Case Study of Xiangzhou, Hubei Province, China

agriculture

Article Application of Three Deep Machine-Learning Algorithms in a Construction Assessment Model of Farmland Quality at the County Scale: Case Study of Xiangzhou, Hubei Province, China

Li Wang , Yong Zhou *, Qing Li , Tao Xu, Zhengxiang Wu and Jingyi Liu

The College of Urban & Environmental Sciences, Central China Normal University, Wuhan 430079, China; [email protected] (L.W.); [email protected] (Q.L.); [email protected] (T.X.); [email protected] (Z.W.); [email protected] (J.L.) * Correspondence: [email protected]

Abstract: Constructing a scientiﬁc and quantitative quality-assessment model for farmland is important for understanding farmland quality, and can provide a theoretical basis and technical support for formulating rational and effective management policies and realizing the sustainable use of farmland resources. To more accurately reﬂect the systematic, complex, and differential characteristics of farmland quality, this study aimed to explore an intelligent farmland quality-assessment method that avoids the subjectivity of determining indicator weights while improving assessment accuracy. Taking Xiangzhou in Hubei Province, China, as the study area, 14 indicators were selected from four dimensions—terrain, soil conditions, socioeconomics, and ecological environment—to build a comprehensive assessment index system for farmland quality applicable to the region. A total

of 1590 representative samples in Xiangzhou were selected, of which 1110 were used as training samples, 320 as test samples, and 160 as validation samples. Three models of entropy weight (EW),

Citation: Wang, L.; Zhou, Y.; Li, Q.; backpropagation neural network (BPNN), and random forest (RF) were selected for training, and the Xu, T.; Wu, Z.; Liu, J. Application of assessment results of farmland quality were output through simulations to compare their assessment Three Deep Machine-Learning accuracy and analyze the distribution pattern of farmland quality grades in Xiangzhou in 2018. The Algorithms in a Construction results showed the following: (1) The RF model for farmland quality assessment required fewer Assessment Model of Farmland parameters, and could simulate the complex relationships between indicators more accurately and Quality at the County Scale: Case analyze each indicator’s contribution to farmland quality scientifically. (2) In terms of the average Study of Xiangzhou, Hubei Province, quality index of farmland, RF > BPNN > EW. The spatial patterns of the quality index from RF and China. Agriculture 2021, 11, 72. BPNN were similar, and both were significantly different from EW. (3) In terms of the assessment https://doi.org/doi:10.3390/ results and precision characterization indicators, the assessment results of RF were more in line agriculture11010072 with realities of natural and socioeconomic development, with higher applicability and reliability. (4) Compared to BPNN and EW, RF had a higher data mining ability and training accuracy, and its Received: 19 November 2020 2 Accepted: 13 January 2021 assessment result was the best. The coefficient of determination (R ) was 0.8145, the mean absolute Published: 16 January 2021 error (MAE) was 0.009, and the mean squared error (MSE) was 0.012. (5) The overall quality of farmland in Xiangzhou was higher, with a larger area of second- and third-grade farmland, accounting for Publisher’s Note: MDPI stays neu- 54.63%, and the grade basically conformed to the trend of positive distribution, showing an obvious tral with regard to jurisdictional clai- pattern of geographical distribution, with overall high performance in the north-central part and low ms in published maps and institutio- in the south. The distribution of farmland quality grades also varied widely among regions. This nal affiliations. showed that RF was more suitable for the quality assessment of farmland with complex nonlinear characteristics. This study enriches and improves the index system and methodological research of farmland quality assessment at the county scale, and provides a basis for achieving a threefold

Copyright: © 2021 by the authors. Li- production pattern of farmland quantity, quality, and ecology in Xiangzhou, while also serving as a censee MDPI, Basel, Switzerland. reference for similar regions and countries. This article is an open access article distributed under the terms and con- Keywords: farmland quality; machine-learning algorithms; comprehensive assessment index system; ditions of the Creative Commons At- random forest; county scale; Xiangzhou tribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Agriculture 2021, 11, 72. https://doi.org/10.3390/agriculture11010072 https://www.mdpi.com/journal/agriculture Agriculture 2021, 11, 72 2 of 23

1. Introduction Farmland is an important natural resource for people to carry out agricultural production, and it plays an important role in ensuring food security, promoting sustainable economic development, and maintaining social harmony and stability [1–3]. However, since the reform and opening up of China, which was accompanied by accelerated urbanization, the amount of farmland area has decreased sharply, especially high-quality farmland around the cities, which has been occupied to a larger extent, creating certain challenges to food security. In order to strictly adhere to the red line of 120 million hectares of farmland, the Chinese government has formulated a very stringent farmland protection system. In the past 10 years, the area of farmland in China has increased to some extent, but the quality has decreased overall, with only 27.3% corresponding to high-quality farmland [4]. In addition, the input of new technologies and varieties has brought hidden security risks—such as pollution—to farmland, which seriously threatens sustainable socioeconomic development and the ecological environment [5]. Therefore, actively carrying out studies on farmland quality assessment; effectively improving and protecting farmland quality; implementing a threefold production pattern of farmland quantity, quality, and ecology; and firmly guarding the red line of farmland are reliable ways to achieve coordinated and sustainable economic and social development and address food security. The quality of farmland refers to the status and condition of the land [6]. Farmland quality has extended from a single dimension of basic land fertility to a comprehensive quality involving suitability, production potential, soil environmental quality, and sustainability [7–9]. Scholars in other countries mainly study the temporal and spatial evolution and sustainable use of farmland, while those in China are more concerned with assessing farmland quality and its connection to food production capacity [10]. At present, there is no standard definition of farmland quality. It is generally accepted that farmland quality is the mixed result of natural, social, economic, and technological progress [11], including four aspects: soil, spatial, management, and ecological quality [12]. The assessment of farmland is a quantitative measure of its state or the extent to which it meets functional needs [13]. Traditional farmland quality assessment mainly considers the natural production potential of land, selects natural factors to construct an index system, and evaluates the suitability of farmland [14,15]. In recent years, environmental factors such as social development, economic level, and utilization patterns have received attention in farmland quality assessment, and the focus has shifted to considering many factors, such as natural, ecological, social, and economic factors [16–18]. In the assessment of farmland quality, different methods are chosen due to differences in the objectives and selection of indicators. The increasing research and application of “3S” technology which refers to remote sensing (RS), geography information systems (GIS), and global positioning systems (GPS), computer technology, and mathematical models has led to the development of more methods for farmland quality assessment, ranging from simple qualitative description to quantitative analysis. However, no matter which method is used, the selection of assessment indicators, classification criteria of indicators, and determination of indicator weights will affect the accuracy of assessment. Therefore, constructing a reasonable index system and exploring effective methods have become important elements of the current research on farmland quality assessment. Currently, the main methods commonly used for farmland quality assessment are entropy weight (EW) [19], fuzzy assessment [20], analytic hierarchy process (AHP) [21], grey correlation analysis [22], and geostatistical methods [23]. EW assigns objective weights based on the physical characteristics of the data, but does not introduce human cognitive judgment of the assessment indicators [19]. The AHP is more subjective, and it is difficult to balance the sensitivity of indicator weights [21]. Fuzzy assessment, grey correlation analysis, and geostatistical methods are less stable in handling high-dimensional data, and have difficulties in drilling down into nonlinear information [24]. In recent years, the use of artificial intelligence algorithms has become a hot method for constructing Agriculture 2021, 11, 72 3 of 23

models for farmland quality assessment, among which artificial neural network and support vector machine (SVM) are typical representatives [25]. Qie et al. [26], Gao [27], and Shekofteh et al. [28] applied a self-organizing neural network, a backpropagation neural network (BPNN), and a genetic algorithm–BP neural network, respectively, to evaluate farmland quality. Gayen et al. [29] used support vector machine to evaluate the sensitivity of farmland soil erosion. Lai et al. [30] used rough sets and support vector machine to evaluate the quality of high-standard farmland. The artificial neural network has strong robustness, memory ability, nonlinear mapping ability, and self-learning ability. It is capable of digging deep into the intricate relationships between indicators and autonomously constructing a mathematical model between the assessment object and the indicators. However, there are shortcomings, such as overfitting, large generalization errors, and difficulty in determining indicator weight [31]. Compared to the artificial neural network, the support vector machine has strong adaptive and generalization capability for high-dimensional and nonlinear data [32]. It can convert assessment problems into convex optimization problems with equal value, such that the weight of indicators can be determined according to the data distribution [33]. It overcomes the problem of local overfitting, but also suffers from problems such as difficult sample partitioning, subjective parameter selection, and overlearning [34]. Thus, how to effectively unify multisource environmental variables into the same assessment unit remains a key and difficult issue in the assessment of farmland quality. Random forest (RF) is a machine-learning algorithm for constructing multiple decision trees for combinatorial classification based on a random resampling technique (bootstrap) and random node splitting technique, and has strong anti-noise and model generalization abilities [35]. RF can achieve a higher prediction rate with optimal parameters and minimum error based on smaller training samples [36]. It can evaluate the importance of indicators during the training process, and thus solve the problem of overfitting indicators within a nonlinear complex system [37]. Farmland quality is a complicated system combining natural, socioeconomic, and ecological factors that requires superior intelligent algorithms to overcome the challenges of nonlinearity, high dimensionality, and missing values. At the same time, RF, as a nonparametric decision tree model, combines all the advantages of previous assessment methods, and has the advantage of dealing with the nonlinear relationships and weight dynamics of high-dimensional data. High prediction accuracy is maintained even when training samples have noise and are missing values. Thus, it is feasible to evaluate farmland quality using a random forest model [38]. Khamoshi et al. [39] used random forest for mapping soil and evaluating the suitability of farmland. Bhowmik et al. [40] evaluated the quality of soil environment based on random forest. Chen et al. [41] used random forest to measure the importance of assessment indicators and construct an assessment index system for farmland capacity enhancement potential. Lin et al. [42] used random forest and correlation analyses to select indicators of quality farmland and determine their weights. Taking Xiangzhou of Hubei Province in China as an example, an assessment index system applicable to this region was constructed, and three models of entropy weight (EW), backpropagation neural network (BPNN), and random forest (RF) were selected for farmland quality assessment to avoid the subjectivity of determining index weights, and to compare the assessment accuracy of the three methods to verify the reliability and superiority of the RF model in this study. This study aimed to explore new ways to evaluate farmland quality, expand the methodology of farmland quality assessment, and improve the accuracy of assessment, in addition to serving as a reference for similar studies.

