Catena 88 (2012) 37–45

Contents lists available at SciVerse ScienceDirect

Catena

j ourna l homepage: www.e lsev ie r .com/ locate /catena

Temporal-spatial variability of desertification in an agro-pastoral transitional zone ofnorthern Shaanxi Province, China

Yanbing Qi a,⁎, Qingrui Chang a, Keli Jia a,b, Mengyun Liu a, Jing Liu a, Tao Chen a

a College of Resources and Environment, North-West A&F University, 3 Taicheng Road, Yangling, 712100, Chinab College of Resources and Environment, Ningxia University, Yinchuan, 712100, China

⁎ Corresponding author. Tel.: +86 29 87082912; fax:E-mail address: ybqi@nwsuaf.edu.cn (Y. Qi).

0341-8162/$ – see front matter © 2011 Elsevier B.V. Alldoi:10.1016/j.catena.2011.08.003

a b s t r a c t

a r t i c l e i n f o

Article history:Received 20 August 2010Received in revised form 22 March 2011Accepted 22 August 2011

Keywords:DesertificationSpatio-temporal variationAgro-pastoral transitional zone

Desertification has been widely treated as one of the major environmental hazards in the world by scientificcommunities and the public. Monitoring the dynamics and causes of desertification is essential to provide im-portant instruction for desertification control strategies and rational planning of land use in arid and semi-arid areas. In order to assess temporal–spatial variability of desertification from 1986 through 1993 to2003 in an agro-pastoral transitional zone of northern Shaanxi, China, satellite images were interpretedand analyzed along with meteorological and socio-economic data. During the intervening 17-year period, de-sertification has decreased in severity and increased in extent. Temporally, severe desertification has de-creased in area by 26.8% from 1986 to 2003. Desertification landscape patch quantities and fragmentationindexes have increased gradually. Spatially, desertification has increased in extent north of the Great Walland decreased in extent in the Wuding river valley and north-western Shenmu County. The geographic centersof deserts in the study area have moved to the south-west and north-east. Climate change and human activitieswere somewhat responsible for the decrease in desertification severity but significantly affected the increase indesertification extent. The “Three North Shelterbelt Program” and “Grain-for-Green Project” carried out by thegovernments were the dominant contributors to the desertification severity reversal. Desertification control isa difficult and gradually process and the government should continue to play and enhance the leading role inthis process.

+86 29 87080055.

rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Increasingly, human activities are changing the natural ecology andenvironment of the planet. One such change, desertification, is widelyrecognized as one of the most critical environmental hazards (Liu andDiamond, 2005; Luo, 2003; Mainguet, 1991; Zhu and Chen, 1994; Zhuand Liu, 1981). The increasing rate of desertification implies a clearmanifestation of human influenced, climatic-changing processes onthe environment (Collado et al., 2002). There is an increasing need forresearch into the dynamics and causes of desertification to provide im-portant instruction for desertification control strategies and rationalplanning of land use in arid and semi-arid areas (Wu and Ci, 2002).Therefore, considerable research and numerous studies have been de-voted to desertification assessment and monitoring. Additionally, thisresearch has informed policy-makers and raised awareness among thepublic.

Many researchers have givenmore attention to the dynamics of de-sertification in northern China by using aerial photographs and satelliteimages and predicted their developing trends (Huang et al., 2009; Liu

et al., 2008; Sun et al., 2006; Wu and Ci, 2002; Yan et al., 2009; Yanget al., 2007; Zhao et al., 2005). Hanafi and Jauffret (2008) assessed thedesertification processes based on vegetation dynamics from 1975 to2000 in the arid steppe of southern Tunisia. Rasmussen et al. (2001) ob-served the desertification reversion in northern Burkina Faso from 1955to 1994. However, few integrated analyses of desertification develop-ment and its causes have been undertaken (Wu and Ci, 2002).

Desertification has been defined as land degradation in arid,semi-arid, and dry sub-humid areas resulting from various factors,including climatic variations and human activities (INCD, 1994). Un-fortunately, no consensus has been reached regarding the exact roleof these two factors (Mainguet and Da Silva, 1998; Sun et al., 2006).Natural variables played a major role in historic desertification analysis(Sun et al., 2006), but seemingly, anthropogenic factors were prior tonatural factors regarding desertification variation, as pointed out bymany studies. Improper agricultural practices, overgrazing, mining,fire management schemes, recreation practices, deforestation, urbani-zation, and the introduction of exotic species have been suggested ascontributing factors of desertification in the United States (Mcclure,1998). Rapid population increase, severe soil erosion, deforestation,low vegetative cover, and unbalanced crop and livestock productionhave been suggested as contributing factors of land desertification inThiopia (Girma, 2001). The main causes of desertification in the Mu

38 Y. Qi et al. / Catena 88 (2012) 37–45

Us Desert seem to be intensified and irrational human activities, such asover-reclaiming, over-grazing, and over-cutting (Wu and Ci, 2002). In-creasingly, human population pressure and rapidly economic develop-ment have shown to be the leading cause of desertification (Ayoub,1998; Darkoh, 1998; Liu and Ci, 2000).

The northern Shaanxi Province, as a representative of agro-pastoraltransitional zone in China, has become seriously degraded in recentlyyears. Since the 1950's, a project to control desertification, the “ThreeNorths Forest Shelterbelt Program”, has been implemented and a sandcontrol station has been constructed. Since 1998, a nation-wide ecologicalrestoration project, named the “Grain-for-Green Project”, was launchedand regulations were enacted by the central government. Meanwhile,rich coal and petrol resources have been exploited since 1990s, whichhas brought economic prosperity to the region. Therefore, the desertareas of northern Shaanxi are a sensitive topicwith regards to desertifica-tion research due to the conflict between desertification controlmeasuresand resource exploration (WuandCi, 2002). Research concerning the for-mation and evolution of deserts, the effects of resource exploitation onsandy soil, and the management and utilization of the farmland providea solid background for further research in this area. However, few studieson desertification in this sensitive area provides an integrated assessmentof the spatial and temporal dynamics and driving forces.

