Evaluating the Dynamic Response of Cultivated Land Expansion and Fallow Urgency in Arid Regions Using Remote Sensing and Multi-Source Data Fusion Methods

Abstract

In order to cope with the ecological pressure caused by the uncontrolled expansion of cultivated land in arid areas and ensure regional food security, the implementation of a cultivated land fallowing system has become an effective way to restore the ecology, alleviate the pressure on cultivated land, and increase productivity. In view of this, this paper takes the Tarim River Basin, located in the arid zone of China’s agricultural continent, as the research object. Using a land use transfer matrix and a gravity center migration model, the paper analyzes the spatiotemporal characteristics of cultivated land expansion in the Tarim River Basin from 2000 to 2020. Through remote sensing and the integration of multi-source data, the paper constructs an arable land fallow urgency index (SILF) from multiple dimensions such as human activity intensity, ecological vulnerability, output value, water resources status, and terrain conditions. The research results show that (1) cultivated land in the Tarim River Basin expanded by 15,665.133 km² in general, which is manifested by spreading around based on existing cultivated land, mainly from the conversion of grassland and unused land; the center of gravity of cultivated land moved 37.833 km to the northeast and 7.257 km to the southwest first. (2) The area of not urgently fallow (NUF) in the watershed showed an overall downward trend, decreasing by 10%, while the area of very urgently fallow (VUF) increased by 16%. VUF is mainly distributed in the marginal areas of cultivated land close to the desert and is gradually expanding into the interior of cultivated land. (3) The overall ecological environment of cultivated land in the watershed is showing a deteriorating trend, and the deterioration is gradually spreading from the edge of the cultivated land to the interior. (4) There are significant differences in the SILF values of different land use types after conversion to cultivated land. The urgency of fallowing cultivated land converted from unused land is the highest, followed by grassland, forest land, water bodies, and construction land. The expanded cultivated land has a higher SILF value than the original cultivated land. The research results can provide insights into regional land resource management, the formulation of cultivated land protection policies, and the ecological restoration of cultivated land.

1. Introduction

The impacts of land use change on soil, climate, water resources, and biodiversity challenge agricultural sustainability and the stability of terrestrial ecosystems [1,2,3]. Cropland, as the most important type of utilization, has an irreplaceable food security carrying function and ecological regulatory value [4,5,6]. Driven by factors such as climate change, increasing human activity, and economic development, cultivated land is showing a sustained expansion trend worldwide [7,8,9]. For example, from 2003 to 2019, the global cultivated land area increased by 9% [8], and the rate of cultivated land expansion has approximately doubled, and such changes have strongly disturbed ecological environments and ecosystem services [10,11,12]. Several studies have shown that this change is particularly evident in arid zones, where the dramatic expansion of arable land, which generates economic gains, may have led to a series of problems such as declining productivity of arable land, intensified soil erosion, increased consumption of water resources, increased ecological degradation, and a tendency to abandonment of arable land [13,14,15,16,17]. These factors pose a huge potential threat to future food security.

Due to the continuous expansion of cultivated land, the global ecosystem is gradually deteriorating, leading to serious ecological problems, as well as forest loss [18], land desertification [19], biodiversity loss [20], increased greenhouse gas emissions [21], and further changes in land cover and carbon–water–energy balance [14]. This conflicts with important development goals for the 21st century, such as protecting biodiversity and reducing greenhouse gas emissions [7]. Under these circumstances, it has been argued that maximizing crop yields and improving the efficiency of cultivated land use is one of the effective ways to alleviate ecological degradation, rather than continuously expanding the area of cultivated land [22,23,24]. As an effective measure to protect cultivated land, fallowing cultivated land can not only promote the restoration and protection of the cultivated land ecological environment but also effectively alleviate the pressure on cultivated land use, thereby improving cultivated land productivity [25]. However, the term fallowing is used to cover a very wide range of practices, in different agroclimatic zones and land use types [26,27,28]. Both effects on yields and environmental impacts will, therefore, depend not only on local conditions, such as topography, soil type, or weather patterns, but also on the specific crops and cropping systems that are applied [29,30]. Assessments of impacts and suitability of areas for fallowing will, in turn, depend on which aspects are considered [31], as well as on the time frame and geographical scope of the assessment [25]. At a time of rapid climate change, the effects of fallowing for climate resilience and soil carbon storage need to be taken into account [32,33], but depending on local conditions, decision-makers may also, for instance, wish to prioritize soil health, biodiversity, or water availability [34,35], just as food security or farmer livelihoods [36,37] remain important considerations in the long and shorter term. Recommendations regarding cropland fallowing are, therefore, far from straightforward and are highly dependent on context.

Cropland fallowing, as an agricultural management and land use measure, is also one of the important means to solve the degradation of cultivated land ecosystems. For example, the CRP (Conservation Reserve Program) plan in the United States aims to reduce soil erosion, improve water quality, and protect the habitat of wildlife [38,39]; the EU’s set-aside program began in 1992 to cultivate soil fertility and reduce the pressure of agricultural production on the environment [40,41]; and the Japanese government also set the improvement of the ecological environment as the goal of cultivated land set-aside in 1993 [42]. China started its fallow work relatively late. The “Pilot Program for Exploring the Implementation of the Fallow System for Cultivated Land” issued by the Chinese government in June 2016 officially started the exploration and construction of the fallow system for cultivated land. It emphasizes the pilot implementation of fallow in groundwater funnel areas, heavy metal pollution areas, and areas with serious ecological degradation, with the aim of promoting ecological restoration and governance, improving the quality of cultivated land, and improving the agricultural ecological environment [43].

