基于机器学习方法的宁夏南部土壤质地空间分布研究

doi:10.3864/j.issn.0578-1752.2022.15.008

摘要/Abstract

摘要：

【目的】基于历史数据,利用机器学习方法分析宁夏南部土壤质地空间变异规律及其与环境因素之间的关系。【方法】基于宁夏回族自治区南部428 个20世纪80年代第二次土壤普查土壤剖面点数据,采用分类回归树（CART）和随机森林（RF）两种机器学习方法,结合地形因子、土壤类型、归一化植被指数,探究与宁夏南部地区土壤质地分布相关性较强的环境因素,并用两种机器学习预测该区土壤质地类型的空间分布,用剖面点验证集数据以及宁夏回族自治区海原县实测样点数据验证模型精度。【结果】（1）RF和CART对剖面点验证集土壤质地类型的预测正确率分别为 62.36%、55.29%,接收者操作特性（receiver operating characteristic,ROC）曲线下面积（area under roc curve,AUC）分别为0.7515、0.6933,对海原县122个实测样点的预测正确率分别为54.10%、48.36%,AUC分别为0.6599、0.5981,RF的预测精度高于CART。（2）该区土壤类型（ST）是与土壤质地空间分布相关性最强的环境因素,其次是高程（Ele）,高程越高,土壤质地越黏重。风力作用指数（WEI）和坡度（Slo）对土壤质地的影响较小。（3）研究区土壤质地类型以轻壤土为主,空间分布格局基本呈现为南部土壤质地黏重,北部土壤质地较轻。【结论】RF更适合预测宁夏南部地区土壤质地的空间分布,且充分利用历史数据,结合新的野外采样,可以达到预测制图的精度要求;土壤类型（ST）和高程（Ele）是与土壤质地空间分布相关性较强的环境因素。

关键词: 土壤质地, 空间分布, 因素分析, 随机森林, 分类回归树

Abstract:

【Objective】Based on historical soil data, this paper studied the spatial variability of soil texture and its relationship with environmental factors in southern Ningxia by using machine learning.【Method】Classification and regression tree (CART), random forest (RF) and traditional statistical methods were used to explore the main environmental factors that affected the soil texture types and predict the spatial distribution of soil texture types in southern Ningxia, based on 428 soil profiles from the second soil survey in the 1980s, combined with topographic factors, soil types, and normalized vegetation index. The accuracy of the models were verified by the validating set of soil profiles and the soil samples in Haiyuan County, Ningxia.【Result】(1)The accuracy rates of RF and CART on the soil texture type of the verification set of soil profiles were 62.36% and 55.29%, respectively; the area under the receiver operating characteristic (ROC) curve of them (area under roc curve, AUC) were 0.7515 and 0.6933, respectively; the accuracy rates of them on soil samples in Haiyuan County were 54.10% and 48.36%, respectively; the AUC of them were 0.6599 and 0.5981 respectively. (2) Soil type (ST) was the most important predictor variable, followed by elevation (Ele). The higher elevation was, the heavier the soil texture was. The effects of wind exposition index (WEI) and slope (Slo) on soil texture were lower. (3)The results predicted by two methods showed a spatial distribution trend that the soil texture was heavy in the southern area but light in the northern area of southern Ningxia.【Conclusion】The prediction accuracy of RF for soil texture type in southern Ningxia was higher than CART. Making full use of historical data, combined with field sampling, could meet the accuracy requirements of digital mapping. In the loess region, soil types and elevation were the environmental factors which had strong correlation with spatial variation of soil texture.

Key words: soil texture, spatial distribution, factor analysis, random forest (RF), classification and regression tree (CART)

申哲,张认连,龙怀玉,徐爱国. 基于机器学习方法的宁夏南部土壤质地空间分布研究[J]. 中国农业科学, 2022, 55(15): 2961-2972.

SHEN Zhe,ZHANG RenLian,LONG HuaiYu,XU AiGuo. Research on Spatial Distribution of Soil Texture in Southern Ningxia Based on Machine Learning[J]. Scientia Agricultura Sinica, 2022, 55(15): 2961-2972.

