2.3.3. Evaluate Appropriate Tillage Sandy

We divided the research ideas of this article into the following three parts (Figure 4). First, this study used ENVI 5.0 and ArcGIS 10.8 as research platforms to extract the vegetation coverage of Hangjin Banner in 2013 by using the band calculation and raster classification functions of ENVI 5.0 and obtained the final range of extremely-low-coverage grassland combined with the spatial intersection function of ArcGIS. Based on the strong linear relationship between Normalized Difference Vegetation Index (NDVI) values and vegetation coverage, we employed a pixel-by-pixel bipartite model to estimate vegetation coverage (VFC). In practical applications, the NDVI of Vegetation (NDVIV) and NDVI of Soil (NDVIS) were not fixed, which makes the estimation of vegetation coverage more

difficult. Therefore, we used the maximum and minimum NDVI values during the plant growth season in August to approximately replace NDVIV and NDVIS.

**Figure 4.** Research framework for the evaluation of appropriate tillage of sandy land in arid sandy area.

Second, this study used the restriction factor method to screen sandy land suitable for cultivation. We digitized the collected indicator (including land use type, vegetation cover, irrigation conditions, ecological conditions, and soil properties) data into various evaluation index factors, which were used as limiting factors to obtain the relevant evaluation index information of the evaluation object by using the spatial analysis function of GIS. In addition, we used the single factor restriction and exclusion method to evaluate the evaluation units. Among the evaluation indicators of the evaluation unit, if any index item is unsuitable for farming, it was classified as unsuitable for farming, and the rest were suitable for farming sandy land.

Finally, this study used the spatial analysis function of GIS to obtain the evaluation objects of suitable arable sandy land in the process of dividing the evaluation objects of suitable arable sandy land. Based on the analysis of the survey and evaluation results of suitable arable sandy land, we counted the number of different types of suitable arable sandy lands and analysed the spatial distribution of different types of suitable arable sandy lands.

#### **3. Results**

#### *3.1. Analysis of the Results of Suitable Arable Sandy Land from an Overall Perspective*

The area of sandy land of suitable arable sandy land in this survey was 1,274,935.9 hm2, accounting for 67.5% of the total area of Hangjin Banner. After the screening of four limiting factors of land use or cover, ecological conditions, irrigation degree, and salinization, the sandy land area suitable for reclamation was 97,550.1 hm2, accounting for only 7.7% of the sandy land area, and 92.3% of evaluation unit is not suitable for farming. According to the different types of sandy land, the suitable ploughing sandy land in Hangjin Banner can be divided into explicit suitable visible sandy land and suitable invisible sandy land. The dominant land type suitable for arable sandy land was sandy land, and the land use types of invisible sandy land are cultivated land, inland tidal flats, saline-alkali land, natural and artificial grasslands with very low coverage, and other grasslands with low coverage (Table 2).


**Table 2.** Result of survey evaluation for appropriate tillage sandy land.

The area of visible sandy land suitable for cultivation was the largest at 23,344.2 hm2, accounting for 23.9% of all sandy land suitable for cultivation. The invisible sandy land suitable for cultivation was 74,205.9 hm2, accounting for 76.1% of the sandy land suitable for cultivation. Among the land use types of the invisible sandy land suitable for cultivation, the main land use types are the extremely-slow-coverage natural grassland and artificial grassland, accounting for 5668.1 hm2, accounting for 5.8% of the sandy land suitable for cultivation. However, the decrease was also the largest. First, less than 3% of natural pastures and artificial pastures were reserved as sandy land suitable for cultivation. Second, the area of inland tidal flats suitable for cultivation was 2010.9 hm2, accounting for 2.1% of all sandy land suitable for cultivation, and 57.6% of the inland tidal flats were suitable for development and utilization. Third, the area of other grassland suitable for cultivation was 1292.0 hm2, accounting for 1.3% of the suitable arable sandy land. The decrease was very large, and only 4.2% of other grassland was suitable for development and utilization. Finally, although the suitable arable saline-alkali land only accounted for 0.2% of the suitable arable sandy land, the decrease was the smallest, and 68.1% of the saline-alkali land was suitable for development and utilization (Table 2).

