Introduction

Aeolian desertification can be defined as a land degradation process that occurs in arid, semi-arid, and dry subhumid regions, as a result of various factors, including climate change and unsustainable human activities (UNEP 1994; Wang and Zhu 2003; Wang et al. 2012; Gad and Abdel-Samie 2000; Huang and Siegert 2006). It is one of the most serious environmental and socioeconomic problems currently observed especially in arid, semi-arid, and dry subhumid zones. In these zones, aeolian desertification destroys land resources, reduces ecosystem productivity, depresses ecosystem services, and exacerbates poverty. It also induces the instability of political systems in many developing countries. China is one of the more seriously affected countries by the hazard of aeolian desertification. The occurrence and development of aeolian desertification causes serious harm to the ecological environment, the natural resources, the socioeconomic activities, and the people’s lives in China’s vast desert areas (Li et al. 2007a; Zhang et al. 2012; Runnstrom 2000; Michael and David 1994).

Wind erosion has expanded desertification rapidly throughout northern China in the past 50 years. Wang et al. (2004a) monitored and assessed the dynamic change process of aeolian desertification using remote sensing and field investigation. Their results showed that (1) the area of aeolian desertified land developed at an increasing rate of 1,560 km2 year−1 from the late 1950s to the mid-1970s, and the average annual increasing rate reached 2,100 km2 year−1 from 1976 to 1988 and 3,600 km2 year−1 from 1988 to 2000; (2) aeolian desertified land has been under accelerated development during the last 50 years; and (3) the area of aeolian desertified land expanded to 38.57 × 104 km2 by the end of 2000 in northern China. The area of aeolian desertified land decreased at the rate of 1,280 km2 year−1 after 2000, which was reduced to 37.59 × 104 km2 by the end of 2010 (Wang et al. 2011). Although the desertification showed a reverse trend after 2000, the desertification situation remains severe in northern China.

The Horqin Sandy Land is a typical agropastoral transitional zone and is an ecologically fragile region in northern China. As one of the four larger sandy lands and the most seriously desertified areas in China, it is also a typical area of the sandstorm source control project around the Beijing–Tianjin region. In the recent 50 years, desertification rapidly aggravated in the Horqin Sandy Land because of increasing drought and irresponsible human activities, such as over-reclaiming, overgrazing, and overcutting (Wu 2003; Chang et al. 2003). Wang et al. (2004b) showed that the fast desertification expansion occurred between the late 1950s and the mid-1970s and the continuous increase between the mid-1970s and the late 1980s. However, desertification showed a reverse trend in the 1990s. Although the trend of desertification exacerbation has been controlled, and the desertification area has been gradually reduced, the desertification status remains severe in 2000. In some regions, the desertification process and its causes still need to be continually given higher attention. However, few studies have provided integrated assessments on the spatial and temporal dynamics as well as the driving forces of desertification in the Horqin Sandy Land after 2000. Therefore, continuous monitoring for the desertification dynamic process is particularly important in the following decades in this region.

Multitemporal remote-sensing data are the perfect source for extracting spatially and temporally explicit information that can be used to monitor and assess regions threatened by aeolian desertification (Gao and Liu 2010; Hostert et al. 2001; Yan et al. 2009). High-resolution satellite images provide unique opportunities to study long-term environmental changes with high monitoring frequency and to cover areas appropriate for monitoring the environment in a large geographic zone (Runnstrom 2003; Yan et al. 2007; Dawelbait and Morari 2012). Several previous studies have investigated the temporal and spatial changes of desertification on different scales using multispectral satellite imagery, such as Landsat Multispectral Scanner (MSS), Thematic Mapper (TM), and Enhanced Thematic Mapper (ETM) images (Liu et al. 2008; Li et al. 2007b).

