Evaluation of Regionalization of Soil and Water Conservation in China

Abstract

This paper evaluates the regionalization of soil and water conservation in China, decades after its widespread implementation across the country. The authors mainly address two questions. First, to what extent could achievements in soil erosion and water management be attributed to regionalization? Statistics and cases show that enhancement of research on soil erosion and ecology, popularization of technologies and mathematical methods and more complete databases are the main improvements to theoretical research; the distribution of region-specific measures and monitoring sites and the development of a responsibility system are the main impacts on practice. Second, is there any weakness to regionalization? Econometric and management methods are currently not well integrated into the process of regionalization; indicators used for division lack standardization, thus weakening the accuracy of regionalization; also, there are limits to its implementation. Using scenario analysis, the authors discuss the possibility of involving economics and management science in the process of regionalization and the approach to combine qualitative analysis with quantitative analysis, while also arguing the importance of establishing an effective cooperation mechanism between different government departments and between government and actors. For government, the evaluation could be helpful to recognize the success, strengths and weaknesses of regionalization of soil and water in China and hence to take further steps to formulate region-specific policies dealing with complex environmental and economic problems in different regions.

1. Introduction

Worldwide efforts at geographical regionalization have had a long history. Several efforts of regionalization developed by government agencies and scholars for improved understanding of nature or social principles are widely known to the public. Examples are global climate zones and agricultural belts of the United States [1,2,3]; among which regionalization of soil and water conservation in China has existed for nearly 70 years and serves as a framework for soil erosion and water management over the whole country.

China spans nearly 50 degrees in latitude and has almost all existing landform patterns; types, rates and controls of soil erosion vary substantially from one region to another [4,5,6,7,8]. Chinese society is undergoing dramatic change. The eastern developed regions attract immigrants from the middle and the western underdeveloped regions, causing a higher urbanization level in the eastern regions (19.17% (calculated from the urbanization level of individual provinces recorded in the China Statistical Yearbook 2016) higher than the middle and western regions in 2015), which results in a change in the use of land from woodland, farmland and lakes to industrial or residential buildings [9]. Due to overexploitation of natural resources, severe environmental problems have been emerging including soil degradation and water quality deterioration [10]. Meanwhile, although the western underdeveloped regions have been less disturbed by human activities, the vulnerable environments are more likely to cause severe ecological problems, especially in the Loess Plateau and the Tibet Plateau [11,12]. Under the background of escalating regional imbalance, it is urgent to formulate region-specific policies to deal with complex environmental and economic problems in different regions. Moreover, after the concept of ecological civilization was first raised at the 17th National People’s Congress, 2007, and the Ministry of Natural Resources and the Ministry of Ecology and Environment were established in 2018, emphasis on the balance between development and conservation also calls for region-specific plans and measures [13]. Considering these, regionalization of soil and water conservation in China was aimed at dividing a certain area into several functional regions, with the hope that region-specific management rather than site-specific management could be applied without too much adjustment in each region [14,15]. The first regionalization scheme of soil and water conservation in China, the Water and Soil Conservation Plan of the Middle Yellow River, was developed by the Yellow River Conservancy Commission in 1955 [16]. By this scheme, the Loess Plateau was divided into eight regions, and each region was assigned different jobs for soil erosion and water management in the next five years. Yet, this original effort was interrupted by the Great Leap Forward (1958–1961) and then by the Great Proletarian Cultural Revolution (1966–1976) until Chinese economic reform was implemented in 1978. At the Fourth National Conference on Soil and Water Conservation held in Beijing 1982, the State Council issued the Regulations of Soil and Water Conservation, defining that the departments of water resources at all administrative levels undertake the task of regionalization. After that, works have been revived. Especially after 2011, the first national regionalization of soil and water conservation (Appendix A), which divided the country into eight super-regions (first-level), 40 regions (second-level) and 115 sub-regions (third-level), was approved by the Ministry of Water Resources [14], and finer distinctions based on this scheme are being made by local governments and institutions within their administrative areas [17,18,19,20].

