Precipitation characteristics and its impact on vegetation restoration in Minqin County, Gansu Province, northwest China

Authors

  • Jianhui Du,

    1. School of Geographical Science and Planning, Sun-Yet San University, Guangzhou, China
    2. State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing Normal University, Beijing, China
    3. College of Resources Science and Technology, Beijing Normal University, Beijing, China
    Search for more papers by this author
  • Ping Yan,

    Corresponding author
    1. State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing Normal University, Beijing, China
    2. College of Resources Science and Technology, Beijing Normal University, Beijing, China
    • State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing Normal University, No.19, Xinjiekouwai Street, Beijing, 100875, China.
    Search for more papers by this author
  • Yuxiang Dong

    1. School of Geographical Science and Planning, Sun-Yet San University, Guangzhou, China
    Search for more papers by this author

Abstract

Annual, seasonal, monthly and daily precipitation records of Minqin County over 1953–2007 were analysed. In addition, the effects of changing precipitation patterns on vegetative restoration in this area were examined. Annual average precipitation in this area showed a non-significant (p > 0.05) increasing trend, and 76.9% of the rainfall occurred from June to September. Light rain events accounted for a large percentage of the total rainfall event frequency and total annual precipitation. Rainy days have increased by 5.2 days over the past 55 years, with rainfall events of 0–5 and 5–10 mm showing an increasing trend. The 0–5-mm rainfall events contributed 82.3% of the annual precipitation events from 1953 to 2007. These events occurred consistently across different years and gradually became the dominant precipitation type in the study region. Other higher precipitation size classes did not show any apparent trend across the months or years studied, occurring 0–2 times per year. Precipitation in the spring, autumn and winter showed an increasing trend, while a decreasing trend was observed in the summer. The number of 1- and 2-day and longer than 2-day rainfall events showed an increasing trend, with the 1- and 2-day events being more frequent. In addition, the number of short and long drought periods also increased, with the number of long droughts showing an increasing trend. Precipitation variation patterns had a great impact on vegetative succession in this area. Species that mainly use topsoil water may be more adaptive future light precipitation patterns. Thus, plants with shallow roots, especially with adventitious roots, and resistant to sand burial should be used for future ecological restoration in this area. Copyright © 2010 Royal Meteorological Society

1. Introduction

Minqin County is located in the lower reaches of the Shiyang River, Gansu Province, northwest China. Although there were human activities before the Han Dynasty (206 BC to AD 220), almost no destruction occurred to the local environment. After the Han Dynasty, intensive human activities led to gradual desertification and environmental deterioration (Chang and Zhao, 2006). Since the foundation of the People's Republic of China (AD 1949), the government has invested substantial money in the Shiyang River basin for ecological restoration. Large areas of shrubs were planted for the prevention of desertification, and the ecological environment was initially improved in some areas. However, in recent years, the ecological health of the entire Shiyang basin has worsened, especially in Minqin County along the lower reaches of the river. Because of the mismanagement of water resources, the health of the environment and local population is greatly threatened by rapid reductions of groundwater, degeneration of large areas of vegetation and more frequent sandstorms (Chang and Zhao, 2006; Meng and Lu, 2007).

Minqin County has been studied extensively by many researchers. Most of these scientists conclude that human activities, like overgrazing animal stock, over-planting crops and mismanaging water resources, are the main causes for vegetative degeneration and land desertification, with climate change serving only as a backdrop (Lee and Zhang, 2004; Sun et al., 2006). Numerous studies examining vegetative succession in this area have shown that Populus euphratica, Haloxylon ammodendron, Tamarix ramosissima, Elaeagnus angustifolia and Calligonum mongolicum have been seriously degraded over the past several decades. Nitraria tangutorum has also gradually become the dominant species with the largest distribution in this area. Thus, the species that should be adopted for ecological restoration in this region is becoming a great problem. The declining water table has undoubtedly damaged vegetative growth (Yang, 1999; Lee and Zhang, 2004). However, not all shrubs are predicted to be adversely affected by reduced water resources, and restoration of vegetation is still possible, with some species adapting to such environmental changes. The average groundwater level was 5.20 m in the study area in 1978, which was impossible for most plants to use (Chang and Zhao, 2006), especially young plants without deep roots. Reduced surface water flow from the Shiyang River is mainly used for agriculture irrigation (Lee and Zhang, 2004), so vegetative succession over the past 30 years may have some correlation to variations in precipitation (Yang and Gao, 2000).

