Assessment of sandy desertification trends in the Shule River Basin from 1978 to 2010

Xiang Song , ChangZhen Yan, Sen Li, JiaLi Xie

Key Laboratory of Desert and Desertification, Cold and Arid Regions Environmental and Engineering Research Institute,Chinese Academy of Sciences, Lanzhou, Gansu 730000, China

1 Introduction

Land desertification has been widely recognized as one of the most serious environmental problems in China(Yanget al., 2007), and it has kept expanding since the 1950s, exerting severe impacts on regional socio-economic development and environmental security(Guoet al., 2010). Sandy desertification, that is, wind erosion and sand accumulation, is one of the main forms of land desertification in China, especially in northern China. An authoritative report cosponsored by the State Environmental Protection Agency of China (SEPA) and the United Nations Environment Program (UNEP) concluded that a total area of 8.62×105km2had been desertified in China (Zhu and Wu, 1998). Within this frame, an area of 6.64×105km2is distributed in arid, semi-arid, and dry sub-humid regions, accounting for 6.92% of the total Chinese land territories. An area of 3.69×105km2has been directly affected by sandy desertification (Zhu and Wu, 1998).

The Shule River is located in western Gansu Province and the Inner Mongolia Autonomous Region, and it originates in the Qilian Mountains. Land use in the Shule River Basin has an increasingly important effect on population support and sustained economic growth. Unfortunately, the environment of the Shule River Basin is very fragile, with harsh physiographic conditions (sparse vegetation coverage, sandy soil, and water deficiency),and irrational land-use practices and population increases have led to continual expansion of sandy desertification.Therefore, sandy desertification assessment and monitoring are of great concern for researchers, the public, and policy makers.

Since information on land cover and land use is critical to understanding environmental issues and changes in arid and semi-arid regions, remote sensing and GIS are very useful for monitoring land desertification (Hostertet al., 2001). Remote-sensing data are a valuable source for extracting spatially and temporally explicit information(Yanet al., 2007). High-resolution satellite images acquired over long periods provide unique opportunities to study long-term environmental changes (Runnstr?m,2003). In particular, multi-spectral satellite imagery, such as that provided by Landsat MSS, TM, and ETM images,is a precious resource for long-term analyses of surface processes, such as changes in landforms and land desertification, at a regional scale (Libertiet al., 2009). The integration of remote sensing with GIS techniques (Ehlerset al., 1989; Zhou, 1989, Staret al., 1997) is becoming increasingly important for the assessment of environmental changes such as land desertification control (Carretero, 1993).

In this study, we used Landsat MSS, TM and ETM+images, which we obtained from five years (1978, 1990,2000, 2005, and 2010), to develop a sandy desertified land (SDL) database at a 1:100,000 scale through visual interpretation of false-color composite images. We then reconstructed the process of change in SDL in the Shule River Basin during the past 30 years. The purpose of this study was to provide useful information for sandy desertification control and environmental management of the Shule River Basin.

2 Study area

The Shule River in the western part of the Hexi Corridor originates in the Shule Nanshan within the Qilian Mountains, with total length of 670 km. The Shule River Basin, the focus of our study, is located at 38°10′N–42°50′N and 92°35′E–99°50′E, with an area of about 14.68×104km2(Figure 1). According to regional differences in topography, the drainage can be divided into the upstream Qilian Mountains, the middle reaches of the corridor plain (also known as the South Basin), and downstream of the Anxi-Dunhuang Basin (also known as the North Basin).