As efforts on regionalization of soil and water conservation are still burgeoning, questions arise such as: To what extent could progress in soil erosion and water management be attributed to regionalization? Is there any weakness as to the formulation and applying of regionalization? Addressing these questions, this article evaluates the regionalization of soil and water conservation in China from three aspects: (i) How is theoretical research on soil erosion enhanced by regionalization? (ii) How is practical management of soil erosion promoted by regionalization? (iii) Is there any drawback of regionalization? By evaluation, this article could be helpful for decision makers to improve regionalization schemes and formulate more region-specific policies.

2. Methods

Quantitatively evaluating regionalization is particularly difficult for two reasons. First, comparative trials are unavailable. It is not practicable to apply another scheme in an area and then compare it with the already existing one. Second, the contribution of regionalization to soil and water conservation is not possible to separate from other factors such as technological advance, financial support and labor input, because the effect of regionalization is always indirect. Considering these difficulties and the fact that so far, to our knowledge, there is no published information on this issue, this paper evaluates the achievements of regionalization in a manner combining statistical analysis of literature with qualitative analysis.

Specifically, to study the effect of regionalization on theoretical research, first we searched the keyword “regionalization of soil and water conservation” on the Chinese National Knowledge Infrastructure (CNKI, the most comprehensive knowledge database that stores over 90% of information resources including Chinese journals, newspapers, conference proceedings, yearbooks, patents, standards, laws and regulations) and gathered the top 10 most highly cited articles (Appendix B) about regionalization. Since almost all research advances about regionalization of soil and water conservation can be accessed on this database, citations of these top 10 would cover most of the theoretical influences of regionalization. Second, we classified these citations according to both the research field they belonged to and their content and then calculated the citation number of each category. Third, according to the citation results and compared to the research advances promoted by regionalization with already exiting research before the implementation of regionalization, we detailed the effect of regionalization on theoretical research.

To study the effect of regionalization on practice, we combined on-the-spot investigation with information retrieval from research articles and government documents. The authors travelled to seven provinces (Shannxi, Gansu, Hubei, Hunan, Jiangxi, Fujian, Guangdong) in 2016–2017 and collected information including region-specific measures and the distribution and operation of soil erosion monitoring sites. Moreover, by consulting local ministries of water resources and local residents, the author acquired their attitude towards and comments on the implementation of regionalization.

To identify the weaknesses of regionalization itself, we selected a total of 19 regionalization schemes at different scales (Appendix C) and focused on the process of regionalization, especially the indicator systems they used. We compared the indicators used for regionalization at the same and at different scales to identify the drawbacks of indicator systems. Additionally, on-the-spot investigation helped to identify deficiencies of regionalization in practice.

3. Results

3.1. Effect on Theoretical Research

By keyword searching, the top 10 most highly cited articles were selected. As Figure 1 illustrates, the top 10 articles were collectively cited 204 times, excluding self-citations and repeated citations. Classified by discipline, they were cited139 times by articles on soil science, among which 32 articles were on the regionalization of soil and water conservation at different scales (e.g., province, city or watershed), another 107 were research on soil conservation, including 83 theoretical research works mainly concerning the benefit of soil conservation, and 24 empirical articles mostly were on the experience of soil erosion management. The other 65 articles of the total of 204 were on regionalization in general use within different research fields, especially in ecology (mainly concerning ecological regionalization and ecological compensation). In addition, 24 articles of the total 204 concerned RS technology and mathematical methods, while the other 180 did not.

Hence, before detailed analysis, these numbers present general viewpoints: (i) regionalization of soil and water conservation influences not only soil science (both theoretically and practically), but also other related disciplines; and (ii) although technologies and mathematical methods such as RS/GIS have been employed, there is still a long way to go to systematically integrate quantitative approaches into the process of regionalization.

3.1.1. Driving the Research on Soil and Water Conservation

Regionalization stimulates the generation of new concepts related to soil erosion. New concepts and formulas have been proposed in the process of regionalization [21,22,23,24]. For example, the potential risk of soil loss (I_e) and the comprehensive index of soil erosion (E) were initially raised in 2007 by the Chinese Academy of Engineering to establish a technical framework for regionalization of soil and water conservation [24]. By calculating the I_e and E value of different regions in the targeted area, a general distribution of soil erosion could be established and further help the regionalization. These two concepts were not only used for regionalization, but also have been employed in other research concerning soil erosion [25,26,27], such as evaluating soil erosion susceptibility, capturing the spatial variation of soil erosion.