Remote-sensing data are a valuable source for extracting spatiallyand temporally explicit information that can be used to monitor andassess regions threatened by desertification (Hostert et al., 2001;Yan et al., 2007). High-resolution satellite images acquired over longperiods provide unique opportunities to study long-term environ-mental changes (Runnstrom, 2003; Yan et al., 2007). In particular,multi spectral satellite imagery, such as the Landsat MSS and TM orETM data, are a precious resource for long- term analyses of surfaceprocesses, such as changes in landforms and land degradation, at a re-gional scale (Liberti et al., 2009). The comparison of databases ofsandy land for different dates is a straight-forward way to assess de-sertification trends.

Fig. 1. The location of the agro-pastoral transi

The objectives of this paper are to: (1) monitor the temporal andspatial dynamics of desertification from 1986 to 2003 in the sensitivearea of an agro-pastoral transitional zone of northern Shaanxi prov-ince, China, and (2) identify potential causal factors associated withspatio-temporal changes in desertification risk. This study intends toprovide useful information for controlling and managing desertifica-tion in the area. It is hoped that through a better understanding of de-sertification risk at the community level, a better set of correctiveactions and policies can be put in place.

2. Materials and methods

2.1. Study area

The study area is located in Shaanxi Province in northwest China(ranging from 107°35′ to 111°29′E and from 37°35′ to 39°02′N) withan approximate area of 36,136 km2 (Fig. 1). The desert areas of northernShaanxi province lie in an agro-pastoral transitional zone. Elevation inthe region ranges from 800 to 1800 m, sloping generally downwardsfrom the northwest to the southeast. The Mu Us Desert lies to thenorth and a loess plateau lies to the south of the region. The area has atypical continental semi-arid climate and the annual average tempera-ture is about 7.0–9.0 °C with monthly mean temperatures of 24 °C inJuly and −8.5 °C in January. The accumulated active temperature≥0 °C is 3524 to 4541 °C and ≥10 °C is 2847 to 4147 °C. Mean annualprecipitation in the area is 250–450 mm, 70% of which is concentratedbetween June and September. Annual average evaporation is 1152 to1290 mm, or 3–5 times annual mean precipitation. Over 150 dayshave wind speeds of more than 5 m s−1.

The main vegetative cover type is sandy grassland, which coversmore than 80% of the desert area and Artemisia Ordosica is a dominantspecies (Zhang, 1994). Other natural vegetative cover types includemeadows and shrubs. In addition, there is farmland distributedalong the river and interspersed with sandy grassland and artificial

tional zone of northern Shaanxi Province.

39Y. Qi et al. / Catena 88 (2012) 37–45

forests and shrubs (Department of Geography of Peking University,1983).

Arid agriculture is the dominant land use type in thewest (Dingbianand Jingbian), and semi-arid agriculture and semi-arid pastoral are thedominant land use types in the north-east (Yuyang, Jiaxian, Shenmu,Fugu). The main soil types are Orthic Entisols, Sandic Entisols, StagnicAnthrosols and Ustic Cambosols (Fig. 2) (CRGCST, 2001).

2.2. Image processing and data generation

2.2.1. Image and geometric correctionRemote sensing images used in this study are from the Landsat 5

Thematic Mapper (TM) (with path/row 127/33), acquired on 2 August1986, 17 July 1993, and 17 August 2003, respectively, totaling 9 scenes(3 scenes for each period). The spatial resolutions of the TM data are30 m. In order to undertake comparative analysis, the images were co-registered by the Image Analyst function of the Module Geography En-vironment based on ERDAS 8.7 (Yan et al., 2009). Geometric correctionwas carried out using ground control points from topographic mapswith a scale of 1:5000 produced in 1983 to geo-code the image from2003. This image was used based on the quadratic polynomial method(Liu et al., 2008) to register the images from 1986 and 1993. Meticulous,per-pixel assessment showed that root mean-square errors among thethree dates were less than 1 pixel, which is sufficient for multi-temporalcomparisons.

2.2.2. Radiometric correction and spectral transformationIn order to remove or normalize the reflectance variation between

images acquired at different times, relative was performed to yield nor-malized radiometric data on a common scale (Paolini et al., 2006). Here,the histogram normalization, a simpler and more effective technique,was used to carry out the relative radiometric correction based onERDAS 8.5 image processing systems. (Liu et al., 2008).

Fig. 2. Soil map of the agro-pastoral transitio

The K–T (Kauth–Thomas) transformation method was used forthe spectral transformation of the 3 scenes image (Liu et al., 2008)to reduce multi-spectral data volume with minimal information lossand generate a new image which loads main information from theoriginal data, which is an effective technique to improve classificationaccuracy and change detection (Liu et al., 2008; Seto et al., 2002). Anew image was obtained with using synthesis of image band colors(TM4 (R), 3 (G), 2 (B)) for different scenes in the same path, whichwas used to directly identify specific physical attributes and can beeasily interpreted for land cover.

2.2.3. Supervised classification and mappingA human–computer interactive approach was applied for sample se-

lection, area identification, and template and discrimination function es-tablishment (Liu et al., 2008; Zhu and Chen, 1994). Then, supervisedclassification was conducted based on the maximum likelihood classifi-cation method (Paolini et al., 2006). Some classes were spectrally con-fused and could not be separated well by supervised classification andhence visual interpretation was required to separate them. The areawas classified into six main classes: desert, arable land, grassland, forest,civil and industrial, unusable land (Table 1). The initial distribution mapof deserts in different periods of time was generated. After small patcheswere removed and non-type error pixels were marked, a vector desertgrade map was obtained (Fig. 3). The accuracy assessment was per-formed using 150 points in 2003, 120 points in 1993 and 1986 fromfield data, respectively. Errormatrixwas derived and the overall accuracyreached 93.25%, 92.56%, and 90.41%, respectively.