Gaiser et al. [44] have argued that cropland fallowing should not be undertaken randomly but rather involves the genuine problematic cropland, such as low cropland productivity and a deteriorating ecological environment. Therefore, “where to fallow” and “in what order to fallow” are key points in the implementation of the cropland fallowing system. Decisions on the selection of suitable areas for fallowing, consequently, need to be supported by adequate mapping and assessment. However, Zhao et al. [45] point to the conceptual and methodological challenges in assessing the spatial details of fallow land from a multi-dimensional perspective based on statistical data alone. To overcome such limitations, Oliphant et al. [46] have argued that remote sensing technology could provide an effective means for the identification of cropland fallow. For example, Tong et al. [47] used high-resolution Sentinel-2 imagery to map fallow land in the Sahel. Nevertheless, this approach also entails methodological challenges, and due to the large processing workload of high-resolution image data, it generally cannot be used for the identification of arable fallow land in a large area. Song et al. [48], therefore, instead used low- and medium-resolution Landsat images to identify arable fallow land in Yuanyang County. Some scholars combine remote sensing technology with algorithms to improve the accuracy and timeliness of fallow land mapping. For example, Xie et al. [49] developed an automated method for fallow land mapping based on Sentinel-2 imagery using mathematical algorithms based on the turning point detection technique. Although these studies have made progress in presenting spatial details, they are still limited to single-dimensional remote sensing image analysis. In recent years, multi-source data fusion methods have provided a new research direction for the identification of fallow land. Shi et al. and Li et al. evaluated the suitability of fallow land in fragile areas of rocky desertification by integrating multi-source data from a multi-dimensional perspective [50,51].

Although previous studies on cropland fallowing in China have achieved remarkable results in identifying the spatial details of fallow land based on a multi-dimensional perspective, the following methodological and empirical deficiencies still exist. First, due to the heterogeneity of the ecological environment in different regions, a unified classification and evaluation system has not yet been established. Second, conditions that are relevant to the selection of areas for fallowing vary greatly across the country. For example, in Southwest China, the ecological degradation of cultivated land, mainly due to the increase in karst stone desertification and population pressure, has exacerbated the urgency of cultivated land fallow [50], while in the arid northwest, the Tarim River Basin is at increased risk of an ecological crisis in cultivated land due to intensive development and utilization of water and land resources [52], which may exacerbate the urgency of fallowing cultivated land. At the same time, most existing studies focus on the ecologically fragile stone desertification area in Southwest China [36,51,53,54,55], while relatively little attention has been paid to cropland fallowing in the arid ecological fragile area in Northwest China. Finally, most studies are limited to analysis at a single point in time, and there is a lack of in-depth discussion on the long-term spatiotemporal evolution of the fallow land pattern, in particular a systematic study of the dynamic response relationship between cultivated land expansion and fallow land distribution.

Considering these gaps in the research basis, as well as in view of the national significance of the Tarim River Basin for agricultural production and as one of the world’s largest oasis agricultural areas [56], we have selected the case of this significant river basin for the present study. The Tarim River Basin is currently facing severe ecological challenges, notably due to imbalances in the utilization of water resources [57,58,59,60], and the implementation of cropland fallow to protect arable land resources, restore the ecological environment, and improve the water balance has, therefore, become an urgent task. This study aims (1) to contribute to addressing the problems of fallow urgency of cropland in the Tarim River Basin; (2) solve the problems of “where to fallow” and “in what order to fallow” cultivated land in the Tarim River Basin through the fallow urgency index (SILF); and (3) elucidate the temporal and spatial dynamics of cropland expansion and fallow urgency in the basin over the same period of time to investigate the influence of cropland expansion on the fallow urgency of cropland.

To achieve these aims, this study first systematically analyzed the spatiotemporal patterns of cultivated land expansion in the Tarim River Basin from 2000 to 2020 using land use transition matrices and centroid migration models. Secondly, through remote sensing and multi-source data fusion methods, a fallow urgency index (SILF) was constructed from multiple dimensions, including human activity intensity, ecological vulnerability, water resource status, production value, and topographic conditions. The spatiotemporal evolution characteristics of the SILF in the basin from 2000 to 2020 were also revealed. Finally, the intrinsic relationship between cultivated land expansion and fallow urgency was thoroughly explored.

2. Study Area and Data Sources

2.1. Overview of the Research Area

The Tarim River Basin is located in the southern part of the Xinjiang Uygur Autonomous Region (75°06′~92°50′ E, 36°30′~42°10′ N) and is China’s largest endorheic river basin. The basin has a typical temperate continental climate. It is dry, with an annual average temperature of 10.7–11.5 °C and scarce rainfall of about 51.2 mm per year. Annual evaporation is as high as 2123.7 mm. This makes it an agricultural water-shortage area with a fragile ecosystem [59]. The basin is a typical oasis agricultural production area in China. According to the 2021 Xinjiang Statistical Yearbook, the cultivated land area is 3.09 × 10⁶ (ha), accounting for 44% of the cultivated land area in Xinjiang. The main crops are cotton, wheat, walnuts, dates, etc., among which cotton has the widest distribution [61]. Today, the watershed is mainly developed as an area oriented towards an agricultural economy [62]. In the context of global warming, previous studies on the Tarim River Basin suggest that extreme hydrological events have increased, exacerbating water resource uncertainty and highlighting water resource conflicts [63]. Agricultural production is based on the principle of “no irrigation, no planting” [64]. Farmland that is far from water sources mainly relies on limited surface water and groundwater recharge. Water resources are also lost during transportation through evaporation and leakage [65], reducing the sustainable use of water resources. In recent years, with the rapid expansion of arable land in the watershed and a significant increase in the intensity of human activities, the area of high-water-consuming crops has further increased, and the proportion of agricultural water consumption has increased to more than 90%, exacerbating the conflict between water scarcity and agricultural water use and putting a huge strain on the regional ecological environment [57,58,59]. Moreover, the basin is located in an arid region with scarce surface water resources. The over-reliance of irrigated agriculture on groundwater has exacerbated the overexploitation of groundwater. For example, the groundwater level in the Kuqa Oasis fell by 4.1 m from 2000 to 2018 [66]. In addition, the fragmentation of arable land, desertification, salinization, land degradation and soil infertility, and ecosystem vulnerability in the watershed have seriously constrained the regional arable land ecosystem construction and sustainable agricultural development [67,68,69,70]. In short, in view of these challenges, a transition to more sustainable agricultural practices is, therefore, the foundation for maintaining the stability and development of the watershed. An overview of the study area is shown in Figure 1.