Precipitation trends in different areas of the world over the past several years have been studied extensively, and some extreme precipitation events, like floods or droughts, have been observed and are likely to have an impact on natural and human systems (Luis et al., 2000; Gonzalez-Hidalgo et al., 2001; Houston, 2006; Costa and Soares, 2009; Nel, 2009). Thus, effective water management strategies are essential to the sustainable development of the study area. It is necessary to analyse precipitation variations over the past several decades to assess their impact on the environment. Numerous studies have analysed precipitation trends in China over the past several decades. These studies note increasing trends in average precipitation associated with global warming but with differences across seasons and regions (Gong and Wang, 2000; Gemmer et al., 2004). Precipitation in northwest China is increasing, which may improve the ecological environment of this area (Ye et al., 2004), and slightly decreasing precipitation is observed in the agropastoral transition zone of northern China (Gong et al., 2004). However, there is more concern about monthly, seasonal mean conditions. Little is known about precipitation patterns in Minqin County and its effects on vegetative restoration. We analysed daily precipitation data from 1953 to 2007, including annual, seasonal, monthly and daily precipitation characteristics. We also examined how these precipitation patterns could affect vegetative restoration in Minqin County. This analysis will enhance our understanding of the influence of climate change on regional vegetative restoration and land desertification.

2. Material and methods

2.1. Study area

Minqin County is a 16 016-km2 area located along the lower reaches of the Shiyang River, Gansu Province, northwest China (latitude 101°50′–104°10′; longitude 38°10′–39°20′N). Desert or wasteland covers 83.7% of the region (Lee and Zhang, 2004), and the elevation varies between 1295 and 1460 m. The region is surrounded by the Badain Jaran Desert to the west and north and the Tengger Desert to the east. The Wuwei oasis lies to the south as shown in Figure 1 (Chang and Zhao, 2006).

Figure 1.

Location of Minqin County in China. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

The study area has an arid continental climate with an average annual temperature of 7.8 °C and effective accumulated temperatures ≥ 0 °C of 3655 °C. The annual daily mean sun duration is 3028 h, and the average annual precipitation is 110 mm. The average annual evaporation is 2664 mm. The only surface water source for irrigation in the oasis comes from the Shiyang River because of over-consumption of water resources in river's upper reaches. Annual surface water flow has decreased from agriculture irrigation in Minqin County, and no excess water is available for vegetative restoration. Groundwater levels were up to 16.43–22.22 m in 2004, with an intensive reduction rate of 0.69 m per year (Chang and Zhao, 2006).

2.2. Data description and analysis

All precipitation data were gathered from observations taken at the Minqin Meteorological Bureau, Gansu Province, China. There are six kinds of readings that are described by Gong et al. (2004). Since precipitation associated with frost, fog and dew is low, this type of precipitation is regarded as non-precipitation in our analysis. A rainy day is considered one with ≥ 0.1 mm per day of precipitation (Gong et al., 2004). Precipitation data during 1953–2007 were selected, and annual, seasonal, monthly and daily precipitation data were analysed. The following six precipitation classes are described in our study: light rain (0–5 and 5–10 mm); moderate rain (10–15, 15–20 and 20–25 mm) and heavy rain (25–50 mm). We calculated the percentage of rainy days per year and per month among each of the precipitation classes. All the data were analysed using SPSS 13.0 (Chicago, IL, USA).

3. Results

3.1. Yearly and monthly precipitation in the Minqin study area

Annual precipitation in Minqin County was 34.6–202.0 mm (Figure 2(a)) with an average precipitation of 110.5 mm. While there is a linear increasing trend in precipitation of 4.1 mm per 10 years from 1953 to 2007, these increases were not statistically significant over different years (p > 0.05). There were 20–53 rainy days per year (average 37.7 days per year) from 1953 to 2007 (Figure 2(b)), with an increasing trend observed over different years that was not statistically significant (p > 0.05). Monthly precipitation ranged from 0.43 to 31.17 mm (Figure 2(c)), while average monthly precipitation was 9.21 mm with increasing rainfall from January to August. The highest precipitation was recorded during August, accounting for 28.2% of the annual rainfall. After August, the peak precipitation decreased, with the lowest rainfall recorded in December, which only accounted for 0.4% of annual precipitation. When the amount and rainy days of precipitation were considered (Figure 2(d)), light precipitations of 0–5 mm accounted for a large percentage of both yearly precipitation and rainy days and were vital to vegetative restoration in this region (Table I).

