**1. Introduction**

China is a huge nation with significant regional variations in land and water resources. Divided by the Qinling Mountains–Huaihe River line, a unique pattern of arable land use of "paddy fields in the south and drylands in the north" has formed. In the south, paddy fields account for the majority of the total area of paddy fields in China. The soil types in the south are mainly yellow–brown soil and red soil (on the Middle-Lower Yangtze Plain), lateritic red soil and latosol soil (on the Pearl River Delta Plain), purple soil and paddy soil (in the Sichuan Basin), and yellow soil (in the hilly southeastern region). The climate in the south is dominated by a tropical and subtropical monsoon climate, with an accumulated temperature of 4500–8000 ◦C, abundant precipitation, and sufficient heat, which forms a pattern of aquatic-based paddy farming. In the north, drylands account for the majority of the total drylands in China. The soil types are mainly Huangmian soil and Heilu soil (in the Loess Plateau), gray–brown desert soil (in Xinjiang), brown soil and cinnamon soil (in the North China Plain), and black soil and dark-brown soil (in the Northeast Plain). The warm-temperate continental monsoon climate in the north, with an

**Citation:** Xie, S.; Yin, G.; Wei, W.; Sun, Q.; Zhang, Z. Spatial–Temporal Change in Paddy Field and Dryland in Different Topographic Gradients: A Case Study of China during 1990–2020. *Land* **2022**, *11*, 1851. https://doi.org/10.3390/ land11101851

Academic Editors: Yongsheng Wang, Qi Wen, Dazhuan Ge and Bangbang Zhang

Received: 8 September 2022 Accepted: 14 October 2022 Published: 20 October 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

accumulated precipitation of approximately between 400 and 800 mm, forms scarce water resources, which contribute to a pattern of wheat- and maize-based dryland farming.

However, the change in paddy field and dryland is continuously dynamic and implicit, and its long-term accumulation may lead to implicit differences in arable land use. Taking the quantity and structure of paddy field and dryland in each province of China as an example, the increase in drylands and the decrease in paddy fields occurred simultaneously over the past 30 years. The province with the largest increase in paddy fields is Heilongjiang, with a cold-temperate climate in the northeast (+27,030 km2). The province with the largest increase in drylands is Xinjiang, with a dry climate in the northwest (+23,306 km2). The province with the largest decrease in both paddy fields and drylands is Guangxi Province, with fertile soil in south China (paddy fields and drylands decreased by 23,306 and 14,719 km2, respectively). In addition, relevant studies also pointed out that arable land has decreased in south China and increased in North China [1,2]. Specifically, the paddy fields increased rapidly in the northeast, whereas the drylands increased rapidly in the northwest. The spatial barycenter of paddy field and dryland in China moved northward, which indicates that the arable land moved from better agricultural areas to worse ones [3,4]. Thus, paddy fields and drylands are mostly reduced in areas with a better natural environment, while the newly added drylands and paddy fields are mostly located in areas with worse natural endowments. For example, benefited by the warm and humid climate, the Middle-Lower Yangtze Plain contributed one of highest yields of rice, wheat, and cotton in China. On the other hand, in inland Northwest China, surface water only accounts for approximately 8% of the total runoff in China because of the dry climate. The decrease in arable land in the Yangtze Plain and the increase in arable land in Northwest China shows a spatial mismatch of natural endowment and land resources. As a result, a serious spatial mismatch exists between the distribution of paddy field and dryland and the distribution of high-quality natural resources in China [5–9].

Research on the distribution patterns of paddy field and dryland has increased rapidly over the past 30 years. Previous studies mostly focused on local areas with unique geomorphological characteristics [10,11] and were devoted to analyzing the spatiotemporal change, landscape characteristics, or ecological service efficiencies of paddy field and dryland from the perspective of a single topographic element [10,12,13]. In Heilongjiang Province, Li [14] and Chen [15] found that the spatial expansion of arable land showed a strong directional trend, and the topography and geomorphology conditions were the key factors influencing the changes in paddy field and dryland. Gao [16] found that new arable and lost arable land were mainly concentrated in the plains, followed by the tablelands and hills. It was found that the drylands and paddy fields in the Loess Plateau and Chongqing city were mainly distributed at lower elevations [17,18]. By comparing land-use images from 1933, 1955, 1990, and 2005, Liu [19] found that new paddy fields in the Jinjing River of Hunan Province mainly came from woodland, which was mainly affected by topography conditions (especially elevation). By using the logit model, Zhong [20] analyzed the low mountain hilly area in southeastern China from 1999 to 2006 and found that the loss of agricultural land was the highest at low altitudes, followed by medium and high altitudes. The above studies indicate the importance of topographic factors when analyzing arable land change. Topographic differences impact the configuration of surface water, fertilizer, air, and heat, which forms the spatiotemporal differences between paddy field and dryland. Therefore, the change in paddy field and dryland on different topographic gradients should become a new tool for studying the mechanism of arable land utilization. However, in a vast area with great geographical differences, such as in China, the differentiation of long-term spatial–temporal changes in paddy field and dryland based on topographic gradients is missing.