2.2. Data Sources

In this paper, the remote sensing data of the MODIS series were obtained from the database of the Google Earth Engine (GEE) platform, and the remote sensing images from June to September of each target year were pre-processed with cloud removal, mosaicing, cropping, and median synthesis. The Land Use/Land Cover Remote Sensing Monitoring Dataset (CNLUCC) categorizes land into six types based on the Chinese Academy of Sciences’ primary land use classification standards: arable land, forest land, grassland, water bodies, construction land, and unused land [71]. The food production value dataset, raw material production value dataset, annual precipitation dataset (PRE) (2000, 2010, and 2020), water body data (2005), China’s Digital Elevation Model (DEM) spatial distribution data, and net primary productivity (NPP) dataset were sourced from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences (DCRES) (http://www.resdc.cn/) (accessed on 25 July 2024). The Groundwater Storage Anomaly (GWSA) dataset [58], annual precipitation datasets (2005 and 2015), and water body data (2000) were obtained from the National Tibetan Plateau/Third Pole Environment Data Center (TRDCAC) (https://data.tpdc.ac.cn/) (accessed on 27 July 2024). The Human Footprint (HF) dataset was derived from the figshare repository (https://doi.org/10.6084/m9.figshare.16571064) (accessed on 3 August 2024) [72]. The 2008 water body data were acquired from the National Ecosystem Science Data Center, National Science & Technology Infrastructure of China (NESDC) (http://www.nesdc.org.cn) (accessed on 6 August 2024) [73]. The 2015 and 2020 water body data were sourced from OpenStreetMap (OSM) (http://www.openstreetmap.org/) (accessed on 9 August 2024). Administrative boundary data were obtained from the National Platform for Common Geospatial Information Services (NPCGIS) (https://www.tianditu.gov.cn/) (accessed on 9 August 2024). The DEM data exhibit no significant temporal characteristics. Population data of counties (cities) in the study area were obtained from the China Statistical Yearbook of Population and Employment and were used to calculate the ecological carrying capacity (TEC) of cropland. To ensure the spatial consistency and comparability of the data, the above data were reprojected to the Mollweide equal-area projection and resampled to 250 m (CNLUCC dataset excluded) using bilinear interpolation. Detailed parameters are listed in

Table 1. Data used in this study.

Data Types	Time	Spatial Resolution	Purpose of the Data	Data Sources
MOD13A1	2000, 2005, 2010, 2015, 2020	500 m	Measuring the cropland ecological vulnerability	National Aeronautics and Space Administration (https://www.nasa.gov/) (accessed on 13 July 2024)
MOD11A2	2000, 2005, 2010, 2015, 2020	1000 m	Measuring the cropland ecological vulnerability	National Aeronautics and Space Administration (https://www.nasa.gov/) (accessed on 13 July 2024)
MOD09A1	2000, 2005, 2010, 2015, 2020	500 m	Measuring the cropland ecological vulnerability	National Aeronautics and Space Administration (https://www.nasa.gov/) (accessed on 13 July 2024)
FOV	2000, 2005, 2010, 2015, 2020	1000 m	Measuring the output value of arable land	DCRES (http://www.resdc.cn/) (accessed on 25 July 2024)
MOV	2000, 2005, 2010, 2015, 2020	1000 m	Measuring the output value of arable land	DCRES (http://www.resdc.cn/) (accessed on 25 July 2024)
CNLUCC	2000, 2005, 2010, 2015, 2020	30 m	Measured cropland area	DCRES (http://www.resdc.cn/) (accessed on 25 July 2024)
Water body	2005	/	Measuring the water resources status of arable land	DCRES (http://www.resdc.cn/) (accessed on 25 July 2024)
PRE	2000, 2010, 2020	1000 m	Measuring the water resources status of arable land	DCRES (http://www.resdc.cn/) (accessed on 25 July 2024)
DEM	/	250 m	Measuring the topographic relief of arable land	DCRES (http://www.resdc.cn/) (accessed on 25 July 2024)
NPP	2001, 2005, 2010, 2015, 2020	1000 m	Validation of the SILF	DCRES (http://www.resdc.cn/) (accessed on 25 July 2024)
HF	2000, 2005, 2010, 2015, 2020	1000 m	Characterizing the intensity of human activity	Figshare (https://figshare.com/) (accessed on 3 August 2024)
PRE	2005, 2015	1000 m	Measuring the water resources status of arable land	TRDCAC (https://data.tpdc.ac.cn/) (accessed on 27 July 2024)
GWSA	2002, 2005, 2010, 2015, 2020	0.05°	Measuring the water resources status of arable land	TRDCAC (https://data.tpdc.ac.cn/) (accessed on 27 July 2024)
Water body	2000	/	Measuring the water resources status of arable land	TRDCAC (https://data.tpdc.ac.cn/) (accessed on 27 July 2024)
Water body	2015, 2020		Measuring the water resources status of arable land	OSM (http://www.openstreetmap.org/) (accessed on 9 August 2024)
Water body	2008	/	Measuring the water resources status of arable land	NESDC (http://www.nesdc.org.cn) (accessed on 6 August 2024)
Population data	2020	/	Measuring the ecological carrying capacity of arable land	China Population and Employment Statistics Yearbook
Administrative division data	/	/	Determination of the extent of the study area	NPCGIS (https://www.tianditu.gov.cn/) (accessed on 9 August 2024)

.

3. Research Methodology

In the first step, our study constructed an arable land fallow urgency index (SILF) based on multiple dimensions, such as the intensity of human activities, ecological fragility, water resources, production value, and topographic conditions, through a thorough review of the literature related to arable land fallow combined with field research and the actual situation of arable land in the watershed. In the second step, the accuracy of the SILF was verified by randomly selecting areas and visually comparing them with high-resolution images, correlating them with net primary productivity (NPP), and verifying them with the cultivated land ecological carrying capacity model. Finally, based on land use data, the trend of cultivated land expansion was analyzed using a land use transfer matrix model and a cultivated land gravity center migration model, and the response of this disordered expansion trend of cropland to the SILF was explored. The flowchart is shown in Figure 2.

3.1. Land Use Transfer Matrix

Under the influence of social and natural factors, land use patterns within a region change over time, and the speed and direction of changes vary among different land use types [74]. The land use transfer matrix is a method used to represent the conversion areas between different land use types over a specific period [75,76]. The calculation formula is as follows:

$$X_{ij}=\left[\begin{array}{ccc}{X}_{11}& \cdots & X_{1n}\\ ⋮& \ddots & ⋮\\ X_{n1}& \cdots & X_{nn}\end{array}\right]$$

(1)

In the formula, $X_{ij}$ represents the area of the $i$-th land use type converted to the $j$-th land use type from the initial period to the terminal period and n denotes the number of land use types.