2.2.4. Classification system for desertificationAfter reviewing the classification criteria proposed in previous stud-

ies (e.g., Dong et al., 1995; Feng, 1985; Sun et al., 2008; Wu, 2001; Zhuand Feng, 1984), we chose the proportions of the area covered by veg-etative cover, species composition, and structure of plant communities

nal zone of northern Shaanxi Province.

Table 1Area and percentage of change of different land cover classes of 1986, 1993 and 2003 classified images in an agro-pastoral transitional zone of northern Shaanxi Province.

Class name 1986 (area in ha) Percentage 1993 (area in ha) Percentage 2003 (area in ha) Percentage

Arable 954,523 26.8 867,691 24.4 799,317 22.4Forest 526,844 14.8 720,979 20.2 788,567 22.1Grassland 848,519 23.8 825,871 23.2 779,498 21.9Civil and Industrial 5276 0.1 4933 0.1 46,195 1.3Unusable 284,604 8.0 178,087 5.0 171,471 4.8Desert 909,683 25.5 922,266 25.9 955,990 26.8

40 Y. Qi et al. / Catena 88 (2012) 37–45

as the main indices used to describe the severity of desertification. Thedesertification intensities were defined as severe, moderate, and mild(Table 2). Then, ancillary data and the result of visual interpretationwere integrated with the classification result using ArcGIS in order toimprove the classification accuracy of the classified image.

Using ArcGIS software, a spatial database was established after thevector classification map was edited and processed, and topologicalrelations were created. Temporal and spatial evolution informationwas extracted and thematic maps were generated through spatialanalysis and overlay of the vector classification maps.

2.3. Desert landscape characteristic variation model

Landscape fragmentation analysis is frequently used to interpretthe impact of land cover changes within a landscape, by calculating,for each land cover class, a range of metrics to describe fragmentationand spatial distribution, often using satellite-based land cover classi-fications (Southworth et al., 2004). Analyzing changes in spatial pat-terns over time facilitate the identification of social and biophysicalprocesses that drive these changes (Brown et al., 2000). The land-scape fragmentation index (LFI) is frequently used to indicate land-scape fragmentation degree. The LFI formula as follows.

LFIi ¼ Ni=Ai ð1Þ

where LFIi is the landscape fragmentation index of the i land type,Ni andAi are the patch number and the total area of the i land type, respective-ly. Here, higher LFI indicates a higher landscape fragmentation degree.

Landscape variation in space can also be identified by a change inthe spatial center. This type of analysis was introduced from geo-graphical field of population distribution and the formula has beendescribed by Wang and Bao (1999):

Xt ¼ ∑m

i¼1Cti×Xið Þ=∑

m

i¼1Cti ð2Þ

Yt ¼ ∑m

i¼1Cti×Yið Þ=∑

m

i¼1Cti ð3Þ

Fig. 3. Desertification spatial distribution in 1986, 1993 and 2003 in the agro-pastoral transit

where Xt, Yt, are the latitude and longitude centers of the landscape inthe year t, respectively; Cti is the patch area of landscape i in the yeart,m is the landscape type; and Xi and Yi are the latitude and longitudecenters of the landscape i, respectively.

2.4. Correlation and variance analyses

Natural and society-economic data, such as wind, precipitation,temperature, population, cultivated land area, livestock population,gross industrial output value, which are related to desertification dy-namics were obtained from statistical year books (1949 to 2003) oflocal government. To identify the effects of the possible affecting fac-tors, correlation and variance analyses between desert area variationand natural and anthropogenic variables was conducted using SPSSsoftware.

3. Results

3.1. Spatio-temporal variation of desert land

In generally, desert severity has decreased over the past 17 years,even though the extent of desertification has increased in the area by5.1% (Table 1). Severe desert land area has decreased by 26.8% from1986 to 2003 (Fig. 4), and is no longer the most dominant grade.Area of the moderate desertification decreased from 1986 to 1993and increased again from 1993 to 2003. The extent of moderate de-sert land area has increased significantly from 1986 to 2003 and be-came the most dominant desert grade in 2003. Inverse to themoderate desert land, the area of mild desert increased from 1986to 1993 and then decreased from 1993 to 2003.

Number of desert patches and LFI of the three desertificationgrades have increased significantly from 1986 to 2003 (Figs. 5 and 6).Especially, for mild desert, number of patches increased by 549%from 1986 to 2003. Meanwhile, the average area of the patches de-creased by 63.9%, 65%, and 79.2% for severe, moderate, and mild de-sert grades, respectively. Which indicates an intense cutting degreeof the landscape, obvious fragmentation, and an overall reductionin desertification.

ional zone of northern Shaanxi Province. The white area is the land of no desertification.

Fig. 5. Changes of desertification landscape patches quantity from 1986 to 2003 in theagro-pastoral transitional zone of northern Shaanxi Province.

Table 2Grading system of desertification land in an agro-pastoral transitional zone of northernShaanxi Province.

Desertificationgrade

Description

Mild Sand area is less than 10% of the patch area, vegetation ispredominantly shrubs, vegetation coverage is more than 50%,fixed dunes are the dominant landscape feature.

Moderate Sand area is 10% to 50% of the patch area, vegetation is moderatewith slight shrub predominance, vegetation coverage is 10 to 50%,semi-fixed dunes or several fixed dune dispersions are the dom-inant landscape features.

Severe Sand area is more than 50% of the patch area, vegetation is sparsewith little shrub presence, vegetation coverage is less than 10%,moving dunes or several semi-fixed dune dispersions are thedominant landscapes.

