研究论文 正式出版 版本 3 Vol 9 (3) : 245-256 2018
下载
中国南方喀斯特地区石灰土容重传递函数模型及影响因素研究
A study on the pedo-transfer functions and influencing factors for prediction of soil bulk density for limestone soil in Karst area of South China
: 2018 - 05 - 18
224 1 0
摘要&关键词
摘要:摘, 要:土壤容重, ( BD), 数据的缺失严重影响了我国南方喀斯特地区土壤碳储量的估算, 亟待利用已有数据来构建容重传递函数模型, (PTFs)。, 本文利用南方喀斯特分布省份的石灰土土壤普查数据, 首次较为系统地研究了喀斯特地区石灰土的容重传递函数模型及影响因素, 。, 研究结果表明, :( 1), 国内外已发表的容重, PTFs, 对中国南方喀斯特地区非地带性石灰土的适用性较差, 需要进行优化或重新建立新的容重函数预测模型, ;(2), 优化后的模型, )*、, 韩光中等, (, )-, a, *, 和韩光中等, (, )-, c, *, 的预测精度得到明显提高, ;(3)基于石灰土亚类建立的, PTFs, 具有很高的拟合度和预测精度, 比优化模型更加适宜于喀斯特石灰土的土壤容重预测, ;(4)不同石灰土亚类容重预测的影响因素存在差异, 其中土壤有机质含量是石灰土容重预测的关键因素, 与各亚类土壤的, BD, 都有很高的相关性, ;(5) 土壤容重传递函数模型的适用性不仅与研究区域有关, 同时也与研究的土壤类型有关, 。, 建议在今后喀斯特地区土壤容重预测模型研究中充分考虑地域差异性和土壤类型因素, 。
关键词:关键词:石灰土; 土壤容重; 传递函数模型 ;影响因素
Abstract & Keywords
Abstract: Background, aim, and scope Soil bulk density is an important physical property of soils and is a basic parameter in many soil mass-volume conversion models. The conventional soil BD determination method is a time-consuming, labor-intensive, expensive and tedious, so most soil databases in our country are lack of soil BD data. The lack of bulk density data has seriously affected the estimation of soil carbon storage in the karst area of south China. It is urgent to construct the bulk density pedo-transfer function models (PTFs) using the available data. Materials and methods Based on the data of limestone soil in the south Karst distribution province, this paper mainly covers 15 soil properties except for the bulk density of four limestone sub-classes. Firstly, the applicability test of published soil bulk density transfer function was carried out. The function model with relatively higher prediction ability was optimized by 1stOpt software to improve its accuracy. In order to further improve the accuracy of the prediction results of bulk density of karst limestone soil, soil types were classified by limestone soil sub-class, and the regression analysis was carried out by SPSS software to establish a new function model. Results The results show that: (1) The published PTFs at home and abroad have poor applicability to non-zonal limestone soil in karst area of South China, and need to be optimized or rebuilt to predict the new bulk density function; (2) The prediction accuracy of the optimized models Shiri et al (2017)*, Han et al (2016) -a* and Han et al (2016) -c* has been significantly improved. However, compared with the newly established PTFs that based on limestone soil sub-class, the precision is not as good as the latter. Therefore, PTFs based on limestone soil subclasses are more suitable for predicting soil bulk density of limestone in this study area. (3) There are differences in the factors affecting the prediction of bulk density of different limestone soil subclasses. Soil organic matter content is the key factor for predicting limestone bulk density, which is highly correlated with BD in all sub-classes of limestone soils. Discussion The application of soil bulk density pedotransfer function has not only the limitation of geographical area but also the limitation of soil type, and the finer soil type classification, the higher prediction accuracy of bulk density pedo-transfer function. Correlation analysis and bulk density prediction model proved that soil organic matter content (OM) has an important influence on the bulk density prediction of limestone soil sub-class. Due to the difference of soil-forming conditions and soil-forming process between different types of limestone soil, there are great differences in soil properties, resulting in different correlations between soil bulk density and other soil properties. Therefore, it is necessary to classify different soil subtypes when predicting limestone soil bulk density to improve the prediction accuracy of the models. Conclusions This paper discusses the prediction model of karst limestone soil bulk density and its influencing factors. Established bulk density pedotransfer function models applicable to different limestone soil sub-classes in the karst area of the South China, and provides convenience for later soil-related work. Recommendations and perspectives We suggest that in the future, the study of soil bulk density forecasting models in karst area should take full account of the regional differences and soil type factors. Meanwhile, due to the limitation of the data, the spatial distribution of soil bulk density is not taken into account. This is one of the factors to be considered in the future to further accurately establish soil bulk density prediction models.