0
/ / 推荐

导出引用管理器 EndNote|Reference Manager|ProCite|BibTeX|RefWorks

链接本文: https://www.chinaagrisci.com/CN/10.3864/j.issn.0578-1752.2022.15.008

https://www.chinaagrisci.com/CN/Y2022/V55/I15/2961

图/表 12

图1

表1

表2

图2

图3

表3

表4

表5

表6

图4

表7

图5

参考文献 38

[1]	吴克宁, 赵瑞. 土壤质地分类及其在我国应用探讨. 土壤学报, 2019, 56(1): 227-241.
	WU K N, ZHAO R. Soil texture classification and its application in China. Acta Pedologica Sinica, 2019, 56(1): 227-241. (in Chinese)
[2]	BORMANN H. Towards a hydrologically motivated soil texture classification. Geoderma, 2010, 157(3/4): 142-153. doi: 10.1016/j.geoderma.2010.04.005. doi: 10.1016/j.geoderma.2010.04.005
[3]	刘剑刚, 张华, 朱岩, 朱夏夏, 何红, 刘玉国, 王颖, 马明军. 辽东山地冰缘地貌表层土壤粒度特征. 中国水土保持科学, 2016, 14(1): 36-45. doi: 10.16843/j.sswc.2016.01.005. doi: 10.16843/j.sswc.2016.01.005
	LIU J G, ZHANG H, ZHU Y, ZHU X X, HE H, LIU Y G, WANG Y, MA M J. Grain size characteristics of overlying soil on periglacial landforms in mountainous region of eastern Liaoning. Science of Soil and Water Conservation, 2016, 14(1): 36-45. doi: 10.16843/j.sswc.2016.01.005. (in Chinese) doi: 10.16843/j.sswc.2016.01.005
[4]	张娜, 张栋良, 屈忠义, 王丽荣. 内蒙古河套灌区区域土壤质地空间变异分析: 以解放闸灌域为例. 干旱区资源与环境, 2015, 29(12): 155-163. doi: 10.13448/j.cnki.jalre.2015.417. doi: 10.13448/j.cnki.jalre.2015.417
	ZHANG N, ZHANG D L, QU Z Y, WANG L R. The spatial variation of soil texture in Hetao Irrigation District in Inner Mongolia. Journal of Arid Land Resources and Environment, 2015, 29(12): 155-163. doi: 10.13448/j.cnki.jalre.2015.417. (in Chinese) doi: 10.13448/j.cnki.jalre.2015.417
[5]	张世文, 黄元仿, 苑小勇, 王睿, 叶回春, 段增强, 龚关. 县域尺度表层土壤质地空间变异与因素分析. 中国农业科学, 2011, 44(6): 1154-1164. doi: 10.3864/j.issn.0578-1752.2011.06.010. doi: 10.3864/j.issn.0578-1752.2011.06.010
	ZHANG S W, HUANG Y F, YUAN X Y, WANG R, YE H C, DUAN Z Q, GONG G. The spatial variability and factor analyses of top soil texture on a County scale. Scientia Agricultura Sinica, 2011, 44(6): 1154-1164. doi: 10.3864/j.issn.0578-1752.2011.06.010. (in Chinese) doi: 10.3864/j.issn.0578-1752.2011.06.010
[6]	李宗善, 杨磊, 王国梁, 侯建, 信忠保, 刘国华, 傅伯杰. 黄土高原水土流失治理现状、问题及对策. 生态学报, 2019, 39(20): 7398-7409. doi: 10.5846/stxb201909021821. doi: 10.5846/stxb201909021821
	LI Z S, YANG L, WANG G L, HOU J, XIN Z B, LIU G H, FU B J. The management of soil and water conservation in the Loess Plateau of China: Present situations, problems, and counter-solutions. Acta Ecologica Sinica, 2019, 39(20): 7398-7409. doi: 10.5846/stxb201909021821. (in Chinese) doi: 10.5846/stxb201909021821
[7]	LIEß M, GLASER B, HUWE B. Uncertainty in the spatial prediction of soil texture: comparison of regression tree and Random Forest models. Geoderma, 2012, 170: 70-79. doi: 10.1016/j.geoderma.2011.10.010. doi: 10.1016/j.geoderma.2011.10.010
[8]	CEDDIA M B, VIEIRA S R, VILLELA A L O, MOTA L D S, ANJOS L H C D, DE CARVALHO D F. Topography and spatial variability of soil physical properties. Scientia Agricola, 2009, 66(3): 338-352. doi: 10.1590/s0103-90162009000300009. doi: 10.1590/s0103-90162009000300009
[9]	江厚龙, 王新中, 刘国顺, 胡宏超, 刘清华. 豫西典型烟田土壤颗粒组成的空间变异性分析. 中国烟草科学, 2012, 33(2): 62-67, 73.