This paper aims (1) to monitor the spatiotemporal patterns and dynamics of aeolian desertification with Landsat MSS, TM, and ETM data from 1975 to 2010 in the Horqin Sandy Land and (2) to elucidate the driving force of aeolian desertification and its sensitivity to climate change and human activities. The results will help regional managers to comprehend the complex desertification process and its dynamic mechanisms. Further, the results aim to provide useful information in controlling and managing desertification in this region.

Materials and Methods

Study area

The Horqin Sandy Land is located in the western part of the Northeast Plain, in the transition zone between the Inner Mongolian Plateau and the Northeast Plain in China. The study area extends from east of Qilaotu (in the Yanshan mountains) to west of the Songliao Plain and from south of the Nuluerhu Mountains to north of the Daxinganling Mountains (41° 41′–46° 05′ N, and 117° 49′–123° 42′ E), with an approximate area of 12.51 × 104 km2 (Fig. 1). It belongs to the semi-arid continental monsoon climate in the temperature zone (Du et al. 2009; Li et al. 2003; Li et al. 2004). The annual mean temperature is about 3–7 °C, and the mean annual precipitation is about 350–500 mm. The annual potential evaporation ranges from 1,500–2,500 mm, which is more than five times that of the mean annual precipitation. The mean annual wind velocity is 3.4–4.4 m s−1, and the prevailing wind directions are northwest in winter and spring, southwest to south in summer and autumn (Ren et al. 2004; Zhao et al. 2000; Liu et al. 2009). The days of threshold wind velocity for sand movement (≥5 m s−1) is seen in about 210–310 days per year (mainly in spring and in winter). Gales (wind velocity ≥17 m s−1) occur 25–40 days per year, and sand dust storms occur about 10–15 days per year, mostly in spring (Zhang et al. 2004; Li et al. 2009).

Fig. 1
figure 1

Sketch map of the study area

Full size image

Zonal soils are primarily chestnut soil, chernozem, and chestnut cinnamon soil. Due to the impact of aeolian desertification, however, a large part of the soil has been degraded to aeolian sandy soil (Wang 2011). The native vegetation is the special sparse-tree grassland. Due to human destruction and aeolian desertification, most of the original vegetation has evolved to be sandy vegetation at different degradation stages over the past several decades (Zhao et al. 2009; Li et al. 2002). Under the impact of strong winds, the study area is vulnerable to threats of sand dust storms and desertification.

Data and processing

The remote-sensing data used in this study included 80 scenes (20 scenes from each period) of four sets of Landsat MSS and TM/ETM remote sensing data in 1975, 1990, 2000, and 2010. The remote sensing data used in 1990 and 2010 were TM images, and those used in 2000 were ETM images with spatial resolution of 30 m. MSS images were mainly acquired in 1975 with spatial resolution of 79 m. The principle of selecting images mainly includes two aspects. One is imaging time. July to September is the most suitable period for the research. During these months, vegetation grows more abundantly, making the distinction between aeolian desertified land and other land use types easy. The other aspect is the presence of clouds. The selected images must have less than 10 % cloud coverage, so that they are conducive to identification and extraction of desertification information. The data used in this study mainly came from the site http://glovis.usgs.gov/. Obtaining cloud-free images that cover the entire study area within a given year was difficult because of the large geographical area covered by the region under investigation. Consequently, some images from previous or subsequent years were chosen to replace unsuitable images. In addition to the remote-sensing data, ancillary materials, such as climatic zone data, vegetation maps, land cover maps, topographic maps, and field survey reports, were used as supplementary data sources to assist in labeling the degree of desertification in each mapped area during the interpretation process.

The images were georeferenced to the Albers Equal Area Conic coordinate system using 30 to 40 ground control points based on 1:100,000 scale topographic maps, using the nearest-neighbor resampling approach. The root mean square error was less than one pixel, which is sufficient for multitemporal comparisons. Landsat images were combined with synthesized image band colors (TM4 (R), 3 (G), 2 (B)) for different scenes along the same path. They were used to identify specific physical attributes directly, and they may be easily interpreted for aeolian desertified land. The relationship between surface desertified land features of different degrees and desertification types of the remote-sensing images was established using a portable global positioning system (GPS) receiver. On the other hand, the interpretation marks of aeolian desertified land were established.