2. Materials and Methods 2.1. Study Area Xiangzhou is a county-level city in the northwestern part of Hubei Province in China (Figure1a) near the middle stream of the Han River, located between 111 ◦440–112◦230 E and 31◦46–32◦280 N. This county is a hilly and mountainous area surrounded by low Agriculture 2021, 11, x FOR PEER REVIEW 4 of 23

2. Materials and Methods 2.1. Study AREA Agriculture 2021, 11, 72 Xiangzhou is a county-level city in the northwestern part of Hubei Province in China4 of 23 (Figure 1a) near the middle stream of the Han River, located between 111°44′–112°23′ E and 31°46–32°28′ N. This county is a hilly and mountainous area surrounded by low mountains, with basins and plains in the middle (Figure 1b). The landscape pattern is mountains, with basins and plains in the middle (Figure1b). The landscape pattern is around 70% hillocks, 10% mountains, and 20% plains. Xiangzhou belongs to a subtropical around 70% hillocks, 10% mountains, and 20% plains. Xiangzhou belongs to a subtropical humid monsoon continental climate zone that is affected by alternating cold and warm humid monsoon continental climate zone that is affected by alternating cold and warm air. TheThe countycounty hashas fourfour distinctdistinct seasons,seasons, withwith thethe samesame hothot andand rainyrainy seasons,seasons, moderatemoderate precipitation,precipitation, and an an average average annual annual temperature temperature of of15.3–15.8 15.3–15.8 °C.◦ AsC. Asof 2018, of 2018, the total the total area areaof arable of arable land landin Xiangzhou in Xiangzhou was 165,257 was 165,257 ha, accounting ha, accounting for 66.98%. for 66.98%. Xiangzhou Xiangzhou has excel- has excellentlent natural natural and ecological and ecological conditions conditions and strong and strong food-production food-production capacity capacity in Hubei in Hubei Prov- Province,ince, with withwheat, wheat, rice, rice,and corn and cornas the as main the main grain grain crops crops [43].[ 43The]. Thetotal total grain grain output output was was1,172,800 1,172,800 tons tons in 2018. in 2018. However, However, in recent in recent years, years, with with the the acceleration acceleration of ofurbanization, urbanization, a alarge large amount amount of of high-quality high-quality farmland farmland around around towns towns has has been occupied, and frequentfrequent human disturbances havehave causedcaused somesome harm to the farmland quality and soil ecology [44]. [44]. Therefore,Therefore, itit is is of of great great importance importance to to scientiﬁcally scientifically evaluate evaluate the the quality quality of farmland of farmland toward to- improvingward improving food-production food-production capacity capacity and security. and security. The natural The natural environment environment and socioeco- and so- nomiccioeconomic conditions conditions of Xiangzhou of Xiangzhou are representative are representative of the assessment of the assessment of farmland of farmland quality and,quality thus, and, this thus, area this was area selected was selected as the study as the area. study area.

FigureFigure 1.1. Study area:area: (a)) locationlocation inin HubeiHubei Province,Province, China;China; ((bb)) digitaldigital elevationelevation mapmap (DEM)(DEM) ofof Xiangzhou.Xiangzhou.

2.1.1. Data Collection Considering the reliabilityreliability andand accessibilityaccessibility ofof data,data, thethe multisourcemultisource environmentalenvironmental data involved in this study mainly included map data, soil sampling data, and socioeconomic statistics. The Land Use Change Survey Database of Xiangzhou in 2018 (1:10,000)(1:10,000) waswas usedused to obtain administrative maps, farmlandfarmland patches, rural settlements, rural roads, highways, ditches,ditches, andand rivers.rivers. The Survey and Assessment ofof the Farmland Quality GradeGrade of XiangzhouXiangzhou inin 20182018 (1:10,000)(1:10,000) (containing(containing 255255 soilsoil samplingsampling points)points) waswas usedused toto extractextract thethe soilsoil fertilityfertility index.index. TheThe FarmlandFarmland SoilSoil EnvironmentalEnvironmental QualityQuality CategoryCategory ClassiﬁcationClassification Database ofof Xiangzhou Xiangzhou in in 2018 2018 (1:10,000) (1:10,000) (containing (containing 326 soil 326 heavy-metal soil heavy-metal sampling sampling points) waspoints) used was to obtainused to the obtain soil cleanlinessthe soil cleanlin index.ess The index. slope The was slope extracted was extracted using a 30using m reso- a 30 lutionm resolution digital digital elevation elevation model model(DEM ()DEMfrom) from the Resource the Resource and Environmentand Environment Science Science and Dataand Data Center Center (http://www.resdc.cn (http://www.resdc.cn).). The The 2018 2018 Statistical Statistical Yearbook Yearbook and and related related agricultural agricul- statisticstural statistics in Xiangzhou in Xiangzhou were were used toused extract to ex socioeconomictract socioeconomic and other and other statistical statistical data. data.

2.1.2. Data Processing (1) Soil index data processing Following the principles of comprehensiveness, representativeness, objectivity, equi- librium, comparability, and accessibility [5], 255 soil sampling points (Figure2a) were selected at the center of representative plots in the study area and sampled before soil fer- tilization following the autumn harvest of 2018 crops, using a handheld GPS for coordinate Agriculture 2021, 11, 72 5 of 23

positioning, to collect the surface layer of 0–20 cm of cultivated soil. Each soil sample was mixed in a star or S shape according to the size of the plot, and 1 kg was collected in a bag for analysis. More than 30 items of information related to the sampling point and its surrounding environment were recorded, such as topographic site, surface soil texture, texture conﬁguration, tillage-layer thickness, barrier factors, biodiversity, reticulation of agricultural land and forestry, and other soil properties, among which tillage-layer thickness was obtained by ﬁeld measurement, and biodiversity was obtained by calculating the Agriculture 2021, 11, x FOR PEER REVIEWpercentage of earthworms at the sample point. 6 of 23

FigureFigure 2.2. SamplingSampling distribution:distribution: (aa)) samplessamples forfor selectionselection ofof thethe soilsoil fertilityfertility indicator;indicator; ((bb)) samplessamples forfor selectionselection ofof thethe soilsoil pollutionpollution indicator.indicator.

2.2. MethodsSoil characteristics such as pH, moisture, available phosphorus, available potassium, and organic matter were obtained by soil sample assay analysis. In addition, there were 2.2.1. Building a Comprehensive Index System for Farmland Quality Assessment 326 heavy-metal sampling points (Figure2b) in the farmland of the study area, and the contents(1) Preliminary of five soil Selection polluting of elements—Pb,Indicators Cd, Cr, Hg, and As—were obtained by sample assayFarmland analysis. Onis a thiscomplex basis, system the Kriging formed interpolation by the interaction value with of various less error factors, was selected such as byclimate, cross-validation topography, comparison physical and of inversechemical distance properties weight of the and soil, values and from human Kriging, activities and polynomial[3], and its quality methods. is influenced Kriging optimal by multiple interpolation factors.was Therefore, performed in the on selection the selected of indica- index datators, inthe this influence study according of each factor to the on optimal farmland semi-covariance quality should function be considered model [45 ].comprehen- Regarding thesively, accuracy and the of dominant test results factor after interpolationshould be sele ofcted soil as pH, the the assessment mean absolute indicator. error In (MAE) terms wasof scale 0.056, effect, the rootthe selection mean square of assessment error (RMSE) indi wascators 0.078, should and be the compatible agreement with coefficient the assess- (AC) wasment 0.913, spatial indicating scale. In thatterms the of interpolation demand function, effect wasthe intrinsic ideal. The attributes remaining reflected indicators by farm- were validatedland quality in ashould similar serve way, not and only spatial as the distribution economic maps value of function each indicator of agricultural were generated produc- aftertion, testbut correction.also as the ecological service function, as an important part of a terrestrial eco- (2)system;Socioeconomic for example, index controlling data processing soil nutrient loss, regulating carbon storage capacity, storing and regulating water balance capacity, shaping landscape patterns, and maintaining ecologicalUsing ArcGIS balance 10.2 [46]. software In addition, developed the fundamental by the Environmental purpose of Systems the assessment Research is In- to stituteprovide in a RedLands, scientific basis California, for carrying USA, out we quality obtained regulation socioeconomic based on indicators an accurate of farmland grasp of quality, such as farming distance, ease of farming, traffic accessibility, drainage capacity, farmland quality and obstacle factors, so the selection of assessment indicators also needs and irrigation capacity, using neighborhood analysis tools. Farming distance was obtained to identify obstacles to better serve farmland quality regulation. by calculating the distance from the farmland patch to the rural settlement; ease of farming Based on the above theoretical analysis, the framework of a farmland quality assess- was obtained by calculating the distance from the farmland patch to the rural road; traf- ment index system was established based on element–function–regulation in this study fic accessibility was obtained by calculating the distance from the farmland patch to the (Figure 3), which shows the connotation of farmland quality at different levels. In this highway; drainage capacity was obtained by calculating the distance from the farmland framework, the term “element” refers to influencing factors, reflecting the influence of the patch to the ditch; and irrigation capacity was obtained by calculating the distance from coupling of factors on farmland quality at a certain spatial scale. Soil conditions are the the farmland patch to the river. basis of high agricultural yield, and are the most important factors affecting farmland Based on the Land Use Change Survey database of Xiangzhou in 2018, the arable quality, giving priority to soil pH, available phosphorus, available potassium, organic land patches were extracted as assessment units, and there were 16,072 found. Meanwhile, matter, and other indicators reflecting the fertility [47]. The term “function” refers to the ArcGIS 10.2 was used to realize the projection transformation and vectorization of each indicator.functional role of farmland quality in agricultural production, ecological landscape, and environmental protection, reflecting the intrinsic needs of human exploitation and ecological processes, with an emphasis on indicators reflecting the sustainable use of farmland, such as biodiversity and soil environmental quality [48,49]. The term “regulation” refers to the implementation of appropriate regulation measures to enhance farmland quality. Topography, field drainage and irrigation facilities, and transportation locations are obstacle factors that limit the improvement of farmland quality [50] and need to be regulated through soil management, so these factors should also be considered in the assessment of farmland quality. In this study, a preliminary assessment indicator system of farmland quality consisting of 19 indicators was constructed from four aspects—terrain, soil conditions, socioeco- Agriculture 2021, 11, 72 6 of 23

2.2. Methods 2.2.1. Building a Comprehensive Index System for Farmland Quality Assessment (1) Preliminary Selection of Indicators Farmland is a complex system formed by the interaction of various factors, such as climate, topography, physical and chemical properties of the soil, and human activities [3], and its quality is influenced by multiple factors. Therefore, in the selection of indicators, the influence of each factor on farmland quality should be considered comprehensively, and the dominant factor should be selected as the assessment indicator. In terms of scale effect, the selection of assessment indicators should be compatible with the assessment spatial scale. In terms of demand function, the intrinsic attributes reflected by farmland quality should serve not only as the economic value function of agricultural production, but also as the ecological service function, as an important part of a terrestrial ecosystem; for example, controlling soil nutrient loss, regulating carbon storage capacity, storing and regulating water balance capacity, shaping landscape patterns, and maintaining ecological balance [46]. In addition, the fundamental purpose of the assessment is to provide a scientific basis for carrying out quality regulation based on an accurate grasp of farmland quality and obstacle factors, so the selection of assessment indicators also needs to identify obstacles to better serve farmland quality regulation. Based on the above theoretical analysis, the framework of a farmland quality assessment index system was established based on element–function–regulation in this study (Figure3), which shows the connotation of farmland quality at different levels. In this framework, the term “element” refers to influencing factors, reflecting the influence of the coupling of factors on farmland quality at a certain spatial scale. Soil conditions are the basis of high agricultural yield, and are the most important factors affecting farmland quality, giving priority to soil pH, available phosphorus, available potassium, organic matter, and other indicators reflecting the fertility [47]. The term “function” refers to the functional role of farmland quality in agricultural production, ecological landscape, and environmental protection, reflecting the intrinsic needs of human exploitation and ecological processes, with an emphasis on indicators reflecting the sustainable use of farmland, such as biodiversity and soil environmental quality [48,49]. The term “regulation” refers Agriculture 2021, 11, x FOR PEER REVIEW 7 of 23 to the implementation of appropriate regulation measures to enhance farmland quality. Topography,nomics, and fieldecological drainage environment and irrigation (Table facilities, 1)—according and transportation to the abovementioned locations are obsta- ele- clement–function–regulation factors that limit the improvement model of farmland of farmland quality quality assessment [50] and based need on tonational be regulated stand- throughards [51–53] soil management,and the regional so thesecharacteristics factors should of Xiangzhou, also be considered while following in the assessmentthe principles of farmlandof comprehensiveness, quality. dominance, and difference.