	JIANG H L, WANG X Z, LIU G S, HU H C, LIU Q H. Spatial variability of soil particle composition based on geostatistics. Chinese Tobacco Science, 2012, 33(2): 62-67, 73. (in Chinese)
[10]	王冬冬, 高磊, 陈效民, 彭新华. 红壤丘陵区坡地土壤颗粒组成的空间分布特征研究. 土壤, 2016, 48(2): 361-367. doi: 10.13758/j.cnki.tr.2016.02.023. doi: 10.13758/j.cnki.tr.2016.02.023
	WANG D D, GAO L, CHEN X M, PENG X H. Spatial distribution characteristics of soil particle composition of slope land red soil region, China. Soils, 2016, 48(2): 361-367. doi: 10.13758/j.cnki.tr.2016.02.023. (in Chinese) doi: 10.13758/j.cnki.tr.2016.02.023
[11]	ADHIKARI K, KHEIR R B, GREVE M B, BØCHER P K, MALONE B P, MINASNY B, MCBRATNEY A B, GREVE M H. High- resolution 3-D mapping of soil texture in Denmark. Soil Science Society of America Journal, 2013, 77(3): 860-876. doi: 10.2136/sssaj2012.0275. doi: 10.2136/sssaj2012.0275
[12]	AKPA S I C, ODEH I O A, BISHOP T F A, HARTEMINK A E. Digital mapping of soil particle-size fractions for Nigeria. Soil Science Society of America Journal, 2014, 78(6): 1953-1966. doi: 10.2136/sssaj2014.05.0202. doi: 10.2136/sssaj2014.05.0202
[13]	FORKUOR G, HOUNKPATIN O K L, WELP G, THIEL M. High resolution mapping of soil properties using remote sensing variables in south-western Burkina Faso: A comparison of machine learning and multiple linear regression models. PLoS ONE, 2017, 12(1): e0170478. doi: 10.1371/journal.pone.0170478. doi: 10.1371/journal.pone.0170478
[14]	申哲, 张认连, 龙怀玉, 王转, 朱国龙, 石乾雄, 喻科凡, 徐爱国. 基于3种空间预测方法的黄土区土壤颗粒组成空间分布研究: 以宁夏海原县为例. 中国农业科学, 2020, 53(18): 3716-3728.
	SHEN Z, ZHANG R L, LONG H Y, WANG Z, ZHU G L, SHI Q X, YU K F, XU A G. Research on spatial distribution of soil particle size distribution in loess region based on three spatial prediction methods-Taking Haiyuan County in Ningxia as an example. Scientia Agricultura Sinica, 2020, 53(18): 3716-3728. (in Chinese)
[15]	赵明月, 赵文武, 刘源鑫. 不同尺度下土壤粒径分布特征及其影响因子: 以黄土丘陵沟壑区为例. 生态学报, 2015, 35(14): 4625-4632. doi: 10.5846/stxb201311272828. doi: 10.5846/stxb201311272828
	ZHAO M Y, ZHAO W W, LIU Y X. Comparative analysis of soil particle size distribution and its influence factors in different scales: A case study in the Loess Hilly-gully area. Acta Ecologica Sinica, 2015, 35(14): 4625-4632. doi: 10.5846/stxb201311272828. (in Chinese) doi: 10.5846/stxb201311272828
[16]	王德, 傅伯杰, 陈利顶, 赵文武, 汪亚峰. 不同土地利用类型下土壤粒径分形分析: 以黄土丘陵沟壑区为例. 生态学报, 2007, 27(7): 3081-3089. doi: 10.3321/j.issn:1000-0933.2007.07.050. doi: 10.3321/j.issn:1000-0933.2007.07.050
	WANG D, FU B J, CHEN L D, ZHAO W W, WANG Y F. Fractal analysis on soil particle size distributions under different land-use types: A case study in the loess hilly areas of the Loess Plateau, China. Acta Ecologica Sinica, 2007, 27(7): 3081-3089. doi: 10.3321/j.issn:1000-0933.2007.07.050. (in Chinese) doi: 10.3321/j.issn:1000-0933.2007.07.050
[17]	姜赛平, 张认连, 张维理, 徐爱国, 张怀志, 谢良商, 冀宏杰. 近30年海南岛土壤有机质时空变异特征及成因分析. 中国农业科学, 2019, 52(6): 1032-1044. doi: 10.3864/j.issn.0578-1752.2019.06.007. doi: 10.3864/j.issn.0578-1752.2019.06.007