The human–computer interactive visual interpretation is one of the most popular information extraction methods, which based on the freehand drawing assistance of the ArcMap 9.3 software to obtain desertification information. According to the established interpretation marks and the geography comprehensive analysis method, the aeolian desertified land vector databases of different periods were acquired, topology relationship was established, and error correction was accomplished in an ArcInfo 9.3 environment.

After interpretation, we performed field verification on the results from August 7 to September 3, 2011, to ensure accuracy of the results. After verification, we used 120 random samples to assess accuracy. The overall accuracy was greater than 95 %, which satisfies research requirements. If the overall accuracy did not satisfy the mapping standards, the map was corrected and rechecked until the mapping accuracy reached the mapping principles. To reveal the change trends and the characteristics of aeolian desertified land, we used the overlay analysis module of ArcGIS in obtaining the dynamic patches.

The meteorological data used in this study came from China Meteorological Data Sharing Service System (http://cdc.cma.gov.cn/), According to observations from local meteorological stations, the wind speed, temperature, and precipitation data recorded at six stations are the typical factors that affect the climate in the study region from 1970 to 2010. Using the average value from the six stations, we studied and analyzed the climatic overall change trends in the past 40 years.

Since the wider scope of the study area, it is difficult to acquire the statistical data of each county in the recent 40 years. Therefore, we selected six typical counties to analyze the causes of desertification by screening statistics.

Classification system for desertification

The classification system for desertified land is an objective reflection of the extent of land degradation in regions where desertification occurred. Establishing a scientific classification system for desertified land is the key to monitor the dynamic changes during desertification (Zhang et al. 2009). The proportion of mobile sand dunes or wind-eroded lands accounts for the total area, and the vegetation coverage and surface landscape can visually reflect the desertification status. Referring to the existing classification observable characteristics of land desertification in the Horqin Sand Land in previous studies (Wu 2001; Han et al. 2010), we adopted the three criteria above to define the aeolian desertification intensities as slight, moderate, severe, and extremely severe (Table 1 and Fig. 2). At the same time, based on the judging characteristics (such as tone or color, shadow, size, shape, texture, site, and association) of different aeolian desertification types in the remote sensing image, the landscape of the aeolian desertified land is divided into three categories according to cause of aeolian desertification (Wang et al. 2008; Han et al. 2010): (1) dune reactivation or shifting sand intrusion, (2) shrub encroachment desertification, and (3) non-irrigated farmland desertification (Table 1).

Table 1 Classification system of aeolian desertification by remote-sensing monitoring in the Horqin Sandy Land
Full size table
Fig. 2
figure 2

Landsat image features of different desertification intensity in the Horqin Sandy Land (a–d correspond to the types of desertification intensity in Table 1; a slight desertified land, b moderate desertified land, c severe desertified land, and d extremely severe desertified land)

Full size image

Desertified land center of gravity migration model

With reference to the distribution principle of population gravity center in demographic geography, relevant scholars introduced the center of gravity migration model in research on farmland, soil erosion, land use, and economic and demographic change to analyze the spatial characteristics of land use change and human activities (Gao et al. 1998; Wulantuya 2000; Huang et al. 2006). The centroid migration of desertified land is a major aspect of the study of desertification spatial changes (Bao et al. 1998; Jia and Chang 2007). In this paper, the center of gravity migration model was used to calculate the centroid coordinate of different degrees of desertified land and to describe the overall change trends and spatial variation characteristics of different degrees of desertified land through migration direction and distance of centroids. The formula for the model was described by Wang and Bao (1999) as follows:

