Introduction

The term dust and sandstorms denotes a disastrous meteorological process that occurs in arid and semi-arid regions that have occurred throughout the recent history and ancient geological time (Awadh 2012). It is also an important indicator for evaluating desertification levels at a given site (Ye et al. 2000; Song et al. 2005). Dust and sandstorms are a meteorological process (Goudie 1983; Littmann 1991; Pye 1987) and are triggered by strong airflow movement. The storms enact a large-scale transport process of fine particulate matters produced by surface wind erosion. Generally, there are three necessary conditions for dust and sandstorm development, namely strong wind, sufficient sand–dust source and unstable air stratification. More specifically, the necessary meteorological and thermal conditions include loose and dry dust on the earth’s surface, the addition of strong and continual wind and unstable air stratification (Wang et al. 2001).

A large amount of research demonstrates that air current in the source area is responsible for the movement of dust and sand particles, but the turbulent fluctuation of air current with dust particles is mainly affected by the features of the underlying surface (Judd et al. 1996). McTainsh et al. argue that the meteorological conditions for dust and sandstorms are primarily related to soil moisture and vegetation conditions, and then wind (McTainsh 1998; Miller et al. 2004). Yang’s research (2010) shows the correlation between the occurrence frequencies of floating dust with the wind speed, temperature and transpiration rate in the Shapotou region of North China, and the reduction of the surface average relative humidity aggravates the frequency of floating dust weather. In the formation process of a dust and sandstorm, dust-released flux is proportional to the surface soil saltation flux, while the surface soil saltation flux is closely linked to the underlying surface types (Alfaro and Gomes 2001; Liao and Seinfeld 1998). The study of Zhao et al. (2011) shows that the transportation of sand–dust at the desert and edge of oasis sites mainly occurs in the near-ground layer, while at the oasis site, the transportation is mainly concentrated in the upper layer. Their research also indicated that vegetation, especially the windbreak shelterbelt, plays an important role in reducing sand–dust flux; for example, compared with the desert site, the edge of the oasis site has 42.0 % less sand–dust and the oasis site has 74.4 % less. The different features of the underlying surface in each region impact the intensity of the storms. Iwasaka (1983) argues that at the regional scale, land use, land cover condition and productivity level within growing seasons and vegetation remnants coverage in the non-growing seasons also exert significant influence on the dust and sandstorm process.

In natural grasslands, the increasing vegetation coverage can increase the surface roughness, which lessens the occurrence of dust and sandstorms. Observation shows that vegetation coverage between 15 and 25 % on sandy soil reduces ground wind erosion by 21–31 % of bare, shifting dunes. In contrast, wind erosion accounts for only 0.95 % of the shifting dunes when vegetation coverage reaches 40–50 %, demonstrating why grassland with high vegetation coverage is not likely to produce dust and sandstorms. Additionally, the thick roots of vegetation can consolidate soil, so they are susceptible to wind erosion. Vegetation adapted to shifting sand could reverse the desertification process and reduce the occurrence of dust and sandstorms (Roels et al. 2001). In sandy regions, particularly areas with shifting sand, low vegetation coverage and low roughness decreases the start wind speed of soil erosion and increases soil saltation flux and dust transport quantity. In addition, the composition of eolian sandy soil also heavily impacts dust and sandstorm occurrence. When the wind speed is certain, the quantity of fugitive dust depends on the amount of dust, sand and clay particles (<0.063 mm) in the surface soil, while the eolian fine sand with a high content of soil-erodible particles such as fine silt and clay can provide plenty of sand–dust sources for dust and sandstorm occurrence (Mei et al. 2004a, b).

In the arid and semi-arid regions of Northern China, there are three major dust and sandstorm invasion paths (Zhu and Zheng 2002). The first path is the north path, originating from the desert–steppe area of Southeast Mongolia (source region), and passing through Ulanchabu, Xinlin Gol, Chifeng and Tongliao in Inner Mongolia to the center of North China, Beijing, Tianjin, Tangshan and the southern part of Northeast China. The second path is the northwest path, originating from the desert and desert–steppe at the central southern part of Mongolia (source region), and passing through the Alxa League, Hexi Corridor, Maowusu Sandy Land, Ulan Buh Desert, Hohhot of Inner Mongolia, Zhangjiakou and then to Beijing, Tianjin and Tangshan. The third path is the west path, originating from the edge of Taklimakan Desert at Tarim Basin (source region) and passing through Dunhuang, Jiuquan, Zhangye, Minqin, Yanchi, Etuoke Banner and Datong, to Beijing, Tianjin and Tangshan. The Horqin region is located at the north path. This paper analyzed the changes in land use and coverage in recent decades within the typical regions of Horqin and evaluated the role of Horqin sandy land in dust and sandstorm development in this region, providing the foundation for dust and sandstorm pre-warning and control.