41Y. Qi et al. / Catena 88 (2012) 37–45

Over the past 17 years, the desert areas have mainly been distribut-ed north of the Great Wall, which includes north Dingbian and JingbianCounties, north-west Hengshan and Shenmu Counties, the majority ofYuyang County, and fragmentary distribution in Fugu and Jiaxian.From 1986 to 1993 the severe desert areas decreased, themoderate de-sert areas shifted to the north, and the mild desert areas are fragmen-tarily distributed in Dingbian and Jingbian Counties. Large areas ofmild desertification have become severe in north of the reservoir inYuyang County. Moderate desertification decreased and surface vegeta-tion cover increased in the Wuding river valley and in north-westShenmu County, however, severe desertification significantly increasedin the town of Daliuta. Hengshan, Dingbian, and Shenmu Counties havethe most active desertification areas, where desert land has expandedand is threatening the natural ecology of the loess plateau in the south-east. From 1993 to 2003, large areas of moderate desertification in-creased in the Dingbian–Jingbian belt, and expanded to the south by1993. Severe desertification shifted to the east and large areas of milddesertification, covered by drift dunes, increased in Hengshan County.

From 1986 to 2003, the center of the desert landscape migrated0.18° to the west and south for the severe desert, and the desert landhas expanded to the south-west for 24,293 m totally (Fig. 7). From1986 to 1993, the center of the landscape migrated 0.44° to the eastand 0.34° to the north for moderate desert, and the desert land has ex-panded to the north-east for 64019 m totally. From 1993 to 2003, mod-erate desert has expanded to the south-west for 53,559 m. From 1986to 2003, the center of the landscape migrated 0.13° to the east and0.11° to the north for moderate desert, and 11,693 m migrated to thenorth-east. During the same time period, the center of the landscapemigrated 0.24° to the west and 0.22° to the south for mild desert, and

Fig. 4. Changes in the areas of desertified land in the agro-pastoral transitional zone ofnorthern Shaanxi Province from 1986 to 2003.

32,891 m migrated to the north-west. The migration of the landscapecenters indicates desertification was intensively affected by local natu-ral and anthropogenic factors, which is threatening the environmentalsafety of loess plateau (Jia and Chang, 2007).

3.2. Transformation of land types

Since 1986, increased desert extent has mainly affected farm landand forest land, which accounted for 55.3% and 25.7% of all land areatransformation, but the grassland have not been significantly affected(Table 3). From 1993 to 2003, more than 110,000 ha forest has beentransformed into desert, which accounts for 37% of desert land area in-crease (Table 3). During this period, massive amounts of desert landhave been transformed into other land types through comprehensivereclamation. From 1986 to 1993, desert land was controlled by foresta-tion and development of farm land, while from 1993 to 2003, desertland was reclaimed through vegetative restoration and farming.

During the research period, more than 300,000 ha of severe desertwere transformed into moderate and mild desert, and about 87,000 hamoderate desert were transformed into mild desert (Table 4). Mean-while, more than 170,000 ha moderate and mild desert were trans-formed into severe desert, and about 5,000,000 ha mild desert weretransformed into moderate desert, particularly from 1986 to 1993. Also,about 67,000 ha mild desert were transformed into severe desert.

3.3. Correlation between desertification and affecting factors

The correlation and variance analyses should that significant cor-relations were found between desert extent and both natural and

Fig. 6. Changes of desertification landscape fragmentation from 1986 to 2003 in theagro-pastoral transitional zone of northern Shaanxi Province.

Table 3Transformation matrix of land use type in an agro-pastoral transitional zone of north-ern Shaanxi Province (104 ha).

Land use type Land converted to desert Desert converted to otherland use types

1986–1993 1993–2003 1986–1993 1993–2003

Arable 20.256 13.125 8.992 14.35Forest 4.337 11.177 12.398 2.549Grassland 2.304 1.76 2.777 2.17Civil and Industrial 0.01 0.007 0.039 0.057Unusable 3.175 4.151 4.59 7.727Total 30.082 30.22 28.791 26.853

Fig. 7. Changes of spatial centroid of desertification landscape patches from 1986 to2003 in the agro-pastoral transitional zone of northern Shaanxi Province.

42 Y. Qi et al. / Catena 88 (2012) 37–45

anthropogenic factors. For natural factors, wind velocity, maximumwind velocity and gale weather had the lower correlation coeffi-cients (less than 0.75) and F values (less than 13). Comparatively,annual precipitation and temperature had higher correlation coeffi-cients (−0.908 and 0.987, respectively) and F values (more than15). For anthropogenic factors, correlation coefficients were 0.985,0.981, 0.776, and 0.964 for population, cultivated land area, livestockpopulations, and gross industrial output value, respectively. The cor-relation coefficients were significant at pb0.01 for population, culti-vated land area, and gross industrial output value, other factors weresignificant at pb0.05 except gale weather with pN0.1 (Table 5).

4. Discussion

4.1. Developing trends of desertification

It was recorded during the Tang Dynasty (1400 years ago) that theregion near the Great Wall was a beautiful area with vast pasture andclear streams (Department of Geography of Peking University, 1983).Since then, it has gradually become a mosaic of shifting, semi-fixed,and fixed deserts, due to its semi-arid climate. This has been exacer-bated by war between the Han and minority nationalities living in

Table 4Transformation matrix of desert type in an agro-pastoral transitional zone of northern Shaa

Desertgrade

1986–1993

Severe Moderate Mild Tota

Severe 20.449 7.079 7.768 35.2Moderate 4.84 4.526 6.467 15.8Mild 6.741 1.534 2.776 11.0Total 32.03 13.139 17.011 63.1

northern China (Wu and Ci, 2002), who, following turbulent timeswould send troops and immigrant farmers to reclaim grasslands.Now, the agro-pastoral transitional zone in semi-arid area in northernChina is the fastest developing region of desertification in China andhas received considerable attention (Huang et al., 2009; Wu et al.,1997; Wu and Ci, 2002).