Figure 2.

Precipitation characteristics in Minqin County from 1953 to 2007. (a) Annual precipitation, (b) rainy days, (c) monthly precipitation and (d) frequency of different precipitation size classes. Dashed lines are the linear trends

Table I. Frequency of different precipitation size classes
Precipitation size classes (mm)0–55–1010–1515–2020–2525–50
Percent of annual rainy days (%)82.311.23.41.40.70.9
Percent of annual precipitation (%)36.126.714.58.05.49.2

3.2. Variation of rainy days among different precipitation size classes

The precipitation size class of 0–5 mm had the most rainy days in comparison to other precipitation size classes, and average rainy events were 31.1 days (Table II). Rainy days of 5–10 mm occurred 1–4 times over 58.2% of the year, and rainy days of 10–15 mm occurred 1–2 times over 56.4% of the year. Rainy days of 15–20 mm occurred only once over 34.5% of the year. No-rain events of 20–25 mm were recorded over 74.5% of the year, while one event at this level occurred over 23.6% of the year.

Table II. Rainy days of different precipitation size classes
Precipitation size class (mm)0–55–1010–1515–2020–2525–50
Number of rainy daysLeast1800000
 Most46115323
 Average31.14.21.30.50.30.3

The number of light rain days showed an increasing linear trend over different years, but these trends were not statistically significant (p > 0.05). Moderate and heavy rain events did not change over time (Figure 3).

Figure 3.

Number of rainy days with different precipitation size classes during 1953–2007. (a) Rainy days of daily rainfall of 0–5 mm, (b) rainy days of daily rainfall of 5–10 mm, (c) rainy days of daily rainfall of 10–15 mm, (d) rainy days of daily rainfall of 15–20 mm, (e) rainy days of daily rainfall of 20–25 mm and (f) Rainy days of daily rainfall of 25–50 mm. Solid lines are the linear trends

3.3. Rainy day percentages by precipitation size class relative to total rainy days

Rainy days of 0–5 mm accounted for 82.3% of the annual rainy days on average (Table III), and precipitation at this level is observed throughout the year and the most typical in the study area. Precipitation of 5–10 mm occurred with relative regularity across the year, and precipitation fell into this size class 96.4% of the year. Precipitation events that were > 10 mm were infrequent in the study area. Specifically, no rains were recorded in 29.1, 58.2, 74.5 and 72.7% of the year at 10–15, 15–20, 20–25 and 25–50 mm, respectively. Thus, not only does light rain account for most rainy days in the study area, but these events are also very regular in most years. Moderate and heavy rains, conversely, contribute only a small percentage of rainy days and are not regular events over any given year (Figure 4).

Figure 4.

Percentage of rainy days with different precipitation size classes relative to the total number of rainy days. (a) Daily rainfall of 0–5 mm, (b) daily rainfall of 5–10 mm, (c) daily rainfall of 10–15 mm, (d) daily rainfall of 15–20 mm, (e) daily rainfall of 20–25 mm and (f) daily rainfall of 25–50 mm

Table III. Percentage of rainy days of different precipitation size classes relative to the total annual rainy days
Precipitation size class (mm)0–55–1010–1515–2020–2525–50
Percent of annual rainy days to the total (%)Least65.800000
 Most10028.211.89.74.47.7
 Average82.311.23.41.40.70.9

3.4. Variation in rainfall among precipitation size classes

Up to 36.1% of the annual precipitation fell into the 0–5-mm size class. Precipitation events of 0–5, 5–10 and 10–15 mm had increasing linear trends from 1953 to 2007, and the increases in the 0–5 mm size class were statistically significant (p < 0.05) (Figure 5(a)–(c)). While a decreasing trend was observed for precipitation of 15–20 mm (Figure 5(d)), no trend was measured for precipitation > 20 mm (Figure 5(e) and (f)). Light rain contributed 62.9% of the annual precipitation, and the total amount of moderate rain gradually decreased over the years, which may have a negative effect on vegetation that relies on deep soil water.

Figure 5.