3.2. Centroid Migration Model

The centroid migration model has been widely applied in the field of land use [77,78]. In this study, it is used to identify the movement trajectory and distance of cultivated land. The calculation formula is as follows:

$$X_t=\sum _{i=1}^n{X}_i{S}_i/\sum _{i=1}^n{S}_i$$

(2)

$$Y_t=\sum _{i=1}^n{Y}_i{S}_i/\sum _{i=1}^n{S}_i$$

(3)

In the formula, $X_t$ and $Y_t$ are the longitude and latitude coordinates of the cultivated land centroid; $S_i$ denotes the area of the i-th patch; and $X_i$ and $Y_i$ represent the central coordinates of the i-th patch.

The migration distance of the centroid between adjacent years is calculated as follows [79]:

$$L=\sqrt{{\left(X_{t+1}-X_t\right)}^2+{\left(Y_{t+1}-Y_t\right)}^2}$$

(4)

3.3. Entropy Weight Method

The entropy weight method is an objective weighting approach based on information entropy theory, which determines the weights by calculating the information entropy values of each indicator [80]. Compared with subjective weighting methods, the entropy weight method has significant advantages in objectivity, effectively avoiding the influence of decision-makers’ subjective preferences on weight allocation, thereby ensuring the scientific and rational nature of evaluation results [81]. Moreover, compared with other objective weighting methods, the entropy weight method does not require the construction of complex linear relationship models, offering stronger applicability and a wider range of applications [82]. In the entropy weight method, the determination of indicator weights depends on their contribution to information entropy; the higher the contribution, the greater the weight [83]. Additionally, this method can effectively screen out key indicators that significantly impact evaluation results while ensuring evaluation accuracy [84]. In this study, the entropy weight method was implemented through programming on the Matlab platform to calculate the weight values of each evaluation indicator.

3.4. Indicator Construction of the Cultivated Land Fallow Urgency Index

3.4.1. Human Activity Intensity

In agricultural ecosystems, human interventions tend to prioritize short-term benefits for yields and economic profit. However, unsustainable practices undermine the capacity of these ecosystems to ensure food security in the longer term, as well as causing detrimental impacts on surrounding ecosystems [85]. Among these, arid land ecosystems are more vulnerable to human disturbances due to their inherent fragility [86]. Therefore, areas with higher human activity intensity should be prioritized when formulating cultivated land fallow strategies. To comprehensively characterize human activity intensity (HAI), this study employs the Human Footprint (HF) dataset. Compared to traditional single indicators, such as nighttime light data or population density, the HF dataset integrates eight pressure variables, including built environment, population density, nighttime lights, cropland, pastureland, roads, railways, and navigable waterways. This provides a more comprehensive and accurate representation of the cumulative impacts of human activities on the ecological environment of regional cultivated land [72].

3.4.2. Ecological Vulnerability

The Tarim River Basin has witnessed intensified ecological vulnerability under the influence of climate and human activities [86]. Cultivated lands with exacerbated ecological vulnerability in the basin should undergo fallow periods to recuperate, restore the ecological environment, and enhance production potential. Since Xu et al. proposed the Remote Sensing Ecological Index (RSEI), some scholars have used the RSEI to evaluate the ecological quality in arid regions, which can accurately reflect the ecological conditions of the study area [87,88]. Therefore, this paper uses the RSEI to characterize the ecological vulnerability (CVI) of the basin. Based on Google Earth Engine (GEE), this paper constructs the RSEI by integrating four indicators: greenness (NDVI), wetness (WET), dryness (NDBSI), and heat (LST). Before construction, cloud removal and water removal using the Modified Normalized Difference Water Index (MNDWI) [89] were performed on the images. NDVI and LST are derived from the terrestrial vegetation product (MOD13A1) and surface temperature product (MOD11A2), respectively [88]. Calculate WET and NDBSI using MOD09A1 with the following formulas:

$$\mathrm{W}\mathrm{e}\mathrm{t}=0.1147{\rho }_1+0.2489{\rho }_2+0.2408{\rho }_3+0.3132{\rho }_4-0.3122{\rho }_5-0.6416{\rho }_6-0.5087{\rho }_7$$

(5)

$$\mathrm{I}\mathrm{B}\mathrm{I}={\displaystyle \frac{2{\rho }_6/\left({\rho }_6+{\rho }_2\right)-\left[{\rho }_2/\left({\rho }_2+{\rho }_1\right)+{\rho }_4/\left({\rho }_4+{\rho }_6\right)\right]}{2{\rho }_6/\left({\rho }_6+{\rho }_2\right)+\left[{\rho }_2/\left({\rho }_2+{\rho }_1\right)+{\rho }_4/\left({\rho }_4+{\rho }_6\right)\right]}}$$

(6)

$$\mathrm{S}\mathrm{I}={\displaystyle \frac{\left({\rho }_2+{\rho }_1\right)-\left({\rho }_2+{\rho }_3\right)}{\left({\rho }_2+{\rho }_1\right)+\left({\rho }_2+{\rho }_3\right)}}$$

(7)

$$\mathrm{N}\mathrm{D}\mathrm{B}\mathrm{S}\mathrm{I}={\displaystyle \frac{IBI+SI}{2}}$$

(8)

In the formula, ${\rho }_1–{\rho }_7$ represent the red, near-infrared, blue, green, shortwave infrared, mid-infrared 1, and mid-infrared 2 bands in the MOD09A1 dataset.

Based on the improved model proposed by Li et al., the RSEI is constructed by determining the eigenvector sign of the first principal component (PC₁) [90]. The calculation formula is as follows:

$$RSE{I}_0=\left\{\begin{array}{l}P{C}_1\left[f\left(NDVI,WET,NDBSI,LST\right)\right],V_{NDVI},V_{WET}>0,\\ 1-P{C}_1\left[f\left(NDVI,WET,NDBSI,LST\right)\right],V_{NDVI},V_{WET}<0.\end{array}\right.$$

(9)

$$RSEI=\left(RSE{I}_{0max}-RSE{I}_0\right)/\left(RSE{I}_{0max}-RSE{I}_{0min}\right)$$

(10)

In the formula, $RSE{I}_0$ is the initial RSEI value; $P{C}_1$ is the first principal component; $f$ represents the standardization of the four indicators; $V_{NDVI}$ and $V_{WET}$ are the $P{C}_1$ eigenvectors of NDVI and WET, respectively; and $RSE{I}_{0min}$ and $RSE{I}_{0max}$ are the minimum and maximum values of $RSE{I}_0$, respectively. Note that the RSEI and SILF are negatively correlated and are calculated using Equation (10).