Comparative research has been promoted because of regionalization. As differences in both the natural environment and society between different areas have been emphasized by division, researchers are inspired to reveal more information with comparative studies. For example, Cheng et al. (2008) [28] compared the different characteristics of splash erosion on loess soil sampled in IV-2-2jt (the loess and hilly sub-region of Northwest Shanxi) and purple soil in VI-1-4tw (the Dabashan mountainous sub-region) and found that soil properties, especially particle size distribution, resulted in the different erosion rates in different regions. Ma et al. (2009) [29] calculated the economic losses caused by soil and water loss on slope cropland in four regions, and the result showed that the red soil region of Southern China suffers the most severe losses (84.63 billion yuan), followed by the Loess Plateau (64.96), the hilly and mountainous region of Northern China (26.78) and the black soil region of Northeast China (16.08); also, the different influences of soil erosion in different regions on sustainable development were analyzed.

Long-standing problems have been solved in the process of regionalization. The boundary between karst area of southwest China and red soil area of southern China had been an issue of controversy for a long time. However, this was settled in 2011 when the regionalization of soil and water conservation of the Pearl River drainage basin was carried out by the Pearl River Water Resources Commission. After detailed inspection and consultation with local departments and inhabitants, six counties (Hechi, Baise, Chongzuo, Longan, Mashan and Yicheng) in Guangxi province were determined as the northeast boundary between karst area and red soil area [30].

3.1.2. Impacting Other Research Fields

Regionalization restricts plans and designs in other fields [31]. In planning the development of the Xinhua-Ziquejie scenic area, the exploitation of resources was severely restricted in high I_e regions according to the regionalization of soil and water conservation of Xinhua county; different types of agriculture were applied in different regions (e.g., sightseeing agriculture and stereoscopic agriculture) [32]. Another case is the afforestation plan of Ningxia Hui Autonomous Region [33]. In generating the plan, regionalization of soil and water conservation of Ningxia was used to determine vegetation species and coverage, with the purpose to both restore native vegetation and conserve soil and water. Furthermore, ecology research has been enriched by regionalization. For example, indicators used for regionalization of soil and water conservation of Beijing were used in a P-S-R model for quantitatively evaluating the ecological security of Beijing [31].

Regionalization serves as a framework for interdisciplinary research [34,35]. Traditional research on soil science was mostly issue-oriented [36,37]. After the appearance of regionalization schemes, however, this manner has been replaced by the region-oriented one to some extent. One can easily find that more research conducted by workers from two or more disciplines, especially ecology, economics and management science, is centering on complex issues within a particular region rather than viewing questions as isolated. For example, in 2010, a cost-benefit analysis of soil and water conservation in different soil and water conservation regions was accomplished by the cooperation between the Ministry of Water Resources (MWR) and local departments of economy. The investment cost, operating cost, total/net benefit and the average time to become profitable of different soil conservation measures in different regions were calculated and then served as the guide for distributing region-specific measures with the hope to balance economy and ecology. For example, in the third-level region, V-4-6tr (the Center Hunan sub-region), the investment cost, operating cost, total benefit and net benefit of land terracing were 3510, 4510, 22,570 and 14,550 million yuan, respectively, and the average time to become profitable was one year [24].

3.1.3. Popularizing Advanced Methods and Completing Databases

Compared with experience, remote sensing (RS) and image interpretation are more accurate and convenient for acquiring the knowledge of how soil, vegetation, terrain and landform are integrated in a certain area, while this knowledge is the basis of regionalization [38]. The first use of RS in regionalization of soil and water conservation was by 56 technicians from the agriculture department of Shanxi province in October 1981, when they interpreted 200 images of the 23 quadrats picked out from six land types (i.e., farmland, woodland, grassland, water, eroded soil and check dams) in east Shanxi province and obtained 80 spectral curves [39].

Regionalization in turn spreads RS technology. Once the advantage of this new approach had been realized, it was widely applied in divisions. For example, in January 1982, just three months after the very first use, a training course in RS technology was organized by the Department of Soil and Water Conservation in Ganzhou, Jiangxi province [40]. Trainees learned how to interpret 1972 MSS satellite images and how to map according to the interpretation. Using this knowledge, they acquired the soil erosion distribution of Xingguo (a county in Ganzhou), which was then mapped [41].

