Ecological Land Suitability for Arid Region at River Basin Scale: Framework and Application Based on Minmum Cumulative Resistance (MCR) Model

Ecological land suitability evaluation (ELSE) refers to the assessment and classification of specific areas of land in terms of their features according to the specific types of land use (Naughton et al., 2015; Wei et al., 2020; Ma et al., 2021). Ecological land refers to the land use that can provide higher ecological services and keep balance between natural land and human activities. Ecological land usually includes low-intensity farmland, woodland, and water bodies, which are important for ecological management. Coordinating the allocation of ecological land and other land types to improve the use efficiency of the land resources without affecting ecosystem services is a challenge for land planners and scholars (Yeh and Huang, 2009; Guo et al., 2020; Zhou et al., 2020). At present, the various and multiple criteria make ecological land suitability evaluation (ELSE) complicated because of the diversity of influencing factors and regional difference (Elsheikh et al., 2013; Zhang et al., 2015; Ma et al., 2021).

At present, in the research at home and abroad, several methods were proposed and have been used to determine weights of ELSE, such as analytic hierarchy process (AHP), fuzzy analytical hierarchy method (FAHP), the artificial neural network method (ANN) and the entropy weight method (EWM). Besides, classical statistics combined GIS technology has been wide used to classify the spatial area of ecological land. For example, analytic network process (ANP) method can calculate the factor weights associated with criterion map layers under help of spatial overlay analysis and weights matrix (Bojórquez-Tapia et al., 2013; Ferretti and Pomarico, 2013; Azizi et al., 2014). Moreover, as an important data resource, remote sensing images are often combined with GIS technique to determine the ecological land and other land types (Li et al., 2015; Guo et al., 2020).

However, there are two aspects need improve for ELSE: firstly, the assessment systems of ELSE need demonstrate a logical framework for ELSE in a complete and clear way. Because of the complexity and multi scale in different region, most previous assessment systems have placed more emphasis on the number of factors, while failing to propose a uniform framework and application. Besides, most assessment systems simply improve the comprehensiveness of ELSE through adding ecological factors into the model, rather than investigating the limiting or inaccuracy of natural and social ecological factors in land use development and management. Secondly, a certain degree of subjectivity cannot be avoided in the process of weight determination used AHP, ANP or EWM, which need more experience and basic knowledge about study area. In addition, some assessment models are more stringent on input data set (Knaapen et al., 1992; Guo et al., 2020). Therefore, it is necessary to explore a simple and easy-understand ELSE method to improve the objectivity and portability of the ELSE model.

The minimum cumulative resistance (MCR) model is widely used to allocate and optimize the land use and land cover, which is originated by Knaapen et al. (1992). The MCR can reflect the degree of resistance in landscape development and can be used reflect the ecological changes. Because of its simple theory, a small amount of data support, fast algorithms, and visual results, the MCR is regarded as one of the best tools in assessing landscape connectivity at the landscape level, and it is also seen as one of the potential methods in ELSE study. The MCR model has been combined to GIS technology to determine the ecological land pattern and ecological planning (Liu et al., 2010; Guo et al., 2020; Wei et al., 2020). Its application is extended to specific ecological process such as ecological management and habitat planning (Li et al., 2015). In this paper, we developed an approach through MCR model to expand research and scope of the existing ELSE models at regional scales, and provide scientific support for ecological planning and management in arid region.

Shiyang River Basin has a large area, and the eco-environment is very fragile in Northwest China because of the severe natural environment and large population (Wei et al., 2013). This area plays a crucial role in maintaining local human survival, activities and social development. Due to the large population, limited water resources, and unreasonable human activities, the natural supply of groundwater has been reduced. The regional underground water level has dropped, resulting in the serious loss of natural vegetation, gradual soil salinization and desertification. This serious ecological crisis challenges the sustainable development of the economy and agriculture in this region. So Shiyang River Basin is a very typical area for ecological land suitability evaluation. This paper aims to assess the suitability for ecological land in arid region using GIS methods. It intends to explore the MCR model in the context of ecological processes and to determine the partitions of ecological land suitability based on the fundamental theory of structural and functional dynamics from landscape ecology.

