1. Introduction

Snow cover has a profound impact on the energy balance over the land surface and in the atmosphere due to its high albedo and low thermal conductivity. It is also important hydrologically when it melts (Reference Barnett, Dümenil, Schlese, Roeckner and LatifBarnett and others, 1989). Thus, the effects of snow cover on regional climate have attracted attention for many years (Reference BlanfordBlanford, 1884). The Tibetan Plateau is an important snow distribution region in the Eurasian continent. The thermal effect of snow cover can reach upwards to the middle of the troposphere (Reference Wei, Huang, Chen and DongWei and others, 2002). Numerous studies of the spatial distribution, variability and effects of snow cover on regional climate have been undertaken (e.g. Reference Chen and YanChen and Yan, 1981; Reference Guo and WangGuo and Wang, 1986; Reference LiLi, 1993, Reference Li1995, Reference Li1996; Reference Xu and LiXu and Li, 1994; Reference Wei and LuWei and Lu, 1995; Reference Wei, Huang, Chen and DongWei and others, 2002). As shown in Figure 1, compiled from the measured data at 71 meteorological stations, the snow cover on the Tibetan Plateau is unevenly distributed. It is especially heavy in the northern foothills of the Himalaya and in the eastern parts of the Nyainqentanggula Shan. Other areas of comparatively thick snow cover are the eastern parts of the Tanggula Shan, the A’nyêmaqên Shan, the Bayan Har Shan and some areas lying along the eastern edge of the Tibetan Plateau. However, snow-cover variations in these latter areas mimic those on the rest of the Tibetan Plateau. Thus, when snowfall is abundant (or low) in these areas, snowfall on the whole Tibetan Plateau is also abundant (or low). These latter areas are noteworthy on the Eurasian continent for the variability of their snow cover (Reference Vernekar, Zhou and ShuklaVernekar and others, 1995). The Tanggula Shan, the A’nyêmaqên Shan and the Bayan Har Shan form the boundaries of the source regions of the Yangtze and Yellow Rivers (Fig. 2). Thus, the hydrological regimes of these rivers are affected by this snow-cover variability.

Fig. 1. Sketch map showing the spatial distribution of average maximum snow depth on the Tibetan Plateau. (Contour interval 40 mm.)

Fig. 2. Geographical location of the study region in China and meteorological stations near to and in the source areas of the rivers. Arrows show direction of flow of rivers.

Snow cover is a product of climate, and snow-cover variations are necessarily influenced by climate change. Reference Ke, Li and WangKe and others (1997) analyzed variation of snow cover using daily snow-depth data from 1957 to 1990 at 60 meteorological stations across the Tibetan Plateau. Their study showed that annual snow depth increased by ~1.7% between 1957 and 1990, and that this snow-cover increase resulted from an increase in precipitation during the cold season (October to May).

We examine whether snow cover continued to increase into the 1990s, and how it has changed with progressive climatic warming. We analyze the variability and trends in maximum snow depth, accumulated snow depth, snow-cover duration, air temperature and precipitation in the source regions of the Yangtze and Yellow Rivers from 1960 to 1999.

2. Data Source and Observation Program

2.2. Snow cover, meteorological data and the observation program

Locations of six meteorological stations in the study region are shown in Figure 2, and further data are given in Table 1. Because snow-cover and meteorological data from the Qumalai and Zhongxinzhan stations are discontinuous, they are not included in the regionally averaged time series. The remaining six stations are part of the permanent meteorological station network of China, and were not relocated during the period 1960–99 (CMA, 2003). Furthermore, as these stations are all located at lower elevations, the resulting records all contain a low-elevation bias. There are no meteorological stations in higher mountain areas of the Plateau. Despite this bias, these stations provide the best data available. The snow-cover and meteorological data from these six stations are complete; there are no missing values during the period 1960–99.

Table 1. Main physical characteristics at the six meteorological stations

The data used in this study are daily snow depth, monthly mean air temperature and monthly precipitation between 1960 and 1999. According to the instructions for taking surface meteorological observations (CMA, 2003) in the visual field from the point where the station is located, if more than 50% of the ground is snow-covered, snow depth should be measured, and a day of snow cover is counted. New snow depth, excluding accumulated old snow, is measured to the nearest 0.01 m in a flat and open area near the meteorological observation field. Daily snow depth is an average of three measurements in this area. Air temperature is observed with temperature sensors placed in a ventilated shutter. Both solid (snow) and liquid (rain) precipitation is measured with gauges (CMA, 2003) which, however, are not equipped with wind shields. Thus, we assume precipitation is underestimated. Observing instruments for snow depth, air temperature and precipitation have not changed during the study period. Thus, these data series have good homogeneity. Annual snow-cover duration is the total number of days with snow cover each year.