#### *3.2. Analysis of the Results of Suitable Arable Sandy Land from a Local Perspective*

The cultivated land suitable for cultivation was mainly distributed in Jirigalangtu town (33.1%), Duguitala town (32.2%) and Huhemudu town (15.3%). There was a small amount of distribution in Balagong town and Yihewususumu, while there was no cultivated land distribution in Xini town. From the spatial distribution, the arable land is mainly concentrated along the Yellow River in the north, where the irrigation conditions are relatively favourable (Table 3; Figure 5).

**Figure 5.** Type and spatial distribution of appropriate tillage sandy land.


**Table 3.** Statistics on spatial distribution of sandy soil suitable for ploughing in Hangjin Banner.

Natural grasslands and artificial grasslands suitable for cultivation were mainly distributed in Yihewususumu (40.3%) and Duguitala Town (28.5%). Except for Jirigalangtu town, where the distribution was only 11.06 hm2, the distribution in other towns was between 590 and 2286 hm2. From the spatial perspective, extremely-low-coverage grassland suitable for cultivating sandy land was intertwined with the dominant sandy land suitable for ploughing and sporadic inlaid in the dominant sandy land suitable for ploughing (Table 3; Figure 5).

The inland tidal flats suitable for farming are mainly distributed in Yihewususumu (23.7%), Xini town (17.4%), and Duguitala town (43.3%). In terms of spatial distribution, there was only a small area of Bayin Wendur Gacha in the Huhemudu Township in the northern Yellow Irrigation District. However, in the Liangwai District, it was distributed in strips along the inland rivers near Arishan Gacha and Baiyinbugacha (Table 3; Figure 5).

Other grasslands suitable for cultivation were mainly distributed in Huhemudu town (57.3%) and Duguitala town (21.9%), and the distribution in the other four towns was not large, ranging from 0.9% to 9.0%. This sandy land was mainly divided into two parts in space: one part was concentrated in Chagannur Gacha in Huhemudu town, and the other part was relatively concentrated in Sharizhao Gacha in Duguitala town. In general, other grassland pattern areas suitable for cultivation were small and fragmented (Table 3; Figure 5).

The saline-alkali land suitable for cultivation was distributed in the other four towns except Xini town and Yihe Wususumu town, and the most distributed was in Huhemudu town and Jirigalangtu town, accounting for 65.1% and 30% of the saline-alkali land suitable for cultivating sandy land, respectively. The sandy land suitable for cultivation in salinealkali land was concentrated in Chagannur in Huhemudu town and along the river in Bayinwenduer in Jirigalangtu town, with an area of approximately 180.8 hm<sup>2</sup> (Table 3; Figure 5).

The visible sandy land suitable for cultivation was mainly distributed in Huhemudu town (14.2%) and Duguitala town (51.6%), and the distribution in other towns was relatively small. Among them, the dominant sandy land in Huhemudu town was relatively concentrated and contiguous, and mainly concentrated in Chagannur Gacha in Huhemudu town; the dominant sandy land in Jirigalangtu town and Duguitala town was in the shape of a concentrated and continuous strip in space, and the span extends from Gegenzhao Gacha in Jirigalangtu town to Huhemudu Shari Zhao Gacha; the visible sandy land in the other four towns was relatively small in number and small in size, but it was relatively concentrated in space (Table 3; Figure 5).

#### **4. Discussion**

In this paper, Hangjin Banner was taken as the research area, and the evaluation object of sandy land research was expanded from the traditional soil type of aeolian sandy soil or the land type of sandy land to all sandy land [36]. According to the different types of land use, specifically vegetation coverage, the sandy land with low vegetation coverage in the current land use survey was regarded as the visible sandy land, and other land types with sandy soil but land use types classified as cultivated land, forestland, grassland, garden land, swamp, water surface, etc., were regarded as invisible sandy land. Although these land types have various types of cover, they have the risk of desertification, and even desertification due to the characteristics of sandy soil, so they need to be taken as the evaluation object [37]. In another method, limiting factors such as land use type, irrigation conditions, and salinization degree are screened and eliminated one by one, and an evaluation system of limiting factors suitable for cultivated sandy land is constructed [38]. This method abandons the conventional comprehensive evaluation method of index factor scoring, avoids the superposition of factors, and ignores the influence of dominant control factors [39]. Our result also shows that 95.7% of the land units were screened by this method (compared with the sandy land units before screening), which means our research method is more effective.

