4.2.1 Atmospheric circulation indices and temperature anomalies
To identify the most suitable indices for snow cover analysis, we first looked at their correlation with temperatures, by considering aggregated winter (DJF) and spring (MAM) indices. The AO and SH emerge as the strongest indices: both the winter AO and SH are correlated with temperature anomalies from all three stations and across basins in the reanalysis data, although the AO is non-significantly (p > 0.05) correlated at some grid points (Fig. 6a). Specifically, the winter AO appears positively correlated with winter temperatures across Siberia exerting its influence north to the Barents Sea and south to the Altay Mountains (see Fig. 6a). Conversely, the Siberian high exerts a more local impact in winter, influencing temperatures north of the Tibetan plateau, across Northern China, East Kazakhstan and Mongolia, with its southernmost tip over southeast China (Fig. 6c).
Spring SH and winter AO are also strongly correlated with spring temperature anomalies at all stations and grid cells in the Upper Irtysh (see Table 2 and Fig. 6b,d). The correlation between the winter AO and ERA-Interim spring temperature anomalies is significant (p<0.05) across the Altay Mountains, between eastern Kazakhstan, northern Mongolia and southern Russia, including the study area (see Fig.7b), with patterns of significant correlation with temperatures displaced to the east compared to the winter. The correlation between spring SH and spring temperature anomalies from ERA-Interim shows a similar pattern to the correlation between winter temperature and winter SH, although the extent of the influence is limited to a latitudinal band between 70° E and 120 ° E (see Fig. 6b,d).
Other indices show moderate correlation at either individual stations or grid cells, e.g. the winter NAO is also correlated with winter and spring temperatures, although the strength of the relationship is generally lower than for the AO, except for the winter index - winter temperature anomalies at ERA grid cells. The winter SCAND and EA also correlate with winter temperatures. The first correlation however is only significant at weather stations, while the latter at two stations and limited grid cells (see Table 2).