*4.3. Simulation and Prediction of Land Use in the Ecotone and Its Basin in 2030* 4.3.1. Accuracy Verification

Since the shrinkage of the ecotone is closely related to land-use type changes, we also predicted future land-use changes based on the CA-Markov model in order to understand the future changes of the transition zone. To simulate future land-use changes in the basin, we first validated the accuracy of the 2015 land-use data. Since the desert-oasis ecotone of the Tarim River Basin is included in the whole Tarim River Basin area, we directly validated the accuracy of the simulation of the entire basin. The land-use structure of the study area was simulated and predicted based on the CA-Markov model. Land-use maps in 1990 and 2000 were defined as input data to simulate land use in 2015. Effectiveness of the simulation was assessed using spatial raster contrast in which the land-use type in specific spatial locations was compared to the actual 2015 land-use map.

Figure 7 shows that the simulation error, as determined by inconsistent land-use locations, was 3.62% of the total number of raster cells. Most inconsistencies appeared adjacent to water bodies and forest land. A total of 96.38% of the regions were consistent with the actual map in 2015. The Kappa index of the simulation result was 0.9551, also indicating the high reliability of the result and the CA-Markov model in predicting land-use

types. The Kappa index of each sub-basin in the Tarim River Basin exceeded 0.80, which meets the accuracy requirement of the Kappa index.

**Figure 7.** Comparisons of measured (**a**) and simulated (**b**) land-use maps of the Tarim River Basin in 2015, along with verification of forecast accuracy; (**c**) The simulation error between 1990 and 2015; (**d**) the Kappa index for each river basin.

The quantitative accuracy of the area of each land type in the simulation was also evaluated by comparing it with the actual area in 2015 (Table 2). The prediction error in 2015 was expressed as the absolute value of the error between the predicted and actual values of each land-use type area. Except for forest land and water bodies, where the error was 6% and 8.56%, respectively, the error in predicting the land-use types was within 5%. The error associated with industrial land and unused land was less than 1%, indicating that the simulation method had a high precision and credibility. Therefore, the CA-Markov model was able to effectively simulate the land-use changes in the study area and can be used to simulate future land use.


**Table 2.** Comparison between simulated and actual land use within the study area in 2015.

4.3.2. Forecast of Changes in the Desert-Oasis Ecotone in the Tarim River Basin

Figure 8 predicts the spatial distribution of land-use change in the Tarim River Basin in 2030 using the CA-Markov model. The prediction suggests that past land-use trends in the Tarim River Basin will continue in 2030 (Table S7). The land type with the greatest change will be arable land, whose area will increase from 31,647 km<sup>2</sup> in 2015 to 34,909 km<sup>2</sup> in 2030, an increase of 10.31%. The land-use type exhibiting the largest decrease will be grassland (it will decrease by 12,497 km<sup>2</sup> compared to 2015), while forest land area, industrial land, water bodies, and unused land will exhibit a small increase. Future projections and simulations of land types within the Tarim River desert-oasis ecotone can infer its ecological status

and development trends. This study simulated and predicted land-use changes within the ecotone by 2030, using the 2015 ecotone extent as the boundary. Land-use changes within the ecotone show a similar trend to the entire area. The simulation found that the arable land area in the ecotone will increase significantly, from 1033 km<sup>2</sup> in 2015 to 2599 km<sup>2</sup> in 2030, while the areas of forest and grassland will decrease by 318 km<sup>2</sup> and 833 km<sup>2</sup> , respectively. As the area of natural vegetation decreases and the arable land increases within the ecotone, the quality of the future ecotone habitats will further deteriorate.

**Figure 8.** Land-use in Tarim River Basin in 2030.

#### **5. Discussion**

#### *5.1. Criteria for the Classification of the Desert-Oasis Ecotone*

There are different approaches to the current delineation of the desert-oasis ecotone. Most scholars define the ecotone as a zone of limited width along the edge of the oasis [48], while others extend the ecotone to the entire foothills or define it as terrain without a clear spatial location [49]. In this study, we found that the usual criteria for delineating the desert-oasis ecotone do not work well in the Tarim River Basin, which consists of a large area composed of many watersheds. As a result, the ecotone is not uniformly distributed. Therefore, the NDVI contour data calculated from the TM images (30 m resolution) were used here, in conjunction with the remote sensing images of land-use/land-cover change in 1990, 2000, and 2015. The land-use type corresponding to the NDVI values in the range between 0.05 and 1 were interpreted to be forest and grassland, whereas the land-use type corresponding to NDVI values between 0.35 and 0.95 were mostly artificial oases organized in neat patterns of blocks and strips. Thus, NDVI values in the range between 0.05 and 0.35 were selected as the desert-oasis ecotone in the Tarim Basin. The results were verified by using ENVI and ACRGIS to interpret, classify, and digitize the images from the three analyzed years, and the results were verified by fieldwork.