Because quantitative methods could eliminate subjective judgment, which is considered as unreliable, as much as possible, mathematical methods were initially applied by Zhang (1990) [42], who introduced fuzzy clustering into the process of regionalization of soil and water conservation of Ansai county, located in the central Loess Plateau. After that, more and more mathematical methods, including the stepwise method, constellation graphical method, etc., have been generally applied for regionalization of soil and water conservation [43,44,45,46,47].

Because generating regionalization demands a large database of geographic and socioeconomic data, it is certain that these data, some of which were not collected or organized before, were completed in the course of regionalization. For example, the national soil erosion distribution map, drew in 2011 at the request of the workgroup for water census released in January 2010, with its detailed illustration, acted as an important basis for developing the national regionalization of soil and water conservation. Another case is that those 200 images interpreted by the agriculture department of Shanxi province, after being used for regionalization in 1981, were compiled and published in 1982 for further use [39]. Moreover, regionalization improved laws and regulations concerning soil erosion and water management. When the Law of the People’s Republic of China on Water and Soil Conservation was first adopted in 1991, no articles were involved with regionalization. When it was revised in 2010, regionalization of soil and water conservation was written into the law as an article in a separate chapter (Chapter II, planning of soil and water conservation).

3.2. Effect on Practice

3.2.1. Distributing Region-Specific Measures

Regionalization serves as the basis for the distribution of region-specific measures. The proportion of the seven measures (terrace, tree, shrubbery, grassland, energy forest, non-tillage and check dam) commonly used in China varies between different first-level regions (Figure 2). Region V has the largest proportion of terraced field (26.53%), almost eight-times higher than Region I (3.45%). This is because compared with Northeast China, which is mostly flatlands, hills and mountains dominate Southern China. Trees are more than shrubberies in Regions I/III/V/VI/VII, while Regions II/IV/VIII have the contrary cases because the lesser rainfall and hot environment are not suitable for trees. For the same reason, grassland has the largest proportion in Region VIII (23.5%) than other regions. Regions VII, III and V have more energy forests (20.73%, 13.7% and 12.93%, respectively) for two reasons: more appropriate climate to grow fast-growing trees and more integrated industry for efficient wood processing. Both Regions VI and VIII have of larger proportion of non-tillage land (30.27% and 28.26%, respectively) than other regions, but for different reasons: in the purple soil region, plants could grow due to the advantage of abundant rainfall and sunlight without tillage, while in the Tibet Plateau, human activities may disturb the fragile ecosystem. Check dams are unique to Regions II, III and IV; especially Region IV, because of the severe soil erosion over, 95% of check dams in China are constructed in the Loess Plateau.

3.2.2. Distributing Monitoring Sites

Soil erosion monitoring sites are distributed according to regionalization [49]. In China, the existing monitoring system of soil erosion consists of one national monitoring center run by MWR, seven regional monitoring centers run by the Conservancy Commissions of the seven main rivers and 30 provincial monitoring centers run by provincial water resources departments. The 483 monitoring sites under the supervision of these 38 centers are generally evenly distributed in the 41 sub-regions divided by the national regionalization of soil and water conservation. Another case is as illustrated in Figure 3: in order for soil erosion to be systematically monitored, the Huaihe River drainage basin was divided into seven monitoring regions, the boundaries of which are exactly similar to the boundaries of the soil and water conservation third-level regions [50].