The Shiyang River Basin (101°22′E–104°16′E and 36°29′N–39°27′N) is one of the three inland river basins in the Hexi Corridor, lying in the east of the corridor, Gansu Province in the northwestern China (Fig. 1). The basin occupies an area of 4.16 × 10⁴ km² and includes seven counties (Wei et al., 2013). The Shiyang River originates from the Qilian Mountain with eight tributaries, which are mainly fed by rainfall, snowmelt and glacier melt in the Qilian Mountain. The main land use types including farmland, woodland, grassland and distributes large area of desert, bare soil and bare rock. The terrain of basin is high in south and low in north, and tilt from the southwest to northeast. The basin has a continental drought climate, the sun radiation of Shiyang River Basin is strong and sunshine time is very long. It is short and very hot in summer, while it is long and very cold in winter. Influence by the climate, ecohydrology, landscape and other natural conditions, the soil and vegetation types of the basin form an obvious vertical spectrum of soil-vegetation (Wei et al, 2013).

Figure 1.  The location of Shiyang River Basin, Gansu Province, China

Landsat OLI images in August 24, 2019 were acquired to obtain the land use classes, the spatial resolution of the images is 30 m × 30 m. Google earth screenshots (September 2018) were obtained to identify land use class and the interpretation accuracy. Other data included Digital Elevation Model (DEM) with a spatial resolution of 30 m × 30 m, and the drainage maps at a scale of 1∶100 000 of 2019. DEM is provided by Geospatial Data Cloud, Computer Network Information Center, Chinese Academy of Sciences (http://www.gscloud.cn). Drainage maps come from Cold and Arid Regions Sciences Data Center at Lanzhou (http://westdc.westgis.ac.cn). Rivers and reservoirs were updated according to the drainage maps in 2019 and Google earth screenshots of 2018. The normalized difference vegetation index (NDVI) was calculated by near infrared wave band and red band through Landsat OLI images.

All the pre-processed (e.g., geo-referenced, rectified and cropped pertaining) images were manually classified using a geographical information system through grid-based visual interpretation (Ruelland et al., 2011). Land use was classified into six categories, including farmland, woodland, grassland, water body, urban and residential land and unused land according to the land use classification standard by Chinese Academy of Sciences (Wang et al., 2000). After that, accuracy assessment (Mitrakis et al., 2008; Ngigi et al., 2008) was done with the help of training samples and field data by testing the statistical significance of a difference, computation of kappa coefficients (Sha et al., 2008). The kappa coefficients were 83.84%, which indicated that the classified results were suitable for the research.

In the processes of ecological change, a heterogeneous landscape can be divided into a ‘source’ or ‘sink’ landscape (Adriaensen et al., 2003; Li et al., 2015). The source and sink landscapes have opposite qualities, but a source landscape in one process may become a sink landscape in another (Chen et al., 2006). The MCR can reflect the degree of resistance in landscape development and can be used reflect the ecological changes. It can reflect the minimum cumulative resistance in different direction and determine the suitable direction of ecological land. We used the following formula to determine the minimum cumulative resistance value, which was first stated by Knaapen (1992):

where MCR is the minimum cumulative resistance. m represents numbers of MCR value. R_k denotes the resistance coefficient of resistance surface k. D_k is the spatial distance of resistance surface k.

The resistance is derived from the cost distance associated with the cells at each end of the link (from the cost surface) and from the direction of movement (eight directions). The resistance value of each cell in the grid is based on the resistance value of the landscape attributed. The calculated equation of resistance value is:

where W_i is the power of resistance value of landscape unit i; Y_i is the comparative obstruction of landscape unit i; n is the number of landscape unit.

The cumulative cost distance is the sum of the spatial distance of landscape units on each path between the landscape unit and the ecological source. It can be described cost distance based on an eight-neighbor-cell algorithm, and the algorithm utilizes the node/link cell representation. In the node/link cell representation, each center of a cell is considered a node and each node is connected to its adjacent nodes by links. Every link has a resistance associated with it.

The calculated equation of D_k is as follows.

where D₁ and D₂ are the cumulative cost distance at the horizontal or vertical direction, and at the diagonal direction, respectively. C_i is the cost value of the landscape unit i, C_i+1 is the cost value of neighbor landscape unit, n is the number landscape unit.