FigureFigure 3.3. Framework ofof thethe farmlandfarmland qualityquality assessmentassessment indexindex system.system.

(2) Correlation Analysis Minitab 18 was used to analyze the correlation of primary indicators of the same dimension, with only one indicator being significantly correlated and reserved (Table 1). Finally, 14 indicators were selected from 19 indicators for the farmland quality assessment indicator system of Xiangzhou applicable to EW, BPNN, and RF simultaneously. (3) Validity Test Because the system of indicators constructed above (Table 1) was comprehensive, it was not possible to determine whether other factors would affect farmland quality. Thus, the validity of the assessment indicator system needed to be tested by calculating the residual coefficient, as follows in Equations (1) and (2):

n − 2 ()YYij 2 =−i=1 R 1 n (1) − 2 ()YYim i=1

eR=−1 2 (2)

where R2 is the decision coefficient, e is the residual coefficient, n is the number of samples, Yi is the actual value for each indicator, Yj is the predictive value for each indicator, and Ym is the mean of the actual value for each indicator. The residual coefficient (e) was calculated to be 0.2030, which is a sufficiently small value, thus indicating that other factors have a negligible effect on farmland quality. Therefore, the assessment index system constructed in this study (Table 1) is highly credible and representative.

Table 1. System of indicators for evaluating the quality of farmland.

Normative Layer Indicator Layer Basis of Calculation Explanation Based on digital elevation model (DEM) Slope Reservation Terrain calculations Topographic site (TS) Based on sample point assay data from Reservation Surface soil texture (SST) survey and assessment of farmland Reservation Soil conditions Texture configuration (TC) quality grade of Xiangzhou in 2018 Deletion Agriculture 2021, 11, 72 7 of 23

In this study, a preliminary assessment indicator system of farmland quality consisting of 19 indicators was constructed from four aspects—terrain, soil conditions, socioeconomics, and ecological environment (Table1)—according to the abovementioned element–function– regulation model of farmland quality assessment based on national standards [51–53] and the regional characteristics of Xiangzhou, while following the principles of comprehensiveness, dominance, and difference.

Table 1. System of indicators for evaluating the quality of farmland.

Normative Layer Indicator Layer Basis of Calculation Explanation Based on digital Slope elevation model Reservation Terrain (DEM) calculations Topographic site (TS) Reservation Surface soil texture Based on sample Reservation (SST) point assay data from Texture configuration survey and Deletion (TC) assessment of Tillage layer thickness Soil conditions farmland quality Reservation (TLT) grade of Xiangzhou Barrier factor (BF) Deletion in 2018 Soil pH Reservation Soil moisture (SM) Deletion Soil available Reservation phosphorus (SAP) Soil available Reservation potassium (SAK) Soil organic matter Reservation (SOM) Drainage capacity Distance from Reservation (DC) farmland to ditches Irrigation capacity Distance from Socioeconomic Reservation (IC) farmland to rivers Distance from Farming distance farmland to rural Deletion (FD) residential locations Distance from Ease of farming (EF) farmland to rural Reservation roads Traffic accessibility Distance from Reservation (TA) farmland to highway The percentage of Biodiversity Reservation Ecological earthworms environment Based on farmland soil environmental quality category Cleanliness Reservation classification database of Xiangzhou in 2018 Based on survey and Reticulation of assessment database agricultural land of farmland quality Deletion and forestry (ALRF) grade of Xiangzhou in 2018

(2) Correlation Analysis Minitab 18 was used to analyze the correlation of primary indicators of the same dimension, with only one indicator being signiﬁcantly correlated and reserved (Table1). Finally, 14 indicators were selected from 19 indicators for the farmland quality assessment indicator system of Xiangzhou applicable to EW, BPNN, and RF simultaneously. Agriculture 2021, 11, 72 8 of 23

(3) Validity Test Because the system of indicators constructed above (Table1) was comprehensive, it was not possible to determine whether other factors would affect farmland quality. Thus, the validity of the assessment indicator system needed to be tested by calculating the residual coefﬁcient, as follows in Equations (1) and (2):

n 2 ∑ (Yi − Yj) 2 i=1 R = 1 − n (1) 2 ∑ (Yi − Ym) i=1 p e = 1 − R2 (2) where R2 is the decision coefficient, e is the residual coefficient, n is the number of samples, Yi is the actual value for each indicator, Yj is the predictive value for each indicator, and Ym is the mean of the actual value for each indicator. The residual coefficient (e) was calculated to be 0.2030, which is a sufficiently small value, thus indicating that other factors have a negligible effect on farmland quality. Therefore, the assessment index system constructed in this study (Table1) is highly credible and representative.

2.2.2. Determining the Classification of Indicators and Their Affiliation To eliminate the influence of the evaluation indicator scale, the indicators were graded according to their influence on farmland quality, and their affiliations were determined based on the grade of indicator for farmland quality; the larger the affiliation, the better the quality, and vice versa. In this study, of the 14 determined assessment indicators (Table1), slope, topographic site, surface soil texture, drainage capacity, irrigation capacity, biodiversity, and cleanliness were qualitative indicators, and the remaining seven indicators were quantitative indicators. The affiliation of the seven qualitative indicators was determined based on the Farmland Quality Grade [51] and National Arable Land Quality Grade Assessment Indicator System [52] developed by the Ministry of Agriculture and Rural Development in China (Table2).

Table 2. Classiﬁcation of qualitative indicators and their afﬁliation.

Indicator Classiﬁcation and Afﬁliation Name <5◦ 5–15◦ 15–25◦ >25◦ Slope 1 0.8 0.6 0.4 Under Upper Middle Plain Broad Plain Plain moun- Middle Under Mountain mountain mountain Upper hill high valley middle low Topographic tain hill hill basin slope slope order basin order order site slope 0.3 0.45 0.6 0.68 0.7 0.8 0.8 0.9 0.95 0.95 1 Medium Heavy Sand Light soil Sand soil Clay Surface soil soil soil soil texture 1 0.9 0.85 0.95 0.6 0.6 Full satis- Basic satis- No satis- Drainage Satisfaction faction faction faction capacity 1 0.8 0.6 0.3 Full satis- Basic satis- No satis- Irrigation Satisfaction faction faction faction capacity 1 0.8 0.6 0.3 No enrich- Enrichment General Biodiversity ment 1 0.8 0.6 No Cleaning Cleanliness cleaning 1 0.8 Agriculture 2021, 11, 72 9 of 23

In the Farmland Quality Grade [51], 16 well-known, experienced experts in relevant fields from the National Agricultural Technology Extension Service Center, Beijing Soil and Fertilizer Workstation, Shandong Soil and Fertilizer General Station, Jiangsu Farmland Quality and Agricultural Environmental Protection Station, Shanxi Soil and Fertilizer Workstation, and South China Agricultural University scored the values in the qualitative indicator affiliation table. The quantitative indicators were based on the positive and negative effects of the indicator on farmland quality, and a suitable affiliation function was selected. According to the characteristics of the effect of the assessment indicator and farmland quality, the affiliation function can be divided into five types: upper precept, lower precept, peak, positive linear, and negative linear [54]. Among them, the larger the values of the four indicators of soil organic matter, tillage layer thickness, soil available phosphorus, and soil available potassium, the more beneficial to the improvement of farmland quality, so the upper precept affiliation function was chosen. The effect of soil pH on farmland quality presented a peak effect, so the peak affiliation function was chosen. The influence of ease of farming and traffic accessibility on farmland quality exhibited distance decay, so the negative linear affiliation function was chosen. With reference to the Farmland Quality Grade [51], the parameter of the affiliation function of each evaluation indicator was determined based on the value range of each evaluation indicator in Xiangzhou and, thus, the affiliation function of the evaluation indicator was established (Table3).

Table 3. Afﬁliation function of the qualitative indicators.

Indicator Function Function Lower Limit Upper Limit a b c Name Type Formula of u of u Soil organic Upper y = 0.001842 33.656446 0 33.7 matter precept 1/(1 + a(u − c)2) Tillage layer Upper y = 0.000205 99.092342 10 99 thickness precept 1/(1 + a(u − c)2) y = Soil pH Peak 0.221129 6.811204 3.0 10.0 1/(1 + a(u − c)2) Soil available Upper y = 0.002025 33.346824 0 33.3 phosphorus precept 1/(1 + a(u − c)2) Soil available Upper y = 0.000081 181.622535 0 182 potassium precept 1/(1 + a(u − c)2) Ease of Negative y = b − au 0.000376 0.90188 5 1600 farming linear Traffic Negative y = b − au 0.000318 0.90636 15 2400 accessibility linear Note: y is the degree of affiliation, a is the coefficient, b is the intercept, c is the standard indicator, and u is the measured value. When the function is an upper precept type, u is less than or equal to the lower limit value, y = 0, and u is greater than or equal to the upper limit value, y = 1. When the function is a peak type, u is less than or equal to the lower limit value, y = 0, and u is greater than or equal to the upper limit value, y = 1. When the function is a negative linear type, u is less than or equal to the lower limit value or greater than or equal to the upper limit value, y = 0.

2.2.3. Construction of Farmland Quality Assessment Model RF Assessment Model The key to constructing an RF model for farmland quality assessment is to analyze the correspondence rules between assessment indicators and quality grades of farmland, which is achieved through decision trees [35]. A binary decision tree consists of a root node, a child node, and a leaf node. The root node represents the observation value of indicators. The path from the root node to the leaf node corresponds to the assessment rule. The leaf node corresponds to the assessment results. The fundamentals of RF model construction for farmland quality assessment are shown in Figure4. First, different sets of training samples are randomly selected using bootstrap to be input into each decision tree to form different classiﬁers. The attribute value of each indicator that has a nonlinear relationship Agriculture 2021, 11, 72 10 of 23

Agriculture 2021, 11, x FOR PEER REVIEW 10 of 23 with the assessment object is decomposed into leaves with a linear relationship using the randomusing the split random node split technique. node technique. Then, the Then, weight the of weight each indicator of each indicator is calculated is calculated through thethrough analysis the ofanalysis the leaf of node the leaf structure. node structur The sete. of The relationships set of relationships between thebetween indicator the indi- and thecator weight and the according weight toaccording the linear to rules the correspondinglinear rules corresponding to the decision to treethe decision is formed, tree and is theformed, RF model and the is outputted.RF model is The outputted. average The of the aver sumage ofof thethe indexsum of values the index at each values leaf at node each multipliedleaf node multiplied by their weights by their is weights the ﬁnal is assessment the final assessment result. The result. RF model The isRF constructed model is con- as follows.structed as follows.

FigureFigure 4.4. Fundamentals ofof randomrandom forestforest (RF)(RF) modelmodel constructionconstruction forfor farmlandfarmland quality.quality.