Precipitation amounts at different precipitation size classes during 1953–2007. (a) Daily rainfall of 0–5 mm, (b) daily rainfall of 5–10 mm, (c) daily rainfall of 10–15 mm, (d) daily rainfall of 15–20 mm, (e) daily rainfall of 20–25 mm and (f) daily rainfall of 25–50 mm. Dashed lines are the linear trends

3.5. Relationships between precipitation amounts and days

Precipitation > 1 mm is assumed to be one precipitation day (Tarhule and Woo, 1998; Gong et al., 2004; Seleshi and Zanke, 2004). After studying the number of precipitation days exceeding 1 mm in Minqin County, we found that both precipitation amounts and days all showed an increasing trend, with the increase in precipitation days being statistically significant (p < 0.05). In addition, high correlations between precipitation amounts and days were measured (Figure 6), which is consistent with results by Kwarteng et al. (2009) in Oman and indicate that high numbers of rainy days often result in high precipitation. The frequency of usable rain is more important than rain amount (Dougherty et al., 1996), and this variable may have a positive effect on vegetative restoration in this area.

Figure 6.

Precipitation amount and days from 1953 to 2007

3.6. Precipitation characteristics across seasons

Precipitation in the spring, autumn and winter showed increasing linear trends, while a slightly decreasing linear trend in precipitation was observed in the summer. From 1953 to 2007, precipitation in the spring ranged from 0.2 to 57.8 mm (average = 16.7 mm) and showed a significant (p < 0.05) increasing linear trend (Figure 7(a)). Precipitation in the summer ranged from 10.4 to 163.1 mm (average = 67.6 mm) and showed a slightly decreasing trend from 1953 to 2007 that was not statistically significant (p > 0.05) (Figure 7(b)). Autumn precipitation ranged from 0.3 to 70.7 mm (average = 24.3 mm) during 1953–2007, with an increasing trend that was not statistically significant (p > 0.05) (Figure 7(c)). Finally, in the winter, precipitation ranged from 0.0 to 7.7 mm (average = 1.9 mm) with an increasing trend that was not statistically significant (p > 0.05) (Figure 7(d)).

Figure 7.

Precipitation amounts across seasons. (a) Precipitation amount in the spring, (b) precipitation amount in summer, (c) precipitation amount in autumn, (d) precipitation amount in winter. Dashed lines are the linear trends

3.7. Monthly percentage of different precipitation size classes

Precipitation in arid areas is often limited and variable, and reliable rainfall is vital to the vegetative production in these regions, especially during the growing season. If precipitation of one size class occurred in each of the past 55 years in the same month, then its probability of occurring in this month was defined as 100%; likewise, if this scenario occurred for half of the years, then its probability was 50%. Percent contribution of precipitation size class per given month was defined as the average percentage of accumulated rainfall of a certain precipitation size class to the total monthly rainfall.

Precipitation events of 0–5 mm occurred in all months, and the event probability for this size class was up to 100% from June to August (Table IV) with the lowest probability in December. Precipitation from December to February was completely from precipitation for the 0–5-mm precipitation size class (Table V). No rainfall from 5 to 10-mm size class occurred during November to February; this precipitation class also constituted only 3.8% of the rainfall for March. The month of May received the most rainfall (up to 27%) at the 5–10-mm size class. Precipitation of 10–15 mm was not observed from November to February, while the probability of rainfall of this size class was the highest in September at up to 30.9%. March received the lowest rainfall (1.7%) at the 10–15-mm precipitation size class, and September had the highest rainfall (19.3%) at this size class. Precipitation of 15–20 mm was not observed from December to March, and the probability of this precipitation class was the highest in August at up to 14.5%. When the contribution to different months was considered, the average contribution in November was the highest at up to 5.1%. Precipitation of 20–25 mm mainly occurred from July to September. The probability of this precipitation size class was the highest in August at up to 12.7%, and its contribution to total precipitation in August was up to 5.3%. Precipitation of 20–25 mm was the lowest in May at only 1.1%. Precipitation of 25–50 mm only occurred from June to August, and its probability was the highest in August at up to 16.4%. Its contribution to this month was up to 11.4%.

