Arctic Oscillation and Climate of China in Winter

Arctic Oscillation and Climate of China in Winter

Gong Daoyi ()

Key Laboratory of Environmental Change and Natural Disaster, Institute of Resources Science,Beijing Normal University,Beijing,100875, China

Wang Shaowu ()

Department of Geophysics, Peking University, Beijing 100871, China

Submitted to Advances in Atmospheric Sciences

August 2000

Abstract

A growing body of evidence indicates that the Arctic Oscillation (AO) has wide-ranging effects in the northern hemisphere. In this manuscript the relationship between the Arctic Oscillation and climate of China in boreal winter are investigated using NCEP/NCAR Reanalysis monthly mean sea level pressure, 500 hPa geopotential heights, two Arctic Oscillation indices, and observed temperature and precipitation.

Correlation analysis for the last 41yr show that the winter temperature and precipitation both change in phase with AO. Higher positive orrelation between temperature and AO are above +0.4, appearing in northern China. High correlation coefficients for precipitation to AO cover the center China east to ~100E, and south to 40N, with the values varying between +0.3 and +0.4.

The correlation between the 160-station averaged temperature and the simultaneous sea level pressure show that the winter temperature of China is strongly connected to the sea level pressure over the high-Eurasia continent. The center locates in Siberia with values lower than -0.6. The partial correlation between the intensity of Siberian High and averaged temperature for China remains -0.58, when AO keeps constant. But the partial correlation for temperature and AO is 0.14 when the influence of Siberian High is excluded. The relationship between AO and precipitation is significant. The partial correlation between AO and 160-station-mean precipitation is 0.36. But when the AO's influence is excluded, the partial correlation between the intensity of Siberian High and precipitation is only -0.16. This suggests that during the recent several decades the AO affects the precipitation strongly, but for the temperature the Siberian High plays more important role. Whereas AO and Siberian High correlate at -0.51, according to the data for period 1958/59-94/95.

Using the long-term series spanning 1899/1900-1994/1995, the long-term variations of AO, Siberian High and the connections to climate of China are analyzed. At the interdecadal time scale the AO shows significant influence on both temperature and precipitation. Partial correlation between AO and temperature is 0.66, and between AO and precipitation is 0.70. Multivariate regression analysis demonstrates that the AO and Siberian High related variance in temperature and precipitation is 35% and 11% respectively. For precipitation, the portion is low. Some other factors may be responsible and the further investigation is needed.

Key words: Arctic Oscillation, Climate of China, Atmospheric circulation

1. Introduction

The planetary and regional scale climate changes have been paid close attention during the past several decades with concerning of the so called "global warming" which is supposed to have occurred in winter and spring (IPCC,1996). However, it is noteworthy that there are significant association between the surface climate and atmospheric circulation (Hurrell, 1995;1996; Gong and Wang, 1999a). Thompson and Wallace (1998a) pointed out that the leading empirical orthogonal function of the wintertime northern hemisphere sea level pressure field resembles the Northern Atlantic Oscillation but with more zonally symmetric appearance. This annular-like mode in the northern extratropical circulation, which has an equivalent barotropic structure from the surface to the lower stratosphere, is called "Arctic Oscillation (AO)" (Thompson and Wallace,1998a). This mode is found to exist in both hemispheres (Thompson and Wallace,1998b; Gong and Wang,1998;1999b). The North Atlantic Oscillation is usually regarded as the regional manifestation of the AO. They are largely the same things, and the Northern Atlantic Oscillation is part of the AO (Wallace, 2000, Kerr, 1999). Fluctuations in the AO create a seesaw pattern in which atmospheric pressure at northern polar and middle latitudes alternates between positive and negative phase.