When moving from a unit to one of its four directly (horizontal or vertical direction) connected neighbors, it adopts the Eq. (3), and if the movement is diagonal, it adopts the Eq. (4). From this model, it is conceivable that the cumulative spatial distance is the sum of the cost distances from one landscape unit to another that passes through the each unit location. There is only one least-cost path which is guaranteed to be the least resistance value from landscape unit to the source unit over the resistance surface (Li et al., 2015).

Ecological source refers to the landscape type that promotes the development of ecological processes and has very important roles in ecological landscpe protection (Li et al., 2015). Generally, the source can be composed of areas that need special protection or areas that have potential protected functions (e.g., water source, woodland, grassland and wetland). In the arid inland river basin in northwestern China, water body, woodland, grassland, wetland and the protected area are particularly scarce, and have very important contributions to the development of the ecological processes and have the key roles for local ecological security. In Shiyang River Basin, wood land mainly distributs in the Qilian Monutain, upstream of the basin,where is the national water source conservation zone of China and is also the core area of national natural reserve (Wei et al., 2012; Wei et al., 2015). So it is the focus area of ecological protection and the main ecological source landscape. In addition, water body plays an very important role in maintaining ecosystem service and social-economic development for dry inland region, and it is called the ‘lifeline’ of the Shiyang River Basin. So in this paper we chose water body, woodland, grassland, wetland and the protected area as the ecological source. During the input of the MCR model, the single ecological source was overlayed as one layer, which can be grid or raster file types. The living source mainly refers to all kinds of urban and rural construction land. In this paper, all kinds of construction land were extracted from the land use map of Shiyang River Basin in 2019 and were chosen as the living source. The ecological sources and living source were showed in Fig. 2.

Figure 2.  Single ecological source (a, b, c and d) , integrated ecological source (e) and living source (f) of Shiyang River Basin, Gansu Province, China

The interactions between adjacent landscape sources and landscape edge effect have important influence on landscape function and ecosystem structure (Costanza et al., 1997). The interactions can be viewed as competitive and control of the ecological energy and species among the components of ecosystem, which must be achieved by overcoming resistance. So resistance can be used to describe the interactions between adjacent landscape sources. The resistance surface contains location and direction factors of landscape sources and a resistance coefficient of landscape sources. The resistance coefficient represents the level of difficulty of some land patch becoming other land type through the effect of land structure (e.g., farmland, woodland, grassland etc.) (Li et al., 2015; Wei et al., 2015).

We used lowest, lower, medium, higher and highest to describe the resistance levels of the resistance factors, which is a subjective determination according to the local specific circumstance (Liu et al., 2001; Wei et al., 2013). Nine aspects for ecological source and seven aspects for living source were selected to constructed resistance surface factors, and the resistance assignments were listed in Table 1 and Table 2.

By calculating the minimum cumulative resistance value of ecological source and living source, ecological and living suitability zones can be established based on the difference value of MCR for the two sources, the modeling formula is as follow:

where MCR_eco and MCR_liv represent the MCR value of ecological land and living land. MCR_dif < 0 inflects that the resistance of ecological source is less than the resistance of living source, which indicating that the protection of ecological land is easier, so this land unit should be classified as suitable ecological land. Inverse if MCR_dif > 0, it should be classified as living land.

The threshold value was determined by MCR difference and the abrupt change point, which can be seen in Fig. 3. It was found that it had a turning point A (–207 731.62) in the negative territory, while it had a turning point B in the positive territory. There was also had a zero-value point C (0). Besides, it had a peak value point between point A and B (28 975.35), which was marked as F (11 942.11). According to the difference of MCR value and threshold value, the land suitability was further divided into five partitions: prohibited development region, restricted development region, key development region, optimized development region and ecological restoration region. The threshold value of suitability partitions and the partition names were showed in Table 3.

Figure 3.  Relationship between MCR value and grid numbers

Through the calculation of resistance surface of ecological source (Fig. 4a) and living source (Fig. 4b), it can be found that the resistance of ecological source changes higher from the southern Qilian Mountain to the northern Minqin County along the elevation, and the boundary between high and low resistance is obvious. The area of higher and highest levels in the northern part were more than that of lower and lowest levels in the southern part of study area. In the lowest resistance areas, the average elevation is higher than 4000 m, and distributes eight rivers, where densely covers with forest and grassland. In addition, there is no construction land, roads and other artificial facilities, and have a few human activities, so the ecological and energy flow are relatively vigorous. The areas of high resistance are mainly concentrated in the northeast of Minqin County, which is surrounded by Badain Jaran Desert and Tengger Desert, and the ecological environment is extremely fragile.