In our research, desertification in the agro-pastoral transitional zonein northern Shaanxi, China is decreasing in severity from 1986 to 2003by much severe desert land area has been converted into moderate andmild grades (Table 3). This is agreement with the results reported byHuang et al. (2009) and Jia and Chang (2007) in the research area.Many researches got the similar results in the near regions of researcharea in the northern China, such as in the Longyangxia Reservoir from1975 to 2005 (Yan et al., 2009), in Minqin County from 1985 to 1997(Sun et al., 2006), in the Otindag sandy land from 1987 through 2000to 2006 (Liu et al., 2008). Many researches reported the rehabilitationof desertification in the desert area abroad, such as Tucker et al. (1991)reported the expansion and contraction of the Sahara Desert from 1980to 1990, Rasmussen et al. (2001) observed the desertification reversionin northern Burkina Faso from 1955 to 1994.

We find the expansion of desertification area from 1986 to 2003 inthe research area, this is agreement with the results reported by Huanget al. (2009) and Jia and Chang (2007) in the research area. But, many re-searches reported the decrease of desertification area, such as in MinqinCounty from 1985 to 1997 (Sun et al., 2006), in Inner Mongolia from1992 to 1996 (Zhao et al., 2005), and in the Otindag sandy land from1987 through 2000 to 2006 (Liu et al., 2008). Maybe the reason is the ex-ploration of coal and petrol resources since the 1990s, which induced thedestruction of surface vegetation. This is also can be seen from the in-crease of landscape fragmentation (Fig. 6).

The decreasing in severity and expansion in area of desertification inthe agro-pastoral transitional zone of northern Shaanxi Province indicatesthat desertification is being controlled, to some degree, through large-scale ecological and environmental government programs. Many desertareas have been covered with higher vegetation coverage. However, inthis fragile area, land surface and vegetation cover is unstable, precipita-tion is decreasing, and forests and grasslands suffering from lack ofwater. This, combined with irrational human activity leaves the ecologyof the study area unstable.

4.2. Affecting factors of desertification

Desertification is a complicated process due to the variety of naturalconditions and human activities involved, and it is difficult so far toreach an agreement about the causes of desertification because themechanism and process of desertification still remain unclear (Wu andCi, 2002) at the level of regional landscapes. In earlier eras, climatechange was considered to be the dominant cause for desertification. Es-pecially during the 1970's, middle-term climate changes and short-term droughts were considered to be the major causes of desertificationin Europe (Rubio and Bochet, 1998). Desertification can, however, be ex-acerbated or triggered by climate variability (Pickup, 1998), even thoughit is difficult to quantify the relative contributions of climate change andhuman activities. Since the 1970's, more researchers have recognizedthe role of human activities on desertification, and the impact of land

nxi Province (104 ha).

1993–2003

l Severe Moderate Mild Total

96 19.667 9.169 6.242 35.07833 2.402 11.014 2.242 15.6595 3.683 3.424 7.529 14.63679 25.754 23.607 16.013 65.373

Table 5Coefficients of correlation and variance analyses of desert extent and affecting factors in northern Shaanxi, China.

Coefficient Windvelocity

Maximum windvelocity

Galeweather

Annualprecipitation

Annualtemperature

Population Cultivated landarea

Livestocknumber

Gross industrialoutput value

R −0.748⁎ −0.738⁎ −0.529 −0.908⁎⁎ 0.987⁎⁎ 0.985⁎⁎ 0.981⁎⁎ 0.776⁎ 0.964⁎⁎

F 12.35 11.76 3.28 31.89 34.58 35.62 38.12 15.51 32.28p 0.002 0.002 0.124 0.002 0.001 0.001 0.001 0.0324 0.002

⁎⁎ Denotes significance at pb0.01; ⁎ denotes significance at pb0.05. Gale weather means the days that wind velocity higher than 17 m/s.

43Y. Qi et al. / Catena 88 (2012) 37–45

use practices on land degradation has been acknowledged (Wu and Ci,2002). A study in Karoo, South Africa shows that land use, not climate,has driven the decline of grasses in this area (Bond et al., 1994). In theCanary-Islands of Spain, awide-range of factors, both natural and anthro-pogenic, induced fragile ecosystems, agricultural desertification and pro-gressive degradation of soil productivity (Rodriguez et al., 1993). In theKalahari Desert of Botswana, boreholes and wells used for supplyingwater to livestock are considered central to the spread of desertification(Perkins and Thomas, 1993). As investigated by the Lanzhou Instituteof Desertification of the Chinese Academy of Sciences in the 1980's,94.5% of desert expansion was caused by the anthropogenic factors,and unreasonable land use by humans was the dominant contributor todesertification in northern China. Sun et al. (2006) reported that a num-ber of agricultural activities are causing desertification inMinqin County,China.

In this study, we analyzed the affecting factors in three major cat-egories: climate, human activities, and political measures.

4.2.1. Climate changeAmong the climate factors, precipitation, temperature, and wind are

the most important contributors to desertification (Lancaster and Helm,2000; Wang et al., 2004, 2005). The agro-pastoral transitional area is lo-cated at the transition zone of the Ordos plateau and the Loess plateau,as well as at the junction of the arid and semi-arid climate regions. Inthe winter and spring, due to powerful cold air masses moving south-east from Siberia, strong winds develop easily. Annual mean wind veloc-ity, maximum wind velocity and gale weather were higher from1970'sto1985's than from1986's to 2003's (Fig. 8). Because there is a cubic rela-tionship between the wind's erosive power and wind velocity and fre-quency, that suggests the influence of wind on desertification wasdecreasing in the study period. The wind change may be responsiblefor the decreasing in severity of desertification in study region.