3. Variability of Snow Cover in the Study Region

3.1. Variability of the regionally averaged time series

The regionally averaged annual maximum snow depth decreased gradually from 1960 to 1964 (Fig. 3a). After 1964, it fluctuated, but generally increased until 1987. The 1985 snow cover was exceptional, the highest for the 40 year period. The maximum snow depth for a single snowfall, 39 cm, was recorded at Tuotuohe station on 18 October 1985. Thereafter, the daily average snowfall exceeded 200 mm until 28 February 1986. This unusual snow accumulation accompanied the lowest mean annual air temperature for the 40 year period. From 1987 to 1992 the annual maximum snow depth decreased. After 1992, it increased. Linear regression calculations indicate that it increased ~0.23% a^–1 (P > 0.25) during 1960–99.

Fig. 3. Time series of regionally averaged (a) annual maximum snow depth, (b) annual accumulated snow depth and (c) annual snow-cover duration in the source regions of the Yangtze and Yellow Rivers during 1960–99. Straight lines indicate linear trends.

The annual accumulated snow depth, i.e. the accumulation of daily snow depth throughout the year, increased gradually until 1999 (Fig. 3b). In 1985 it was 422.7 cm, the highest for the 40 year period. It increased ~2.4%a^–1(P < 0.01) between 1960 and 1999.

The annual snow-cover duration exceeded 270 days in the study region (Fig. 3c), but it was highly variable from 1962 to 1967 and from 1983 to 1985. The long-term trend is an increase of ~06%a^–1 (P > 0.25) over the 40 year period.

Summarizing, the variability of annual snow-cover duration was small. The regionally averaged annual maximum and accumulated s increased from 1960 to 1999. However, the annual maximum snow depth was reduced between 1988 and 1999. That is, heavy snow cover decreased in the source regions of the Yangtze and Yellow Rivers over the Tibetan Plateau after 1987.

3.2. Variability of the station time series

Time series for the annual maximum and accumulated snow depth from the six meteorological stations are given in Tables 2 and 3, respectively. Tuotuohe station, in the source region of the Yangtze River, and Qingshuihe station, near the source region of the Yellow River, are located in central regions of relatively thick snow cover (Fig. 1). These stations were therefore selected for detailed study in order to show inter-station variability (Fig. 4). As shown in Figure 4a and b, annual maximum snow depth had the same trend as annual accumulated snow depth over the 40 year period. However, the specific variations at Tuotuohe were different from those at Qingshuihe. At Tuotuohe, they fluctuated, but generally increased to 1985, decreased from 1985 to 1994, and increased again after 1994 (Fig. 4a). At Qingshuihe (Fig. 4b), they both declined irregularly from 1960 to 1969. Then the annual maximum snow depth increased from 1970 to 1985 and from 1986 to 1999, while the annual accumulated snow depth was highly variable from 1970 to 1976 and from 1996 to 1999.

Fig. 4. (a, b) Time series of annual maximum snow depth and annual accumulated snow depth at Tuotuohe (a) and Qingshuihe (b). (c, d) Annual snow-cover duration at Tuotuohe (c) and Qingshuihe (d). Straight lines indicate linear trends.

Table 2. The annual maximum snow depth (cm) at each station for each year of the study

Table 3. The time series for the annual accumulated snow depth (cm) at each station for each year of the study

Using linear regression, at Tuotuohe the annual maximum and accumulated snow depths increased ~0.35%a^–1(P > 0.25) and ~10.6%a^–1 (P > 0.25), respectively, during 1960–99 (Tables 2 and 3). In contrast, at Qingshuihe the annual maximum and accumulated snow depths increased ~0.42% a^–1 (P >0.25) and ~2.3%a^–1 (P < 0.05), respectively (Tables 2 and 3).

Annual snow-cover duration was highly variable between 1973 and 1990 at Tuotuohe station (Fig. 4c). It increased ~0.08%a^–1 (P > 0.25) over the 40 year period. In contrast, at Qingshuihe, except in 1966, 1982, 1993 and 1996 when snow-cover duration was exceptionally low, it fluctuated regularly (Fig. 4d). Over the 40 year time-span, it decreased 0.12% a^–1 (P < 0.25).

At the station scale, the variability of annual snow-cover duration is small. The significance level is low, generally P > 0.25. Except at Dari where the annual maximum snow depth decreased, and at Wudaoliang and Maduo where the annual accumulated snow depths decreased, all annual maximum and accumulated snow depths increased over the 40 year study period. The variation trends in the annual maximum and accumulated snow depths at Tuotuohe and Qingshuihe are consistent with those of the regionally averaged maximum and accumulated snow depths during 1960–99. Values of annual maximum and accumulated snow depths at Qingshuihe show an opposite direction to the regionally averaged values during 1985–99. Although not discussed here, the variations in these parameters at the other four meteorological stations were also analyzed (Tables 2 and 3). We find that among the six stations the situation at Tuotuohe played a dominant role in the regionally averaged snow depth decrease between 1985 and 1999.