Figure 1 Location of the study area

The Shule River Basin has a typical continental monsoon climate and there are distinct climate zones within the basin; different regions have different climatic characteristics. South of the basin is the Qilian Mountains alpine area, the climate of which is characterized by low temperatures, low evaporation, and relatively abundant precipitation. The annual average temperature there ranges from 0 °C to 4 °C, the annual precipitation is 150–500 mm, and the annual evaporation is 1,300–1,700 mm. In the river basin, which is in a warm, temperate,arid region, the dry climate is characterized by scarce rainfall and strong evaporation. The annual average temperature there ranges from 6 °C to 8 °C, the annual precipitation is 50–250 mm, and the annual evaporation is 2,200–2,800 mm (Qu and Yu, 2007). According to the results of our previous study, the landscape of the Shule River Basin is characterized by bare land, desert, grass land,farm land, and forest land (Figure 2). Bare land and desert comprised 79.69% of the area in 2000 and thus was the dominant land cover. In contrast, grass land accounted for 17.97%, farm land was 0.90%, water areas and wetlands were 0.80%, forest land was 0.54%, and 0.11% was built-up land. The main source of the whole river basin system is the water originating from the Qilian Mountains.

Figure 2 Land use and cover type map of the Shule River Basin in 2000

3 Data and methods

3.1 Classification system for desertification

Sandy desertification is defined as a land degradation process that occurs in arid, semi-arid, and sometimes even semi-humid climate zones as a result of various factors, including climate variations and unsustainable human activities (Wanget al., 1999). Evaluation of sandy desertification includes not only the sum area of SDL but also the extent of degradation (Liu and Wang, 2007).

According the classification criteria proposed in previous studies (Zhu and Liu, 1984; Feng, 1985; Donget al., 1995; Wu, 2001; Yanet al., 2009), we chose the proportions of the area covered by shifting sand dunes (sand sheets), wind-eroded areas, and vegetation as the main indices to describe the severity of sandy desertification.The sandy desertification intensities were classified into four levels: extremely severe, severe, moderate, and slight. Details of these classification indices are provided in table 1.

3.2 Landsat images and preprocessing

In this study, we obtained five sets of Landsat MSS,TM and ETM+ remote sensing images provided by the U.S. Geological Survey (http://glovis.usgs.gov) from 1978 to 2010, totaling 56 scenes (14 scenes for each period), which we used to monitor and map the processes and spatial patterns of sandy desertification in the Shule River Basin. Because it was difficult to acquire cloud-free images that covered the whole study area within a given year, due to the large area of study region, some images from the year earlier or later than the specific years were chosen to replace images that could not be obtained in the required years. We also used the Chinese Brazil Earth Resources Satellite images that had spectral characteristics and spatial resolution similar to the Landsat images that had to be replaced. The MSS images were mainly acquired from 1978, the TM images were from 1990,2005 and 2010, and the ETM+ images were from 2000.

Using the Image Analyst function of the Module GIS Environment (MGE) software (Intergraph Corp., Huntsville, Alabama), we obtained the false-color images by stacking near the infrared, red, and green bands, and then we georeferenced and orthorectified the 2000 ETM+images using 50 to 60 ground control points (GCPs) derived from a 1:100,000 topographic map. The mean location error for this georectification was less than 1 pixel(i.e., <30 m). The 1990 TM images were matched with the 2000 images by means of an image-to-image matching method. During the image-matching process, we randomly selected 40 to 50 GCPs in 1990 and in 2000 in order to cover most of the area represented by the two sets of images. The root-mean-square (RMS) error of the geometrical rectification between the two images was 1 to 2 pixels (i.e., <60 m) for plains areas and 2 to 3 pixels(i.e., <90 m) for mountainous regions. We used the same method to process the 2005 and 2010 TM images and the 1978 MSS images, and obtained similar results.

Table 1 Indices used for the classification of desertified land in the Shule River Basin

3.3 Mapping and data processing

We used visual interpretation to derive the SDL information. Although the visual interpretation of Landsat images is labor-intensive and time-consuming, the mapping accuracy of this method is higher than that of image classification using only the algorithms provided by image-processing software because of the low spatial and spectral resolution of Landsat images (Liu, 1996; Zhuanget al., 1999; Yanet al., 2007).