Table IV. Probability of precipitation of different size classes occurring in different months
Precipitation size class (mm)0–55–1010–1515–2020–2525–50
Probability of precipitation in different months (%)January27.300000
 February32.700000
 March43.61.81.8000
 April81.823.61.81.800
 May87.341.810.93.61.80
 June10045.520.07.305.5
 July10049.121.87.39.15.5
 August10067.321.814.512.716.4
 September98.241.830.97.31.80
 October78.227.35.53.600
 November34.5001.800
 December45.500000
Table V. Contribution of precipitation of different size classes occurring in different months
Precipitation size class (mm)0–55–1010–1515–2020–2525–50
Contribution of precipitation in different months (%)January10000000
 February10000000
 March94.43.81.7000
 April74.521.81.91.800
 May61.527.07.62.71.10
 June57.823.711.14.003.3
 July53.323.710.12.96.04.0
 August39.026.912.35.05.311.4
 September53.021.619.33.52.60
 October66.225.46.02.400
 November94.9005.100
 December10000000

3.8. Persistence of daily precipitation

Most precipitation falls as occasional showers in arid areas, but there are still rainfalls occurring over successive days. To examine the persistence of these rainfalls, we used the definition by Gong et al. (2004) as follows: 1- and 2-day events were defined as short duration; and > 3-day events were defined as long duration. The number of 1- and 2-day events during 1953–2007 showed an apparent increasing linear trend of 0.3 times per 10 years, but these increases were not statistically significant (p > 0.05) (Figure 8(a)). The number of rainfall events longer than 2 days displayed a slightly increasing linear trend of 0.06 times per 10 years, but these increases were also not statistically significant (p > 0.05) (Figure 8(b)).

Figure 8.

Changes in the daily precipitation persistence. (a) Number of short duration rainfall events and (b) number of long duration rainfall events. Dashed lines are the linear trends

3.9. Number of dry spells

Precipitation is the only water resource that can be used for vegetative restoration in Minqin County. Extended droughts will contribute to vegetative stress and mortality. For these drought periods, we used the following classes established by Gong et al. (2004): short droughts were defined as no-rain periods of not more than 10 days; and long droughts were defined as no-rain periods longer than 10 days. From 1953 to 2007, we calculated that short droughts had a slightly increasing linear trend of 0.1 times per 10 years, but this trend was not statistically significant (p > 0.05) (Figure 9(a)). Likewise, a linear increasing trend of long droughts was observed at 0.2 time per 10 years but was also not statistically significant (p > 0.05) (Figure 9(b)).

Figure 9.

Changes in the frequency of dry spells. (a) Number of short dry spells (≤10 days) and (b) number of long dry spells (>10 days). Dashed lines are the linear trends

3.10. Variation of temperatures in different seasons during 1953–2007

Average temperatures across different seasons had an increasing trend (Figure 10). Specifically, average temperatures in the spring, autumn and winter showed an obvious change and were statistically significant over different years (p < 0.01). However, average temperature changes in the summer were not significant (p > 0.05). Precipitation and temperatures in the spring, autumn and winter were positively related to each other, but these relationships were not statistically significant (p > 0.05). Conversely, summertime precipitation and temperature showed a significant (p < 0.01) negative relationship over different years.

Figure 10.

Temperature characteristics across seasons from 1953 to 2007. (a) Temperature in the spring, (b) temperature in summer, (c) temperature in the autumn and (d) temperature in the winter. Dashed lines are the linear trends

4. Discussion

Average precipitation in Minqin County showed an increasing trend from 1953 to 2007, but these increases were not statistically significant (p > 0.05) over the years. These findings are consistent with a study by Ye et al. (2004) in this same area. They concluded that precipitation in northwest China showed an increasing trend, especially after 1999, which probably improved environmental conditions (Ye et al., 2004).

Rainy days of 0–5 and 5–10 mm accounted for 82.3 and 11.2% of the annual rainy days, respectively, while the other precipitation size classes only accounted for 6.5% of the total precipitation events. Thus, light precipitation events are the dominant precipitation type in Minqin County, which may have a substantial impact on vegetative restoration. Precipitation characteristics in Oman previously studied by Kwarteng et al. (2009) indicate that precipitation < 10 mm contributes 66–95% of the yearly precipitation, which is consistent with our findings. The average annual precipitation for all 31 stations monitored in Oman was 117 mm, which is close to the average precipitation in Minqin County.