It is found that AO strongly coupled to surface air temperature fluctuations over the Eurasian continent (Thompson and Wallace, 1998a; 2000a; 2000b). The positive phase brings wetter weather to Alaska, Scotland and Scandinavia, and drier conditions to California, Spain, and the Middle East (Cutlip, 2000). Some regional climate association with AO are highlighted, for example, Cavazos (2000) reported that the wintertime extreme precipitation events in Balkans are modulated by changes in the circulation associated with the AO. Wang and Ikeda (2000) demonstrated the significant relationship between the sea-ice cover in the Arctic and subpolar regions and the AO. The surface air temperature changes over the Arctic Ocean are strongly related to the AO too, which accounts for more than half of the surface air temperature trends over Alaska, Eurasia and the eastern Arctic Ocean during the last about two decades (Rigor et al., 2000). Variability for some regional circulation systems such as Aleutian Low also shows apparent relation to AO (Overland et al.,1999).

In this manuscript we focus the investigation of AO's climate influence on the domain of China in wintertime. In Section 2 the data used here are described. The influence of AO on the surface air temperature and precipitation in China are investigated in Section 3. Then, long-term variations in AO, temperature and precipitation in climate and their co-variability are discussed in Section 4. Concluding remarks are given in Section 5 finally.

2 Data

The main surface climate data set for this study consists of the monthly precipitation and mean air temperature data of 160 stations in China compiled by the China Meteorological Administration (CMA). These data cover 49 years, from 1951 to 1999, although 26 stations' data are from 1953 or 1954. Monthly mean sea level pressure (SLP) data, and 500 hPa geopotential heights (H500) for northern hemisphere are taken from National Center of Environmental Prediction / National Center for Atmospheric Research (NCEP/NCAR) Reanalysis data set (Kalnay et al., 1996). Here we pick out the sub-data set on the 55 box from the original 2.52.5 grids for both the SLP and H500 on the purpose of reducing data downloading time and quickening the calculation task. This spatial resolution for these two fields are supposed can satisfy our research.

The Arctic Oscillation indices used here are kindly provided by Dr. David Thompson of University of Washington, one longer time series begin in January 1899 (ended in April 1997) which is derived from the empirical orthogonal function analysis of the northern hemisphere sea level pressure field observations (Thompson and Wallace, 1998a). This longer AO index is hereafter referred to as AO1. And other monthly AO records are also available over the 1958 to 1999 period, which is derived from the NCEP/NCAR Reanalysis SLP field. This shorter AO index is hereafter referred to as AO2. Thesetwo AOs correlate at 0.99 for the period 1958-1997 for all four seasons. Both above AO indices can be accessed via internet at ftp://ftp.atmos.washington.edu/pub/jisao/davet/indices. Regarding of the concerning season, all above mentioned data are rearranged by averaging (temperature, sea level pressure and 500hPa geopotential heights) or summing (precipitation) the data of three wintertime months (i.e., December, January and February).

3 Influence of AO on the climate in China

3.1 Temperature and precipitation

Figure 1(a) shows the correlation coefficients between AO index (AO2) and temperatures over China for wintertime (1958-98/99). It is apparent that the possitive relationship exists everywhere in China except in the small regions over the southwestern Tibet Plateau, where the correlation coefficients vary from 0 to -0.2. The most significant areas cover the northern territory of China north to about 40N, where the correlation coefficients are above 0.4. There are 16%~36% of variance associated with the AO. Thompson and Wallace(1998a; 2000a) have regressed northern hemispheric surface air temperature anomalies onto the standardized AO for January, February and March. They found that the positive phase of the winter AO is associated with positive surface air temperature anomalies throughout high latitudes of Eurasia. Regression coefficients vary from about 0.25 to 0.5K per standard deviation of AO index over northern China. Our results presented here is consistent with those previous findings but with more regional details.

Figure 1(b) shows the correlation coefficients between AO index (AO2) and precipitation for the same period. It is interesting to note that the positive phase of AO also associate with positive precipitation anomalies generally. The most significant regions cover the center China east to ~100E, and south to 40N, with the values varying between ~0.3 and 0.4. This means there are about 10%-15% variance of winter precipitation can be explained by AO. Taking whole China as one, the correlation becomes 0.47, it is significant at the 95% confidence level too. Also see Table 1.

Fig. 1.Correlation between AO index (AO2) and temperature (a) and precipitation (b) for winter (1958/59-98/99). Areas above 95% significance level are shaded.