3.2.3. Developing the Responsibility System

Regionalization enables the household responsibility system to operate. Before the 1980s, the period when the centrally-planned economy and equalitarianism prevailed, confusion over roles and responsibilities were one of the top reasons for less-than-ideal performance of soil erosion and water management. However, this has been changed after China’s economic reform and decollectivization of agriculture. In 1984, the household responsibility system was applied to soil management in Sichuan province. According to the regionalization scheme, 19.6 million yuan was allocated to each region for practices [24]. The principle that “whoever has the land management right shall be responsible for its erosion management; whoever is responsible for erosion management shall be entitled to its benefit” encouraged farmers to build water conservation projects and terraced fields with labor and capital inputs. Another case is that from 1986–2005, the crop yield and per capita annual net income of rural households had increased from 4305 kg/hm² to 5600 km/hm² and from 504 yuan to 3630 yuan in eight southern China provinces where the responsibility system of soil erosion and water management was also implemented [24]. Furthermore, this system has been evolved into several new forms in the 21th Century. In Jiangxi province, the pattern “government-company-household”, in which local governments offer preferential policies to companies who offer job opportunities to farmers, has achieved remarkable success. For example, a series of policies such as including Gardenia jasminoides planting into the “Returning Farmland to Forest” (RFFP) program [51,52], providing free nurseries and five years’ tax exemption on agricultural specialty products attracted 264-million-yuan in capital investment in fiscal year 2004 alone, and over 4670 hm² of Gardenia jasminoides were planted. In addition, because villages within one region were more likely to take joint actions, regionalization made massive mechanization possible. In 2000–2004, under the supervision of a government workgroup, over 16,700 hm² of terraced fields were constructed by bulldozers in Tongwei county, Gansu province [53].

3.3. Weaknesses of Regionalization

3.3.1. Weaknesses of Regionalization Itself

First, because socio-economic factors are of great importance for regionalization, it is necessary to have them integrated into the process of regionalization. However, the current situation is opposite. Methods and models of economics and management science have always been absent from regionalization. The proof is that, as in Figure 1, only 3% of the total 204 articles concerned these two disciplines. However, the common practice of merely including some socio-economic factors into indicator system is not enough to reflect the effect of society and economy on regionalization. Second, economic and administrative departments fail to participate in the development of regionalization. Most of the regionalization schemes are generated merely by local and national departments of water resources without taking advice from others, e.g., the department of agriculture/transport/industry. The lack of cooperation between disciplines and between departments impairs the accuracy of division and could hinder its implementation once generated.

Second, building an indicator system by including core factors concerning geographic features, socio-economic condition and soil erosion (e.g., mean annual precipitation, population density and soil erosion rate) is the essential first step of regionalization [54,55]. However, the indicator system has drawbacks. For a detailed statistical analysis of the indicator system, as in

Table 1. Statistical analysis of the indicator system.

Scale	Area	The Number of Indicators Used	Total/Excluding Repetition	Most Repeated Indicators/Repeated Times	Total/Excluding Repetition	Most Repeated Indicators/Repeated Times
drainage basin	the Pearl River	10	25/20	population density/3 mean annual precipitation/2 vegetation coverage/2 proportion of eroded soil/2	31/13	population density/5 vegetation coverage/5 mean annual precipitation/4
	the Huaihe River	6
	the Taihu Lake	9
region	the mountainous region of Zhejiang and Fujian	12	36/23	population density/4 vegetation coverage/4 average elevation/3 soil erosion modulus/2 mean annual precipitation/2 accumulated temperature above 10 °C/2 GDP per capita/2 farming intensity/2
	the hilly-gully region of the Loess Plateau	6
	the plateau region of Yunnan, Guizhou and Sichuan	10
	the mountainous region of Yanshan	8
province	Liaoning	21	71/36	mean annual precipitation/4 vegetation coverage/4 average elevation/3 mean annual temperature/3 population density/3 GDP per capita/3 secondary industry production/3 proportion of eroded soil/3 proportion of forestland/3
	Jiangsu	22
	Shanxi	8
	Hubei	20
small watershed	Tuweihe	7	51/46	soil erosion modulus/3 gully density/2 vegetation coverage/2 population density/2
	Kuyehe	10
	Xiannangou	5
	Lishui	29
county	Fangshan	5	29/18	population density/4 vegetation coverage/4 soil erosion modulus/4 farming intensity/2 gully density/2 mean annual precipitation/2
	Ansai	8
	Shaowu	9
	Xigu	7

, in total, 19 regionalization schemes have been selected, including three sets at the drainage basin scale, four sets at each regional scale, provincial scale, small watershed scale and country scale.

