4.2.2 Atmospheric circulation indices and snow cover variability
The relationship between the two strongest indices (i.e. winter AO and spring SH), spring temperature anomalies from reanalysis data and Altay weather station and DPSCD in the Upper Irtysh are plotted in Fig. 7. All variables show significant moderate to strong cross-correlation, particularly temperature anomalies from ERA-Interim and Altay stations (0.94). Additionally, high interannual variability exists for the AO and temperatures, while the SH is mostly negative from 2000 to 2017. The largest anomalies in all series occur in 2010, with negative values for the AO and temperatures and positive ones for the SH and DPSCD. In contrast, positive anomalies in temperatures in 2008 are synchronous with a positive AO and negative SH and DPSCD. In spite of the general agreement between the variables, some exceptions occur, such as positive temperature anomalies in 2004 followed by a late DPSCD and the negative winter AO in 2013 with above average temperatures the following spring.
Among the two indices, the winter AO shows the strongest (negative) correlation with DPSCD in all basins, while Spring SH is significantly positively correlated with DSPCD only in Kara Erys and Narym basins. Both indices also correlate with Upper Irtysh basin as a whole (see Fig. 8 and Table 3). Across different elevation ranges, winter AO shows moderate to strong correlation with DPSCD up to 1500 m a.s.l. in Kara Ertis, Bukhtarma and Narym basins, with generally higher values in the Kara Ertis basin (max |r| 0.73) compared to the others (see Fig.8a,c). As regards spring SH, the highest (positive) correlations are again seen in Kara Ertis basin up to 2000 m a.s.l.. In the other basins, correlation is either not significant or significant in fewer elevation ranges compared to the AO (see Fig. 8b,d).
The per-pixel correlation between the AO and SH and DSCD shows stronger correlations compared to DPSCD across multiple elevation ranges (see Fig. 9). Spatially, the correlation with the winter AO shows elevation dependence, with the strongest values below 2000 m a.s.l. south of the Altay Mountains in most basins. The main exception is an area of lower correlation in the south-eastern part of Kara Ertis, on the Mongolian side, corresponding to the dry steppes of the Gobi Desert (see Fig. 9a). Across elevation ranges, moderate to strong significant negative correlation with winter AO is seen up to 2000 m a.s.l. (3500 m a.s.l. in Kara Ertis basin), with the strongest values all between 1000 and 1500 m a.s.l. (see Fig. 9c). The spring SH shows generally lower correlation values, mainly limited to the Kara Ertis basin, while very few pixels correlate in the smaller basins, limited to the lower elevations. The highest correlations are seen in the Gobi Desert, the same area that showed lower correlation with the winter AO, and west of lake Ulungur in the southern Kara Ertis basin (see Fig. 9b). Moderate significant positive correlation with DSCD is seen up to 2000 m a.s.l. in this basin. Significant correlation is also seen up to 1000 m a.s.l. in Narym and 500 m a.s.l. in Bukhtarma and Uba basins (see Fig. 9d).
Both snow cover variables show significant correlations with spring temperature anomalies from weather stations in the relevant elevation band, i.e. 500-1000 m a.s.l. for Altay and Fuyun and 1000-1500 m a.s.l. for Baitag. For DPSCD, the strongest correlation is seen in Kara Ertis basin with a max. |r| of 0.79 at Altay; this basin has the highest correlations at all three stations, while no significant correlation with temperatures from any weather station is seen in Uba basin (see Table 4). For DSCD, all correlations with temperatures from weather stations are statistically significant: the strongest correlation is between DSCD in Kara Ertis basin and Baitag spring temperature anomalies, with high values also in Narym and Uba basins (see Table 4).