3.4.3. Water Resource Status

In view of the severe water scarcity in the Tarim River Basin, improved management of water resources is an urgent priority for the area. Fallow farming on cultivated land with serious groundwater overdraft, insufficient precipitation, and a long distance from the water source [69] has the potential to save water resources at the source, alleviate irrigation pressure, and achieve sustainable water resource utilization. In this study, based on the entropy weight method, the water resource status (SWR) was constructed using the groundwater storage anomaly (GWSA) dataset, annual precipitation (PRE), and distance to water sources (DWSs). The calculation formula is as follows:

$$\mathrm{S}\mathrm{W}\mathrm{R}=W_{GWSA}\times {\alpha }_1+W_{PRE}\times {\alpha }_2+W_{DWS}\times {\alpha }_3$$

(11)

In the formula, ${\alpha }_1$, ${\alpha }_2$, and ${\alpha }_3$ are the entropy weights of the GWSA, PRE, and DWS and $W_{GWSA}$, $W_{PRE}$, and $W_{DWS}$ are the standardized values of the GWSA, PRE, and DWS. Here, the DWS has a negative correlation with the SWR.

3.4.4. Value of Outputs

Cultivated land is the “lifeline” of Chinese farmers [91]. The willingness of farmers to fallow is primarily determined by their perception of value, particularly economic value, which directly influences their decision to fallow, i.e., the value of the output from cultivated land [92,93]. This paper constructs the Cultivated Land Output Value (COV) based on the Food Output Value (FOV) and the Material Output Value (MOV). The calculation formula is referenced from [94] as follows:

$$\mathrm{C}\mathrm{O}\mathrm{V}=FOV+MOV$$

(12)

3.4.5. Topographic Conditions

Topographic conditions are one of the factors affecting the quality of cultivated land and are closely related to the process of soil and water erosion. In areas with significant topographic relief, cultivated land is prone to dual effects of water erosion and wind erosion, leading to the loss of topsoil and nutrients, thereby resulting in a decline in land quality [95,96,97]. Additionally, they impact farmland irrigation, resulting in poor water retention capacity [98]. The Tarim River Basin is not only a major region for agricultural irrigation but also has over 40% of its area affected by soil wind erosion [99]. Therefore, cultivated land with significant topographic relief should be prioritized for fallow. This paper employs the moving window method to calculate the Relief Amplitude (RA), and the calculation formula is as follows:

$$RA=H_{\hspace{0.25em}max}-H_{\hspace{0.25em}min}$$

(13)

In the formula, $RA$ represents the topographic relief value of the central grid within the window, $H_{\hspace{0.25em}max}$ is the maximum elevation within the window, and $H_{\hspace{0.25em}min}$ is the minimum elevation within the window.

3.4.6. Construction of the Suitability Index for Land Fallow Urgency

The cultivated land system is a complex ecosystem integrating “nature-economy-society” [100], playing a crucial role in ensuring food security, ecological security, and sustainable development [101]. In this study, the Tarim River Basin in the arid zone was taken as the study area, and the urgency index of arable land fallow (SILF) was constructed by the entropy weighting method based on remote sensing and multi-source data fusion methods, with the comprehensive consideration of the regional ecological environment characteristics. The construction of SILF is based on a multi-dimensional indicator system, including human activity intensity (HAI), ecological vulnerability (CVI), water resource status (SWR), output value (COV), and topographic conditions (RA). Prior to constructing SILF, each indicator was standardized based on its positive or negative correlation with the urgency of fallow. The “Natural Breaks Classification Method” was used to divide SILF into four levels: Not Urgent Fallow (NUF), generally urgent fallow (GUF), Moderately Urgent Fallow (MUF), and very urgent fallow (VUF). Among these, the urgency of fallow increases from low to high, indicating that cultivated land at the VUF level should be prioritized for fallow. The indicator system of SILF is shown in Figure 3, and its calculation formula is as follows:

$$\mathrm{S}\mathrm{I}\mathrm{L}\mathrm{F}=G_{HAI}\times {\gamma }_1+G_{CVI}\times {\gamma }_2+G_{SWR}\times {\gamma }_3+G_{COV}\times {\gamma }_4+G_{RA}\times {\gamma }_5$$

(14)

In the formula, ${\gamma }_1$, ${\gamma }_2$, ${\gamma }_3$, ${\gamma }_4$, and ${\gamma }_5$ are the entropy weights of HAI, CVI, SWR, COV, and RA, respectively, and $G_{HAI}$, $G_{CVI}$, $G_{SWR}$, $G_{COV}$, and $G_{RA}$ are the standardized values of HAI, CVI, SWR, COV, and RA, respectively.

3.5. Ecological Carrying Capacity Model for Cultivated Land

The ecological carrying capacity of cultivated land is the area of biologically productive cultivated land owned within a region. It is a real cultivated land area that objectively reflects the level of coordination between the pressure of human activities on the ecosystem and the ecosystem’s service capacity [102].

$$TEC=N\times ec$$

(15)

$$ec=a\times r\times y$$

(16)

where $ec$ is the ecological carrying capacity of arable land per capita, $N$ is the total population, and $a$ is the per capita production area of arable land. $r$ is the equilibrium factor and $y$ is the yield factor, which were assigned the values of 3.83 and 0.96, respectively [102].

4. Results

4.1. Characteristics of Land Use Change

From 2000 to 2020, unused land and grassland were the dominant land use types in the basin, together accounting for more than 90% of the total basin area (Figure 4). Among these, unused land is widely distributed across the entire basin, with a relatively stable spatial distribution pattern, while grassland is mainly concentrated in the peripheral areas of the basin but shows a significant declining trend overall (Figure 4). Cultivated land and water bodies are secondary land use types in the basin, each accounting for more than 2% of the total area. Spatially, cultivated land is mainly concentrated in the northern and western regions of the basin, with its area showing a continuously increasing trend, while water bodies are primarily distributed in the peripheral areas of the basin, dominated by snow-capped mountains and glaciers, but their area exhibits a declining trend (Figure 4). Forest land and construction land are the least dominant land use types in the basin, both showing a scattered distribution pattern (Figure 4A). Among these, the area of forest land continued to decrease during the study period, while the area of construction land exhibited a phased change, showing an increasing trend from 2000 to 2015 and a decreasing trend from 2015 to 2020 (Figure 4B).