	JIANG S P, ZHANG R L, ZHANG W L, XU A G, ZHANG H Z, XIE L S, JI H J. Spatial and temporal variation of soil organic matter and cause analysis in Hainan island in resent 30 years. Scientia Agricultura Sinica, 2019, 52(6): 1032-1044. doi: 10.3864/j.issn.0578-1752.2019.06.007. (in Chinese) doi: 10.3864/j.issn.0578-1752.2019.06.007
[18]	马秉春, 康凯剑, 李世虎. 宁夏南部山区苦水资源运用分析. 中华建设科技, 2011(6): 70, 79.
	MA B C, KANG K J, LI S H. Ningxia southern mountain bitter analysis of water use. Construction Technology of China, 2011(6): 70, 79. (in Chinese)
[19]	BÖHNER J, ANTONIĆ O. Chapter 8 land-surface parameters specific to topo-climatology. Developments in Soil Science. Amsterdam: Elsevier, 2009: 195-226. doi: 10.1016/s0166-2481(08)00008-1. doi: 10.1016/s0166-2481(08)00008-1
[20]	BREIMAN L. Random forests. Machine Learning, 2001, 45(1): 5-32. doi: 10.1023/A:1010933404324
[21]	曹铭昌, 周广胜, 翁恩生. 广义模型及分类回归树在物种分布模拟中的应用与比较. 生态学报, 2005, 25(8): 203-212.
	CAO M C, ZHOU G S, WENG E S. Application and comparison of generalized models and classification and regression tree in simulating tree species distribution. Acta Ecologica Sinica, 2005, 25(8): 203-212. (in Chinese)
[22]	GENUER R, POGGI J M, TULEAU-MALOT C. Variable selection using random forests. Pattern Recognition Letters, 2010, 31(14): 2225-2236. doi: 10.1016/j.patrec.2010.03.014
[23]	冯盼峰, 温永仙. 基于随机森林算法的两阶段变量选择研究. 系统科学与数学, 2018, 38(1): 119-130. doi: 10.12341/jssms13325
	FENG P F, WEN Y X. Two-stage stepwise variable selection based on random forests. Journal of Systems Science and Mathematical Sciences, 2018, 38(1): 119-130. (in Chinese) doi: 10.12341/jssms13325
[24]	赵北庚. 基于R语言randomForest包的随机森林建模研究. 计算机光盘软件与应用, 2015, 18(2): 152-153.
	ZHAO B G. Research on randomForest modeling based on R language random forest package. Computer CD Software and Applications, 2015, 18(2): 152-153. (in Chinese)
[25]	周志华. 机器学习. 北京: 清华大学出版社, 2016: 24-27.
	ZHOU Z H. Machine Learning. Beijing: Tsinghua University Press, 2016: 24-27. (in Chinese)
[26]	汪云云, 陈松灿. 基于AUC的分类器评价和设计综述. 模式识别与人工智能, 2011, 24(1):64-71.
	WANG Y Y, CHEN S C. A survey of evaluation and design for AUC based classifier. Pattern Recognition and Artificial Intelligence, 2011, 24(1): 64-71. (in Chinese)
[27]	FAWCETT T E. An introduction to ROC analysis. Pattern Recognition Letters, 2006, 27(8): 861-874. doi: 10.1016/j.patrec.2005.10.010
[28]	吕喜玺, 沈荣明. 土壤可蚀性因子K值的初步研究. 水土保持学报, 1992, 6(1): 63-70. doi: 10.13870/j.cnki.stbcxb.1992.01.010. doi: 10.13870/j.cnki.stbcxb.1992.01.010
	LÜ X X, SHEN R M. A preliminary study on the values K of soil erosibility factor. Journal of Soil and Water Conservation, 1992, 6(1): 63-70. doi: 10.13870/j.cnki.stbcxb.1992.01.010. (in Chinese) doi: 10.13870/j.cnki.stbcxb.1992.01.010
[29]	刘超, 卢玲, 胡晓利. 数字土壤质地制图方法比较: 以黑河张掖地区为例. 遥感技术与应用, 2011, 26(2): 177-185.
	LIU C, LU L, HU X L. Comparison analysis on digital soil texture mapping in an area of Zhangye, Heihe River basin. Remote Sensing Technology and Application, 2011, 26(2): 177-185. (in Chinese)