Keywords: Limestone soil; Soil bulk density; Pedotransfer function model; Influencing factors
BD是土壤有机碳储量估算模型的基本参数之一,由于传统的土壤BD测定方法是一项费时、费力、耗资且繁琐的过程(Benites et al, 2007; Suuster et al, 2011; Brahim et al, 2012),而且在一些特殊环境下(如岩石广布、植物根系较多等),传统采样方法更是难以实施(Huntington et al, 1989; Wiesmeier et al, 2012; Xu et al, 2016),因而导致我国乃至全球大多数土壤数据库中土壤BD数据的缺失,这给土壤有机碳储量的估算及相应研究带来了很大的困难。因此,需要寻求替代的方法来获得土壤BD数据。目前,有效替代方法有两大类:一类是均值(或中值)替代法,主要是根据剖面结构相似和土壤类型相近原则,利用同一数据库中已有剖面容重数据或该类型土壤容重数据的中值或均值来代替(潘根兴, 1999;解宪丽等, 2004;倪九派等, 2009;Milne et al, 2007; Wen and He, 2016);另一类是传递函数模型法(Pedo-Transfer Functions, PTFs)(Nanko et al, 2014; Behzad et al, 2015; Yi et al, 2016; Wang et al, 2014),通过输入相关的较容易获取的属性值来拟合得出BD的预测值。
中国南方喀斯特地区是我国陆地生态系统的重要组成部分,前人对该地区土壤有机碳密度和储量进行了估算,但是估算值大小不一、不确定性大(张勇等, 2008; 倪九派等, 2009; 陈曦等, 2012; 张珍明等, 2017)。原因何在?我们认为,喀斯特地区由于岩石裸露率高、空间异质性大、非地带性等特征(张美良和邓自强, 1994; 刘方等, 2008),导致已有的土壤学研究基础相对薄弱,数据库中的土壤BD缺失更为严重,且很难通过中值或均值来替代,是该地区土壤有机碳密度及储量估算存在很大不确定性的原因之一。基于此,本文拟利用已有的全国第二次土壤普查数据资源,选择喀斯特地区主要的土壤类型石灰土(非地带性)为研究对象,检视已有土壤BD传递函数模型对非地带性土壤石灰土的适用性,并在此基础上探索优化已有模型或新建适宜于喀斯特地区石灰土模型的可能性,为更可靠地估算中国南方喀斯特地区土壤有机碳密度和储量奠定基础。
1   材料与方法
1.1   数据源
文章所用数据主要来自“国家地球系统数据共享平台-土壤科学数据中心(http://soil.geodata.cn)”、《贵州土种志》、《四川土种志》中记载的全国第二次土壤普查的资料。受资料收集难易度和数据本身完整性等原因影响,仅以贵州、四川、重庆、湖南、江西的石灰土典型剖面样点数据为代表,来探讨南方喀斯特石灰土类的容重预测模型及影响因素。本文数据涵盖了黑色石灰土、棕色石灰土、黄色石灰土、红色石灰土四种亚类石灰土的16种土壤属性(表1)。其中,土壤粒度划分等级是根据国际制土壤质地分类标准:砂粒(Sand)2-0.02 mm、粉粒(Silt)0.02-0.002 mm、粘粒(Clay)<0.002 mm。本文为方便定量分析,将土地利用类型(Land use)和土壤质地类型(Soil texture)分别进行了数字定量化转换处理(易小波等, 2017; Wang et al, 2014)。
1.2   研究方法
1.2.1   容重传递函数模型适用性验证方法
模型的表现取决于统计标准的类型(Moreels et al, 2003),预测方法的评估应基于一组相应的验证指标(Han et al, 2012)。参考前人研究经验,本文选取的验证指标包括平均预测误差(MPE)、预测误差的标准偏差(SDPE)、均方根误差(RMSPE)、拟合优度判定系数(R2)四类指标(Han et al, 2012; Nanko et al, 2014)。
1.2.2   土壤容重传递函数模型(PTFs)的优化和新建
对已发表函数模型的优化主要依靠1stOpt软件(国产数学优化分析综合工具软件包)进行,其方法是基于【通用全局优化算法】(Universal Global Optimization – UGO)自动进行迭代计算,找出最优解。新函数模型的建立利用SPSS 22软件,主要应用了简单线性回归、逐步多元线性回归、多元非线性回归等方法。
2   结果与分析
2.1   石灰土土壤属性统计特征