Land use transfer data for the four periods, 2000–2005, 2005–2010, 2010–2015, and 2015–2020, based on the land use transfer matrix, were plotted on a Sankey diagram (Figure 5). An analysis of land use transfer in each period reveals the dynamic characteristics of land use change in the watershed. During the period 2000–2005, the cultivated land area in the watershed increased significantly, with a net increase of 3917.625 km² (16.7%), mainly due to the transfer of grassland, and 2905.707 km² (49.2%) was transferred; forest land and unused land decreased by 378.455 km² (2.7%) and 1170.99 km² (0.2%), respectively, and was mainly converted to cultivated land. From 2005 to 2010, the cultivated land area continued to increase, with a net increase of 3917.625 km², and it was mainly transferred from grassland, unused land, and forest land. The forest area decreased by 468.387 km² (17.5%) and was mainly converted to grassland and cultivated land. From 2010 to 2015, the cultivated land area continued to increase, with a net increase of 4745.722 km² (13.2%), and it was mainly converted from grassland, with 3713.251 km² (72.7%) transferred; the area of unused land decreased by 1224.43 km² (0.2%) and was mainly transferred to water bodies and cultivated land. From 2015 to 2020, the area of cultivated land increased by 1647.683 km² (4.1%) and was mainly transferred from grassland, with 3713.251 km² (66.7%) transferred.

Through the analysis of land use change characteristics across the four periods from 2000 to 2020, it is evident that land use changes within the basin exhibit significant spatiotemporal differentiation. The continuous expansion of cultivated land exhibits a spatial pattern of outward diffusion from existing cultivated areas, showing a clear phased transition trend in space (Figure 4A). Specifically, during 2000–2015, the expansion of cultivated land was primarily concentrated in the northeastern part of the basin, with a rapid expansion rate; whereas, during 2015–2020, the focus of cultivated land expansion gradually shifted to the western part of the basin (Figure 4A). This spatial shift is also reflected in the movement of the center of gravity of cropland; during the period of 2000–2015, the center of gravity of cropland shifted 37.833 km to the northeast, while during the period of 2015–2020, it shifted 7.257 km to the southwest, with the fastest movement of the center of gravity of cropland occurring during the period of 2005–2010 (Figure 6).

4.2. Spatiotemporal Variation Characteristics of Cultivated Land Fallow Urgency

4.2.1. Construction of SILF

The weight order of SILF is CVI > COV > SWR > HAI > RA (Figure 7). Among these, CVI, COV, and SWR are the primary contributing indicators, while RA has the smallest contribution, likely because most of the cultivated land in the basin is located in flat areas with minimal relief.

4.2.2. Spatiotemporal Variation Trends of Cultivated Land Fallow Urgency

This study fitted the mean SILF values for the basin from 2000 to 2020, revealing a slope of 0.00429, indicating an upward trend, which suggests a gradual deterioration of the ecological environment of cultivated land in the basin (Figure 8). Specifically, the mean SILF values decreased during 2000–2005, indicating an improvement in the ecological environment of cultivated land in the basin (Figure 8).

In terms of temporal changes, the fallow urgency of cultivated land in the basin from 2000 to 2020 was predominantly at the NUF and GUF levels (Figure 9). Among these, the area of NUF showed an increasing trend from 2000 to 2005 and a decreasing trend from 2005 to 2020, with an overall decline of 10%. The areas of GUF and MUF exhibited fluctuating changes during the study period, with the area of GUF decreasing by 11% overall and the area of MUF increasing by 5% overall. The area of VUF showed a significant upward trend from 2000 to 2020, with an increase of 16% (Figure 9B). Spatially, from 2000 to 2020, NUF and GUF were mainly concentrated in the interior regions of cultivated land in the basin; NUF and VUF were primarily distributed in the marginal areas of cultivated land, near desert regions, and gradually expanded toward the interior of cultivated land (Figure 9A). This spatial distribution pattern indicates that the ecological degradation of cultivated land in the basin exhibits a trend of gradual deterioration from the edges toward the interior.

Figure 10 and

Table 2. Changes in SILF level grades in the Tarim River Basin from 2000 to 2020.

SILF Changes Category	2000–2005		2005–2010		2010–2015		2015–2020
	Area (km2)	Ratio (%)	Area (km2)	Ratio (%)	Area (km2)	Ratio (%)	Area (km2)	Ratio (%)
NUF to NUF	9130.59	35.21	9554.74	38.43	11,226.65	33.28	9565.87	25.83
NUF to GUF	628.54	2.42	1725.51	6.94	2594.89	7.69	3944.22	10.65
NUF to MUF	9.17	0.04	43.53	0.18	91.16	0.27	291.35	0.79
NUF to VUF	1.13	0.00	2.62	0.01	10.47	0.03	47.47	0.13
GUF to NUF	2963.17	11.43	2396.15	9.64	2350.23	6.97	1323.34	3.57
GUF to GUF	5577.56	21.51	4183.97	16.83	8256.92	24.48	6924.38	18.70
GUF to MUF	1464.81	5.65	249.99	1.01	898.39	2.66	5007.23	13.52
GUF to VUF	7.31	0.03	10.05	0.04	69.44	0.21	742.63	2.01
MUF to NUF	46.54	0.18	334.81	1.35	79.96	0.24	58.58	0.16
MUF to GUF	570.46	2.20	2660.48	10.70	1924.02	5.70	337.37	0.91
MUF to MUF	2988.39	11.52	1599.59	6.43	2979.93	8.83	2169.91	5.86
MUF to VUF	554.41	2.14	150.22	0.60	319.60	0.95	3493.89	9.44
VUF to NUF	1.86	0.01	16.32	0.07	7.45	0.02	5.92	0.02
VUF to GUF	19.54	0.08	201.06	0.81	210.58	0.62	23.22	0.06
VUF to MUF	244.52	0.94	919.78	3.70	1030.42	3.05	125.25	0.34
VUF to VUF	1723.13	6.65	815.80	3.28	1678.92	4.98	2967.27	8.01