(1)(1) Training set generation BasedBased onon TableTable1 1 and and the the results results of of the the national national farmland farmland quality quality grades, grades, the the original original samplesample setset waswas generatedgenerated fromfrom randomlyrandomly selectedselected farmlandfarmland patchespatches ofof differentdifferent qualityquality gradesgrades inin Xiangzhou,Xiangzhou, andand adjustedadjusted accordingaccording toto thethe spatialspatial locationlocation ofof thethe samplessamples toto achieveachieve uniform distribution, distribution, with with a atotal total of of 1590 1590 samples samples selected. selected. Then, Then, 70% 70% of the of data the datawas randomly was randomly selected selected as a training as a training sample sample (1110), (1110), 20% as 20% a test as sample a test sample (320), and (320), 10% and as 10%a validation as a validation sample sample(160), as (160), shown as in shown Figure in 5. Figure First, 51590. First, training 1590 trainingsets equal sets to equalthe num- to theber numberof samples of sampleswere chosen were randomly chosen randomly with playback with playback from the from training the trainingsample. sample.Second, Second,the indicator the indicator attribute attribute values were values sampled were sampled and learned and learnedmultiple multiple times until times the until average the averageerror rate error was rate stabilized was stabilized using R usingto access R to th accesse random the random forest library. forest library. Then, based Then, on based the onmapping the mapping of quality of quality index indexto attribute to attribute values values for each for indicator, each indicator, a decision a decision tree was tree wascon- constructedstructed to output to output the therules rules for predicting for predicting the quality the quality index index of farmland. of farmland. The above The above steps stepswere repeated were repeated to construct to construct different different decision decision trees for trees training for trainingthe RF model the RF for model farmland for farmlandquality assessment, quality assessment, which were which applied were to applied the assessment to the assessment of Xiangzhou of Xiangzhou in 2018. inThe 2018. re- 2 Thesults results showed showed that the that coefficient the coefﬁcient of determination of determination (R2) (Rfor )the for training the training sample sample was was in the in therange range of 0.7727–0.9483 of 0.7727–0.9483 (Table (Table 4)4 and) and was was significant signiﬁcant at at the the 5% level, withwith highhigh predictionprediction accuracy.accuracy. Agriculture 2021, 11, 72 11 of 23

Agriculture 2021, 11, x FOR PEER REVIEW 11 of 23

Figure 5. Spatial distribution of 1590 farmland quality samples.

2 Table 4. Corresponding errors forfor differentdifferent valuesvalues ofof nn andand theirtheir RR2..

2 Number ( n()n) ErrorError R R2 11 0.0002110.000211 0.77270.7727 22 0.00008300.0000830 0.910.91 3 0.0000620 0.9332 43 0.00005480.0000620 0.94090.9332 54 0.00005280.0000548 0.94320.9409 65 0.00004990.0000528 0.94620.9432 7 0.0000480 0.9483 86 0.00004950.0000499 0.94660.9462 97 0.00004940.0000480 0.94680.9483 108 0.00005020.0000495 0.94590.9466 11 0.0000493 0.9469 9 0.0000494 0.9468 12 0.0000482 0.948 1310 0.00004970.0000502 0.94640.9459 1411 0.00005070.0000493 0.94540.9469 12 0.0000482 0.948 (2) Parameter13 optimization 0.0000497 0.9464 The random14 forest package in R was used0.0000507 to apply out-of-bag (OOB) data0.9454 for unbiased estimation of the accuracy of the RF model in this study. After debugging and parameter (2)sensitivity Parameter analysis, optimization the OOB error tended to stabilize when the number of decision trees wasThe greater random than forest 200 (Figure package6). in The R was overall used error to apply of the out-of-bag model and (OOB) the coefﬁcientdata for unbi- of 2 aseddetermination estimation (R of )the were accuracy compared of the by RF adjusting model in the this number study. ofAfter assessment debugging indicators and pa- 2 rameter(Table4) tosensitivity determine analysis, the optimal the OOB segmentation error tend nodeed to ( nstabilize). The error when was the minimized number of and deci- R sionwas maximizedtrees was greater when thethan number 200 (Figure of node 6). segmentsThe overall (n )error was 7.of the model and the coefficient Forof determination BPNN, the neural-net (R2) were compared package in by R adjusting was used the to number train neural of assessment networks indica- using torsbackpropagation. (Table 4) to determine In the main the parameteroptimal segmentation settings, the node hidden (n). layer The waserror 5, was the minimized number of andrandom R2 was seeds maximized was 2000, when the initial the number weight of range node was segments 0–1, the (n weight) was 7. adjustment rate was 0.1, theFor minimum BPNN, the target neural-net error waspackage 0.01, in the R learningwas used rate to train was neural 0.05, the networks maximum using number back- of iterations was 1000, the number of training repetitions was 1, the impulse coefﬁcient propagation. In the main parameter settings, the hidden layer was 5, the number of ran- was 1.2, the error type was sse, the activation function was tanh, and the output form was dom seeds was 2000, the initial weight range was 0–1, the weight adjustment rate was 0.1, linear. the minimum target error was 0.01, the learning rate was 0.05, the maximum number of iterations was 1000, the number of training repetitions was 1, the impulse coefficient was Agriculture 2021, 11, x FOR PEER REVIEW 12 of 23

Agriculture 2021, 11, 72 1.2, the error type was sse, the activation function was tanh, and the output form12 ofwas 23 linear.

Figure 6. RF decision tree for the out-of-bag (OOB) error. Figure 6. RF decision tree for the out-of-bag (OOB) error. (3) Determination of weighted indicators (3) Determination of weighted indicators To avoid overﬁtting assessment indicators, the importance of evaluating indicators was determinedTo avoid overfitting in the construction assessment of theindicators, RF model the by importance calculating of the evaluating reduced value indicators of the wasGini determined index at node in splittingthe construction based on of out-of-bag the RF model (OOB) by data calculating not involved the reduced in constructing value of the Gini decision index trees, at node which splitting is implemented based on out- as follows.of-bag (OOB) data not involved in constructing theThe decision Gini index trees, expresses which is the implemented impurity of as a node,follows. which is deﬁned as the following in EquationThe Gini (3): index expresses the impurity of a node, which is defined as the following in Equation (3): j 2 Gini(n) = 1 − ∑[p(i|n)] (3) i=j1 2 Gini() n=− 1 p ( i| n) where p(i|n) is the probability of farmland quality.  When Gini(n) is 0, it means that the(3) i =1 training data at n nodes are of the same quality grade of farmland. A larger Gini(n) means that the training data at n nodes are more scattered, and the nodes need to be further where pi(|) n is the probability of farmland quality. When Gini(n) is 0, it means that divided. the training data at n nodes are of the same quality grade of farmland. A larger Gini(n) The original split Gini index of the kth decision tree is assumed to be Gini gk. The means that the training datai at n nodes are more scattered, and the nodes need to be fur- new Gini index (Gini gk ) is recalculated by performing random series changes on the therith assessment divided. indicator (Z) of the OOB data. Then the weight of indicator Z in the The original split Gini index of the kth decision tree is assumed to bei Gini gk. The new corresponding single decision tree (WGi) is denoted as Gini gk − Gini gk . The weight of Gini index (Gini gki) is recalculated by performing random series changes on the ith assess- the ith assessment indicator (RFWi) is the average of all decision trees in the RF, which is mentcalculated indicator as follows (Z) of inthe Equations OOB data. (4) Then and (5):the weight of indicator Z in the corresponding single decision tree (WGi) is denoted as Gini gk − Gini gki. The weight of the ith assessment

indicator (RFWi) is the average of all decision trees in the RF, which is calculated as follows m WGi = /N (4) in Equations (4) and (5): (Ginigk − Gini i ) ∑ gk i=1

m m WG RFW()i Gini= i/− GiniWGi (5)  gk ∑ g i = k WG = i =1 i 1 i N (4) In the formulas, WGi is the reduced value of the Gini index, m is the number of assessment indicators, and RFWi is the weight of the ith assessment indicator. (4) Calculation of farmland quality index The best RF model between each assessment indicator of the independent variable and the farmland quality of the dependent variable according to the linear decomposition rules of the decision tree mapping was constructed, and the weight of each assessment indicator was calculated. The test set was then entered into each RF decision tree. The sum Agriculture 2021, 11, 72 13 of 23

of the index values at each leaf node multiplied by their weights is the farmland quality index for that tree. Finally, the farmland quality index was obtained by calculating the average value of the index for each decision tree.

Comparison Methods Current methods of assessing farmland quality can be broadly divided into two categories: conventional weighting and machine learning. Therefore, to more accurately verify the reliability and superiority of the machine learning (RF model) in this study, we compared the performance of EW (represented as the conventional weighting method) and BPNN (represented as the machine learning method). (1) Compared to other conventional weight assessment methods, EW is an objective weight method that is free from human influence and has the most basic and simple model construction, which enables better interpretation of assessment results. There- fore, EW was chosen as a typical representative of conventional weight-assessment methods. The fundamental principle of EW is to calculate objective weight based on the amount of variability in the indicator. The less information entropy there is, the greater the variation and weight of the indicator. Due to the wide application of this method, it was not repeated in this paper concerning the specific calculation procedure [19]. (2) BPNN and SVM are classical representatives of machine-learning methods, both of which are more stable and flexible than conventional weight-assessment methods. However, compared to BPNN, the choice of SVM parameters poses greater constraints to construction, which largely limits the depth of research and breadth of SVM application. Artificial neural networks tend to mature in farmland quality assessment studies, mainly using BPNN or its deformed form model. Therefore, in this study, BPNN was chosen as a typical representative of machine-learning methods, and we attempted to construct a five-layer BPNN structure for farmland quality assessment model using R.

3. Results 3.1. Analysis of Indicator Weights Determined Based on Three Models The RF weight (RFW) for each assessment indicator was calculated based on the reduction in the Gini index produced by RF during the training process. The drainage capacity, irrigation capacity, soil available phosphorus, topographic site, and soil organic matter were identified as the five most important indicators affecting farmland quality in RFW, with a total weight of 82.41% (Figure7). Slope, traffic accessibility, and ease of farming were the three indicators that had the least impact on farmland quality, with a weight of only 1.35%. Meanwhile, the weight of each indicator in BPNN (BPNNW) and EW (EWW) were calculated. The BPNNW identification was more balanced, with drainage capacity and irrigation capacity being more important, and slope and cleanliness the least important indicators. EWW identified soil available phosphorus as the most important indicator, and ease of farming as the least important. Thus, the difference between RFW and BPNNW was small in terms of indicator weight values, while the difference between EWW and BPNNW was significant. Adequate drainage and irrigation capacity are important for farmland productivity, soil available phosphorus and soil organic matter reflect the fertility of farmland, and topography is an important factor limiting farmland productivity in hilly areas. Therefore, they all have an important impact on farmland quality. Regularity among slope, traffic accessibility, ease of farming, and the remaining indicators is not sufficiently evident. Thus, they were identified by RF as the least-important impact factors. From the basic principle of the method, EWW calculates the indicator weight based on the objective trend of the indicator attribute values, while RFW and BPNNW reflect the intrinsic connection between the quality of farmland and each indicator and calculate the indicator weight based on the reduced value of the Gini coefficient, which is more in line with the complexity and nonlinearity of the farmland quality system. Therefore, RFW Agriculture 2021, 11, x FOR PEER REVIEW 14 of 23

and BPNNW was small in terms of indicator weight values, while the difference between EWW and BPNNW was significant. Adequate drainage and irrigation capacity are important for farmland productivity, soil available phosphorus and soil organic matter reflect the fertility of farmland, and topography is an important factor limiting farmland productivity in hilly areas. Therefore, they all have an important impact on farmland quality. Regularity among slope, traffic accessibility, ease of farming, and the remaining indicators is not sufficiently evident. Thus, they were identified by RF as the least-important impact factors. From the basic principle of the method, EWW calculates the indicator weight based on the objective trend of the indicator attribute values, while RFW and BPNNW reflect Agriculture 2021, 11, 72 the intrinsic connection between the quality of farmland and each indicator and calculate14 of 23 the indicator weight based on the reduced value of the Gini coefficient, which is more in line with the complexity and nonlinearity of the farmland quality system. Therefore, RFW andand BPNNW BPNNW should should have have more more explanatory explanatory power power for for the indicators. Meanwhile, from from thethe results results of of the the run run program, program, the the RF RF aver averageage error error was 0.0019, the average prediction accuracyaccuracy was was 93.01%, 93.01%, and and the the convergence convergence time time was was 4.43 4.43 s; s; while while the the BPNN BPNN average average error error waswas 0.0057, 0.0057, the the average average prediction prediction accuracy accuracy was was 88.19%, 88.19%, and and the the convergence time time was 5.875.87 s. s. It It can can be be seen seen that that compared compared to to BP BPNN,NN, RFW RFW was was more more interpretable, interpretable, more more accurate accurate inin predicting predicting indicators, indicators, and and faster faster at at proce processingssing data, data, and and its its weights weights were were more more accurate accurate inin fitting ﬁtting the the interrelationships interrelationships between between indicators indicators and and assessment assessment objects. objects.