Study area

Naiman Banner is located in the east of Inner Mongolia Autonomous Region, and the south edge of the Horqin sandy region in the east farming–pastoral ecotone of the arid and semi-arid areas in Northern China with a total area of 8,120 km2 (E120°19′40″–121°35′40″, N42°14′10″–43°32′20″) (Fig. 1). The geomorphology changes from tectonic mountains to alluvial wavy plain and alluvial valley plain. The topography is high in the south and low in the north. The south comprises low hills and a cinnamon soil platform with an average altitude of 400–500 m and is situated at the longitudinal tectonic belt of the Yinshan Mountains and the uplift belt of New Cathaysian System’s third system. The central area is alternated with sand bog and dunes with an altitude of 300–400 m. The main landscape comprises fixed dune, semi-fixed dune, shifting dune and marshy grassland, which are distributed alternatively. In the north is the Kailu Basin in the twisted and Songliao compressed tectonic belt. At this location, the land is flat with fertile soil with an average altitude below 300 m.

Fig. 1
figure 1

Location of Naiman Banner

Full size image

Naiman Banner falls within temperate continental and semi-arid climate zones with average temperatures of 6.8 °C and annual precipitation of 366.3 mm. Rainfall is mainly concentrated in June, July and August, accounting for 67.4 % of the annual precipitation. The annual evaporation is 1,935 mm, which is 5.3 times the average precipitation. The days over 8 grade wind are 33.5 days per annum, occurring mainly in spring (from March to May) and accounting for 55–61 % of the annual total days. The annual average days with dust and sandstorms are 7.6 days, while the maximum reaches 23 days.

The soil can be categorized into five types, namely cinnamon soil, chestnut soil, meadow soil, eolian sandy soil and bog soil. Cinnamon soil is distributed at the south, accounting for 16 % of the total area; chestnut soil is generally dispersed at the central south covering 65 % of the total area, most of which has already developed into eolian sandy soil with coarse texture and loose structure; while the meadow soil and bog soil are concentrated along the rivers, sand bogs, lakes and other lowlands. Due to intense human interference, zonal vegetation in this region becomes secondary vegetation to a large extent. In most parts, zonal vegetation has been replaced by semi-recessive vegetation such as sand, shrub and recessive meadow vegetations. The main plant species in this region are shrub, perennial herbaceous plants and annual herbaceous plants including caragana microphylla, artemisia halodendron, artemisia frigida, mellissitus ruthenicus, cleistogenes squarrosa and setaria viridis.

Method

Collection of land use/land cover information

Field investigation was combined with the visual interpretation of two MSS images (1975 and 1985) and three Landsat TM images (1995, 2005 and 2008). ERDA’s Erdas8.6 software was adopted to undertake geometric calibration and projection transformation to achieve accurate spatial registration of images. The research methodology used in this paper is shown in Fig. 2. Images interpretation and topology was completed by ArcGis9.1 with display accuracy of the interpretation screen between 1: 50,000 and 1:100,000. According to the features of the landscape reflected from the images and actual conditions of Naiman Banner, the northern region of Naiman Banner is selected as the study area and is categorized into four types of sandy land based on Zhu’s sandy land division principle (1981): shifting dunes, semi-shifting dunes, semi-fixed dunes and fixed dunes.

Fig. 2
figure 2

Research methodology

Full size image

Collection of land productivity information

Analysis on the regional productivity was based on the 16-day synthetic NDVI data of EOS-Modis/Terra provided by American USGS to global customers free of charge with a resolution of 250 m. A data time frame of 11 years was selected (2000–2010) and acquired on June 26th. Image calibrations were completed based on the TM images in the same region.

Determination of ground spectrum and annual aboveground net primary productivity (ANPP)

Spectrometric determination of meadow vegetation was undertaken from June 18 to 20, 2011 using American ASD Company’s Fieldspec 1 spectroradiometer. In the process, the sensor was set vertically and 0.5 m from the canopy, and a standard light calibration board was used every 10–15 min. This work was conducted under sunny day with clear, low wind, as well as intense and stable sunlight. The field spectrometric determination time ran between 10:00 am and 3:00 pm.

Five groups of spectral data were collected from each quadrant. Plants were neatly cut along the ground. After fresh weight was weighed, they were put in cloth bags and taken back to dry at 80 °C constant temperature for 10–12 h and then the dry weight was determined. Since NPP included the underground growth increment and the amount eaten by herbivores, it was impossible to obtain entirely accurate data. Therefore, NPP in this paper was replaced by ANPP (Tilman 1996, 2001).