4. Climate Background of Snow-Cover Variation

Mean annual air temperature decreased from 1960 to 1963 and increased from 1963 to 1970. It changed little during the 1970s, increasing over that decade by ~0.1 °C. Temperature was highly variable from 1982 to 1985, and increased sharply after 1985 (Fig. 5a). The average increase during the 1980s was again ~0.1°C. During the 1990s however, air temperature increased by ~0.3°C. Thus, the air-temperature increase occurred primarily during the 1990s. On average over the 40 year study period, air temperature increased ~0.6°C, or 0.015°Ca^–1 (P < 0.05). Anomalous years were 1965, 1983, 1985 and 1997.

Fig. 5. Time series of regionally averaged (a) mean annual air temperature and (b) annual precipitation. Straight lines indicate linear trends.

The precipitation increase occurred largely in the 1980s (Fig. 5b). The average precipitation during the 1980s was 24.3 mm above the long-term mean. It was below the long-term mean in the 1960s and 1970s. The four occurrences (1980, 1983, 1985 and 1987) of >100 mm maximum snow depth were attributed to the heavy precipitation in the 1980s. During the 1990s, though annual precipitation increased, it was still 8.4mm below the long-term mean. Precipitation increased 7.5 mm, or ~0.05%a^–1 (P >0.25), over the 40 year study period.

The regionally averaged annual maximum snow depth, accumulated snow depth, snow-cover duration, air temperature and precipitation all exhibited increasing trends during the 40 year study period. The air-temperature increase was notable. The number of sunshine hours increased ~5.3 hours a^–1 during 1960–99 (Reference Yang, Ding, Shen, Liu and ChenYang and others, 2004), contributing to the increase in air temperature. Prior to the late 1980s, air temperature increased slightly, precipitation increased and snow depth and snow-cover duration increased. From the late 1980s to 1999, especially from 1990 to 1999, while air temperature progressively increased, precipitation was 32.7 mm lower than the average for 1980–89, and snow depth and snow-cover duration also decreased. It seems snow depth variability is primarily determined by precipitation. The precipitation decrease is believed to be the major cause of the decrease in snow depth in the 1990s. With global warming, precipitation in the study region has exhibited an increasing trend. This has been documented by some studies (Reference Liu, Zhang and HuiLiu and others, 1998; Reference Yang, Ding, Shen, Liu and ChenYang and others, 2004). Though the precipitation increase is small, the climate has become gradually warmer and drier during the 40 year study period.

5. Conclusions and Discussion

The source region of the Yangtze and Yellow Rivers encompassed by the Tanggula Shan, the Kunlun Shan, the Bayan Har Shan and the A’nyêmaqên Shan is an area of abundant snow cover on the Tibetan Plateau. The snow-cover variability in this region is unusual on the Eurasian continent. The regionally averaged annual maximum snow depth increased ~0.23%a^–1 over the 40 year study period. The long-term trend in the regionally averaged annual accumulated snow depth was an increase of 2.4% a^–1 during 1960–99. There was an exceptional snow year in 1985, when unusually heavy snow cover was accompanied by an all-time low air temperature. Snow-cover duration exceeded 270 days in the source region. However, its variability was small. It only increased ~0.06%a^–1 during 1960–99.

The annual maximum and accumulated snow depths at Tuotuohe meteorological station increased ~0.35 and ~0.6%a^–1, respectively, over the 40 year study period. The changes at Tuotuohe are consistent with those of the regionally averaged maximum and accumulated snow depths.

At Qingshuihe, annual maximum and accumulated snow depths increased ~0.42 and ~2.3%a^–1, respectively, over the 40 year study period. Between 1985 and 1999, changes in annual maximum and accumulated snow depths at Qingshuihe exhibited the opposite sign to changes in regionally averaged maximum and accumulated snow depths. Of all stations, the situation at Tuotuohe played a dominant role in the regionally averaged snow depth decrease between 1985 and 1999.

Though annual precipitation increased slightly (0.05% a^–1) during the 40 year study period, in the 1980s it was 24.3 mm above average. As a consequence of the heavy precipitation in the 1980s, the regionally averaged maximum snow depth exceeded 100 mm four times during that decade.

Mean annual air temperature increased by ~0.6°C over the 40 year study period. Precipitation in the 1990s was 32.7mm lower than the average precipitation of the 1980s. Precipitation decrease was the major cause of the decrease in maximum snow depth in the 1990s. Air temperature, precipitation, maximum snow depth, accumulated snow depth and snow-cover duration increased over the total study period. However, the air-temperature increase was notable and precipitation increase was small. So, while precipitation increased over the total 40 year study period, in later years the climate tended to become warmer and drier.