It is crucial to study the spatial–temporal evolution and internal changes in paddy field and dryland under different topographic gradients in China. As the two major subtypes of arable land, paddy field and dryland may show different changes, further reflecting the implicit transformation of arable land use. This study refines the research perspective of arable land use by separately observing paddy field and dryland and discusses the following questions:

I. When observing topographic conditions in terms of elevation, slope, and slope aspect, are there any differences in the spatial distribution between paddy field and dryland?

II. What is the difference in the land type changes among paddy field, dryland, and other land under different topographic conditions?

III. What do the landscape characteristics of paddy field and dryland look like from the perspective of different topographic conditions?

To address these problems, this paper took the topographic gradient as the research perspective and analyzed the spatial–temporal differentiation of two types of arable land use, paddy field and dryland, at different elevations, slopes, and slope aspects. The scientific basis for optimizing and reconstructing sustainable arable land use was provided based on the research results.

#### **2. Research Methods and Data Source**

#### *2.1. Data Sources*

This paper used land-use raster data on China with a resolution of 1 km in 1990, 1995, 2000, 2005, 2010, 2015, and 2020. These data came from the Chinese Academy of Sciences, Resource, and Environmental Science Data Center (https://www.resdc.cn/, accessed on 10 February 2021). The land-use raster data were established based on remote sensing satellite imagery data (Landsat 8 OLI and GF-2). Using a high-resolution remote sensing drone ground survey observation system, the spatial distribution of the land-use cover was interpreted artificially by manual visual interpretation. The testing involved a large number of samples and achieved > 94.3% accuracy [1], and such testing has played an important role in national land research. The land-use classification included six land-use types: arable land, woodland, grassland, water area, construction land, and unutilized land. These data extracted the arable land in the data set and included the two major subtypes of arable land used in the analysis: dryland and paddy field. Specifically, dryland included rain-fed land and land with irrigation facilities that grows dryland crops; paddy field contained land with irrigation facilities that grows aquatic crops.

#### *2.2. The Identification of the Topographic Conditions of Paddy field and Dryland*

This study used 250 m DEM raster data to calculate the elevations and slopes of paddy field and dryland using the ArcGIS Toolbox. We extracted by mask the 1 km land-use data by the reclassified DEM information to obtain the different topographic conditions of the paddy field and dryland distribution. Based on the suitability of growing staple crops, the elevations were divided into 7 gradients including 0–200, 200–500, 500–1000, 1000–1500, 1500–2500, 2500–3500, and >3500 m. Similarly, the slopes were divided into 5 grades: 0–2◦, 2–6◦, 6–15◦, 15–25◦, and >25◦. The slope directions were divided into north, northeast, east, southeast, south, southwest, west, and northwest categories.

#### *2.3. The Transformation between Paddy Field and Dryland*

This paper calculated the land-use transformation matrix among paddy field, dryland, and other land-use types based on the toolbox in ArcGIS 10.2. The calculations were as follows:

Firstly, paddy field, dryland, and other land-use types in the Ath year were reclassified as 1, 2, and 3. Paddy field, dryland, and other land-use types in the Bth year were reclassified as 10, 20, and 30.

Secondly, the land-use layers of the Ath year and the Bth year were summed in the raster calculator, and the attribute data of the newly exported layer indicated the number of rasters that changed from one land-use type in the Ath year to another one in the Bth year. For instance, 12 indicates the land-use change from a paddy field in the Ath year to dryland in the Bth year. Detailed attributed data and the implication of the land-use change are listed in Table 1.

Thirdly, the total area of land changed from one land-use type to another was calculated by multiplying the number of the changed rasters by the area of the rasters (i.e., 1 km<sup>2</sup> in this study).

**Table 1.** The attribute data and the implication of the land-use change.