We used a linear regression model to define the trends in annualmean temperature and total precipitation during the study period(Fig. 9). Temperature increased during the 1980's after a gradual

Fig. 8. Changes in annual mean wind velocity, maximum wind velocity and galeweather from 1949 to 2003 in the agro-pastoral transitional zone of northern ShaanxiProvince.

reduction from 1950 to 1985. With an increase of topsoil tempera-ture resulted from the climate warming, thickening of the seasonalthawing layer and thinning of the permafrost during the warm sea-son are increasing (Yan et al., 2009). A decrease of top soil moistureinduced by the loss of soil water and the increasing evaporationmake vegetation degradation and wind erosion intensified (Wanget al., 2002; Xue et al., 2009). Inversely, precipitation significantlydecreased after a gradual increase from 1950 to 1985 (Fig. 9). Be-cause with an especially dry period during the study period in north-ern Shaanxi province and accelerated desertification, that is onereason that desertified land area was increased (Table 2).

4.2.2. Human activityIn the agro-pastoral transitional zone, human activities, including

population increase, over-grazing, inadequate reclamation, andmining,are significant contributors to modern desertification. The populationhas significantly increased from 689,800 in 1951 to 2,129,900 in 2003.The population density is 58.9 per km2, significantly over the standardproposed by the United Nations (Zhu and Chen, 1994). Increasing pop-ulation induced the increasing need for basic resources, such as food,which caused exploitation of available resources. Subsequently, inten-sive land management induces desertification in fragile environments.

Rich natural grasslands have encouraged livestock production, whichplay an important role in economic development of the research area.Livestock populations increased from 673,000 in 1950 to 1,369,000 in2003. Consequently, over-grazinghas causedwide-spread grasslanddeg-radation. The soil surface has also been destroyed by livestock-inducedcompaction (Wu and Ci, 2002).

To meet population demands, grasslands reclaimed for agricul-ture. Reclaimed grasslands were subsequently eroded or buried byshifting sands due to the shortage of protection measures. Therefore,after reclamation, grasslands were often quickly abandoned.

At the same time, the abundant resources underground, such as coal,petroleum, and natural gas, have been explored since the 1990's, whichhas greatly contributed to local economic development, and subsequently

Fig. 9. Annual temperature and precipitation and their trends from 1949 to 2003 in theagro-pastoral transitional zone of northern Shaanxi Province.

44 Y. Qi et al. / Catena 88 (2012) 37–45

induced urbanization. Mining and urbanization often destroys surfacevegetation, soil structure, and water resources (Wu and Ci, 2002).

4.2.3. Political measuresDesertification processes reflect a feedback mechanism between

human activities and the environment. Fortunately, local officialshave recognized the harm of desertification and paid more attentionto controlling desertification. Planting shrubs to elevate vegetationcoverage is the main approach used for desertification control. Twoprojects have been applied to the agro-pastoral region of northernShaanxi province to control desertification: the “Three Norths ForestShelterbelt Program” and the “Grain-for-Green Project”.

In 1950s, the Chinese Central Government decided to begin atremen-dous afforestation exercise, the Three Norths Forest Shelter belt program,in the Three Norths part of China (Northeast China, North China, andNorthwest China). The key goal of this program in the following decadeswas to improve forest coverage in arid and semiarid China from5% to 15%by using this programs the primary method to combat desertificationand to control dust storms. The project is acclaimed as China's GreenGreatWall andWorld's Best Ecological Project both at home and abroad.Since the 1950's, the “Three Norths Forest Shelterbelt Program” has beenapplied by the government of agro-pastoral transitional zone of northernShaanxi Province. Local government re-forested many areas and alsoused aerial seeding in other areas. The re-forested area of research regionincreased from the late 1950's to the early l960's, but decreased in the1960's and early 1970's. Since the 1980's, due to increased investment,the forestation has increased steadily (Wu and Ci, 2002). Statistically,since the 1980's, 93,750 ha desert land has been controlled, in which10,000 hadesert has become farmland and940,00 ha has been convertedinto forest, and 40,000 ha has been converted into artificial grass. Thevegetation cover increased from 24.5% in 1986 to 74.3% in 2003. This islargely responsible for the reduction in desert severity from 1986 to2003.

Since 1998, a nation-wide ecological restoration project, named the“Grain-for-Green Project”, was launched and regulations were enactedby the central government, which have played an important role in con-trolling desertification in China (Liu et al., 2008; Oñate and Peco, 2005).This project aims to reduce croplandwhich is not suitable for cultivationto restore natural vegetation (Liu et al., 2008). In the research area, a se-ries of measures aimed at lessening disturbance and increasing the cov-erage of natural vegetation, were inacted by local governments tocontrol further ecological degradation. These measures included graz-ing prohibition, plantation of high yield forage grasses, aerial seeding,and adjustment of the socio-economic structure. By the end of 2003,72million ha of farmland in pilot areas had been transformed into forestor grassland. In addition, 79.3 million ha of barren grassland had beenplanted with trees (Liu and Diamond, 2005).

Although the area of afforestation is increasing rapidly as a result ofthe above mentioned projects, the area of degraded land has continuedto expand and the severity of desertification has continued to wakenfrom our research. Maybe we can found some reasons from a survey onthe attitudes of 2,000 farms, by 85.6% had priorities that differed fromthose of the government, toward the “Grain to Grain” project in theNorthern Shaanxi province taken by Cao et al. (2009). From this research,these political measures were somewhat effective in controlling landdegradation. However, these improvements are not sufficient enoughto improve the eco-environment in the short time in this arid andsemi-arid area. There is obviously still a long way to go for the projects(Wang and Zhou, 2003), because restoration of the damaged ecosystemsis a slow process.