We used the freehand drawing function of the MGE software to delineate and label regions of the TM and ETM+ images by visually interactive interpretation to establish the SDL databases for the five years. Based on the recognition ability of the MSS, TM, and ETM+ images and the accuracy of the mapping, the manual visual interpretation and digitization of the TM and ETM+images was carried out at a scale of 1:100,000, and the MSS images were carried out at a scale of 1:250,000.During interpretation, we adopted the following mapping principles: (1) the minimum mapping patch was 7pixel×7pixel (approximately 2mm×2mm on the maps);(2) the deviation of the delineating locations was less than 1 pixel on the screen; and (3) the accuracy of the labeling patches was greater than 96% based on our ground-truth results.

In addition to the MSS, TM, and ETM+ images, we collected ancillary materials such as regional land-use maps, topographic maps, climatic zone data, vegetation maps, and field survey reports to assist in labeling the map patches during the interpretation process. After we completed the manual visual interpretation, the increases or decreases in SDL (desertification and rehabilitation of land, respectively) from 1978 to 2010 could be obtained by detecting changes in the desertification degree in a time series of SDL databases.

4 Results and discussion

4.1 Status of sandy desertified land

Figure 3 shows the spatial distribution of SDL in the Shule River Basin in 2010. The SDL was mainly distributed in the corridor plains between the Qilan Mountains and the North Mountains, which are along the edge of the sandy desert and the Gobi desert, and in some areas of sparse active dunes in oasis areas. From the administrative perspective, the SDL is mainly distributed in three counties: Dunhuang, Anxi, and Yumen. The desertified land adjacent to oases jeopardizes the stability and development of these oases.

The total area of SDL had increased to 3,557.88 km2in 2010 (Table 2), which accounts for 2.42% of the total 146,800 km2in the study area. The areas of slight, mod-erate, severe, and extremely severe sandy desertification accounted for 830.16 km2, 1,031.85 km2, 1,215.48 km2,and 480.39 km2(23.33%, 29.00%, 34.16%, and 13.50%,respectively) of the desertified land. The severe sandy desertification land was the major type of sandy desertification in the study area.

Figure 3 Map of the distribution of SDL in the Shule River Basin in 2010. The white areas are the land of no desertification

Table 2 Changes in the area of sandy desertified land (SDL) in the Shule River basin from 1978 to 2010

4.2 Sandy desertification trends

During the years from 1978 to 2010, the area of SDL increased by 79.90 km2, which represents a 2.30% increase compared to the area in 1978 (Table 2). This also indicates that the average annual (linear) increase in the area of sandy desertification was 0.07% in this region.The overall change differed among the classes of SDL:-6.81% for extremely severe, -5.62% for severe,-2.14% for moderate, and 33.84% for slight.

Although the area of desertification land in general shows an increasing trend, the trends of desertification land in the different periods are not same. From 1978 to 1990, SDL expanded by 255.34 km2, representing a 7.34% increase in the total area of SDL. The area of SDL increased in all severity classes, and the area of severely desertified land and slightly desertified land increased faster than that in any other category during this period(Table 3). From 1990 to 2000 there was widespread restoration of SDL. During that time the area of SDL decreased by 113.03 km2, a 3.03% decrease, but the increasing trend of extremely severe desertification was contrary to the decreasing trend of overall desertification. The area of extremely severe SDL expanded by 16.43 km2, a 3.18% increase. From 2000 to 2005, the restoration of SDL was slower; although SDL decreased by only 54.64 km2, the degree of SDL quickly reduced.The areas of severely and extremely severely desertified land showed a larger decrease in trend. At the same time,the slightly desertified land showed an increase in trend,but the increase was mainly due to the decrease of the other degrees of SDL. The trend in 2005–2010 was consistent with that in 2000–2005, only the recovery was further slowed.

4.3 Driving forces behind desertification

Human activities and natural factors can both influence sandy desertification at the level of regional landscapes (Yanet al., 2009). In this study, we analyzed the driving forces in two major categories: climate and human activities (political measures were considered as a form of human activity).