Total annual rainy days had an increasing trend from 1953 to 2007. Days of light precipitation had an increasing trend while days of heavy precipitation did not change. Rainy days increased by 5.2 days from 1953 to 2007. This finding is inconsistent with calculations by Gong et al. (2004) of rainfall trends in semi-arid regions of China. They concluded that rainy days had been reduced by about 8 days in the 1990s (Gong et al., 2004), although significant increases were observed in the number of days with light rain. Rainy days in the study area ranged from 20 to 53 days, which is consistent with conclusions by Noy-Meir (1973) that the number of rainy days in arid regions was between 10 and 50 days.

Gong and Wang (2000) studied the relationship between global warming and summer rainfall over eastern China. They concluded that global warming may play an important role in the recent significant increases in the summer rainfall (Gong and Wang, 2000). These conclusions are inconsistent with our results because atmospheric water vapour over eastern China comes primarily from ocean evaporation. Global warming could enhance ocean evaporation thereby increasing atmospheric water vapour content in the atmosphere. While Minqin County in northwest China is far from the ocean, the source of atmospheric water vapour is mainly from outside the oasis area. Thus, global warming may accelerate drought stress in this study area by reducing water vapour content in the atmosphere.

In the absence of groundwater, precipitation is becoming the only water resource for local vegetation growth (Yang and Gao, 2000). Moreover, small rainfall events have an increasing trend, while large events have a decreasing trend. These trends are probably the main reason for vegetation that depends on deep soil water is greatly degraded, while other species, like N. tangutorum with its adventitious roots, are becoming dominant in the study area (Chang and Zhao, 2006). Water use strategies of desert vegetation in this area were studied with stable isotopes by Chu (2007). This study reveals that N. tangutorum mainly uses shallow soil water in arid summer months and can use precipitation of only 3 mm, while other shrubs with deeper root systems cannot avail themselves of this precipitation size class (Chu, 2007). Field investigations also suggest that N. tangutorum coverage on fixed dunes was smaller than on semi-fixed dune with the same groundwater levels (Yang et al., 2007) because semi-fixed dune can better use small amounts of precipitation than fixed dunes.

The ecological importance of small precipitation events has been pointed out previously by many authors. Wang and Tang (2009) studied rainfall events and plant responses in the southern edge of the Gurbantunggut Desert northwest China. They demonstrated that 89.8% of the rainfall events were not more than 5 mm. Moreover, 1 shrub species and 11 herb species had different degrees of response to 2–5 mm rainfall, while some shrubs, like H. ammodendron, T. ramosissima, did not respond to rainfall events < 5 mm (Wang and Tang, 2009). Utilisation of rainfall was life-form dependent, and changes in precipitation patterns will have a substantial impact on competition and possibly community structure (Ehleringer et al., 1991; Dodd et al., 1998). Sala and Lauenroth concluded that Bouteloua gracilis can use small rainfall events (<5 mm), thereby producing an advantage that makes it a dominant species in the steppe region (Sala and Lauenroth, 1982). Dougherty et al. (1996) also indicate that Cactus can utilise small rainfall events (2.5–5 mm) in North American shortgrass steppes, and its success over the short-term depends more on the frequency of usable rain than rain amounts. Cheng et al. (2006) indicated that more frequent small rainfall events promote the dominance of Stipa bungeana and Cynanchum komarovii, which take advantage of shallow-water sources derived from small (<10 mm) rain events. Precipitation regimes that are altered are likely to accelerate the rates of degradation in northwestern China (Cheng et al., 2006). Even under very dry conditions, a 5-mm rainfall event should wet the root zone and become a potential soil water resource for plants. However, the value of these light rainfall events may vary depending on the species (Sala and Lauenroth, 1985). Light showers may play an important role in plant survival during dry seasons (Glover and Gwynne, 1962).