Figure 4.  Resistance surface of ecological source (a) and living source (b) in Shiyang River Basin, Gansu Province, China

As to the resistance surface of living source, the low-value is mainly distributed in the urban area of Liangzhou District, Minqin County and the artificial oasis area. Moreover, it is distributed along rivers and agricultural irrigation channels, where has large area of farmland and construction land.

The MCR value of ecological source and living source are showed in Fig. 5. It can be found that MCR value of ecological source is between 0 and 299 794.12 (Fig. 5a), most areas have low MCR values, accounting for more than 85% of the total watershed area. The high values are mainly distributed in the northern and eastern desert areas of Shiyang River Basin, where the vegetation cover is very low, the precipitation is little. It is not suitable for human survival and habitation. The areas with high MCR value have low ecological flow and energy, and they are sensitive to the ecological and vegetation changes.

Figure 5.  MCR value of ecological source (a), living source (b) and difference value of MCR between ecological source and living source (c) in Shiyang River Basin, Gansu Province, China

The MCR value of the living source is between 0 and 628 834.41, and the high value is dispersedly distributed in four different regions of Shiyang River Basin (Fig. 5b). It mainly distributed in the south of the Qilian Mountain foothills, where the average elevation is above 5200 m and mainly distributed with glaciers and cold desert. The low-value areas are mainly distributed in Liangzhou District, Jinchuan District and Minqin County, with an area of 28 496 km², which accounts for 68.5% of the total area of the basin. The MCR difference value and the spatial distribution are shown in Fig. 5c. From the distribution of MCR, we found that the desert region in north had high value of resistance, which indicated little area of ecological land. It is very difficult to transform into ecological land. The regions with less resistance, which indicated a high level of ecological protection and development, mainly spread around the river, large area of woodland and grassland, while the others were scattered in ecotone between desert and oasis.

Based on MCR difference values of ecological source and living source, we analyzed the spatial pattern of ecological land suitability, and the partition results were shown in Fig. 6 and Table 4. The results indicate that suitable ecological land includes prohibited development region and restricted development region, with an area of 6269.47 and 9473.63 km², respectively, which accounts for 15.45% and 23.35% of the total land area of the Shiyang River Basin, respectively. These two regions mainly distribute in high altitude region in the south Qilian Mountain and Tianzhu County, where has high density of rivers (eight rivers). The region is covered by a large quantity of glacier, permanent snow, rivers, forests and grassland, which have a positive effect on the local eco-environment, hence should be strictly protected.

Figure 6.  Reclass result of the MCR difference value (a) and ecological land suitability partition (b) in Shiyang River Basin, Gansu Province, China

The key development region is mainly located in urban areas and oasis agriculture areas, including Liangzhou District, Minqin County, Yongchang County and Jinchuan City, which accounts for 24.31% (9864.03 km²) of the total area of study area. It is the core region of Wuwei Oasis and Minqin Oasis, which have large areas of farmland, dense artificial agricultural irrigation system. This region is the important grain producing region in northwest China, with a rapid population growth (7.15‰) and obvious urban expansion in recent year.

The optimized development region mainly distributes in edge of farmland and the transition between farmland and deserts in the downstream and low altitude, with the area of 6606.41 km², where is very sensitive to changes in the oases and water body of the basin, may easily converted into farmland or desert under human activities influence. Therefore, this region should be protected and optimized in the process of ecological management.

The ecological restoration region locates in the north of basin and the edge of boundary, where distributes large area of desert, unused land and bare rock. This region is the convergence of Badain Jaran and Tengger deserts, which is not suitable for human habitation and survival. This region should be paid more attention and take some measures to carry out ecological restoration projection.