Figure 7. Indicator weights for the three assessment methods.

3.2.Figure Analysis 7. Indicator and Comparison weights for of the Results three ofassessment Farmland methods. Quality Assessment According to RFW, BPW, and EWW, the cultivated quality index was calculated for 3.2.Xiangzhou Analysis inand 2018. Comparison Using the of Results natural of fracture Farmland method Quality of Assessment ArcGIS, farmland quality was classifiedAccording into five to grades,RFW, BPW, in descending and EWW, order the fromcultivated first to quality fifth. The index results was of calculated the farmland for Agriculture 2021, 11, x FOR PEER REVIEWXiangzhouquality assessment in 2018. basedUsing onthe RF, natural BPNN, fracture and EW method are shown of ArcGIS, in Figure farmland8 and Table quality5. 15 wasof 23

classified into five grades, in descending order from first to fifth. The results of the farmland quality assessment based on RF, BPNN, and EW are shown in Figure 8 and Table 5.

Table 5. Acreage and proportion of grades for the three methods of evaluating grades of farmland quality.

Model Area/Proportion First Grade Second Grade Third Grade Fourth Grade Fifth Grade Area/hm2 14,680.44 41,365.19 48,912.49 46,704.13 13,594.74 Random forest (RF) Proportion/% 8.88 25.03 29.60 28.26 8.23 Backpropagation neural Area/hm2 19,163.23 40,484.52 43,261.89 45,500.65 16,846.71 network (BPNN) Proportion/% 11.60 24.50 26.18 27.53 10.19 Area/hm2 27,558.40 42,137.51 43,976.13 32,252.10 19,332.85 Entropy weight (EW) Proportion/% 16.68 25.50 26.61 19.52 11.70

Figure 8. SpatialSpatial distribution distribution of of farmland farmland quality quality grades: grades: (a) (Randoma) Random forest forest (RF (RF)—based,)—based, (b) Backpropagation (b) Backpropagation neural neural net- worknetwork (BPNN)—based, (BPNN)—based, and and (c) Entropy (c) Entropy weight weight (EW)—based. (EW)—based.

3.2.1. Analysis of RF Assessment Results From an overall perspective, the average farmland quality index in Xiangzhou was 0.8275, which indicates higher quality. The proportion of second- and third-grade farmland was larger, accounting for 54.63%, and the proportions of first- and fifth-grade farmland were smaller, at 8.88% and 8.23%, respectively, which shows that the farmland quality grade was in a normal distribution (Figure 8a). In terms of the average farmland quality index, Zhangjiaji, Zhangwan, Shuanggou, Guyi, and Longwang had the highest values, at 0.8517, 0.8475, 0.8395, 0.8375, and 0.8356, respectively; while the average values in Yushan, Huopai, and Huanglong were relatively low, at 0.8175, 0.8149, and 0.8094, respectively. As far as different land types were concerned, the average index of farmland quality was 0.8286, 0.8282, and 0.8267 for paddy fields, irrigated land, and dry land, respectively, which shows that paddy fields had the best quality. In addition, the spatial distribution of farmland quality in Xiangzhou was uneven, and was greatly influenced by topography and socioeconomic development, showing an overall distribution pattern of high in the north-central part and low in the south (Figure 8a). Higher-quality farmland was mainly distributed in Zhangjiaji, Zhangwan, and Shuanggou in the central part and Longwang and Guyi in the north, while lower-quality farmland was mainly distributed in Yushan and Huanglong in the south. The central part of Xiangzhou belongs to the alluvial plain of the Han River, with flat terrain and fertile soil, mainly loam and sand. There are certain sources of contamination in the region due to economic development, which affects farmland productivity. However, there are more rivers, better field drainage and irrigation facilities, and richer biodiversity in this area. Therefore, in general, the fertile alluvial plain is suitable for the development of cultivated agriculture. The northern and southern parts of Xiangzhou have hillock landforms and low hilly areas with high terrain and steep slopes that are prone to soil erosion and have relatively poor soil. Although the farmland area is large, the level of land use is low due to fewer water sources and poor drainage and irrigation facilities resulting from extensive farming, which make this farmland suitable for subtropical cash crops.

Table 5. Acreage and proportion of grades for the three methods of evaluating grades of farmland quality.

3.2.1. Analysis of RF Assessment Results From an overall perspective, the average farmland quality index in Xiangzhou was 0.8275, which indicates higher quality. The proportion of second- and third-grade farmland was larger, accounting for 54.63%, and the proportions of first- and fifth-grade farmland were smaller, at 8.88% and 8.23%, respectively, which shows that the farmland quality grade was in a normal distribution (Figure8a). In terms of the average farmland quality index, Zhangjiaji, Zhangwan, Shuanggou, Guyi, and Longwang had the highest values, at 0.8517, 0.8475, 0.8395, 0.8375, and 0.8356, respectively; while the average values in Yushan, Huopai, and Huanglong were relatively low, at 0.8175, 0.8149, and 0.8094, respectively. As far as different land types were concerned, the average index of farmland quality was 0.8286, 0.8282, and 0.8267 for paddy fields, irrigated land, and dry land, respectively, which shows that paddy fields had the best quality. In addition, the spatial distribution of farmland quality in Xiangzhou was uneven, and was greatly influenced by topography and socioeconomic development, showing an overall distribution pattern of high in the north-central part and low in the south (Figure8a). Higher-quality farmland was mainly distributed in Zhangjiaji, Zhangwan, and Shuanggou in the central part and Longwang and Guyi in the north, while lower-quality farmland was mainly distributed in Yushan and Huanglong in the south. The central part of Xiangzhou belongs to the alluvial plain of the Han River, with flat terrain and fertile soil, mainly loam and sand. There are certain sources of contamination in the region due to economic development, which affects farmland productivity. However, there are more rivers, better field drainage and irrigation facilities, and richer biodiversity in this area. Therefore, in general, the fertile alluvial plain is suitable for the development of cultivated agriculture. The northern and southern parts of Xiangzhou have hillock landforms and low hilly areas with high terrain and steep slopes that are prone to soil erosion and have relatively poor soil. Although the farmland area is large, the level of land use is low due to fewer water sources and poor drainage and irrigation facilities resulting from extensive farming, which make this farmland suitable for subtropical cash crops.

3.2.2. Comparison of Assessment Results Overall, as far as the average farmland quality index for Xiangzhou was concerned, RF > BPNN > EW, with values of 0.8275, 0.8271, and 0.7532, respectively. Thus, the farmland quality in Xiangzhou was higher. For the average farmland quality index of each town, the assessment results of RF and BPNN were relatively similar, but there were also significant relative differences in terms of local regions. As can be seen in Table5, the percentage of area of each measured farmland quality grade based on RF was closer to that of BPNN, but differed significantly from that of EW; the most significant difference was for first grade, which was 8.88% for RF and 16.68% for EW. In Figure8a, it can be seen that first-grade farmland based on RF was mainly distributed in Zhangwan, Guyi, and Longwang, while that based on EW was mainly distributed in Shuanggou, Chenghe, Guyi, Longwang, and Shiqiao (Figure8c). Shuanggou and Chenghe are industrial towns in Xiangzhou with high urbanization and high intensity of farmland use, and existing studies have shown that the heavy-metal element content of farmland soil around the towns is high overall, so it is unreasonable for the farmland to be identified as first-grade land [44,55]. The topography of Agriculture 2021, 11, 72 16 of 23

Shiqiao is high and the soil water and fertilizer retention ability is poor, so theoretically, the quality grade of farmland should be lower, which is consistent with the existing research results [56]. RF identified Shuanggou as having high-grade farmland quality (Figure8a), while BPNN identified it as low grade (Figure8b). Shuanggou has a flat terrain, fertile soil, numerous rivers, abundant water sources, and relatively complete irrigation and drainage facilities, and its natural geographical conditions are advantageous for the development of arable agriculture. Therefore, this town should, theoretically, be an area of high-quality farmland. This analysis is also empirically supported by numerous studies [57]. The results of the EW assessment differed greatly from the other two methods. Areas with higher quality farmland were mainly distributed in Zhangwan, Shuanggou, Chenghe, and Prison Farm, while areas with lower-quality farmland were mainly distributed in Huopai, Yushan, and Huanglong (Figure8c).

3.3. RF Model Reasonableness Test To further measure the rationality of the RF model for evaluating farmland quality, the consistency of the results of RF, BPNN, and EW was compared and analyzed in this study. The superiority of RF was revealed by four indices: mean absolute error (MAE), standard mean square error (MSE), coefﬁcient of determination (R2), and signiﬁcance.

3.3.1. Consistency Test The validity of the RF model was verified by analyzing the differences among the assessment results of RF, BPNN, and EW in the following process. First, the five grades of farmland quality were assigned values of 1 to 5, respectively. Second, using the raster calculator tool of ArcGIS, the assigned values of farmland quality grade evaluated by RF were multiplied by 10 and summed with those evaluated by BPNN and EW to obtain a two-digit result. If the two digits were the same, then RF was the same as BPNN and EW; otherwise, it was different. The spatial distribution of the differences between the results of farmland quality assessment for RF and BPNN and between RF and EW are shown in Figure9, where the first number of the difference marker is the RF grade and the second number the BPNN or EW grade. For example, a value of 13 means that RF identified the area as first-grade farmland and BPNN or EW identified it as third grade, while 31 means that RF identified the area as third-grade farmland and BPNN or EW identified it as first grade. In the comparison of results between RF and BPNN, the number of errors as a percentage of the total sample was 22.74%, mainly concentrated in one grade of difference—12, 21, 23, 32, 34, 43, 45, 54—whose ratio was as high as 99.36% (Figure9). From the results of the comparison between RF and EW, the proportion of error points among the total number of samples was significantly higher, reaching 48.24%, and the difference between the two assessments was mainly concentrated in one grade—12, 21, 23, 32, 34, 43, 45, 54—whose error points account for about 79.67%. Thus, the assessment results of RF and BPNN were in high agreement. In these errors, EW identified areas of higher quality that RF and BPNN identified as lower. Some of these areas were relatively high terrain, with poor soil fertility and mediocre drainage and irrigation facilities, where the farmland quality was largely constrained by topography and soil fertility and was, theoretically, relatively low. EW identified areas of lower farmland quality that RF and BPNN identified as higher. Some of these areas were flat, with relatively fertile soil and rich biodiversity and theoretically high-quality farmland. Obviously, the assessment results of RF and BPNN were more convincing compared to EW, so their superiority was further compared. Agriculture 2021, 11, 72 17 of 23

Agriculture 2021, 11, x FOR PEER REVIEW 17 of 23

Figure 9. Spatial distribution of farmland quality grade differences between (a)) RFRF andand BPNNBPNN andand ((b)) RFRF andand EW.EW.