5. Conclusions

With the analysis of 3 Landsat 5 TM images, and natural and society-economic data, desertificationprocesses and its control factorswere eval-uated. Desertification in the agro-pastoral transitional zone in northern

Shaanxi, China is decreasing in severity, but expanding to the south-east and threatening the environmental safety of loess plateau. Thoughthe decreasing trend of wind velocity prevents the development of de-sertification, the especially dry period (shown as increasing temperatureand decreasing precipitation), combinedwith intensive human activities,such as population increase, over-grazing, inadequate reclamation, andmining, accelerate desertified land increase. The “Three Norths ForestShelterbelt Program” and the “Grain-for-Green Project” carried out bythe local and national governments were somewhat effective in control-ling land degradation. Given the region's harsh climate and intensivehuman activities, though some political measures have carried out, de-sertification will continue to threaten the region's eco-environment.Therefore, desertification control is a difficult and gradual process, andthe government should enhance the leading role in this process.

Acknowledgement

The authors are grateful for the funding from the National Foundationof China (No. 30872073).

References

Ayoub, A.T., 1998. Extent, severity and causative factors of land degradation in theSudan. Journal of Arid Environments 38, 397–409.

Bond, W.J., Stock, W.D., Hoffman, M.T., 1994. Has the Karoo spread a test for desertifi-cation using carbon isotopes from soils. South Africa Journal of Science 90,391–397.

Brown, D.G., Duh, J.-D., Drzyzga, S.A., 2000. Estimating error in an analysis of forestfragmentation change using North American Landscape Characterization (NALC)data. Remote Sensing of Environment 71, 106–117.

Cao, S.X., Xu, C.G., Chen, L., Wang, X.Q., 2009. Attitudes of farmers in China's northernShaanxi Province towards the land-use changes required under the Grain forGreen Project, and implications for the project's success. Land Use Policy 26,1182–1194.

Collado, A.D., Chuvieco, E., Camarasa, A., 2002. Satellite remote sensing analysis tomonitor desertification processes in the crop-rangeland boundary of Argentina.Journal of Arid Environments 52, 121–133.

Cooperative Research Groupon Chinese Soil Taxonomy (CRGCST), 2001. Chinese SoilTaxonomy. Science Press, Beijing.

Darkoh, M.B.K., 1998. The nature, cause and consequences of desertification in the drylandsof Africa. Land Degradation and Development 9, 11–20.

Department ofGeographyof PekingUniversity, Commission for Integrated SurveyofNaturalResources of Chinese Academy of Sciences, Lanzhou Institute of Desert Research of Chi-nese Academy of Sciences & Lanzhou Institute of Glacier and Frozen Soil of ChineseAcademy of Sciences, 1983. Natural Conditions and Its Improvement and Utilizationin the Mu Us Sandland. Science Press, Beijing, pp. 210–230.

Dong, Y.X., Liu, Y.Z., Liu, Y.H., 1995. On Some Problems of Desertification. Xi'an MapPublishing House, Xi'an, China, pp. 141–156.

Feng, Y.S., 1985. Compilation of the map of desertification and its indications. Journal ofDesert Research 5, 15–21.

Girma, T., 2001. Land degradation: a challenge to Ethiopia. Environmental Management27, 815–824.

Hanafi, A., Jauffret, S., 2008. Are long-term vegetation dynamics useful in monitoringand assessing desertification processes in the arid steppe, southern Tunisia. Journalof Arid Environments 72, 557–572.

Hostert, P., Roder, A., Jarmer, T., Udelhoven, T., Hill, J., 2001. The potential of remotesensing and GIS for desertification monitoring and assessment. Annals of AridZone 40, 103–140.

Huang, Y.Z., Wang, N.A., He, T.H., Chen, H.Y., Zhao, H.Q., 2009. Historical desertificationof the Mu Us Desert, Northern China: a multidisciplinary study. Geomorphology110, 108–117.

INCD, 1994. Elaboration of an international convention to combat desertification in coun-tries experiencing serious drought and/or desertification and particular in Africa. Finalnegotiations text of the convention. Unites Nations, Geneva.

Jia, K.L., Chang, Q.R., 2007. Dynamic changes of land desertification landscape patternin agriculture and pasturage interlaced zone of northern Shaanxi. Chinese Journalof Applied Ecology 18, 2045–2049.

Lancaster, N., Helm, P., 2000. A test of a climatic index of dunemobility usingmeasurementsfrom the south western United States. Earth Surface Processes and Landforms 25,197–208.

Liberti, M., Simoniello, T., Carone, M.T., Coppola, R., D'Emilio, M., Macchiato, M., 2009.Mapping badland areas using LANDSAT TM/ETM satellite imagery and morpholog-ical data. Geomorphology 106, 333–343.

Liu, Y., Ci, L.J., 2000. Driving action of population increase on desertification. Journal ofArid Land Resources and Environment 14, 28–33.

Liu, J., Diamond, J., 2005. China's environment in a globalizing world. Nature 435,1179–1186.

Liu, H.J., Zhou, C.H., Cheng, W.M., Long, E., Li, R., 2008. Monitoring sandy desertificationof Otindag Sandy Land based on multi-date remote sensing images. Acta EcologicaSinica 28, 627–635.

45Y. Qi et al. / Catena 88 (2012) 37–45

Luo, Z., 2003. Chinahastens efforts to combat desertification. Beijing Review46 (14), 29–31.Mainguet, M., 1991. Desertification: Natural Background and Human Mismanagement.