In this paper, the water body, woodland, grassland, wetland and natural protection area were determined as the ecological sources which are critical to ecological protection for arid region in northwest China. However, different area may have different factors and ecological sources on the process of selection. To obtain more accurate sources, we compared and analyzed different ecological elements and related study results (Guo et al., 2020; Wei et al., 2020; Ma et al., 2021). The resistance surface was also established based on the ecological sources, which can help with understanding and confirming the spatial locations and boundaries of ecological land, living land, and ecological management land. In arid inland regions, water is the most important factor for the basin to maintain the economic development, population increase, and urban expansion, and also one of the critical factors for land use optimization and ecological security (Ma et al., 2021; Xue et al., 2021). With the gradual increase of population and expansion of urban, the water consumption is increasing year by year, which lead to the serious shortage of water resources (Wei et al., 2015; Xue et al., 2021). Besides, the water conservancy project such as reservoir, pond and irrigation channel changed the natural hydro-geologic cycle of the basin, which leads to the decrease of surface and underground water supply gradually. It intensifies the contradiction between water supply and consumption (Li et al., 2015).

In Shiyang River Basin, woodland and grassland widely distributed in high altitude region, it is the important ecological water conservation and the natural ecological environment area. Moreover, human activity is very frequent in the downstream of the basin (Wei et al., 2015; Wei et al., 2020; Ma et al., 2021). The dry biological protection area and wetland protection area nearing the Shiyang River are created by local government in order to establish isolation between human activity and original environment. The protection areas play an important role in ecological restoration and management. They were all included in the ecological sources in this study. However, some land use categories such as farmland and urban green space were not considered as ecological sources because of the artificial landscape element, which are easily changed with the development of the economy.

Shiyang River Basin is located in the interaction region between desert (Tengger Desert, Badain Jaran Desert) and oasis (Wuwei Oasis and Minqin Oasis), where is very vulnerable and sensitive. With the change of land use types and the degradation of ecological system function, the coordination of ecological land protection and economic development has been broken in the past 30 yr (Wei et al., 2013). It reflects the fundamental conflicts between natural land protection and urban expansion. Therefore, suitable ecological land is very important and necessary for local government to protect and restore ecology. The protection of ecological land requires not only conserving existing ecological land but also focusing on ecological buffer belts around ecological sources and improving ecological land service ability.

We found that farmland change occurred along the ecotone between farmland and desert, the expansion of urban and residential land mainly distributes in the core oasis area, where closed to water resource. Some management measures such as efficient exploitation of land and water resources, improvement of fragile farmland, woodland and grassland are very important to ecological recovery and management in arid inland region in northwest China (Liu et al., 2001; Sha et al., 2008; Pourebrahim et al., 2011). Besides, setting up a protected green corridor along the Shiyang River would be beneficial for the protection of the ecological systems. Ecological networks should be constructed through artificial planting trees around the boundary of oasis and ecological restoration region. Buffer greenbelts should also be established between optimized development region and ecological restoration region. The proportion of surface water supply should be increased and the groundwater mining should be reduced on a massive scale. A large number of anti-wind and anti-sand plants should be plants at the periphery of desert to prevent the desert from expanding further.

Taking Shiyang River Basin of Gansu Province, China as a typical study area, using MCR model, selecting woodland, grassland, water body, the protected area and wetland as the ecological sources, urban and rural construction land as the living source, this paper analyzed and evaluated ecological land suitability and the different partitions.

The results showed that: 1) The high resistance areas of ecological source are mainly concentrated in the downstream Minqin County, while the low resistance areas are mainly concentrated in the southern Qilian Mountain. For the resistance surface of living source, the low value is mainly distributed in the urban area of Liangzhou district, Minqin County and the artificial oasis area. 2) For the MCR, the high values of ecological source are mainly distributed in the northern and eastern desert areas of Shiyang River Basin. The high value of the living source is dispersedly distributed in four different regions of Shiyang River Basin, while the low-value areas are mainly distributed in Liangzhou District, Jinchuan District and Minqin County. 3) The resistance surface for ELSE research were determined according to the actual situation of Shiyang River Basin. Whole study areas were divided into ecological land, living land and ecological management land according to the difference value of MCR between ecological sources and living source.

The MCR model provides an effective approach to suitable ecological land evaluation for worldwide arid area because of the simple parameters and a small amount of data requirement. It can comprehensively consider the horizontal relationship between landscape units in the region, which has good applicability and expansibility. The evaluation framework proposed and practiced in Shiyang River Basin can also be applied to other arid regions by adjusting the resistance surface factors relevant to the variables of concern in the region of interest.