[30]	马冉, 刘洪斌, 武伟. 流域尺度下地形属性对土壤质地类型变异的影响: 以重庆市彭水县一小流域为例. 农业资源与环境学报, 2019, 36(3): 279-286. doi: 10.13254/j.jare.2018.0184. doi: 10.13254/j.jare.2018.0184
	MA R, LIU H B, WU W. Effect of topographic attributes on soil texture class variations at a watershed scale: A case study of a basin in Pengshui County of Chongqing, China. Journal of Agricultural Resources and Environment, 2019, 36(3): 279-286. doi: 10.13254/j.jare.2018.0184. (in Chinese) doi: 10.13254/j.jare.2018.0184
[31]	杨艳丽, 史学正, 于东升, 王洪杰, 徐茂, 王果. 区域尺度土壤养分空间变异及其影响因素研究. 地理科学, 2008, 28(6): 788-792. doi: 10.3969/j.issn.1000-0690.2008.06.013. doi: 10.3969/j.issn.1000-0690.2008.06.013
	YANG Y L, SHI X Z, YU D S, WANG H J, XU M, WANG G. Spatial heterogeneity of soil nutrients and their affecting factors at regional scale. Scientia Geographica Sinica, 2008, 28(6): 788-792. doi: 10.3969/j.issn.1000-0690.2008.06.013. (in Chinese) doi: 10.3969/j.issn.1000-0690.2008.06.013
[32]	邹心雨, 张卓栋, 吴梦瑶, 万缘强. 河北坝上地区坡面尺度土壤机械组成的空间变异. 中国水土保持科学, 2019, 17(5): 44-53. doi: 10.16843/j.sswc.2019.05.006. doi: 10.16843/j.sswc.2019.05.006
	ZOU X Y, ZHANG Z D, WU M Y, WAN Y Q. Spatial variability of particle size distribution at slope scale in Bashang region of Hebei Province. Science of Soil and Water Conservation, 2019, 17(5): 44-53. doi: 10.16843/j.sswc.2019.05.006. (in Chinese) doi: 10.16843/j.sswc.2019.05.006
[33]	曲潇琳, 龙怀玉, 谢平, 曹祥会, 王佳佳. 宁夏中部地区典型灰钙土的发育特性及系统分类研究. 土壤学报, 2018, 55(1): 75-87. doi: 10.11766/trxb201706120097. doi: 10.11766/trxb201706120097
	QU X L, LONG H Y, XIE P, CAO X H, WANG J J. Genetic characteristics and classification of typical sierozem in central Ningxia, China. Acta Pedologica Sinica, 2018, 55(1): 75-87. doi: 10.11766/trxb201706120097. (in Chinese) doi: 10.11766/trxb201706120097
[34]	季耿善. 黑垆土的形成环境. 土壤学报, 1992, 29(2): 113-125.
	JI G S. Genetic environment of dark loessial soil. Acta Pedologica Sinica, 1992, 29(2): 113-125. (in Chinese)
[35]	王吉智. 宁夏土壤的形成作用. 华中农业大学学报, 1989, 8(S1): 38-44. doi: 10.13300/j.cnki.hnlkxb.1989.s1.009. doi: 10.13300/j.cnki.hnlkxb.1989.s1.009
	WANG J Z. Soil formation process for soils in Ningxia. Journal of Huazhong Agricultural, 1989, 8(S1): 38-44. doi: 10.13300/j.cnki.hnlkxb.1989.s1.009. (in Chinese) doi: 10.13300/j.cnki.hnlkxb.1989.s1.009
[36]	LI A D, GUO P T, WU W, LIU H B. Impacts of terrain attributes and human activities on soil texture class variations in hilly areas, south-west China. Environmental Monitoring and Assessment, 2017, 189(6): 281. doi: 10.1007/s10661-017-5997-0. doi: 10.1007/s10661-017-5997-0
[37]	曲潇琳, 龙怀玉, 曹祥会, 谢平. 宁夏山地土壤的发育规律及系统分类研究. 土壤学报, 2019, 56(1): 65-77.
	QU X L, LONG H Y, CAO X H, XIE P. Development rules and taxonomy of the soil in Helan and Liupan mountains of Ningxia Province. Acta Pedologica Sinica, 2019, 56(1): 65-77. (in Chinese)
[38]	STĘPIEŃ M, SAMBORSKI S, GOZDOWSKI D, DOBERS E S, CHORMAŃSKI J, SZATYŁOWICZ J. Assessment of soil texture class on agricultural fields using ECa, Amber NDVI, and topographic properties. Journal of Plant Nutrition and Soil Science, 2015, 178(3): 523-536. doi: 10.1002/jpln.201400570. doi: 10.1002/jpln.201400570