Define the Indicator’s Degree of Aggregation (IDA) as the quotient of the net number of indicators (excluding repeated occurrences) divided by the total number of indicators. Larger IDA values indicate less aggregation and, hence, less uniformity. The IDA value of regionalization at the drainage basin scale, regional scale, provincial scale, small watershed scale and country scale were 0.80, 0.64, 0.51, 0.90 and 0.62, respectively. From these numbers, it is possible to conclude that regionalization at the same scale had no universally accepted standards for selecting indicators. Especially at the small watershed scale, even a total 51 indicators was used in the four schemes, and only four of them occurred more than once.

The lesser aggregation indicates that unreasonable indicators have been used for regionalization. For example, the proportion of fishery production in primary industry, reservoir volume and fertilizer used per year, which are barely related to soil erosion and water management, are considered as indicators for generating the regionalization of Lishui watershed [56]. Furthermore, involving too many less related factors into the indicator system resulted in a noticeably high correlation between indicators. In developing the regionalization of Yunnan province [20], the proportion of nonagricultural population and the proportion of agricultural population were both included in the indicator system. However, the Pearson’s correlation coefficient between the two variables was one, indicating that a liner function, y = 1 − x, could explain their relationship. Considering that all indicators could impact the calculation, there is no doubt that unreasonable indicators and collinearity between variables could damage the accuracy of fuzzy cluster analysis, which serves as a basis of regionalization.

In contract with regionalization at the same scale, works at different scales tended to use more similar indicator systems. The scale effect lost its significance. By adding up the number of the most repeated indicators at each scale and removing repeated occurrences, we obtained the total IDA value of 0.42, which is much lower than that at any scale, indicating that too many factors were repeatedly used at different scales. For example, population density and vegetation coverage were the two most repeated indicators at both the drainage basin scale and small watershed scale. This suggests that for both of the two scales, while a drainage basin could be 20–30-times larger than a small watershed, the population density and vegetation coverage were all key factors for regionalization. However, this was in contradiction to already existing knowledge, that heat and rainfall (controls of vegetation coverage) worked on larger scales, while human activities (partly expressed as population density) were only prominent at scatter ed sites [57,58].

3.3.2. Lack of Implementation

Because of the incomplete communication platforms and the conflict of interest between different regions and between different departments of government, there is a lack of a data sharing mechanism in China. This causes the data of some factors, which may have important influence on division, not to be included in the indicator system for calculation. Consequently, the accuracy of regionalization could be weakened.

Second, in many cases, regionalization schemes are merely formalities. Decision makers do not recognize regionalization schemes as the basis of distributing region-specific measures; they tend to put less emphasis on local-specific soil erosion management, but apply other region’s experience without adjustment. Hence, regionalization only is done to cope with supervisors’ requirements and has no practical effects on practice.

Third, the regionalization of soil and water conservation could be in conflict with other regionalization schemes or development plans generated by departments of transport, housing, land resources, water resources, environment, agriculture and industry. For example, in many cases, a certain area could belong to both a forbidden development region in the regionalization of soil and water conservation where human activities are prohibited to avoid soil disturbance and the development region in the plan of transport where an expressway is intended to be constructed. Under this circumstance, actors would be confused about balancing environment protection and infrastructure construction. This conflict is because the performance assessment of the Department of Transport does not include its responsibility for the environment; likewise, the performance assessment of the Department of Water Resources does not include its responsibility for construction. Consequently, each department acts to its own benefit, and plans and ideas of different departments are hardly exchanged clearly and efficiently. Furthermore, as the national policy of “remain committed to economic development as our central task” is currently valid, the priority of regionalization of soil and water conservation is in doubt.

4. Discussion

4.1. Modifying the Process of Regionalization

4.1.1. Integrating Economics and Management Science

As mentioned in Section 3.1.3, fuzzy clustering has been widely used in the process of regionalization. Cluster analysis assigns every single unit to clusters such that units in the same cluster are as similar as possible, while units belonging to different clusters are as dissimilar as possible [59]. Clusters are identified via the Euclidean distance; a certain region belongs to the category in the regionalization scheme that has the shortest Euclidean distance to it.

However, there are times that the Euclidean distance disables identifying clusters. Consider the scenario in Figure 4: Cities 1~6 located in Regions A, B and X and connected by Roads a~h; A, B are already existing first-level regions, and A-X and B-X have the exactly same Euclidean distance. The question is: should Region X be incorporated into A or B?