对石灰土样本数据进行统计分析,结果表明(表1):石灰土的不同土壤属性均存在较高的异质性,其中,有机质含量、土层深度、全氮含量、全磷含量、海拔以及碳酸钙含量的异质性超过了60%,容重、砂粒、粉粒、粘粒、全钾含量、年降水量以及年均温的异质性超过15%,PH值的异质性最低为6%。这一现象说明喀斯特地区石灰土的土壤属性随着样点位置的改变具有显著差异性,土壤属性较高的异质性也证明了重新以石灰土亚类进行容重函数模型建立的必要性。
表1   石灰土的土壤属性数据的统计特征
土壤属性
Soil property
统计特征 Statistics character
样本数
N
最小值
Minimum
最大值
Maximum
平均值
Mean
标准偏差
SD
变异系数
CV
容重BD(g·cm-3)1040.751.761.280.170.13
有机质OM(%)1040.2052.3710.302.070.73
深度depth(cm)1046.00110.0045.1730.630.68
砂粒sand(%)1009.0064.7031.4611.980.38
粉粒silt(%)10013.9967.2034.719.940.29
粘粒clay(%)1009.0067.2033.8311.790.35
全氮TN(%)1040.030.670.170.120.71
全磷TP(%)1040.020.170.060.040.67
全钾TK(%)930.423.241.710.730.43
酸碱度(H2O)pH1046.008.407.380.460.06
海拔height(m)104120.003450.001086.62823.210.76
年降水precipitation(mm)102620.061609.001124.71236.530.21
年均温temperature(℃)1029.0019.0014.912.370.16
土壤质地Soil texture104
土地利用 Land use102
碳酸钙含量CaCO3 content(%)460.1215.592.973.251.09
单就石灰土容重的统计结果分析可知,其最小值为0.75 g·cm-3,最大值为1.75 g·cm-3,变异系数为0.13,存在中等程度变异。K-S检验统计量为0.082(P=0.083>0.05),其峰度和偏度分布为0.312和-0.342,结果表明样本BD数据在数值分布上符合正态分布规律,但峰值较高(表2,图1)。在垂直剖面上,土壤容重随土层深度的增加有增大的趋势,但其关系并不明显(R2=0.24),因此,本研究没有从深度角度划分,只从土壤亚类进行划分对石灰土的容重函数模型进行了探讨。


图1   南方喀斯特石灰土容重频率分布图
Fig.1 Distribution of soil bulk density in karst limestone soil of South China


图2   垂直剖面上土壤容重分布特征
Fig.2 Distribution characteristics of soil bulk density invertical profile
2.2   已发表土壤容重传递函数(PTFs)的适用性验证
为检验国内外已发表的PTFs是否适用于中国南方喀斯特石灰土的容重预测,本文整理了20世纪60年代至今的国内外发表的部分土壤容重PTFs,利用石灰土样本数据代入原函数模型,依据选取的验证指标进行验证(表2),PTFs对BD变化的解释度从0 %─38.9 %。综合各指标看,只有Shiri et al(2017)、韩光中等(2016)-a和韩光中等(2016)-c三个模型的R2值较高都在0.3以上,且MPE、SDPE和RMSPE的值也相对较小,分析认为可能是这三种模型在土壤类型上接近本研究区的土壤类型的原因。总之,结论认为国内外已发表的土壤容重传递函数模型并不适用于喀斯特石灰土地区,这一结果同时也验证了Brahim et al(2012)、De Vos et al(2005)及Martín et al(2017)的研究结论,土壤容重传递函数的应用具有一定的农业气候条件和地理空间限制。此外,分析发现在国内外土壤容重传递函数模型的建立过程中,土壤有机质(有机碳)、土壤粒度(砂粒、粉砂粒、粘粒)、土层深度等是主要的影响因子(Kaur et al, 2002; Howard et al, 1995; 门明新等, 2008)。
表2   国内外已发表土壤容重传递函数(PTFs)在本研究中数据的适用性
PTFs函数方程Functional equation研究区Research areas土壤类型Soil typeMPESDPERMSPER2
Curtis and Post(1964)美国,富蒙特州灰壤、棕色灰化土-0.1550.1740.0540.194
Jeffrey(1970)澳大利亚、英国多种类0.1790.2110.0770.227
Alexander(1980)加拿大、美国高山土壤-0.0020.1710.0290.219