#### *5.2. Combined Effect of Climate Change and Human Activities on the Desert-Oasis Ecotone*

Analysis of temperature and precipitation in the study area shows that both parameters have increased since 1990 (Figure 1) but the increase in temperature has been much greater than the increase in precipitation. By calculating the SPEI values of the basin for different time scales, we found that there has been an increase in drought conditions in the study area since 2005, especially during the years of 2005, 2006, and 2008, when droughts reaching the moderate level occurred year-round. The increase in droughts, and external climatic changes in general, may be an important reason for the accelerated decrease in the area of the ecotone after 2000. Runoff (flow) in the Tarim River Basin is primarily

generated in high mountain areas where glaciers and snowmelt recharge dominate. The input from these water sources is very sensitive to global climate change. The increased hydrological volatility and water resource uncertainty caused by climate change may lead to more prominent conflicts between water supply and demand in the oasis economy and desert ecosystem in the basin [50].

Because of the expansion of arable land, agricultural water consumption remains high. Agricultural water has long been the main form of water in this area. The proportion of agricultural water is too large, and the water structure is seriously imbalanced. The proportion of agricultural water in the basin has long been as high as about 95%, which is much higher than the Chinese average (65%) and the world average (70%). At present, the development of water resources in the Tarim River Basin has greatly exceeded the carrying capacity of regional water resources [28]. For example, the groundwater level in the Kaxgar River Basin dropped by nearly 1 m between 2004 and 2010 and the water crisis has become more prominent [25]. Groundwater overexploitation has led to the degradation of desert vegetation and damage to ecosystems [51]. For instance, the 321 km river cutoff in the downstream reach of the Tarim River has caused shrinkage and even disappearance of oases [52,53]. The reduction in ecological water has also led to a decrease in surface vegetation cover, NDVI, and the area of the desert-oasis ecotone.

#### *5.3. Applicability of the Land-Use Change Model and Future Work*

Human influence on land use reflects not only natural factors but also economic and social factors. Therefore, predicting land-use changes is extremely important for promoting natural, economic, and sustainable development and protecting ecological balance. Relevant data show that the irrigated area of the Tarim River Basin has increased by 67% in the last 30 years [53] and the future expansion of cultivated land area will decrease ecological space and ecological water, leading to a shrinkage in the desert-oasis ecotone and a decline in its function as an ecological barrier. Therefore, it is necessary to perform quantitative prediction and analysis of future land-use and pattern changes in the Tarim River Basin. The CA-Markov model is widely used in urban land-use pattern simulation and in the assessment of watershed land-use change; however, the model has rarely been applied in arid areas, especially in areas with complex geographical features combining desert-oasis characteristics. This study experimentally applied and optimized the CA-Markov model using relevant, scientifically selected factors, and improved the simulation accuracy to obtain reliable prediction results. As a case study, it provides a good example and basis for the prediction of land-use change in arid areas.

The model also has a few shortcomings related to the quantification of some factors. For example, the influence of road distance, water body distance, and various administrative policies were not considered and the analysis of the social, geographical, economic, and resource environment that affects land-use changes was not fully characterized. Therefore, to improve model accuracy, future research should consider the influence of natural and human factors on the change in the geospatial system. In addition, future research can attempt to establish a model of land-use/land-cover change under the joint action of several different influencing factors and different decision makers by adding weighting elements.

#### **6. Conclusions**

This paper investigated the dynamic evolution of the desert-oasis ecotone in the Tarim River Basin and predicted the near-future land-use change in the desert-oasis ecotone using the CA-Markov model. The main findings of this paper are as follows:

With the decrease in the NDVI (from 0.142 in 1990 to 0.127 in 2015) of the desert-oasis ecotone, the area of the ecotone also shrank from 67,642 km<sup>2</sup> in 1990 to 46,613 km<sup>2</sup> in 2015. In the context of global climate change, the temperature showed a significant increase compared with precipitation, which led to an obvious increase in aridity in the study area. Meanwhile, the increase in arable land area led to a decrease in the groundwater table. The above factors have led to the shrinkage of the desert-oasis ecotone in the Tarim River Basin.

The CA-Markov model was verified to have good applicability in this study area, which was used to predict and simulate the future land dynamics of the study basin. Assuming the present development trend continues without intervention, the arable land area in the ecotone will increase from 1033 km<sup>2</sup> in 2015 to 2599 km<sup>2</sup> in 2030 and the woodland area and grassland area will decrease from 318 km<sup>2</sup> to 833 km<sup>2</sup> , respectively. The main land-use types in the Tarim River Basin in 2030 will be arable land, unused land, and grassland.

In light of the above conclusions, it is necessary to establish reasonable management countermeasures for land-use planning in Tarim River Basin development to achieve sustainable development and protect the ecology of the basin.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2072-429 2/13/4/647/s1: Table S1: The detailed information on Landsat images used in this study; Table S2: Land cover classification system; Table S3: SPEI categories; Table S4: Kappa index and consistency relationship; Table S5: Changes in the ecotone area of each basin in 1990, 2000, and 2015; Table S6: Transfer matrix of land-use area between 1990 and 2015 in the Tarim River Basin; and Table S7: Area of land-use types in the study area in 1990, 2000, 2015, and 2030.

**Author Contributions:** F.S. and Y.W. conceived the original design of this paper. Y.C. and Y.L. improved the structure of the paper. Q.Z., J.Q., and P.M.K. provided comments on this paper. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research was supported by the National Youth Thousand Talents Project (Y771071001).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** We thank LetPub (www.letpub.com) for the linguistic assistance and scientific consultation provided during the preparation of this manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