Apparently, fuzzy clustering is not applicable here because the equal Euclidean distance failed to identify clusters. However, the minimum cost flow model could be used. This model belongs to operational research and management science. As an approach to linear programming problems, it has been widely used in economics [54,60,61,62].

This model views cities as nodes and roads as arcs; goods (flow) transported from a supply city to a demand city (whether via a transshipment city or not) are limited by the cost per unit and the road capacity. The minimum transportation cost between two cities could be calculated by the following formula:

$$\mathrm{Z}={\displaystyle \sum }_{i=1}^n{\displaystyle \sum }_{j=1}^n{C}_{ij}{x}_{ij}$$

(1)

for each node i, subject to:

$${\displaystyle \sum }_{j=1}^n{x}_{ij}-{\displaystyle \sum }_{j=1}^n{x}_{ji}=b_i$$

(2)

and for each arc i → j:

$$0\le x_{ij}\le u_{ij}$$

(3)

where Z represents the minimum cost between node i and j; c_ij represents the cost per unit flow through arc i → j; x_ij represents the flow through arc i → j; u_ij represents the arc capacity for arc i → j; b_i represents the net flow generated at node i.

When applying the minimum cost flow model in this scenario, we first make hypotheses:

(i)

Soil erosion management in the undeveloped Region X demands resources support from developed Regions A and B;

(ii)

In Figure 4, Cities 1 and 4 are supply cities; City 3 is a demand city; Cities 2, 5 and 6 are transshipment cities;

(iii)

Road condition, truck availability and resource abundance limit the road capacity (unit: t) and the cost per unit (unit: dollars per t);

(iv)

In Figure 4, the road capacity and the cost per unit are listed in the brackets;

(v)

Managers want the maximum flow and the minimum cost at the same time.

Hence, for Regions A and X, we have:

• Minimum Z_AX = 2x₁₂ + 3x₁₃ + 4x₂₃
• subject to: x₁₂ + x₁₃ = 14, x₂₃ − x₁₂ = 0, −x₁₃ − x₂₃ = −16
• and 0 < x₁₂ ≤ 5, 0 < x₁₃ ≤ 9, 0 < x₂₃ ≤ 7.

For Regions B and X, we have:

• Minimum Z_BX = 3x₄₅ + 3x₄₆ + 6x₄₃ + 2x₅₃ + x₆₃
• subject to: x₄₅ + x₄₆ = 15, −x₅₃ − x₄₃ − x₆₃ = −14, x₄₅ − x₅₃ = 0, x₄₆ − x₆₃ =0
• and 0 < x₄₅ ≤ 6, 0 < x₄₃ ≤ 6, 0 < x₄₆ ≤ 9, 0 < x₅₃ ≤ 5, 0 < x₆₃ ≤ 6.

Calculating the two linear programming equation sets, we have the results: for Regions A and X, the maximum flow and the minimum cost are 14 t and 57 dollars per t; for Regions B and X, the numbers are 14 t and 67 dollars per t. Since to maintain the same maximum flow, Region A costs less than Region B, it is reasonable for Region X to be incorporated into Region A for efficient management.

There are other models that could be employed in the process of regionalization; examples are the shortest-path model, the maximum flow model and the network simplex method [63]. Applying these models/methods makes it possible to realize better interdisciplinary cooperation in the process of regionalization and, thus, better regionalization schemes.

4.1.2. Combining Quantitative Analysis with Qualitative Analysis

The “3S” technology and mathematical methods have limitations when employed in regionalization [64]. First, indicators selected have been limited to those that are easy to express as exact numbers rather than descriptions; mathematical models could hardly include several important socio-economic factors quantitatively, such as farmers’ enthusiasm for soil conservation. Moreover, mathematical methods and computer calculation occasionally make it possible to use more unreasonable factors. Because computers can process a large volume of data at extremely high speed, some research is encouraged to include as many indicators as possible. For example, in generating the regionalization of soil and water conservation of Yunnan province [20], a total of 86 indicators was included in the indicator system, ten of which were about slope cropland: the area of slope cropland, the area of slope cropland with slopes <5°, 5–8°, 8–15°, 5–15°, 15–25°, >5°, >15°, >25° and 25–35°. These ten indicators are apparently unnecessary to all be used because of their linear correlation.

