50%| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Percent change in PMM5 cloud fraction between La Niña (DJF 1998/99) and El Niño (DJF 1997/98), as described in the text. Observed changes are given in the white boxes next to stations. Positive (negative) changes indicate greater (less) cloud fraction for the El Niño summer. (Contour interval: 25%; areas with change > |50%| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Percent change in PMM5 cloud fraction between La Niña (DJF 1998/99) and El Niño (DJF 1997/98), as described in the text. Observed changes are given in the white boxes next to stations. Positive (negative) changes indicate greater (less) cloud fraction for the El Niño summer. (Contour interval: 25%; areas with change > |50%| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Similar to Fig. 9 but for precipitation. (Contour interval: 50%; areas with change > |100%| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Similar to Fig. 9 but for precipitation. (Contour interval: 50%; areas with change > |100%| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Similar to Fig. 9 but for precipitation. (Contour interval: 50%; areas with change > |100%| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Jul 1997–Jun 1998 minus Jul 1998–Jun 1999 PMM5 surface meridional wind. Observed anomalies are given in the white boxes next to stations. (Contour interval: 0.5 m s−1; areas with change > |1.5 m s−1| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Jul 1997–Jun 1998 minus Jul 1998–Jun 1999 PMM5 surface meridional wind. Observed anomalies are given in the white boxes next to stations. (Contour interval: 0.5 m s−1; areas with change > |1.5 m s−1| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Jul 1997–Jun 1998 minus Jul 1998–Jun 1999 PMM5 surface meridional wind. Observed anomalies are given in the white boxes next to stations. (Contour interval: 0.5 m s−1; areas with change > |1.5 m s−1| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Jul 1996–Jun 1999 seasonal anomalies for the MBL (solid) and WRF (dashed) sectors. Variables shown are (a) surface temperature (K), (b) surface meridional wind speed (m s−1), (c) 500-hPa meridional wind (m s−1), (d) cloud fraction (%), and (e) P minus E (mm). Shaded areas correspond to the months for which the spatial figures in section 4a are plotted
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Jul 1996–Jun 1999 seasonal anomalies for the MBL (solid) and WRF (dashed) sectors. Variables shown are (a) surface temperature (K), (b) surface meridional wind speed (m s−1), (c) 500-hPa meridional wind (m s−1), (d) cloud fraction (%), and (e) P minus E (mm). Shaded areas correspond to the months for which the spatial figures in section 4a are plotted
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Jul 1996–Jun 1999 seasonal anomalies for the MBL (solid) and WRF (dashed) sectors. Variables shown are (a) surface temperature (K), (b) surface meridional wind speed (m s−1), (c) 500-hPa meridional wind (m s−1), (d) cloud fraction (%), and (e) P minus E (mm). Shaded areas correspond to the months for which the spatial figures in section 4a are plotted
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
500-hPa geopotential height anomalies (10-gpm contours) for (a) all DJF El Niños minus all DJF La Niñas between 1979–99 and (b) 1997/98 DJF El Niño minus DJF 1998/99 La Niña
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
500-hPa geopotential height anomalies (10-gpm contours) for (a) all DJF El Niños minus all DJF La Niñas between 1979–99 and (b) 1997/98 DJF El Niño minus DJF 1998/99 La Niña
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
500-hPa geopotential height anomalies (10-gpm contours) for (a) all DJF El Niños minus all DJF La Niñas between 1979–99 and (b) 1997/98 DJF El Niño minus DJF 1998/99 La Niña
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Top-of-atmosphere OLR anomalies for (a) all DJF El Niños minus all DJF La Niñas between 1979 and 1999 and (b) 1997/98 DJF El Niño minus DJF 1998/99 La Niña. Negative anomalies indicate enhanced convection. (Contour interval: 10 W m−2; areas with anomalies > |40 W m−2| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Top-of-atmosphere OLR anomalies for (a) all DJF El Niños minus all DJF La Niñas between 1979 and 1999 and (b) 1997/98 DJF El Niño minus DJF 1998/99 La Niña. Negative anomalies indicate enhanced convection. (Contour interval: 10 W m−2; areas with anomalies > |40 W m−2| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Top-of-atmosphere OLR anomalies for (a) all DJF El Niños minus all DJF La Niñas between 1979 and 1999 and (b) 1997/98 DJF El Niño minus DJF 1998/99 La Niña. Negative anomalies indicate enhanced convection. (Contour interval: 10 W m−2; areas with anomalies > |40 W m−2| are shaded; negative contours are dashed)
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Annual running mean of P minus E (mm) for the West Antarctic sector bounded by 75°–90°S, 120°W–180° calculated from WMO, TOGA, and PMM5, plotted along with the SOI (hPa). Adapted from Bromwich et al. (2000).
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Annual running mean of P minus E (mm) for the West Antarctic sector bounded by 75°–90°S, 120°W–180° calculated from WMO, TOGA, and PMM5, plotted along with the SOI (hPa). Adapted from Bromwich et al. (2000).
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Annual running mean of P minus E (mm) for the West Antarctic sector bounded by 75°–90°S, 120°W–180° calculated from WMO, TOGA, and PMM5, plotted along with the SOI (hPa). Adapted from Bromwich et al. (2000).
Citation: Journal of Climate 17, 1; 10.1175/1520-0442(2004)017<0109:MTEMOA>2.0.CO;2
Table A1. Expanded version of Table 2
Table A1. (Continued )
Table A2. Expanded version of Table 3
Descriptions of the datasets used
Bias, rms error (rmse), and correlation coefficients (Corr) for modeled vs observed surface variables averaged over AWS and manned sites above 500 m (8 total) and below 500 m (10 total). The winter (JJA) and summer (DJF) statistics are weighted averages based on synoptic (3- or 6-hourly) data from Jul 1996–Jun 1999. The annual data are weighted correlations of the monthly mean anomalies over the annual cycle for the same period. Statistics for individual sites used to compile this table are given in the appendix (Table A1). Correlations that are significant from zero at the 99% confidence level after accounting for autocorrelation are indicated in bold type
Byrd Polar Research Center Contribution Number 1280.
Although it is not evident in these plots, (due to using geopotential height), this wave train is generally shown to curve back toward the northeast, over South America, after reaching the southeast Pacific (e.g., Karoly 1989).