Springer Series in Physical Environments, vol. 9. Springer Verlag, Heidelberg.Mainguet, M., Da Silva, G.G., 1998. Desertification and drylands development: what can

be done? Land Degradation and Development 9, 375–382.Mcclure, B.C., 1998. Policies related to combating desertification in the United States of

America. Land Degradation and Development 9, 383–392.Oñate, J.J., Peco, B., 2005. Policy impact on desertification: stakeholders' perception in

southeast Spain. Land Use Policy 22, 103–114.Paolini, L., Grings, F., Sobrino, J.A., 2006. Radiometric correction effects in Landsat

multi- date/multi-sensor change detection studies. International Journal of RemoteSensing 27, 685–704.

Perkins, J.S., Thomas, D.S.G., 1993. Spreading deserts or spatially confined environmentalimpacts-land degradation and cattle ranching in the Kalahari Desert of Botswana.Land Degradation and Rehabilitation 4, 179–194.

Pickup, G., 1998. Desertification and climate change-the Australian perspective. ClimateResearch 11, 51–63.

Rasmussen, K., Foga, B., Madsen, J.E., 2001. Desertification in reverse? Observationsfrom northern Burkina Faso. Global Environmental Change 11, 271–282.

Rodriguez, A.R., Soto, M.C.G., Hernandez, L.A.H., Mendoza, C.C.J., Gonzalez, M.J.O.,Padron, P.A.P., Cabrera, J.M.T., Chavez, G.E.V., 1993. Assessment of soil degradationin the Canary-islands (Spain). Land Degradation and Rehabilitation 4, 11–20.

Rubio, J.L., Bochet, E., 1998. Desertification indicators as diagnosis criteria for desertifi-cation risk assessment in Europe. Journal of Arid Environments 39, 113–120.

Runnstrom, M.C., 2003. Rangeland development of the Mu Us sandy land in semiaridChina: an analysis using Landsat and NOAA remote sensing data. Land Degradationand Development 14, 189–202.

Seto, K.C., Woodcock, C.E., Song, C., Huang, X., Lu, J., Kaufmann, R.K., 2002. Monitoringland-use change in the Pearl River Delta using Landsat TM. International Journal ofRemote Sensing 23, 1985–2004.

Southworth, J., Munroe, D., Nagendra, H., 2004. Land cover change and landscapefragmentation-comparing the utility of continuous and discrete analyses for a west-ern Honduras region. Agriculture, Ecosystems and Environment 101, 185–205.

Sun, D.F., Richard, D., Li, B.G., 2006. Agricultural causes of desertification risk in Minqin,China. Journal of Environmental Management 79, 348–356.

Sun, Y.J., Zhou, Q., Yang, R.H., 2008. A study of land desertification dynamic change inYellow River basin. Remote Sensing for Land and Resources 76, 74–78.

Tucker, C.J., Dregne, H.E., Newcomb, W.W., 1991. Expansion and contraction of theSahara desert from 1980 to 1990. Science 253, 299–301.

Wang, X.L., Bao, Y.H., 1999. Study on the methods of land use dynamic change research.Progress in Geography 18, 81–87.

Wang, H.J., Zhou, H., 2003. A simulation study on the eco-environmental effects of 3 NShelterbelt in North China. Global and Planetary Change 37, 231–246.

Wang, S.L., Zhao, L., Li, S.X., 2002. Interaction between permafrost and desertificationon the Qinghai–Tibet Plateau. Journal of Desert Research 22 (1), 33–39.

Wang, X., Dong, Z., Liu, L., Qu, J., 2004. Sand sea activity and interactions with climaticparameters in the Taklimakan Sand Sea, China. Journal of Arid Environments 57,225–238.

Wang, X., Dong, Z., Yan, P., Qian, G., 2005. Wind energy environments and dune fieldactivity in the Chinese deserts. Geomorphology 65, 33–48.

Wu, W., 2001. Using the TM image for monitoring land desertification. Remote SensingTechnology and Application 16, 86–90.

Wu, B., Ci, L.J., 2002. Landscape change and desertification development in the Mu UsSandland, Northern China. Journal of Arid Environments 50, 429–444.

Wu, W., Wang, X.Z., Yao, F.F., 1997. Applying remote sensing data for desertificationMonitoring in the Mu Us Sandland. Journal of Desert Research 17, 415–420.

Xue, X., Guo, J., Han, B.S., Sun, Q.W., Liu, L.C., 2009. The effect of climate warming andpermafrost thaw on desertification in the Qinghai–Tibetan Plateau. Geomorphology108, 182–190.

Yan, C.Z., Wang, T., Han, Z.W., Qie, Y.F., 2007. Surveying sandy deserts and desertifiedlands in north-western China by remote sensing. International Journal of RemoteSensing 28, 3603–3618.

Yan, C.Z., Song, X., Zhou, Y.M., Duan, H.C., Li, S., 2009. Assessment of aeolian desertificationtrends from 1975's to 2005's in the watershed of the Longyangxia Reservoir in theupper reaches of China's Yellow River. Geomorphology 112, 205–211.

Yang, X., Ding, Z., Fan, X., Ma, N., 2007. Processes and mechanisms of desertification innorthern China during the last 30 years, with a special reference to the HunshandakeSandy Land, eastern Inner Mongolia. Catena 71, 2–12.

Zhang, X.S., 1994. Ecological background, principles of management and optimizedmodels of the Mu Us sandy grasslands. Acta Phytoecologica Sinica 18, 1–16.

Zhao, H.L., Zhao, X.Y., Zhou, R.L., Zhang, T.H., Drake, S., 2005. Desertification processes dueto heavy grazing in sandy rangeland, Inner Mongolia. Journal of Arid Environments62, 309–319.

Zhu, Z.D., Chen, G.T., 1994. Sandy desertification in China. Science Press, Beijing, pp. 20–33.Zhu, Z.D., Feng, Y.S., 1984. The principles and methods for compiling the map of desert-

ification of China. Journal of Desert Research 4, 3–15.Zhu, Z.D., Liu, S., 1981. Desertification processes and their control in northern China.

China Forestry Press, Beijing.