Second, because cluster analysis and the Euclidean distance are commonly used for division, mathematical methods only calculate indicators as numbers and throw their actual meanings away [65,66,67]. This could be illustrated by a scenario analysis assuming that: (i) R2 and R3 are two existing first-level regions; (ii) the indicator system includes only two factors: X1, soil erosion rate, and X2, population density; (iii) μ(X1) and μ(X2) are membership functions determined by the random fuzzy statistics method, and the dependent variables’ values are listed as in

Table 2. Membership functions values of the three regions.

	μ (X1)	μ (X2)
R1	0.8	0.4
R2	0.4	0.8
R3	0.4	0.1

. The work is, by calculating the Euclidean distance with the formulas below, to determine whether the second-level region R1 should be incorporated into R2 or R3.

$$r_{ij}=\sqrt{\frac{1}{n}{\displaystyle \sum }_{k=1}^n{[\mu (X_{ik})-\mu (X_{jk})]}^2}$$

(4)

where r_ij represents the Euclidean distance between areas i and j; n represents the total number of indicators (here, two); X_ik and X_jk are the values of the k-th factor in areas i and j, respectively.

Calculations (r₁₂ = 0.4, r₁₃ = 0.5) demonstrate that, technically, R1 could be incorporated into R2 as they have a comparatively lower value, therefore shorter distance. However, if we thoroughly inspect Table 2 and consider the actual meanings behind those figures, a question would be posed. Both R1 and R3 have a comparatively high soil erosion rate and low population density; on the contrary, R2 has a comparatively low soil erosion rate and high population density. Thus, is it inappropriate to incorporate R1 into a region that has the quite opposite condition? As region-specific soil erosion and water management expect measures could be implemented in a more homogeneous region, perhaps it is more reasonable to incorporate one area into another that shares similar natural characteristics and social conditions. Holding to this thought, qualitative analyses are needed for further study.

4.2. Strengthening the Implementation of Regionalization

To let regionalization of soil and water conservation guide the practice of soil erosion and water management, works should be emphasized on: (i) constructing a data sharing platform between different regions and between different departments of government to guarantee researchers’ access to the data needed for regionalization; (ii) coordinating the work of different departments of government to avoid conflict between different regionalization schemes and plans; and (iii) raising decision makers’ awareness of the effect of regionalization on region-specific soil conservation practice to lead to them formulate policies balancing economy and environment.

Both theoretical research on soil erosion and management practices of soil and water conservation are the bases of regionalization because of the theories and experience they brought. Studies on controls of soil erosion could help to establish a more reliable indicator system; already existing conservation projects are also important for division. At present, there is a trend that both governments and researchers tend to introduce more mathematical methods into the process of regionalization without considering their real effect. In the authors’ view, it is of priority to pay more attention to practice, because it is just meaningless to do regionalization work if actors do not have profound understanding of the importance of regionalization and if government departments do not put regionalization into practice.

5. Conclusions

This article is the first time the achievements and drawbacks of regionalization of soil and water conservation in China after its nearly 70-year implementation have been generally evaluated. Combining statistical analysis and on-the-spot investigation, the effect of regionalization on theoretical research and on practice are evaluated. Results show that regionalization drives the research on soil and water conservation, and other disciplines, especially ecology, popularize the “3S” technology and mathematical methods and help database construction. Regionalization serves as a framework for distributing region-specific measures and soil erosion monitoring sites, while also complementing the responsibility system that could encourage works on soil erosion and water management. On the other side, there exists limits to regionalization itself and its implementation. Addressing these weaknesses, methods of economics and management science should be employed in the process of regionalization, and qualitative methods should be combined with quantitative methods. Above all, for better implementation of regionalization in practice, it is of great importance to establish an effective cooperation mechanism between different government departments and between the government and actors. For researchers in the field of regionalization, deeper interdisciplinary research could be inspired by the idea of integrating operations research into the process of regionalization. For governments, this evaluation could be helpful to recognize the success, strengths and weaknesses of regionalization of soil and water in China and hence to take further steps to formulate region-specific policies dealing with complex environmental and economic problems in different regions.