Federer(1983)美国,新罕布什尔州砂壤土-0.0860.1920.0440.219
Manrique and Jones(1991)-a美国,夏威夷岛、波多黎各多种类-0.0140.1720.0290.219
Manrique and Jones(1991)-b美国,夏威夷岛、波多黎各多种类0.0360.1770.0330.192
Federer et al.(1993)新英格兰、美国砂壤土-0.1780.1810.0650.226
Tomasella and Hodnett(1998)巴西多种类-0.1480.1900.0580.045
Leonavičiute(2000)-a立陶宛多种类0.2050.1870.0770.181
Leonavičiute(2000)-b立陶宛多种类0.2460.1760.0920.141
Kaur et al.(2002)]澳大利亚多种类-0.3160.2380.0240.104
Prévost(2004)-a加拿大,魁北克多种类-0.0490.1800.0350.216
Benites et al.(2007)-a巴西多种类0.2440.1870.0950.117
Benites et al.(2007)-b巴西多种类0.2130.1910.0820.038
Perie and Ouimet(2008)加拿大,魁北克壤土、砂壤土-0.0190.2120.0450.230
Han et al.(2012)-a中国多种类0.1630.1760.0580.230
Han et al.(2012)-b中国多种类-0.0290.1740.0310.221
Han et al.(2012)-c中国多种类0.2560.1690.0940.221
Brahim et al.(2012)-a突尼斯多种类0.1370.1800.0510.197
Brahim et al.(2012)-b突尼斯多种类(土深0-40cm)0.3060.1960.1320.059
Brahim et al.(2012)-c突尼斯多种类(土深>40cm)0.5210.1250.2870.283
Hong et al.(2013)-a韩国火山灰土-0.3240.1720.1350.213
Hong et al.(2013)-b韩国矿物质土-0.0920.1940.0460.000
Al-Qinna and Jaber(2013)-a约旦多种类0.3460.1710.1490.213
Al-Qinna and Jaber(2013)-b约旦多种类0.3190.1670.1290.226
Al-Qinna and Jaber(2013))-c约旦多种类-0.0920.1810.0410.213
Al-Qinna and Jaber(2013)-d约旦多种类0.1290.2160.0630.092
Al-Qinna and Jaber(2013)-e约旦多种类0.1690.2140.0740.000
Al-Qinna and Jaber(2013)-f约旦多种类0.1360.2150.0650.095
Nanko et al.(2014)-a日本火山灰土-0.3640.1670.1610.222
Nanko et al.(2014)-b日本火山灰土-0.3200.1670.1300.220
Abdelbaki(2016)美国多种类0.0910.1730.0380.198
Shiri et al.(2017)伊朗西南部粗骨土、始成土、盐土-0.1480.1910.0580.389
韩光中,等.(2016)-a中国均腐土0.1560.1660.0520.310
韩光中,等.(2016)-b中国淋溶土-0.0100.1820.0330.172
韩光中,等.(2016)-c中国雏形土0.0320.1460.0220.323
注Note:BD:容重 Bulk density(g·cm-3);OM:有机质含量 Organic matter content( :g·g-1;其余单位为% Units in others are %);OC:有机碳含量 Organic carbon content(%);Sand:砂粒Sand(%);silt:粉砂粒Silt(%);clay:粘粒Clay(%); TN:全氮含量 Total nitrogen content(%);depth:土壤深度 Soil depth(cm);PH:土壤酸碱度 Soil pH; CaCO3:土壤碳酸钙含量 Soil calcium carbonate content(%)。
2.3   南方喀斯特石灰土BD预测的影响因子
经Pearson相关分析可知(表3),并非所有土壤属性都与容重之间具有显著的相关性,且石灰土类及不同亚类的容重与各属性变量间的相关性存在差异。基于相关分析结果,在建立石灰土亚类土壤容重传递函数(PTFs)模型时,只选取各类石灰土中与容重(BD)显著相关的属性作为输入变量进行回归分析和拟合,最终筛选出预测精度较高且输入变量相对简单的模型,以提高模型的实用性。