3.3.2. Superiority Test To validate the superiority of RF, thethe testtest datasetsdatasets werewere selectedselected toto constructconstruct the RF and BPNN models for assessing farmland quality and to compare to the results of thethe assessment above. As shown in Table 6 6,, inin RF,RF, thethe MAEMAE waswas 0.0090.009 andand thethe MSEMSE waswas 0.012,0.012, both less than in BPNN. The coefficientcoefficient ofof determinationdetermination (R(R22) of RF waswas 0.8145,0.8145, greatergreater than BPNN. The F-test showed that RF passed the 5%5% levellevel ofof significance,significance, whilewhile BPNNBPNN was not not significant. significant. Thus, Thus, compared compared to toothe otherr assessment assessment methods, methods, RF has RF the has lowest the lowest gen- eralizationgeneralization error, error, highest highest accuracy, accuracy, better better stab stability,ility, and and better better assess assessmentment capability. capability. RF may be preferred over BPNN for farmland quality assessment.assessment.

Table 6.6. Error analysisanalysis ofof assessmentassessment results.results.

ModelModel MAE MAE MSE MSE R2 R2 SignificanceSigniﬁcance RFRF 0.009 0.009 0.012 0.0120.8145 0.8145 0.000 0.000 BPNNBPNN 0.021 0.021 0.026 0.0260.4594 0.4594 0.213 0.213

4. Discussion 4.1. Construction of a Comprehensive AssessmentAssessment Index System for Farmland Quality The construction of the index system is the focus focus of of quality-assessment quality-assessment research research [24]. [24]. How to reasonably reasonably determine determine an an assessment assessment in indexdex system system based based on on different different purposes purposes is theis the primary primary problem problem of farmland of farmland quality quality assessment. assessment. Kong Kong et al. et[17] al. established [17] established a pres- a pressure–state–effect–responsesure–state–effect–response assessment assessment system system based based on changes on changes in farmers’ in farmers’ land-use land-use objectives.objectives. Chen Chen et etal. al. [58] [58 constructed] constructed an an element–demand–regulation element–demand–regulation assessment assessment system for slope farmland based on thethe minimumminimum dataset.dataset. Due to the complexitycomplexity ofof farmlandfarmland quality systems, the topography, physical and chemical properties of the soil, land use, quality systems, the topography, physical and chemical properties of the soil, land use, economic level, and ecological environment should be fully considered in the selection economic level, and ecological environment should be fully considered in the selection of of indicators. However, soil factors and topography as indicators were selected only in indicators. However, soil factors and topography as indicators were selected only in the the traditional farmland quality assessment [47]. The ecological conditions of farmland traditional farmland quality assessment [47]. The ecological conditions of farmland have have not been adequately considered in current research [59]. Therefore, it is particularly not been adequately considered in current research [59]. Therefore, it is particularly im- important to establish an index system that considers both socioeconomic factors and portant to establish an index system that considers both socioeconomic factors and the the ecological environment. In this study, a comprehensive assessment index system ecological environment. In this study, a comprehensive assessment index system was was built to characterize the diversity of farmland ecosystems in terms of their natural, built to characterize the diversity of farmland ecosystems in terms of their natural, socio- socioeconomic, and ecological environment conditions. economic, and ecological environment conditions. Slope, topographic site, surface soil texture, soil organic matter, tillage-layer thickness, Slope, topographic site, surface soil texture, soil organic matter, tillage-layer thick- soil pH, soil available phosphorus, and soil available potassium were chosen to represent ness, soil pH, soil available phosphorus, and soil available potassium were chosen to represent natural conditions based on the natural properties of farmland in this study. Soil is Agriculture 2021, 11, 72 18 of 23

natural conditions based on the natural properties of farmland in this study. Soil is an important natural resource with a diversity of ecological functions, and is the basis for agricultural development [60]. Slope, topographic site, and surface soil texture are chosen to assess the suitability of farmland [61]. Tillage-layer thickness is used as an important indicator of land degradation [62]. Soil organic matter, pH, available phosphorus, and available potassium have been used to evaluate soil fertility quality [63]. Soil organic matter is the main source of crop nutrition, and is an important indicator of soil fertility [64,65]. Soil pH is one of the most important factors affecting the effectiveness of soil nutrients, which is closely related to fertilizer absorption efficiency, and most crops grow best when soil pH is 6.5–7.5 [66]. Soil available phosphorus and potassium play important roles in plant growth, and are the main fertilizers applied to plants in agricultural production [67,68]. In the 1980s, the Land Evaluation and Site Assessment (LESA) system established by the Soil Conservation Service (SCS) in Washington of USA emphasized the important role played by socioeconomic conditions in farmland quality assessment [69]. In this study, drainage capacity, irrigation capacity, ease of farming, and traffic accessibility were chosen as socioeconomic indicators. Water resource is an important factor influencing crop yields [70,71]. A proper drainage and irrigation pattern not only improves drainage capacity, but also maintains the soil moisture of the farmland [72]. Accessibility is regarded as having an influence on the spatial pattern of land use [73], which is calculated by the distance from patches to roads [74,75], and it was reflected in the ease of farming and road accessibility in this study. Over the past decade, China’s agricultural development has shifted from the pursuit of increased yields to a concern for the ecological safety of farmland [76]. Therefore, ecological factors need to be considered in the assessment of farmland quality [77,78]. In this study, biodiversity and cleanliness indicators were selected to measure ecological quality based on previous studies [79]. Biodiversity is an important indicator of farmland ecological values, and is usually composed of genetic, species, and ecosystem diversity [80]. In terms of the data acquisition of biodiversity indicators, due to the limitation of experimental conditions in the study area, it was not possible to analyze them quantitatively, and their levels were only analyzed qualitatively using the percentage of earthworms in the soil, and further in-depth study is needed. Cleanliness reflects the environmental condition of the soil, and it has been used as an important indicator of soil ecological safety [81]. In the acquisition of cleanliness index data, due to the lack of data, we failed to start from the essence of environmental quality and select pollution elements for analysis, and only used a few substitute indicators that were not comprehensive enough for a description of ecological environmental quality, and require further exploration and in-depth study.

4.2. The Influence of Research Scale on Indicator Selection Due to the large differences in the selection of assessment indicators and access to basic data, the focus and accuracy of the results of farmland quality assessment at different scales also differ greatly; i.e., the assessment has spatial scale effects [82]. Kong et al. [83] argued that farmland quality has significant scale variability, with some characteristics acting at the micro level and some at the macro level. The smaller the assessment scale, the more assessment indicators are available and the higher the accuracy of indicators, and thus the finer the degree of assessment, though the applicability of results is often narrower. As the assessment scale is enlarged, relatively fewer indicators are available, and their accuracy is reduced, and the results reflect the macroscopic characteristics of farmland quality, but the applicability of large-scale farmland quality assessment results is also wider [84]. From the selection of assessment indicators at different scales, the small-scale farmland quality assessment index system is based on soil attribute indicators, and its process focuses on assessing the natural quality of farmland, while at large scale, macro factors are selected from natural conditions, land use level, and economic level to build an assessment system [85]. The physicochemical, biological, and health characteristics of soil, as well as topographical conditions, are the main factors that influence the productive Agriculture 2021, 11, 72 19 of 23

potential of farmland. Realistic farmland quality is obtained by analyzing the impact of land use and economic levels on the productive potential. The combination of field surveys and remote sensing is an important tool for evaluating land quality. The county is the basic administrative unit for farmland management in China, and farmland resource management and conservation policies can be formulated based on the results of quality assessment [86]. In this study, the farmland quality assessment system was built from four dimensions of terrain, soil conditions, socioeconomics, and ecological environment in Xiangzhou. The micro-indicators of social production and ecology were selected. In reflecting the public’s demand for diversified farmland function, this research can provide a theoretical reference for national farmland assessment. It could also provide technical support for farmland management in Xiangzhou. More importantly, the method could be applied to similar national and regional cultivated quality assessments. However, as influenced by the scale effect of the assessment, there are problems such as the lack of comparability of different scales of farmland quality assessment index systems and results [83,84]. Research on the construction of a multiscale farmland quality assessment index system and the establishment of an assessment model and scale conversion need to be further developed.

4.3. Construction of Farmland Quality Assessment Model How to reasonably identify the weight of indicators is central to the construction of a farmland quality assessment model [20]. Lin et al. [42] used the random forest algorithm, with the average standard yield as the dependent variable and the influence factor as the independent variable, to construct a regression model to determine indicator weights and compare the accuracy of this determination with that of Delphi. To further verify the reliability and superiority of the RF model, typical samples of the study area were selected, and three deep machine-learning models were selected for training, namely entropy weight (EW), backpropagation neural network (BPNN), and random forest (RF), and the results of farmland quality assessment were output through simulation. Comparing training accuracy in terms of the BPNN and EW measurements of the weights of assessment indicators may not be in line with the actual situation: RF can more effectively decompose indicators that have a nonlinear relationship with the assessment object into leaves with linear relationships and determine the weights of the indicators in the operation process, which is associated with higher indicator interpretation ability, and the assessment results are more ideal and in line with objective reality. In addition, the selection of training samples is crucial; different samples have an impact on the final evaluation results, and this study only used the uniform method to select samples, so the next step can select samples in different ways, and an in-depth comparative analysis can be conducted. Due to the complexity of the factors influencing the quality of farmland, whether the assessment model is properly constructed during actual production will directly affect the accuracy of the comprehensive assessment of farmland quality. Furthermore, the indicator weights are automatically generated during the training process based on the rules of decomposition of the attribute values of real data, whereas the assessment indicators change with socioeconomic development. Therefore, the model constructed in this study is not applicable to the future assessment of farmland quality. Further improvement of the RF model to improve the accuracy of farmland quality assessment will be the focus and challenge of subsequent research, and will take into account the actual conditions of the study area and the natural and socioeconomic factors that affect farmland quality.

5. Conclusions In this study, an RF model for farmland quality assessment was constructed and applied to the assessment of Xiangzhou in Hubei Province in 2018, based on the construction of decision trees using bootstrap. BPNN and EW were used as a comparison to verify the advantages and disadvantages of the three models. The main ﬁndings were as follows: Agriculture 2021, 11, 72 20 of 23

(1) The results showed that as far as the average farmland quality index in Xiangzhou is concerned, RF > BPNN > EW, and the assessment results of RF and BPNN showed more similar spatial distribution, while that of EW differed greatly. From a practical point of view, the assessment result of RF was more in line with the local natural conditions and socioeconomic development and was more objective. In terms of assessment accuracy, the RF model had the advantage of digging deeper into the nonlinear relationship between the indicators and the evaluated object. Its generalization ability was stronger, its assessment accuracy was higher, and its assessment results were more in line with the spatial distribution of farmland quality, which were consistent, typical, and superior to those of BPNN and EW. (2) The quality of farmland in Xiangzhou was generally high, with a large area being of second- and third-grade quality, accounting for 54.63% of the total farmland area, and the grades basically conformed to a positive distribution trend. From the distribution point of view, the spatial distribution of farmland quality in Xiangzhou was unbalanced, inﬂuenced by the topography and socioeconomic development level and showing an obvious geographical distribution pattern, with overall characteristics of high in the north-central area and low in the southern area. The distribution of farmland quality grades also differed greatly among regions. (3) To a certain degree, due to the complexity, uncertainty, and nonlinearity of farmland quality systems, farmland quality from the perspective of the RF model was re- searched as an expansion of assessment methods, based on the theories and methods of artiﬁcial intelligence technology, which can improve the accuracy and quantitative level of farmland quality assessment. This study provides a new assessment method for farmland quality that can support the formulation of rational and effective management policies toward realizing the sustainable use of farmland resources. Moreover, the assessment model constructed in this study could be used as a reference for similar countries and regions.