Rainfalls of 0–5 mm comprise the largest percentage of precipitation events in the study area and were very stable over different years. These events showed a significantly increasing trend and have become the dominant precipitation resource in the study area. Currently, many measures have been undertaken for vegetative restoration in the local area, including drip irrigation for species, like H. ammodendron, and aerial seeding for species, like Artemisia arenaria. However, results to date are not compelling because these species use soil water from deep soils (Chu, 2007). In the absence of groundwater, variations in precipitation, especially increases in light rain, provide a competitive advantage to species that use shallow soil water thereby adversely impacting species that rely on deeper soil water. Surface water from the upper reaches of the Shiyang River is mainly used for agricultural irrigation. Thus, vegetation with shallow roots should be the focus of study in future desertification prevention programs since limited precipitation events are predicted for this area. Vegetation with shallow roots, such as Agriophyllum squarrosum and Bassia dasyphylla, has little biomass and cannot be used for sand fixation. Conversely, N. tangutorum can generate adventitious roots after being buried by sand, and much of this plant's roots extend just 3–5 cm below the soil surface. This shallow root system ensures that N. tangutorum will continue to be a dominant species in the future. N. tangutorum can be used to reduce sand erosion from inter-dunes and prevent sand-related damage of agriculture crops. Therefore, in the absence of groundwater, shrubs with adventitious roots and sand fixation properties should be the focus of future programs to prevent desertification. Vegetation that relies on deep soil water may be improper for future ecological restoration efforts in Minqin County.

5. Conclusions

Our results show that average annual precipitation was 110.5 mm in Minqin County from 1953 to 2007, with an increasing trend measured at a rate of 4.1 mm per 10 years during 1953–2007. Light rains of 0–5 mm were the dominant precipitation type in the study area, accounting for 82.3% of the precipitation events and 36.1% of precipitation amounts. Monthly precipitation ranged from 0.43 to 31.17 mm, with August having the most precipitation and December having the least precipitation. Increasing trends of rainy days from 1953 to 2007 were measured at a rate of 0.9 day per 10 years. When precipitation size classes were considered, light rains displayed an increasing trend. They accounted for 93.6% of the annual rainy days and contributed 62.9% of the annual precipitation. These light rainfalls appear to be the dominant type of precipitation in the study area. Other precipitation size classes accounted for only 6.4% of the annual rainy days and contribute 37.1% of the annual precipitation. These other rainfall size classes were highly variable over the 1953–2007 time period.

Precipitation in the spring, autumn and winter all showed a significant (p < 0.05) increasing trends, while slightly decreasing trends were observed in the summer that were not statistically significant (p > 0.05). Summer is when vegetation activity is at its maximum, and so even small water deficiencies can stress plants, agriculture and the environment (Gong et al., 2004). Thus, rainfall declines over summer months, which were mainly attributable to decreases in moderate rainfall, may be the main reason for declines among species that rely on deep soil water. Moreover, increases in light rainfall may contribute to the extensive distribution of species that use water from shallow soils. These hypotheses are consistent with observations of vegetation in the study area.

Days with 0–5-mm rainfall events occurred in all months, and variations of their probability across different months were opposite to their accumulated contribution to monthly precipitation amounts. Other precipitation size classes only occurred in some months, and variations of their probability across different months were consistent with their accumulated contribution to monthly precipitation. The number of 1-, 2-, and > 2-day rainfall events showed an increasing trend from 1953 to 2007, but the 1- and 2-day events were more frequent. The number of short and long droughts showed an increasing trend from 1953 to 2007, although the long droughts were more frequent. While precipitation in the study area has an increasing trend, the number of drought periods is also increasing. This may hinder vegetative growth and present challenges to desertification prevention in this area.

Groundwater levels in Minqin County are now 20–30 m, and it is difficult to restore groundwater in a short period. Thus, variations in precipitation patterns should be considered prior to other factors for vegetative restoration and desertification prevention. Precipitation is increasing in the study area as other authors have also reported, and these increases may improve the environment. However, only light rains are significantly (p < 0.05) increasing. Thus, while these light rains provide precipitation for some species of vegetation in the study, decreasing trends in heavier rainfall events present problems for vegetation that require more or deeper soil water. Variations in precipitation patterns elicit both positive and negative challenges to vegetative restoration in the study area, although vegetation has greatly degenerated over the past several years. Fortunately, there are still some species, like N. tangutorum, that could grow well and become dominant species in this environment. Thus, vegetation with shallow roots, especially adventitious roots, that is resistant to sand burial should be considered for vegetative restoration and desertification prevention in the study area. We believe that light rainfall events will have a substantial impact on the succession of vegetation in this area, and these changing rainfall patterns should be considered for vegetative restoration in Minqin County.

Acknowledgements

This study was supported by the State Key Laboratory of Earth Surface Processes and Resource Ecology (2008-ZZ-02). We thank Professor Youhao E, National Climate Center, China for providing us with daily rainfall data sets. The authors would also like to thank the editor and referees for their useful and constructive suggestions and comments.

Ancillary