$$ {X}_t={\displaystyle \sum_{i=1}^n\Big(}{C}_{ti}\times {X}_{ti}\Big)\raisebox{1ex}{$$}\!\left/ \!\raisebox{-1ex}{$$}\right.{\displaystyle \sum_{i=1}^n{C}_{ti}} $$
(1)
$$ {Y}_t={\displaystyle \sum_{i=1}^n\Big(}{C}_{ti}\times {Y}_{ti}\Big)\raisebox{1ex}{$$}\!\left/ \!\raisebox{-1ex}{$$}\right.{\displaystyle \sum_{i=1}^n{C}_{ti}} $$
(2)

where X t and Y t are the latitude and longitude centroids of the desertified land in year t, respectively; n is the number of patches of the desertified land; C ti is the patch area of landscape i in year t; and X ti and Y ti are the latitude and longitude centroids of landscape i, respectively.

Results and discussion

Current status and spatial distribution of aeolian desertified land

The current status and spatial distribution characteristics of aeolian desertified land were analyzed using the monitoring results in 2010. The total area of aeolian desertified land was 30,199 km2 in 2010, which accounted for 24.1 % of the total area. The areas of desertified lands according to the categories of slight, moderate, severe, and extremely severe were 13,318, 8,066, 6,223, and 2,592 km2, respectively (Table 2), accounting for 44.1, 26.7, 20.6, and 8.6 % of the desertified land in the study area, respectively. Slight desertified land is the most widely distributed type. The desertification area gradually decreased from slight to extremely severe. Figure 3 shows the spatial distribution of the aeolian desertified land in the Horqin Sandy Land in 2010. Slight and moderate desertified lands were mainly distributed over the mid-eastern regions. Severely desertified land was fragmented and was mainly distributed over Ongniud Qi and Horqin Zouyi Houqi. Finally, extremely severe desertified land was continuously distributed over the mid-eastern part of Ongniud Qi and the northwest region of Naiman Qi, located on both sides of the Laoha River, the area of which accounted for 73.5 % of the total extremely severe desertified land in this region. The extent of the aeolian desertified lands in the midwestern region is much smaller than that in the mid-eastern region, but the severity of desertification was higher than that in the mid-eastern region.

Table 2 Area of aeolian desertified land in the Horqin Sandy Land in 2010
Full size table
Fig. 3
figure 3

Spatial distribution map of aeolian desertified land in the Horqin Sandy Land in 2010

Full size image

Dynamics of aeolian desertified land from 1975 to 2010

During the past 35 years (from 1975 to 2010), the area of aeolian desertified land in the Horqin Sandy Land increased from 1975 to 1990 and then gradually decreased from 1990 to 2010 (Table 3). The total area of aeolian desertified lands in 1975, 1990, 2000, and 2010 was 30,884, 32,071, 31,762, and 30,199 km2, respectively, accounting for 24.7, 25.6, 25.4, and 24.1 % of the total area, respectively. Aeolian desertified land obviously expanded from 1975 to 1990. The average annual rate of linear increase was 79.1 km2 year−1 during this period. In addition to slight desertified land, other types expanded to different degrees. The area under extremely severe desertification increased faster than that of any other types during this period.

Table 3 Changes in desertified land areas in the Horqin Sandy Land from 1975 to 2010
Full size table

Since 1990, the area of desertified land has gradually decreased. From 1990 to 2000, desertified land decreased by 309 km2. However, severe and extremely severe desertified lands both obviously expanded, the severity of desertification was still aggravated in some regions, which are mainly associated with the severe drought event in the north China in 2000. The reduction of the total area of desertification mainly was the result of the activities of human governance. Control over desertification was mainly concentrated in the relatively easy management regions that are mainly dominated by slight desertified land.

The area of aeolian desertified land from 2000 to 2005 decreased more rapidly than that from 1990 to 2000. In particular, desertified lands that are classified as severe and extremely severe began to reverse; these decreased by 670 and 487 km2, respectively. The severity of desertification obviously reduced at the rate of 110.3 km2 year−1.