NDVI calculation of ground spectrum

ASD’s ViewSpec pro software (version 5.6) was used to process the obtained ground spectral data to determine the average spectral reflectivity of the red spectral band (0.62–0.67 µm) and near-infrared waveband (0.841–0.876 µm) which were consistent with the Modis satellite waveband. The NDVI calculation formula was as follows:

$${\text{NDVI}} = \left( {\rho {\text{NIR}} - \rho {\text{Red}}} \right)/\left( {\rho {\text{NIR}} + \rho {\text{Red}}} \right)$$
(1)

where ρNIR and ρRed are the average spectral reflectivity of the near-infrared waveband and red spectral band, corresponding to Modis satellite data.

Construction of NPP estimation spectrum model

Estimation of the aboveground dry matter quantity from ground spectrum model was established based on the correlation analysis of NDVI and aboveground biomass of current year from the collected ground spectral data.

Existing research showed that there were internal links between the actual measured and monitored ground vegetation spectral features by remote sensing. The actual measured ground vegetation spectral features can be used to replace the monitored ground vegetation spectral features by remote sensing (Li et al. 2001; Xu 2008). 68 groups of data were used to develop a high spectrum ANPP Model in Naiman Banner (Fig. 3):

Fig. 3
figure 3

Spectrum model of aboveground biomass in June in Naiman Banner

Full size image
$$y = 5.0829e^{5.3497x}$$
(2)

where y represented NPP of the current year with a unit of g/m2; x represented NDVI value, and the test of significance of correlation coefficient reached the most significant level (R = 0.8639>R 0.01,67 = 0.1480). According to formula (2), the NDVI data were calculated using grid calculation function in the spatial analyst module of Arcgis to produce 250 × 250 m productivity structure maps.

Collection of vegetation coverage in the non-growing seasons in sandy land

Landsat TM data from 1989, 1995, 2000 and 2007 were collected to get the information about vegetation coverage in non-growing seasons in sandy land. Given that waveband 3 (0.62–0.69) of Landsat TM data was the optimal waveband of visible light for extracting information of vegetation, lithology, strata, structure and geomorphology and sensitive to sand–dust and vegetation chlorophyll, the research applied TM data at the non-growing seasons as the data source to conduct bare surface extraction within the largest dune (including all kinds of dunes) region (1975) for interpreting land use cover.

The construction of bare surface extraction model was completed by the Model Maker module of Erdas9.1. This module had key algorithmic language and programming, which can establish the logical and conditional statement of model calculation through this language, and export image data translated by pixel grayscale value of bare land (Valuation 1) and with vegetation cover (Valuation 0). The circle demonstrated the model’s operational set, which was composed of logical language and computational language, while the arrows represented the calculation procedure.

Results

Land use and land cover changes in sandy regions

In the past 30 years, strong increase in greenness was observed over large areas of the Northern China from the NDVI data (An et al. 2013; Duan et al. 2011), demonstrating that the reduction was mainly due to the conversion from sandy land to farm land and forest land. Our result showed that the acreage of sandy land had decreased to 1086.50 km2 from 1975 to 2008 with an average annual decrease of 32.92 km2 (Fig. 4; Table 1). In terms of the general features of the 5 years, the distribution acreage of fixed dunes and shifting dunes were comparatively large (2,158.25 and 643.63 km2, respectively), while semi-fixed dunes and semi-shifting dunes were comparatively small (388.20 and 434.15 km2, respectively).

Fig. 4
figure 4figure 4

The distribution map of different sand dunes in Naiman Banner

Full size image
Table 1 Change in the features of different types of dunes (hectare)
Full size table

Second, we examine the changing features of different types of dunes. The acreage of fixed dunes had grown smaller from 1975 to 1995 and then began to increase following that year. The trend of shifting dunes had increased from 1975 to 1985, and started to decrease since then. The changes in semi-fixed and semi-shifting dunes were comparatively complicated. The acreage of semi-fixed dunes became the biggest in 1995 and the smallest in 2005, while that of semi-shifting dunes was the biggest in 2005 and the smallest in 1975.

Third, we observed the changing features of the maximum patch index. Specifically, we examined the single maximum patch as it accounted for the total acreage of all types of patches (Fig. 5). The maximum index of fixed dunes was the largest of all types with the highest of 48.7 % in 2008 and the lowest 14.0 % in 1985, followed by shifting dunes that reached the highest of 9.5 % in 1995 and the lowest of 3.9 % in 1975. The maximum patch index of semi-fixed and semi-shifting dunes was comparatively small and never surpassed 5 % at their highest levels.