Vegetation coverage can quantitatively characterize the degree of land desertification, but there are large differences in the thresholds of vegetation coverage set by different researchers for the degree of desertification [40,41]. According to the classification of vegetation coverage on desertification degree in "Technical Regulations for Monitoring Desertification" (GB/T 24255-2009), in the investigation and evaluation of arable sandy land in Hangjin Banner, vegetation coverage ≤ 10% is regarded as extremely-low-coverage vegetation, and the corresponding degree of desertification is extremely severe desertification. In addition, taking 10% ≤ vegetation coverage ≤ 30% as low-coverage vegetation, the corresponding degree of desertification is severe desertification; taking vegetation coverage ≥ 30% as medium–high vegetation, the corresponding degree of desertification is moderate–slight desertification [42]. This division is based on the idea of protective development and with reference to the overall situation of local land desertification, and other sandy areas can be selected for threshold selection and related research based on this method [43].

The current situation of land use is an important indicator for the investigation and evaluation of arable sandy land. In the investigation and evaluation of arable sandy land in Hangjin Banner, the vegetation coverage calculated by remote sensing is superimposed with the current situation of land use. Among the sandy land types, extremely severe sandy land (vegetation coverage ≤ 10%) and severe sandy land (10% ≤ vegetation coverage ≤ 30%) account for 96% of the total area of sandy land. In the grassland category, 1% of the grassland had extremely-low-coverage vegetation (vegetation coverage ≤ 10%), and 31% of the other grasslands had medium and high coverage (vegetation coverage ≥ 30%). Overall, the accuracy of sandy land surveying and mapping is relatively high [44,45]. However, for the purpose of research, we should further divide the land types in the current land use situation to meet the accuracy needs of the research. Therefore, on the basis of the current situation of land use, sandy land with vegetation coverage ≥ 10% in sandy land and grassland with vegetation coverage ≥ 10% in invisible sandy land were excluded.

#### **5. Conclusions**

In this study, Hangjin Banner was taken as the research area, and all sandy land was taken as the research object. The restricted factor exclusion method was used to investigate and evaluate arable sandy land in arid and semiarid areas. The results show that the total area of visible sandy land and invisible sandy land in Hangjin Banner is 1,274,935.9 hm2. Among this area, the total area of arable sandy land is 97,550.1 hm2, accounting for 7.7% of all sandy land. On the basis of invisible arable sand land and visible arable sand land, according to the land use type, invisible arable sand land can be divided into arable land, inland beach arable sand land, other grassland arable sand land, saline-alkali arable sand land, and very-low-coverage grassland (natural grazing grassland and artificial grazing grassland), accounting for 66.7%, 2.1%, 1.3%, 0.2%, and 5.8% of the area of arable sand land, respectively.

We believe that the research method of this paper is effective, which can provide a valuable reference for the sustainable development and utilization of arable sandy land and desertification control in arid and semiarid areas. Meanwhile, it should be noted that the development and utilization of arable sandy land is a systematic project, and the close cooperation and overall arrangement of all links of investigation and evaluation, planning and layout and engineering design are very necessary. We hope that this study can provide the arid and semiarid areas with similar development conditions as Hangjin Banner in the world, combined with the local actual situations, to use the restrictive factor exclusion method to determine the development and utilization area of arable sand, and formulate a more practical development and protection scheme of arable sand.

**Author Contributions:** Conceptualization, Y.X. and F.Z.; methodology, Y.X.; software, Y.Z.; validation, K.W., Y.X. and F.Z.; formal analysis, Y.Z.; investigation, Y.Z.; resources, Y.Z.; data curation, K.W.; writing—original draft preparation, Z.C.; writing—review and editing, K.W.; visualization, Y.X.; supervision, F.Z.; project administration, Y.X.; funding acquisition, Y.X. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China (grant number 41301614) and the Special Scientific Research of the Ministry of Land and Resources of China—Key Technology and Demonstration based on Protective Development of Sandy Land in Inner Mongolia (grant number 201411009).

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

**Informed Consent Statement:** All participation was voluntary and verbal consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Data and material are available upon request.

**Acknowledgments:** We are grateful for the comments and criticisms of the journal's anonymous reviewers, as well as our colleagues.

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

#### **References**