The area of aeolian desertified land decreased to 1,012 km2 at the rate of 202.4 km2 year−1 from 2005 to 2010. Among the four aeolian desertified land types, except for the slight desertified land, the area with other degrees of desertification had decreased; moderate desertified land accounted for most of the reduction. Analysis showed that the area of aeolian desertified land had an overall reverse trend during this period and that the extent of the different desertified lands was in a state of stable reversal.

Viewed from the desertification types, the areas that have different aeolian desertified land were determined from 1975 to 2010. Results showed that shrub encroachment desertification is the dominant desertification type. The area of shrub encroachment desertification and non-irrigated farmland desertification increased from 1975 to 1990, and after 1990; both gradually decreased and reached the maximum in 1990. Compared with the data in 1975, the 2010 data showed that areas decreased 757 and 86 km2, respectively. Their change trends were similar to that of the whole desertified land. However, the area of dune reactivation or shifting sand intrusion had volatile increases from 1975 to 2010 (Table 4).

Table 4 Changes in the different desertification types of desertified land in the Horqin Sandy Land from 1975 to 2010
Full size table

Spatial migration of desertified land

Figure 4 shows the spatial pattern of the gravity center of the desertified lands. Extremely severe desertified land was mainly distributed in the southwest of the study area, between the Xilamulun and the Laoha rivers (Fig. 1). Severe and moderate desertified lands were mainly distributed in the mid-east of the study area, which have complex and mixed staggered geographical positions. Slight desertified land was mainly distributed in the northeast of the study area. The spatial distribution trend in the gravity center of the desertified land is similar to the actual distribution of desertified land in space, indicating that the change in gravity center can properly reflect the spatial variation of desertified land.

Fig. 4
figure 4

Spatial changes in the centroid of aeolian desertified lands from 1975 to 2010 in the Horqin Sandy Land

Full size image

From 1975 to 1990, the gravity center of extremely severe and slight desertified land migrated to the northeast 7.76 and 10.49 km, respectively. Severely desertified land migrated 9.32 km to the southeast. However, moderate desertified land had no obvious migration during this period. From 1990 to 2000, the gravity centers of different desertified lands migrated to the northeast, except for slight desertified land that migrated 13.32 km to the northwest. The distance was the maximum compared with the other desertified lands during this period. From 2000 to 2005, the migration distance of the extremely severe desertified land was relatively minor. The gravity centers of the severe, moderate, and slight desertified lands obviously migrated, but they moved in different directions. Severe desertified land migrated 15.91 km to the northeast, moderate desertified land migrated 9.44 km to the northwest, and slight desertified land migrated 14.48 km to the south. From 2005 to 2010, severe and moderate desertified lands migrated more obviously compared with extremely severe and slight desertified lands. Their migration distances were 11.38 and 8.98 km, respectively.

Over the past 35 years, the overall migration of the gravity centers of the extremely severe and severe desertified lands was to the northeast, whereas that of the moderate and slight desertified lands was to the northwest. The migration distances of the extremely severe, severe, moderate, and slight desertified lands were 11.00, 16.58, 14.21, and 4.17 km, respectively. The spatial migrations of the severe and moderate desertified lands were the most obvious. The migration of the gravity centers of the different desertified lands can effectively reflect the spatial variation characteristics of desertification. The centers of gravity of different desertified lands tending to move northward as a whole indicated that the southern desertified lands of the Horqin Sandy Land were improved at some degrees during the last few decades.

Factors affecting desertification

Effects of climate change on aeolian desertification

The development of desertification is controlled by climate change to a large extent (Wang and Ha 2004). Among the climate factors, precipitation, temperature, and wind are the more important contributors to desertification (Lancaster and Helm 2000; Wang et al. 2005). The Horqin Sandy Land is located at the agropastoral transitional zone, which is sensitive to the global climate change. Climatic and environmental changes in the Horqin Sandy Land show a certain degree of particularity.

