Temporal and spatial variability of extreme snowfall indices over northern Xinjiang from 1959/1960 to 2008/2009

Abstract. Plentiful snowfall is an important resource in northern Xinjiang. However, extreme snowfall events can lead to destructive avalanches, traffic interruptions or even the collapse of buildings. The daily winter precipitation data from 18 stations in northern Xinjiang during 1959/1960–2008/2009 were selected for purpose of analyzing long-term variability of extreme snowfall events. Five extreme snowfall indices, Maximum 1 day snowfall amount (SX1day), Maximum 1-weather process snowfall amount (SX1process), Blizzard days (DSb), Consecutive snow days (DSc) and Blizzard weather processes (PSb), were defined and utilized to quantitatively describe the intensity and frequency of extreme snowfall events. Temporal trends of the five indices were analyzed by Mann–Kendall test and simple linear regression, and their trends were interpolated using universal kriging interpolation. Temporally, we found that most stations have upward trends in the five indices of extreme snowfall events, and over entire northern Xinjiang, they were all increasing at the 0.01 significance level (MK test), with the linear tendency rates of 0.49 mm (10 a)⁻¹ (SX1day), 0.89 mm (10 a)⁻¹ (SX1process), 0.024 days (10 a)⁻¹ (DSb), 0.14 days (10 a)⁻¹ (DSc), and 0.069 times (10 a)⁻¹ (PSb) respectively. Meanwhile, obvious decadal fluctuations besides long-term increasing trends are identified. Trends in the intensity and frequency of extreme snowfall events show a~distinct difference spatially. In general, trends of five indices were found shifting from decreasing to increasing from the northeast to the southwest and from the north to the south of northern Xinjiang. Furthermore, the regions covered by increasing or decreasing extreme snowfall events were identified, implying the hot or cold spots for extreme snowfall events changes. These results may be helpful for northern Xinjiang on the regional and local resource and emergency planning.