reflect the changes in the levels of cultivated land fallow urgency in the basin from 2000 to 2020. The areas of the four SILF levels (NUF, GUF, MUF, and VUF) showed significant differences across the four periods (2000–2005, 2005–2010, 2010–2015, and 2015–2020). Among these, the unchanged areas accounted for the highest proportion during 2000–2005, 2005–2010, and 2010–2015, with an average proportion of about 70%. However, the changed area reached 41.6% during 2015–2020 (Figure 10 and Table 2). In terms of area changes due to level transitions, during 2000–2005, the largest transition was from GUF to NUF, accounting for 11.43% (2963.17 km²), while the smallest transition was from VUF to NUF, accounting for only 0.01% (1.86 km²). During 2005–2010, the largest transition was from MUF to GUF, accounting for 10.70% (2660.48 km²), followed by GUF to NUF, accounting for 9.64% (2396.15 km²), while the smallest transition was from NUF to VUF, accounting for only 0.01% (2.62 km²). During 2010–2015, the largest transition was from NUF to GUF, accounting for 7.69% (2594.89 km²), followed by GUF to NUF, accounting for 6.97% (2350.23 km²), while the smallest transition was from VUF to NUF, accounting for only 0.02% (7.45 km²). During 2015–2020, the largest transition was from GUF to MUF, accounting for 13.52% (5007.23 km²), while the smallest transition was from VUF to NUF, accounting for only 0.02% (5.92 km²) (Figure 9 and Table 2). Overall, the changes in the levels of cultivated land fallow urgency mainly occurred between adjacent levels (Figure 10).

5. Discussion

5.1. Evaluation of SILF

To validate the performance of SILF, eight regions were randomly selected (Figure 11) and a comparative analysis was conducted between the 2020 SILF values and high-resolution imagery. Among these, regions a, b, c, and d in Figure 11 were high SILF value areas, while regions e, f, g, and h were low SILF value areas. High-value areas were mostly distributed in regions far from water sources, far from residential areas, with significant topographic relief and severe desertification. In contrast, low SILF value areas were mainly distributed in regions with flat terrain, convenient transportation, proximity to water sources, and low levels of desertification. These regions also tended to be more suitable for crop growth and production.

A study by Li et al. on the effects of topography on agricultural productivity in China concludes that areas with large topographic fluctuations are vulnerable to erosion and leaching, which causes nutrient loss in the soil and exacerbates the deterioration of cultivated land quality. In addition, they have poor water retention capacity, high food transportation costs, and poor farming efficiency [103]. In such areas, priority fallowing can serve the purpose of “trading space for resources” by concentrating scarce water resources in high-producing areas of food to consolidate food production, allowing it to recuperate and restore the ecology. Additionally, previous studies from China, Slovakia, and Nepal have also suggested that fallow should be prioritized in sloping fields, arid lands, areas with poor accessibility, low-quality cultivated land, and regions far from residential areas [104,105,106]. These conclusions are largely consistent with the findings of the present study.

To verify the reliability of the above conclusions with respect to the Tarim River Basin, remote sensing images were resampled using a 3 km × 3 km grid and evaluated the spatial correlation between SILF and net primary productivity (NPP) based on this approach. The time-series analysis results indicate that SILF and NPP exhibited significant negative correlations in five representative years (2000, 2005, 2010, 2015, and 2020) during 2000–2020 (Figure 12). As a key indicator of cultivated land productivity, NPP values are positively correlated with land productivity [107,108], whereas SILF shows an inverse relationship, with its values negatively correlated with land productivity. This finding further validates the reliability and effectiveness of the index.

Shi et al. found that the urgency of farmland fallowing is significantly and positively correlated with the ecological carrying capacity of cultivated land (TEC) [50]. To assess this correlation in the case of the Tarim River Basin, we quantified the ecological carrying capacity (TEC) in 2020 of cultivated land in 46 counties and cities in the watershed, and a correlation analysis with the SILF was performed. It was found that TEC and SILF were significantly positively correlated (p < 0.0001) (Figure 13), which further verified the accuracy of SILF in identifying the urgency of cultivated land fallowing.

5.2. Response of Cropland Fallow Urgency to Cropland Expansion

Our findings showed that cultivated land in the Tarim River Basin continued to expand over the investigated period (Figure 4B), expanding by 15,665.133 km² from 2000 to 2020. Its main source was the conversion of grassland and unutilized land (Figure 14). Previous studies by Wang et al. [109] and by Zhao et al. [110] suggest that population growth and economic growth are the main forces driving the reclamation of grassland and unutilized land into farmland in the basin. Other studies on the Tarim River Basin show that the expansion of cultivated land has further increased the area of high-water-consumption crops, among which the area of high-water-consumption crop cotton has grown rapidly since 2000, and the amount of water (8000–10,000 m³) used to irrigate 1 hectare of cotton is more than twice that used for wheat and maize [111], resulting in an increase in agricultural water consumption, which accounts for as much as 95% of total water consumption, far exceeding the world average (70%) and representing a serious imbalance in the water structure [112]. At present, the development of water resources in the northern and western parts of the basin has exceeded the carrying capacity of regional water resources [66]. For example, the groundwater level in the Kuqa Oasis fell by 4.1 m from 2000 to 2018. This may be related to the regional expansion of cultivated land. Our findings show that from 2000 to 2015, the expansion of cultivated land was mainly concentrated in the northeast of the basin, with a faster expansion rate. From 2015 to 2020, the focus of cultivated land expansion gradually shifted to the western part of the basin (Figure 4A). An earlier study on the Tarim River Basin by Zhang et al. observes that the overexploitation of groundwater leads to the degradation of desert vegetation and exacerbates ecological degradation [113], which may be the main reason for the increased urgency of fallowing, as reflected in our SILF values.

With respect to spatial distribution, our study found that the trend of cultivated land expansion in the basin is mainly to expand outward from the existing area (Figure 4A). This may be the main reason for the upward trend in the area of NUF and VUF, which are mainly distributed near the edge of cultivated land in the desert area (Figure 9). In their study of the Tarim River Basin, Zhang et al. [114] found that this expansion trend has occupied the desert–oasis transition zone, destroyed the ecological barrier, caused land desertification, and exacerbated the deterioration of the ecological environment. There is, therefore, reason to believe that if cultivated land continues to expand, irreversible damage may be caused to the ecological environment.