The change trends in annual maximum wind velocity and mean wind velocity were analyzed in the study area (Fig. 5). Results showed that the annual maximum wind velocity and mean wind velocity underwent overall significant (P < 0.01) decreasing trends in recent 40 years. The rates of decrease were −0.155 and −0.022 m/s year−1, respectively. From 1970 to 1989, the annual mean wind velocity showed an insignificant but slight increasing trend, and the annual maximum wind velocity slowly decreased. The change trends in the two factors were not obvious. Therefore, velocity changes do not play a significant role on the reversion of desertification. After 1989, the annual mean wind velocity and maximum wind velocity showed a significant (P < 0.01) downward trend. The influence of wind on desertification was decreasing during this period, which was entirely consistent with the desertification trends. Land surface was mostly bare and vulnerable to wind erosion in winter and spring. In particular, sandstorm occurs when sand dunes and semimobile dunes are eroded by the wind. Strong winds that accompany sandstorms cause serious harm on land use/cover and agriculture production.

Fig. 5
figure 5

Annual maximum wind velocity and annual mean wind velocity and their trends from 1970 to 2010 in the Horqin Sandy Land

Full size image

The temperature in the study area showed a significant (P < 0.01) increasing trend (Fig. 6). The annual mean growth rate was 0.041 °C year−1, which is higher than China’s average rate in the past 40 years (0.004 °C year−1) and that of the global average warming rate in the past century (0.005 °C year−1) (Yin 2002; Shi et al. 2003; Yang and Yan 2002). The changes in temperature had a change trend similar to that of the global temperature. There is a general trend of fluctuant decrease in precipitation during the last 40 years (Fig. 6). The annual mean precipitation was 359.71 mm, with an annual decrease rate of −1.267 mm year−1. The highest precipitation was 534.3 mm in 1998, and the lowest was 244.02 mm in 2009. The difference was even greater than that in the lowest year. The variation in annual mean precipitation was large, which easily led to frequent droughts and floods. Regional climate was developing toward warmer and drier climate during the study period, as indicated by rising temperature, decreasing precipitation, and drying up of lakes and reservoirs. The ecological environment in the area was intrinsically very fragile and was prone to desertification which was characterized by wind erosion. At the same time, climate warming and drying caused drought that could bring harm to agricultural and livestock production. The combined effect of precipitation and temperature accelerated the development of desertification between 1975 and 1990 in the study area. Desertification rapidly increased during this period. Although the increasing temperature and decreasing precipitation should also have contributed to the increased desertification from 1990 to 2010, desertification in the study area showed a reverse trend after 1990. This is consistent with the results of Wu (2003) and Wang et al. (2004b) in this region. This phenomenon is the remarkable result of reasonable human economic activity and desertification control measures (Wang et al. 2004b). In the background of natural environment, human activity converted the negative effects into positive effects during the process of desertification.

Fig. 6
figure 6

Annual temperature and precipitation and their trends from 1970 to 2010 in the Horqin Sandy Land

Full size image

Effects of human activity on aeolian desertification

From the mid-1970s to the late 1980s, desertification rapidly developed in the Horqin Sandy Land. The main driving factor for desertification is irresponsible human activities, which include over-reclaiming, overgrazing, and excessive firewood cutting that cause rapid desertification expansion. Statistics show that population density had a significant increasing trend in the past 40 years in the typical regions of the Horqin Sand Land. Population density increased from 35.03 people per km2 in 1970 to 58.75 people per km2 (an increase of 67.7 %) in 2010 (Fig. 7). With rapid population growth, food requirement also increased. Farmland areas significantly increased after 1990 (Fig. 7), from 64.11 × 102 to 99.81 × 102 km2 (an increase of 55.7 %) in the past 40 years. To satisfy current demands and to obtain more economic benefits, the local government reclaimed a large area of grassland, which was then abandoned after 1 to 3 years. Surface vegetation and soil were destroyed by intensive human activities and irresponsible economic activity, which led to subsequent serious aeolian erosion and desertification. Desertification exacerbated during the dry and windy periods. Pasture over-utilization and overgrazing caused grassland degradation.