3.2AO and the Siberian High

A plenty of evidence indicated that the most important regional factor affecting winter climate in China is the Siberian High (for example, Tu, 1936;.Wang, 1962; Guo, 1996; Zhu et al.,1997). Gong and Wang (1999c) pointed out that the Siberian High can account for about 43.6% variance of the winter temperature for China in average. Figure 2 shows the correlation between the 160-station averaged temperature and the simultaneous SLP for winter (1951/52-1998/99). It is obvious that the winter temperature of China is strongly connected to the SLP variation over the high-Eurasia continent. Significant negative correlation coefficients center at Siberia with values lower than -0.6. Some previous studies found that the positive phase of AO associate the lower SLP over polar region and much Eurasia continent, when the AO becomes one standard deviation higher, the SLP over Siberia is 1-3 hPa lower than normal. Figure 3 shows the correlation of AO and SLP. It suggests the out-of-phase relationship between AO and Siberian SLP variation again. Are there dynamical connection between planetary scale AO and regional Siberian High? It needs to be clarified further.

Fgiure 2. Correlation between 160-station averaged temperature for China and sea level pressure over northern hemisphere in winter (1951/52-98/99). Areas above 95% significance level are shaded.

Figure 4 shows the plots of the AO, the intensity of Siberian High and the mean temperature of 160-station for winter. Here the intensity of Siberian High is defined as the mean of SLP with the value above 1028 hPa over the middle to higher Asia continent. This index provides a measure of the anomaly of atmospheric mass over the area occupied the atmospheric center (see Gong and Wang, 1999c for details). To facilitate comparison all the intensity of Siberian High, AO and averaged temperature are normalized with respect to 1961-90. As shown in Figure 4, the changes of AO and Siberian High agree with the temperature satisfactorily. The out-of -phase relationship between the AO and the intensity of Siberian High is also clear. The Intensity of Siberian High correlates to AO at -0.51, exceeding the 95% confidence limit. More detailed correlation statistics are summarized in Table 1. The partial correlation analysis is used. Partial correlation of a and b keeping factor c constant is computed using the following formula:

R(a,b)c=

Where R(a,b) indicates the correlation coefficient between factor a and b. R(a,b)c is the partial correlation between factor a and b

Fgiure 3. Correlation between AO and winter sea level pressure over northern hemisphere (1958-98/99). Areas above 0.05 significance level are shaded.

Figure 4. Time series of AO(AO1), the intensity of Siberian High and the mean temperature of 160-station for winter. To facilitate comparison all series are standardized regarding to 1961-90.

Table 1. Correlation statistics for the AO (AO1), the Intensity of Siberian High and wintertime climate in China. The considered epoch is 1958/59-94/95. Correlation coefficients above the significant at the 95% confidence level are bold.

For temperature: / AO / Siberian High
Correlation / +0.43 / -0.67
Partial Correlation / +0.14(Sib. Hi. constant) / -0.58(AO constant)
For precipitation:
Correlation / +0.47 / -0.36
Partial Correlation / +0.36(Sib. Hi. constant) / -0.16(AO constant)
Cor.(AO, Siberian High) / -0.51

It is interesting to note that when the contribution of Siberian High keeps constant, the partial correlation between AO and averaged temperature of China is only 0.14, not significant. But when the AO's influence is excluded, the partial correlation between the intensity of Siberian High and temperature remains -0.58. The regional Siberian High plays more important and direct influence on temperature in China. However, the condition for winter precipitation seems in different way. When the contribution of Siberian High is excluded, the partial correlation between AO and averaged precipitation of 160-station is 0.36. But when the AO's influence is excluded, the partial correlation between the intensity of Siberian High and precipitation is only -0.16.

Figure 5. Correlation between mean temperature of China and winter 500 hPa geopotential heights over northern hemisphere (1951/52-98/99). Areas above 95% significance level are shaded.

Fgiure 6. Correlation between China mean precipitation and winter 500 hPa geopotential heights over northern hemisphere (1951/52-1998/99) in winter. Areas above 95% significance level are shaded.

Fgiure 7. Correlation between AO (AO2) and winter 500 hPa geopotential heights over northern hemisphere (1958/59-1998/99). Areas above 95% significance level are shaded.