Fig. 5
figure 5

Changes in maximum patch index of different types of dunes

Full size image

Changes in productivity pattern

Based on the spectrum-biomass model and ArcGis spatial analysis and mapping calculation, the productivity pattern map, the corresponding spatial distribution data from 2000 to 2010 and regional productivity capacity calculated with mean values were obtained (Fig. 6; Table 2).

Fig. 6
figure 6figure 6figure 6

Distribution map of productivity patterns at different times in Naiman Banner

Full size image
Table 2 Acreage distribution features of different productivity capacities (hectare)
Full size table

From the analysis in Table 2, the acreage of productivity capacity per unit less than 50 g/m2 decreased with fluctuation each year. The actual acreage gradually decreased from 5,468.44 km2 in 2000 to 2,328.91 km2 in 2010 with an average annual decrease of 285.41 km2.

On the contrary, acreage had increased gradually from 584.38, 198.55, 102.50 and 47.83 km2 in 2000 to 1,893.10, 1,465.88, 611.93 and 255.69 km2 in 2010, respectively, in areas with the productivity capacity per unit of 100–200, 200–300, 300–400 and 400–500 g/m2.

The change of total productivity capacity had increased gradually from 469,958 t in 2000 to 1,186,783 t in 2010.

Changes in vegetation remnants coverage in the non-growing seasons in the sandy region

Naiman Banner is located in the farming–pastoral ecotone of the arid and semi-arid areas in North China. This area suffers severe soil erosion and frequent dust and sandstorms due to climate change and human interference. Storms most frequently occur in the dry winter and spring. The most likely factors that influence the occurrence of dust and sandstorms are the features of underlying surface and vegetation remnants coverage in the non-growing seasons (Sun et al. 2001; Zhou and Wang 2002). The analysis of remote sensing showed that vegetation remnants coverage in the non-growing seasons had increased from 1989 to 2007 (Fig. 7). In terms of the changes in vegetation remnants coverage in sandy land, the acreage had grown from 2,686.76 km2 in 1989 to 3,250.39 km2 in 2007 with an increase of 563.63 km2. The changes in the maximum patch index of vegetation remnants covered area were consistent with the acreage change trend, which was 56.5 % in 1989 and 75.8 % in 2007 with an augmentation of 19.3 %.

Fig. 7
figure 7

Coverage changes in vegetation remnants in the non-growing season

Full size image

Conclusion

At the desertification region in the north of the study area, there existed the largest acreage distribution of fixed dunes and shifting dunes, in which the acreage decrease of shifting dunes and the increase of maximum patch index of fixed dunes became the most remarkable characteristics of land use and land cover change. Shifting dunes had decreased 328.99 km2 from 1975 to 2008 with an average annual reduction of 9.97 km2, while the maximum patch index of fixed dunes had increased from the lowest 13.97 % in 1985 to 48.67 % in 2008. This demonstrated that the vegetation area in the north sandy region was characterized by reduced shifting dunes and increased connected fixed dunes. These changes might be directly linked to the large-scale eco-restoration programs, which had been focusing on desertification land control, conversion of farmland to forest and grassland and re-plantation in the Horqin region since 2002. From the perspective of the productivity pattern changes within the growing season (June), the acreage of low-productivity regions (<50 g/m2) had decreased from 5,468.44 km2 in 2000 to 2,328.91 km2 in 2010 with an average annual reduction of 285.41 km2. Meanwhile, the acreage of the productivity level of 200–300 g/m2 had increased from 198.55 km2 in 2000 to 1,465.88 km2 in 2010, with an average annual increase of 115.21 km2. The regional vegetation productivity had grown from 469,958 t in 2000 to 1,186,783 t in 2010 with an average annual increase of 13.9 %. This showed that the vegetation in the Horqin region had improved significantly due to decades of efforts in eco-construction and restoration. In terms of the vegetation remnants coverage changes in the non-growing seasons in Horqin sandy regions from 1989 to 2007, the accumulative total acreage had increased by 563.63 km2 with an average growth of 31.31 km2. This was linked directly to the increase of vegetation coverage in the growing seasons in the designated region. The increase of the vegetation remnants coverage in the non-growing seasons could effectively increase the topsoil’s roughness in sandy land and ease soil wind erosion and desertification, which in turn played an important role in lessening the occurrence and development of dust and sandstorms in the designated region (Iwasaka et al. 1983; Gu et al. 2002).

As the transit zone of the east pathway of dust and sandstorms in North China (Zhu and Zheng 2002), the regional vegetation environment in the Horqin sandy region maintained a benign cycle in the past decade in terms of the changes in land cover, vegetation features in the growing seasons and the remnants features of surface vegetation in the non-growing seasons. In particular, the effective vegetation restoration of the sandy region will play a significant role in lessening the frequency of dust and sandstorms in the Horqin region.