Climate is an important driving force for sandy desertification (Zhu and Liu, 1984). Climate affects sandy desertification mainly through the influence of changes in three key factors: wind velocity, temperature, and precipitation (Lancaster and Helm, 2000; Wanget al., 2004;Wanget al., 2005). In this study, we used data from the Dunhuang, Yumen, and Anxi meteorological stations to analyze trends in wind velocity, temperature, and precipitation during the study period, and these factors changed in different ways from 1971 to 2010 (Table 4). From 1971 to 2010, the annual mean wind velocity decreased significantly in all three areas because there is a cubic relationship between the wind’s erosive power and wind velocity (Yanet al., 2009). This suggests that the influence of wind on sandy desertification was the weakest later in the study period. During the study period, due to the overall increasing aridity of the climate, the mean air temperature of the three areas steadily increased throughout the study area (by 1.2 °C to 1.4 °C) from 1971 to 2010, and precipitation generally decreased at the same time except for an increase in the Dunhuang area from 2001 to 2010.

Although the trends of temperature and precipitation can contribute to the increase of sandy desertification,human activities can either exacerbate sandy desertification or mask an increasing trend of natural sandy desertification. Some human activities will accelerate the expansion of sandy desertification, such as the rapid economic development since 1978 during which inappropriate land use (e.g., overgrazing and over-reclamations)took place in the study area, causing sandy desertification to expand rapidly from 1978 to 1990. Conversely, certain other human activities will reverse sandy desertification,such as combating desertification by planting shrubs and trees to fix sand dunes (e.g., the Three-North Shelter Forest Project and watershed rehabilitation programs in areas such as the Shule River Basin). This has resulted in many SDL being turned into grasslands or forestlands by plantings to increase the vegetation cover at the edges of oases and cities. With population growth, many SDLs have been reclaimed as farm land where water-saving agricultural methods are used, or have been converted into built-up land as the result of urbanization.

Table 3 Changes in the area of desertified land in the Shule River Basin from 1978 to 2010

Table 4 Changes of climate factors in the Shule River Basin from 1970 to 2010

5 Conclusions

Sandy desertified land is concentrated at the edges of oases, and sandy desertification processes in the study area have had three stages: rapid expansion before 1990,generally quick restoration during 1990 to 2000, and slow rehabilitation from 2000 to 2010. Although the decreasing trend of wind velocity reduces the potential of wind erosion, the driving factors responsible for sandy desertification are mainly increases in temperature and decreases in precipitation, resulting in the overall increasing aridity of the climate. Human activities can either accelerate or reverse sandy desertification. Inappropriate human activities, such as the extensive reclamation during 1978 and 1990, could have been responsible for the rapid sandy desertification increase, but some planned conservation measures could have slowed or even reversed sandy desertification because the climatic change is not the main driver of land rehabilitation. Unfortunately,such conservation measures to help slow sandy desertification are dependent on current technology; it is difficult to prevent natural sandy desertification due to the fragility of the local ecosystem.

This research was supported by the China National Key Basic Research Program (No. 2009CB421301). We also thank Geoff Hart and Anda Divine for their detailed edits of our manuscript.

Carretero EM, 1993. Regional development and desertification control through ecological farming: Three examples in Argentina. Desertification Bulletin, 23: 34–38.

Dong YX, Liu YZ, Liu YH, 1995. On Some Problems of Desertification. Xi’an Map Publishing House, Xi’an, China, pp. 141–156.

Ehlers M, Edwards G, Bedard Y, 1989. Integration of remote sensing with geographic information systems: A necessary evolution. Photogrammetric Engineering and Remote Sensing, 55(11):1619–1627.

Feng YS, 1985. Compilation of the map of desertification and its indications. Journal of Desert Research, 5: 15–21.

Guo J, Wang T, Xue X,et al., 2010. Monitoring aeolian desertification process in Hulunbir grassland during 1975–2006, northern China.Environmental Monitoring and Assessment, 166(1–4): 563–571.DOI: 10.1007/s10661-009-1023-5.