From 2000 to 2020, the SILF values of different land use types converted to cropland were different (Figure 14), from low to high, in the order of construction land, water bodies, forest land, grassland, and unused land (Figure 15). In the Tarim River Basin, the SILF value of cultivated land converted from construction land is the lowest, which may be because construction land is usually artificially modified and invested in infrastructure, making it have better soil conditions and water resource management capabilities. In addition, construction land is mostly located in areas with convenient transportation, flat terrain, and sufficient water resources, and these areas may be more likely to receive technical support and management inputs for agriculture. The findings from our study are consistent with a study by Wang et al. [115] on land use and cropland quality in China, where it appeared that construction land was located in areas with similar conditions as high-yield farmland, such as gentle slopes and proximity to settlements and water bodies. With respect to conditions in the Tarim River Basin, it can further be argued that the soil around water bodies usually has a high water content and good fertility conditions. In arid areas, the scarcity of water resources makes the land around water bodies have high agricultural potential, especially if irrigation is guaranteed. Woodlands and grasslands usually have good soil structure and organic matter content, but in arid areas, the soil in woodlands and grasslands may be water limited, resulting in lower quality when converted to cropland than cropland converted from built-up land and water bodies. Unused land is usually located in areas with poor ecological conditions, poor soil, and scarce water resources without human management and improvement measures, resulting in the highest SILF value for the converted cultivated land.

Our investigation additionally showed that the cultivated land after expansion has a higher SILF value than the original cultivated land (Figure 16). These findings are consistent with a study by Cai et al. [116] on cropland expansion in arid regions of China, where it was observed that with the reduction in high-quality land resources, newly cultivated land is mostly located in areas with poor soil conditions (such as sandy land and saline–alkali land); newly cultivated land may lack effective soil improvement measures and sustainable management practices; arid areas have fragile ecosystems; and land reclamation may exacerbate soil degradation and water scarcity, further aggravating the ecological deterioration of cultivated land.

5.3. Recommendations for Fallowing Cropland in Arid Zones

First, the government should formulate strict land use plans to restrict the expansion of cultivated land in ecologically fragile areas. At the same time, the protection of existing cultivated land should be strengthened to avoid excessive reclamation and abuse. For cultivated land that has been expanded, soil improvement and water resource management should be strengthened to ensure its sustainable use. Second, they should be implemented in stages and managed precisely. According to the spatial distribution of the urgency level of fallowing, the very urgent fallow (VUF) cultivated land is located in areas with poor ecological quality of cultivated land, water scarcity, poor irrigation efficiency, low economic benefits of cultivated land output, large topographic relief, and low utilization level of cultivated land. For this type of fallow area, mandatory fallowing should be carried out every year under the condition of ensuring food security. The cultivated land with relatively urgent fallow (MUF) is located in areas with relatively poor ecological quality of cultivated land, water resources, irrigation efficiency, economic benefits of cultivated land output, topographic relief, and other factors, and it is barely suitable for farming. For these types of fallow areas, it is necessary to raise farmers’ awareness and establish incentive mechanisms to encourage and guide them to actively participate in fallowing. Generally urgent fallow (GUF) is located in areas where the various indicators of cultivated land are in a relatively good state and the use of cultivated land is moderately suitable. For this type of cultivated land, voluntary participation by farmers should be the main approach. Non-urgent fallow (NUF) cultivated land is located in areas where the indicators of all cultivated land elements are at their optimal or relatively optimal state. Cultivated land has a relatively good ability to be sustainably cultivated, and cultivated land can have a relatively high productivity. There is currently no need for fallow. Third, the management of fallow land should be strengthened. Fallowing cultivated land does not mean abandoning or leaving it fallow. Green manure crops, such as alfalfa and lentils, should be planted reasonably. For example, alfalfa and lentils have well-developed root systems and high carbon sequestration capacity, and they decompose slowly after being tilled, continuously releasing nutrients. This can achieve a virtuous cycle of “fertilizer nourishes the soil, which in turn increases production, which supports the people”.

5.4. Limitations and Prospects

This paper also has limitations that need to be overcome in future research. First, this paper only considers the output economy of cultivated land and does not consider other socio-economic factors that affect farmers’ fallowing, such as government subsidies for farmers to fallow land. In future research, the impact of other socio-economic factors on the cultivated land fallowing system may be explored through methods, such as questionnaire surveys and interpolation methods, in order to further optimize the indicator system for cultivated land fallowing. Second, this paper mainly uses remote sensing data, which may ignore subtle changes in land use and ecological conditions. In future research, field sampling, surveys, and other methods may be used to focus on the impact of this change on the cultivated land fallow system. Third, future research needs to further extend the time span of the data and improve the resolution in order to provide a scientific basis for refined agricultural management and policy formulation.

6. Conclusions

This study first analyzed the spatiotemporal characteristics of cultivated land expansion in the Tarim River Basin from 2000 to 2020 using a land use transition matrix and centroid migration model. Through remote sensing and multi-source data fusion, we constructed an index of the urgency of arable land fallow (SILF) incorporating human activity intensity, ecological vulnerability, output value, water resource status, and topographic conditions, revealing the spatiotemporal dynamics of SILF during 2000–2020. Key findings include the following. (1) The cultivated land in the Tarim River Basin exhibited continuous expansion, characterized by outward diffusion from existing cultivated areas, primarily through the conversion of grassland and unused land. The centroid of cultivated land shifted 37.833 km northeastward during 2000–2015, followed by a 7.257 km southwestward movement during 2015–2020. (2) Temporally, the Not Urgent Fallow (NUF) areas showed an overall declining trend, while the very urgent fallow (VUF) areas increased significantly from 2000 to 2020. Spatially, NUF and generally urgent fallow (GUF) were predominantly distributed in the interior cultivated zones, whereas VUF concentrated along the cultivated land margins near desertification-prone areas, gradually expanding inward. (3) The ecological environment of cultivated land demonstrated progressive deterioration, with degradation spreading from marginal zones toward interior regions. (4) In the process of arable land expansion, there is a significant difference in the SILF values of different land use types after being converted into arable land, which are, from low to high, constructed land, water bodies, forest land, grassland, and unused land, and the SILF values of expanded arable land are higher than those of the original arable land.