Fig. 7
figure 7

Change in the main anthropogenic factors in the Horqin Sandy Land over the last 40 years

Full size image

According to statistics, the total number of livestock significantly increased from 2.32 million heads in 1970 to 9.50 million heads (an increase of 309.5 %) in 2010 (Fig. 7). The grazing rates far exceeded the stocking capacity of pasture. Zhao et al. (2002) proposed that the mechanism of overgrazing led to the desertification of sandy grassland. The increasing of grazing intensity make the surface baring, and wind erosion appeared rapidly under wind action after ground surface baring, which resulted in the grassland degradation and desertification. The fuel demand continuously increased with the growth of the population. Thus, excessive firewood cutting became common, which is also one of the reasons for the destruction of natural vegetation. Serious firewood gathering not only destroyed the loose structure of the forest and affected the growth of the trees, but also it directly exposed the soil to wind erosion. All of these factors led to the degradation of forests and forest resources. Therefore, overcultivation, overgrazing, and excessive firewood gathering are the three main factors that lead to eco-environmental degradation and desertification in the Horqin Sandy Land.

After the 1990s, climatic factors played a beneficial role in desertification development. However, the desertification process showed a reverse trend during this period, primarily because the Central Government and the Inner Mongolian Autonomous Region had a better understanding of desertification hazards and paid much attention to desertification control. At the same time, a series of antidesertification projects were carried out, including natural forest protection projects, reforestation, the Beijing–Tianjin sandstorm source control project, the eco-environment treatment project, the natural grassland protection and construction project, and the Three-North Shelter Forest Programme. From the central to the local governments, as well as to the local farmers, all stakeholders increased their financial, material, and manpower investments on reasonable adjustments for land use structure and strength. Areas of low-quality farmland were compressed. Arable lands with viable soil and water conditions were intensively managed. Pasture fencing was implemented in large areas. Overgrazing and deforestation were prohibited. The series of positive human activities not only slowed down the rate of aeolian desertification, but also controlled the desertification of a fairly large area. The driving effect of climatic factors on desertification development was offset by the positive of human activities such that desertification showed a reverse trend, especially since 2000.

Conclusions

On the basis of RS and GIS technologies, we monitored and analyzed the temporal and spatial dynamics of desertification and its driving factors using remote sensing images, as well as natural and socioeconomic data. The results showed that aeolian desertified land in the study area covered 30,199 km2 in 2010, accounting for 24.1 % of the total area. The aeolian desertified land obviously expanded from 1975 to 1990 and gradually decreased from 1990 to 2010. The human activities had greater impact on aeolian desertification than climate change. By analyzing the gravity centers spatial–temporal changes of different desertified lands during the past 35 years, the southern desertified lands were improved during the last few decades. Compared with traditional methods that usually discuss the desertification changes in the area, combination of area and gravity centers can easily and visually display the spatial variation characteristics of aeolian desertified lands across different periods.

Although the climatic variation in the past 35 years has been favorable to desertification development in the Horqin Sandy Land, the desertification in the study area still showed a reverse trend after 1990. Compared with other’s area studies (Liu et al. 2008; Yang et al. 2007; Zhang et al. 2003), like the Otindag, Hunshandake, and Mu Us Sandy Land in northern China, the variability of climate had some impact on aeolian desertification, but the processes of aeolian desertification have been forced mainly by human activities, and policies played an important role in controlling aeolian desertification. In this research, desertified land rapidly expanded from 1975 to 1990 because the combination of climate change and unreasonable human activities. After the 1990s, and especially since 2000, a series of antidesertification and ecological restoration projects were enacted by local and national governments which aims to control desertification and restore natural vegetation. Under the impacts of policies, the aeolian desertification was gradually alleviated. Given the fragile ecological environment and intensive human activities, though some political measures have had many successes in controlling desertification, desertification control is a difficult and long process, so desertification will still threat the regional eco-environment. Therefore, we suggest that these effective measures should be implemented continually, and which should include mechanisms both to protect existing achievements and prevent further land desertification.