Above-mentioned AO related changes in temperature and precipitation would be compared and confirmed by calculating the AO associated variations in 500 hPa heights. The corresponding changes in heights to the AO, temperature and precipitation are shown in Figure 5 to 7, by the mean of correlation coefficients. Comparing Figure 6 and 7, they are virtually similar to some degree. Associated with the more precipitation and positive AO, 500 hPa geopotential heights tend to be above normal in far-Asia higher continent, lower normal in west Asia, and above normal over south Europe. But the spatial pattern in Figure 5 is much different. It shows that in the warm-than-normal winters there is higher middle tropospheric height over much of China and much lower height in the regions north to Siberian. These results also provide clues to understand the phenomena of Siberian High-related-temperature and AO-related-precipitation in China. However, the responsible dynamical processes remain to be unraveled.

4 Long-term climate variations

4.1 Interdecadal fluctuation

In this section the long-term variations of AO, Siberian High and the connections to climate of China are analyzed by employing the low-pass filtering. Figure 8 shows the long time series of winter precipitation and temperature in China. Temperature is the mean of Shanghai and Beijing, this 2-staion-mean series correlate to that for 160-station-mean at 0.92 in period 1951-1998, since there are similar temperature changes over China in winter as revealed by empirical orthogonal function analysis (for example, Wang et al.,1999). Precipitation is the mean of 33 stations over eastern China north to 100E (Wang et al., 2000). This 33-station-mean series correlate to that for 160-station-mean at 0.99 in period 1951-1999. The long-term indices of the intensity of Siberian High and AO (AO1) are shown in Figure 9.

Some previous studies demonstrated that there are interdecadal variations in climate of China as well as Siberian High. For example, Gong and Wang (1999c) indicated the variation in Siberian High at 30-40yr time scale is clear. In order to compare the correlation between climate and atmospheric indices at the interdecadal scale, a low-pass filter is employed. Here the filter is designed to remain the variation at 10-40yr (Huang, 1990). The low frequent components for these series are shown in Figure 10. To facilitate comparison, all series are normalized before filtering. Shown here is the results for period 1899-1994 since AO started at 1899/1900 and Siberian High series ended in 1994/1995.

Figure 8. Long-term variation of winter precipitation and temperature. Precipitation is the mean of 33 stations over eastern China north to 100E. Data taken from Wang et al., 2000. Temperature is the mean of Shanghai and Beijing, this 2-staion-mean series correlate to that for 160-station-mean at 0.92 in period 1951-1998. Both normalized.

Figure 9. Intensity of Siberian High and AO (AO1). To facilitate comparison all series are standardized regarding to 1961-90.

In above analysis using 1958/59-94/95 it is found that there are good relationship between AO and precipitation, and Siberian High and temperature. As shown in Figure 10, these relationship seem do work still, but the correlation coefficients suggest that on the interdecadal time scale the AO plays significant role in both temperature and precipitation. Table 2 is the correlation matrix. In parentheses are partial correlation. Both the correlation and partial correlation between AO, and temperature are the top two. This implies that the planetary scale AO have more significant influence in China at the interdecadal time scale.

Figure 10. Interdecadal components for AO (AO1), Siberian High, temperature and precipitation. As shown as the filtered results at the time scale of 10-40yr. Filter used here is taken from Huang (1990).

Table 2. Mutual correlation coefficients. Values in the lower portion of matrix are calculated using the interdecadal components as shown in Figure 10. In parentheses are partial correlation as in Table 1 but for period 1899/00-1994/95.

Temperature / Precipitation / AO / Siberian High
Temperature / 1.00
Precipitation / 0.12 / 1.00
AO / 0.68 (0.66) / 0.72 (0.70) / 1.00
Siberian High / -0.25(-0.11) / -0.35(-0.25) / -0.25 / 1.00

4.2 Regression analysis

The indices of AO and intensity of Siberian High in Figure 9 are regressed on the winter temperature and precipitation respectively. In order to compare, all series are cut into the same period of 1899-1994. The regression model using both AO and Siberian High can explain 35% of the temperature, and 11% of precipitation variance, respectively. See Table 3 for the multivariate regression details.