Hostert P, Roder A, Jarmer T,et al., 2001. The potential of remote sensing and GIS for desertification monitoring and assessment. Annals of Arid Zone, 40(2): 103–140.

Lancaster N, Helm P, 2000. A test of a climatic index of dune mobility using measurements from the southwestern United States. Earth Surface Processes and Landforms, 25(2): 197–207. DOI: 10.1002/(SICI)1096-9837(200002)25:2<197::AID-ESP82>3.0.CO;2-H.

Liberti M, Simoniello T, Carone MT,et al., 2009. Mapping badland areas using LANDSAT TM/ETM satellite imagery and morphological data. Geomorphology, 106(3): 333–343. DOI:10.1016/j.geomorph.2008.11.012.

Liu JY, 1996. Macro-Scale Survey and Dynamic Study of Natural Resources and Environment of China by Remote Sensing. China Science and Technology Press, Beijing, pp. 113–124.

Liu SL, Wang T, 2007. Aeolian desertification from the mid-1970s to 2005 in Otindag Sandy Land, northern China. Environmental Geology, 51(6): 1057–1064. DOI: 10.1007/s00254-006-0375-1.

Qu JX, Yu SB, 2007. The chemical characteristics of surface water and groundwater of the Shule River and transforming the relationship between them. Gansu Science and Technology, 23(4):119–121.

Runnstr?m MC, 2003. Rangeland development of the Mu Us Sandy Land in semiarid China: An analysis using Landsat and NOAA remote sensing data. Land Degradation & Development, 14(2):189–202. DOI: 10.1002/ldr.545.

Star J, Estes JE, McGwire KC, 1997. Integration of Geographic Information Systems and Remote Sensing. Cambridge University Press, Cambridge, UK.

Wang T, Zhao HL, Xiao HL, 1999. Advances in desertification research of China. Journal of Desert Research, 19(4): 299–311.

Wang XM, Dong ZB, Liu LC,et al., 2004. Sand sea activity and interactions with climatic parameters in the Taklimakan Sand Sea,China. Journal of Arid Environments, 57(2): 225–238. DOI:10.1016/ S0140-1963(03)00096-X.

Wang XM, Dong ZB, Yan P,et al., 2005. Wind energy environments and dunefield activity in the Chinese deserts. Geomorphology,65(1): 33–48. DOI: 10.1016/j.geomorph.2004.06.009.

Wu W, 2001. Using the TM image for monitoring land desertification.Remote Sensing Technology and Application, 16(2): 86–90.

Yan CZ, Song X, Zhou YM,et al., 2009. Assessment of aeolian desertification trends from 1975s to 2005s in the watershed of the Longyangxia Reservoir in the upper reaches of China’s Yellow River. Geomorphology, 112(3): 205–211. DOI: 10.1016/j.geomorph.2009.06.003.

Yan CZ, Wang T, Han ZW,et al., 2007. Surveying sandy deserts and desertified lands in northwestern China by remote sensing. International Journal of Remote Sensing, 28(16): 3603–3618. DOI:10.1080/01431160601024218.

Yang X, Ding Z, Fan X,et al., 2007. Processes and mechanisms of desertification in northern China during the last 30 years,with a special reference to the Hunshandake Sandy Land,eastern Inner Mongolia. Catena, 71: 2–12. DOI:10.1016/j.catena.2006. 10.002.

Zhou QM, 1989. A method for integrating remote sensing and geographic information systems. Photogrammetric Engineering and Remote Sensing, 55: 591–596.

Zhu Z, Wu H (eds.), 1998. Study on Combating Desertification/Land Degradation in China. China Environmental Science Press, Beijing.

Zhu ZD, Liu S, 1984. The concept of aeolian desertification and the differentiation of its development. Journal of Desert Research,4(3): 2–8.

Zhuang DF, Liu JY, Liu ML, 1999. Research activities on land-use/cover change in the past ten years in China using space technology. Chinese Geographical Science, 9(4): 330–334.