Author Contributions: Funding acquisition, Y.Z.; conceptualization, L.W., Y.Z.; methodology, L.W.; investigation, T.X., Z.W.; writing—original daft, L.W., Y.Z., Q.L.; writing—review & editing, Q.L., J.L. 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 project- Impact of land use evolution on watershed water resources and ecological response in the Jianghan Plain over 50 years (No. 41271534); Department of Agriculture and Rural Affairs Key Project of Hubei Province in China-Assessment of Farmland Quality Grade and Natural Resource Asset Liability Compilation of Xiangzhou (No. 20180520); and Department of Agriculture and Rural Affairs Key Project of Hubei Province in China-Farmland Soil Environmental Quality Category Classification of Xiangzhou (No. 20181015). Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: No new data were created or analyzed in this study. Data sharing is not applicable to this article. Acknowledgments: We would like to express our deepest gratitude to the Xiangzhou government staff for their help in the research process. We also acknowledge the data support from the Resource and Environmental Science and Data Center (http://www.resdc.cn). Conflicts of Interest: The authors declare no conflict of interest.

References 1. Jiang, G.; Zhang, R.; Ma, W.; Zhou, D.; He, X. Cultivated land productivity potential improvement in land consolidation schemes in Shenyang, China: Assessment and policy implications. Land Use Policy 2017, 68, 80–88. [CrossRef] 2. Kong, X. China must protect high-quality arable land. Nature 2014, 506, 7. [CrossRef][PubMed] 3. Wang, N.; Zu, J.; Li, M.; Zhang, J.Y.; Hao, J.M. Spatial Zoning of cultivated land in Shandong Province Based on the Trinity of Quantity, Quality and Ecology. Sustainability 2020, 12, 1849. [CrossRef] Agriculture 2021, 11, 72 21 of 23

4. Xiao, Q.; Zong, Y.T.; Lu, S.G. Assessment of heavy metal pollution and human health risk in urban soils of steel industrial city (Anshan), Liaoning, Northeast China. Ecotoxicol. Environ. Saf. 2015, 120, 377–385. 5. Burges, A.; Epelde, L.; Garbisu, C. Impact of repeated single-metal and multi-metal pollution events on soil quality. Chemosphere 2015, 120, 8–15. [CrossRef] 6. Jin, X.B.; Shao, Y.; Zhang, Z.H.; Resler Lynn, M.; Campbell, B.J.; Chen, G.; Zhou, Y.K. The evaluation of land consolidation policy in improving agricultural productivity in China. Sci. Rep. 2017, 7, 2792. [CrossRef] 7. Zheng, H.; Huang, H.; Zhang, C.; Li, J. National-scale paddy-upland rotation in Northern China promotes sustainable development of cultivated land. Agric. Water Manag. 2016, 170, 20–25. [CrossRef] 8. Cao, Y.; Bai, Z.; Zhou, W.; Wang, J. Forces driving changes in cultivated land and management countermeasures in the Three Gorges Reservoir Area, China. J. Mt. Sci. 2013, 10, 149–162. [CrossRef] 9. Zhao, X.; Ye, Y.; Zhou, J.; Liu, L.; Hu, Y. Comprehensive assessment of cultivated land quality and sensitivity analysis of index weight in hilly region of Pearl River Delta. Trans. Chin. Soc. Agric. Eng. 2017, 33, 226–235. 10. Yang, J.; Wang, L.; Song, F.; Fan, P. Major Grain-Producing Areas Of Cultivated Land Quality Influence Factors And Grain Production Capacity Relationship Analysis. Chin. J. Agric. Res. Reg. Plan. 2017, 38, 15–22. 11. Yuan, X.; Shao, Y.; Wei, X.; Hou, R.; Ying, Y.; Zhao, Y.; Li, R. Study on the potential of cultivated land quality improvement based on a geological detector. Geol. J. 2018, 53, 387–397. [CrossRef] 12. Song, W.; Pijanowski, B. The effects of China’s cultivated land balance program on potential land productivity at a national scale. Appl. Geogr. 2014, 46, 158–170. [CrossRef] 13. Li, X.; Zhang, W.; Peng, Y. Grain Output and Cultivated Land Preservation: Assessment of the Rewarded Land Conversion Quotas Trading Policy in China’s Zhejiang Province. Sustainability 2016, 8, 821. [CrossRef] 14. Bui, E.N.; Henderson, B.L.; Viergever, K. Knowledge discovery from models of soil properties developed through data mining. Ecol. Model. 2006, 191, 431–446. [CrossRef] 15. Parisi, V.; Menta, C.; Gardi, C.; Jacomini, C.; Mozzanica, E. Microarthropod communities as a tool to assess soil quality and biodiversity: A new approach in italy. Agric. Ecosyst. Environ. 2005, 105, 323–333. [CrossRef] 16. Calabrese, J.; Ladisa, G.; Ladisa, G.; Otekhile, A. Impacts of land cover change on soil quality of manmade soils cultivated with table grapes in the Apulia Region of south-eastern Italy. Catena 2014, 121, 13–21. 17. Kong, X.B.; Liu, L.W.; Qin, J. Arable Land Evaluation Based on the Household Land Use Behavior in Daxing District of Beijing. Acta Geogr. Sin. 2008, 63, 856–868. 18. Zhu, C.M.; Hao, J.M.; Chen, L.; Deng, S.Y.; Liu, P.H. Well-facilitied capital farmland construction based on cultivated land comprehensive quality. Trans. Chin. Soc. Agric. Eng. 2015, 3, 233–242. 19. Gu, Z.; Wang, Y.; Sun, Y.; Ding, J. Application of entropy-weight TOPSIS model in synthetical quality assessment of Angelica sinensis growing in Gansu Province. J. Chin. Med. Mater. 2014, 37, 1549–1553. 20. Nie, Y.; Zhou, Y.; Qian, J.; Huang, J. Evaluation of Cropland Quality in Jingzhou’s Urban Fringe. Res. Sci. 2008, 6, 121–126. 21. Garcia-Melon, M.; Ferris-Onate, J.; Aznar-Bellver, J.; Aragones-Beltran, P.; Poveda-Bautista, R. Farmland appraisal based on the analytic network process. J. Glob. Optim. 2008, 42, 143–155. [CrossRef] 22. Meng, X.; Shi, F. An extended data envelopment analysis for the decision-making. J. Inequal. Appl. 2017, 1, 240. [CrossRef] [PubMed] 23. Ye, Y.; Zhao, X.; Hu, Y. Evaluation of cultivated land Quality in Pearl River Delta Based on GA-BP Neural Network. Ecol. Environ. Sci. 2018, 27, 964–973. 24. Zhu, M.; Liu, S.; Xia, Z.; Wang, G.; Hu, Y.; Liu, Z. Crop Growth Stage GPP-Driven Spectral Model for Evaluation of Cultivated Land Quality Using GA-BPNN. Agriculture 2020, 10, 318–333. [CrossRef] 25. Li, L. Model of Productivity of Cultivated Land Based on Improved-SVM. J. Shenyang Agric. Univ. 2012, 43, 126–128. 26. Qie, R.Q.; Guan, X.; Yan, X.J.; Dou, S.X.; Zhao, L. Method and its application of natural quality evaluation of arable land based on self-organizing feature map neural network. Trans. Chin. Soc. Agric. Eng. 2014, 30, 298–305. 27. Gao, Y. Research on Natural Quality Evaluation of Arable Land Based on BP Neural Network. Master’s Thesis, Central China Normal University, Wuhan, China, 2012. 28. Shekofteh, H.; Masoudi, A. Determining the features influencing the-S soil quality index in a semiarid region of Iran using a hybrid GA-ANN algorithm. Geoderma 2019, 355, 113908. [CrossRef] 29. Gayen, A.; Pourghasemi, H.R.; Saha, S.; Keesstra, S.; Bai, S.B. Gully erosion susceptibility assessment and management of hazard-prone areas in India using different machine learning algorithms. Sci. Total Environ. 2019, 668, 124–138. [CrossRef] 30. Lai, H.S.; Wu, C.F. Productivity Evaluation of Standard Cultivated Land Based on Rough Set and Support Vector Machine. J. Nat. Res. 2011, 26, 2141–2154. 31. Wang, H.; Luo, P.; Zhao, Z.; Nie, K. Establishment of agricultural land appraisal model based on deep belief network. Trans. Chin. Soc. Agric. Eng. 2018, 34, 263–271. 32. Hassoun, M. Fundamentals of Artificial Neural Networks. Proc. IEEE 2002, 84, 906. [CrossRef] 33. Herzog, S.; Tetzlaff, C.; Wrgtter, F. Evolving artificial neural networks with feedback. Neural Netw. 2020, 123, 153–162. [CrossRef] [PubMed] 34. Chen, G.; Du, J.B.; Sun, L.; Zhang, W.; Xu, K.; Chen, X.T.; Reed, G.; He, Z. Nonlinear Distortion Mitigation by Machine Learning of SVM Classification for PAM-4 and PAM-8 Modulated Optical Interconnection. J. Lightwave Technol. 2018, 36, 650–657. [CrossRef] Agriculture 2021, 11, 72 22 of 23