土壤类型 Soil type	土壤质地类型 Soil texture type
土壤类型 Soil type	松砂、紧砂土 Sand	砂壤土 Sandy loam	轻壤土 Light loam	中壤土 Medium loam	重壤土 Heavy loam	合计 Total
风沙土 Aeolian sandy soil	10	1	0	0	0	11
灰钙土 Sierozem	6	55	69	15	4	149
黄绵土 Loessal soil	0	1	3	5	0	9
黑垆土 Dark loessal soil	0	14	51	48	7	120
灰褐土 Grey cinnamon	0	1	32	36	3	72
亚高山草甸土 Meadow soil	0	0	1	2	0	3
新积土 Alluvial soil	0	13	22	7	2	44
盐土 Saline soil	0	2	5	3	1	11
潮土 Fluvo-aquic soil	0	0	1	5	2	8
粗骨土 Fragmental soil	0	1	0	0	0	1
合计 Total	16	88	184	121	19	428

土壤质地类型 Soil texture	<0.01mm （%）	赋值 Assignment
松砂、紧砂土Sand	0-10	1
砂壤土Sandy loam	10-20	2
轻壤土Light loam	20-30	3
中壤土Medium loam	30-45	4
重壤土Heavy loam	45-60	5

项目 Item	土壤质地类型 Soil texture type
项目 Item	松砂、紧砂土 Sand	砂壤土 Sandy loam	轻壤土 Light loam	中壤土 Medium loam	重壤土 Heavy clay loam	合计 Total
验证样点 Validation sample	3	16	37	24	5	85
CART正确数 Right quantity (CART)	1	7	20	18	1	47
CART正确率 Right rate (CART) (%)	33.33	43.75	54.05	75.00	20.00	55.29
RF正确数 Right quantity (RF)	2	10	23	16	2	53
RF正确率 Right rate (RF) (%)	66.67	62.50	62.16	66.67	40.00	62.36

项目 Item	土壤质地类型 Soil texture type
项目 Item	松砂、紧砂土 Sand	砂壤土Sandy loam	轻壤土 Light loam	中壤土 Medium loam	重壤土 Heavy loam	合计 Total
实测样点	0	10	64	44	4	122
CART正确数 Right quantity (CART)	—	3	28	27	1	59
CART正确率 Right rate (CART) (%)	—	20.00	43.75	61.36	25.00	48.36
RF正确数 Right quantity (RF)	—	4	36	23	3	66
RF正确率 Right rate (RF) (%)	—	40.00	56.25	52.27	75.00	54.10