35. Liaw, A.; Wiener, M.; Liaw, A. Classification and Regression with Random Forest. R News 2002, 23, 18–22. 36. Chunrong, M.; Falk, H.; Guo, Y.; Han, X.; Wen, L. Why choose Random Forest to predict rare species distribution with few samples in large undersampled areas? Three Asian crane species models provide supporting evidence. PeerJ 2017, 5.[CrossRef] 37. Lindner, C.; Bromiley, P.; Ionita, M.; Cootes, T. Robust and accurate shape model matching using Random Forest Regression-Voting. IEEE Trans. Pattern Anal. Mach. Intell. 2015, 37, 1862–1874. [CrossRef] 38. Chen, D.L.; Lu, X.H.; Kuang, B. Measurement of cultivated land utilization efficiency: Construction and application of random forest. J. Nat. Res. 2019, 34, 1331–1344. 39. Khamoshi, E.; Sarmadian, F.; Keshavarzi, A. Digital Soil Mapping Using Random Forest Model and Land Suitability Evaluation for Abyek Region, Qazvin Province, Iran. J. Range Watershed Manag. 2019, 71, 885–899. 40. Bhowmik, A.; Kukal, S.S.; Saha, D.; Sharma, H.; Kalia, A.; Sharma, S. Potential Indicators of Soil Health Degradation in Different Land Use-Based Ecosystems in the Shiwaliks of Northwestern India. Sustainability 2019, 11, 3908. [CrossRef] 41. Chen, X.F.; Ren, X.N.; Zhang, C.; Xian, C.L.; Feng, X.K.; Ma, T.; Liu, J.M. Evaluation of medium and large scale farmland capacity enhancement potential based on random forest—An example from Guangdong Province. Southwest China J. Agric. Sci. 2019, 32, 2133–2140. 42. Lin, Z.C.; Ren, X.N.; Zhu, A.X.; Zhao, X.; Hu, Y.M. A study of arable land quality grading index system based on random forest algorithm. J. South. China Agric. Univ. 2020, 41, 38–48. 43. Hao, F.X.; Wu, S.; Zhang, Y.; Xu, Z.Y. Constructing Xiangyang City as “a Five Billion Kilograms” Food Production City. Hubei Agric. Sci. 2014, 53, 969–972. 44. Huang, H.; Zhou, Y.; Liu, Y.; Li, K.; Xiao, L.; Li, M.; Tian, Y.; Wu, F. Assessment of Anthropogenic Sources of Potentially Toxic Elements in Soil from Arable Land Using Multivariate Statistical Analysis and Random Forest Analysis. Sustainability 2020, 12, 8538. [CrossRef] 45. Hu, Y.M.; Jia, Z.L.; Cheng, J.C.; Zhao, Z.; Chen, F. Spatial variability of soil arsenic and its association with soil nitrogen in intensive farming systems. J. Soils Sediments 2016, 16, 169–176. [CrossRef] 46. Yao, S.B.; Li, H. Agricultural productivity changes induced by the sloping land conversion program: An analysis of Wuqi County in the Loess Plateau Region. Environ. Manag. 2010, 45, 541–550. [CrossRef][PubMed] 47. Tan, Y.; Chen, H.; Lian, K.; Yu, Z. Comprehensive Evaluation of Cultivated Land Quality at County Scale: A Case Study of Shengzhou, Zhejiang Province, China. Int. J. Environ. Res. Public Health 2020, 17, 1169. [CrossRef][PubMed] 48. Fan, F.; Liu, Y.; Chen, J.; Dong, J. Scenario-based ecological security patterns to indicate landscape sustainability: A case study on the Qinghai-Tibet Plateau. Landsc. Ecol. 2020.[CrossRef] 49. Wilson, G.A. The spatiality of multifunctional agriculture: A human geography perspective. Geoforum 2009, 40, 269–280. [CrossRef] 50. Caroline, C.; Olivier, E.; Sébastien, S.B.; Florent, H.; Kristof, V.O.; Irène, L.; Joël, D.; Jean-Jacques, M. Quantifying and modelling the impact of land consolidation and field borders on soil redistribution in agricultural landscapes (1954–2009). Catena 2013, 110, 184–195. 51. Ministry of Agriculture and Rural Affairs of the People’s Republic of China. FarmLand Quality Grade; China Standard Press: Beijing, China, 2016. 52. Ministry of Agriculture and Rural Affairs of the People’s Republic of China. National Arable Land Quality Grade Evaluation Indicator System; China Standard Press: Beijing, China, 2019. 53. General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China; Standardisation Administration of the People’s Republic of China. Regulations for Classification on Agriculture Land of the People’s Republic of China; China Standard Press: Beijing, China, 2012. 54. Chen, Z.F.; Shi, D.M.; Jin, H.F.; Lou, Y.B.; He, W.; Xia, J.R. Evaluation on cultivated-layer soil quality of sloping farmland in Yunnan based on soil management assessment framework (SMAF). Trans. Chin. Soc. Agric. Eng. 2019, 35, 256–267. 55. Xiao, L.; Zhou, Y.; Huang, H.; Liu, Y.J.; Li, K.; Li, M.Y.; Tian, Y.; Wu, F. Application of Geostatistical Analysis and Random Forest for Source Analysis and Human Health Risk Assessment of Potentially Toxic Elements (PTEs) in Arable Land Soil. Int. J. Environ. Res. Public Health 2020, 17, 9296. [CrossRef][PubMed] 56. Zhang, C.; Nie, Y.; Ma, Z. Assessment of farmland quality based on BP neural network method—A case study at Xiangyang urban area. China Agric. Inform. 2019, 31, 72–83. 57. Zeng, R.; Zhao, R.; Liang, Y. Farmland quality assessment based on AHP-grey correlation analysis method—Taking Xiangyang city of Hubei province as an example. Sci. Surv. Mapp. 2018, 43, 90–96. 58. Chen, Z.F.; Shi, D. Spatial Structure Characteristics of Slope Farmland Quality in Plateau Mountain Area: A Case Study of Yunnan Province, China. Sustainability 2020, 12, 7230. [CrossRef] 59. Thomas, B.W.; Hunt, D.; Bittman, S.; Hannam, K.D.; Messiga, A.J.; Haak, D.; Sharifi, M.; Hao, X. Soil health indicators after 21 yr of no-tillage in south coastal British Columbia. Can. J. Soil Sci. 2019, 2, 222–225. [CrossRef] 60. Zhong, L.; Wang, J.; Zhang, X.; Ying, L.; Zhu, C. Effects of agricultural land consolidation on soil conservation service in the Hilly Region of Southeast China-Implications for land management. Land Use Policy 2020, 95, 104637. [CrossRef] 61. Raiesi, F.; Salek-Gilani, S. Development of a soil quality index for characterizing effects of land-use changes on degradation and ecological restoration of rangeland soils in a semi-arid ecosystem. Land Degrad. Dev. 2020, 31, 1533. [CrossRef] Agriculture 2021, 11, 72 23 of 23

62. Fresno, T.; Penalosa, J.M.; Flagmeier, M.; Eduardo, M.J. Aided phytostabilisation over two years using iron sulphate and organic amendments: Effects on soil quality and rye production. Chemosphere 2020, 240, 124827.1–124827.11. [CrossRef] 63. Sefati, Z.; Khalilimoghadam, B.; Nadian, H. Assessing urban soil quality by improving the method for soil environmental quality assessment in a saline groundwater area of Ira. Catena 2019, 173, 471–480. [CrossRef] 64. Huang, Y.; Chen, G.; Huang, Y.; Liu, S.; Meng, Y.; Liu, B.; Xie, L.; Sengxua, P.; Chuayeng, M.; Tang, Q. Study on Soil Nutrient Fertility of cultivated land in Laos. Agric. Biotechnol. 2020, 9, 94–96. 65. Rezaei, H.; Saadat, S.; Mirkhani, R.; Bagheri, Y.R.; Ghadbeiklou, J. The state of soil organic carbon in agricultural lands of Iran with different agroecological conditions. Int. J. Environ. Anal. Chem. 2020, 7, 1–17. [CrossRef] 66. Xu, Z.; Zhang, T.; Wang, S.; Wang, Z. Soil pH and C/N ratio determines spatial variations in soil microbial communities and enzymatic activities of the agricultural ecosystems in Northeast China: Jilin Province case. Appl. Soil Ecol. 2020, 155, 103629. [CrossRef] 67. Buthelezi-Dube, N.N.; Hughes, J.C.; Muchaonyerwa, P.; Caister, K.F.; Modi, A.T. Soil fertility assessment and management from the perspective of farmers in four villages of eastern South Africa. Soil Use Manag. 2020, 36, 250–260. [CrossRef] 68. Khadka, D.; Lamichhane, S.; Amgain, R.; Joshi, S.; Ghimire, N.H. Soil fertility assessment and mapping spatial distribution of Agricultural Research Station, Bijayanagar, Jumla, Nepal. Eurasian J. Soil Sci. 2019, 8, 237–248. [CrossRef] 69. Tyler, M.; Hunter, L.; Steiner, F.; Roe, D. Use of agricultural land assessment and site assessment in Whitman County, Washington, USA. Environ. Manag. 1987, 11, 407–412. [CrossRef] 70. Hu, Q.; Yang, Y.; Han, S.; Wang, J. Degradation of agricultural drainage water quantity and quality due to farmland expansion and water-saving operations in arid basins. Agric. Water Manag. 2019, 213, 185–192. [CrossRef] 71. Li, M.; Singh, V.P.; Tortajada, C. Sustainability of water and energy use for food production based on optimal allocation of agricultural irrigation water. Int. J. Water Res. Dev. 2020, 36, 528–546. [CrossRef] 72. Li, J.; Fei, L.; Li, S.; Shi, Z.; Liu, L. The inﬂuence of optimized allocation of agricultural water and soil resources on irrigation and drainage in the Jingdian Irrigation District, China. Irrig. Sci. 2020, 38, 37–47. [CrossRef] 73. Kang, C.D. The effects of spatial accessibility and centrality to land use on walking in Seoul, Korea. Cities 2015, 46, 94–103. [CrossRef] 74. Geurs, K.T.; Wee, B.V. Accessibility assessment of land-use and transport strategies: Review and research directions. J. Transp. Geogr. 2004, 12, 127–140. [CrossRef] 75. Yigitcanlar, T.; Sipe, N.; Evans, R.; Pitot, M. A GIS-based land use and public transport accessibility indexing model. Aust. Plan. 2007, 44, 30–37. [CrossRef] 76. Li, L. Towards a protocol on fair compensation in cases of legitimate land tenure changes: Input document for a participatory process. J. Chin. Gov. 2018, 1, 124–127. [CrossRef] 77. Zhang, C.; Qiao, M.; Yun, W.; Liu, J.; Zhu, D.; Yang, J. Trinity Comprehensive Regulatory System about Quantity, Quality and Ecology of Cultivated Land. Trans. Chin. Soc. Agric. Mach. 2017, 48, 1–6. 78. Liu, Y.; Zhang, Y.; Guo, L. Towards realistic assessment of cultivated land quality in an ecologically fragile environment: A satellite imagery-based approach. Appl. Geography. 2010, 30, 271–281. [CrossRef] 79. Feng, Y.; Yang, Q.; Tong, X.; Chen, L. Evaluating land ecological security and examining its relationships with driving factors using GIS and generalized additive model. Sci. Total Environ. 2018, 633, 1469–1479. [CrossRef] 80. Chu, M.; Singh, S.; Walker, F.R. Soil Health and Soil Fertility Assessment by the Haney Soil Health Test in an Agricultural Soil in West Tennessee. Commun. Soil Sci. Plant. Anal. 2019, 50, 1123–1131. [CrossRef] 81. Ma, J.; Chen, Y.; Antoniadis, V.; Wang, K.; Huang, Y.; Tian, H. Assessment of heavy metal(loid)s contamination risk and grain nutritional quality in organic waste-amended soil. J. Hazard. Mater. 2020, 399, 123095. [CrossRef] 82. Turner, M.G.; O’Neill, R.V.; Milne, G.B.T. Effects of changing spatial scale on the analysis of landscape pattern. Landsc. Ecol. 1989, 3, 153–162. [CrossRef] 83. Kong, X.B.; Zhang, B.B.; Wen, L.Y.; Hu, Y.J.; Lei, M.; Yao, J.T.; Xin, Y.N. Theoretical Framework and Research Trends of Cultivated Land Quality based on Elements-Process-Function. China Land Sci. 2018, 32, 14–20. 84. Shen, R.F.; Chen, M.J.; Kong, X.B.; Li, Y.T.; Wang, J.K.; Li, T.; Lu, M.X. Conception and evaluation of quality of arable land and strategies for its management. Acta Pedol. Sin. 2012, 49, 1210–1217. 85. Zhao, Y.F.; Chen, D.Q.; Chen, J.; Sun, Z.Y.; Zhang, H.N. Problems and analytical logic in building cultivated land productivity evaluation index system. Acta Pedol. Sin. 2015, 52, 1197–1208. 86. Cheng, J.N.; Zhao, G.X.; Zhang, Z.X.; Wang, J. GIS supported comprehensive evaluation of cultivated land quality at small scale: A case study in Dingzhuang town of Shandong province. J. Nat. Res. 2009, 24, 536–544.