表3   石灰土及其各亚类BD与其它属性变量间的相关系数
属性变量
Property variables
石灰土及各亚类容重 BD of lime soil and each sub-category
石灰土
Limestone soil
黑色石灰土
Black limestone soil
棕色石灰土
Brown limestone soil
黄色石灰土
Yellow limestone soil
红色石灰土
Red limestone soil
OM-0.442**-0.650**-0.396*-0.555**-0.634**
Depth0.414**0.561*0.453*0.507**0.529*
Sand-0.237*-0.549*-0.535*-0.079-0.250
Silt0.0510.2250.545*-0.0350.137
Clay0.196*0.3800.3060.1220.006
TN-0.388**-0.631**-0.440*-0.518**-0.473*
TP-0.303**-0.213-0.414-0.253-0.330
TK0.128-0.014-0.3320.498**0.186
pH-0.0700.051-0.3400.1940.130
Height-0.1720.323-0.167-0.1680.056
P-0.015-0.3050.030-0.2660.093
L-0.084-0.3580.076-0.141-0.106*
T0.217*-0.3600.2240.405*-0.048
CaCO30.1830.780*-0.234-0.2440.240
ZD-0.0170.0540.054-0.0170.127
注Note:**. 相关性在 0.01 水平上显著(双尾检验)The correlation was significant at 0.01 level (two-tailed test);*. 相关性在 0.05 水平上显著(双尾检验)The correlation was significant at 0.05 level (two-tailed test);OM:土壤有机质含量 Soil organic matter content;Depth:土壤深度 Soil depth;Sand:砂粒sand;Silt:粉粒silt;Clay:粘粒clay;TN:全氮含量 Total nitrogen content;TP:全磷含量 Total phosphorus content;TK:全钾含量Total potassium content;pH:酸碱度 pH;Height:海拔高度 Altitude;P:年均降水量 Average annual rainfall;L:土地利用类型 Land use type;T:年均温 Average annual temperature;CaCO3:碳酸钙含量 Calcium carbonate content;ZD:土壤质地 Soil texture。
有学者认为,自然状态下的BD在成土母质、成土过程、气候、生物作用及耕作等的环境影响下具有高度的变异性(郑纪勇等,2004)。此外,Calhoun et al(2001)研究认为土壤有机质含量和土壤质地对BD变化的解释度一般不低于50%。本研究结果显示,在与BD显著相关的属性因子中,土壤有机质和土壤深度对BD的解释度较高,石灰土中分别到达44.2%和41.4%,黑色石灰土亚类中分别达到65%和56.1%,棕色石灰土亚类中分别达到39.6%和45.3%,黄色石灰土亚类中分别达到55.5%和50.7%,红色石灰土亚类中分别达到63.4%和52.9%。土壤有机质会对土壤的孔隙度和土壤持水量产生影响,从而对BD产生影响(王益等,2005)。土壤深度增加,土壤有机质含量会逐渐降低(Hobley et al, 2013; Jobbágy et al, 2000)以及土体在超负荷压力下变得越来越紧实(Tranter et al, 2007),是造成BD随土壤深度增加而增大的主要原因。
2.4   喀斯特石灰土最优容重传递函数模型(PTFs)的建立
2.4.1   已发表PTFs的优化
根据本研究区石灰土属性数据,对Shiri et al(2017)、韩光中等(2016)-a和韩光中等(2016)-c三个模型重新进行优化拟合,以使其预测精度进一步提高。对比表3、表4和图3发现,优化后的模型精度均得到提高,其中模型“韩光中等(2016)-a*”提高程度最大,而Shiri et al(2017)*与韩光中等(2016)-c*的精度提高相对较少。
表4   优化后的土壤容重传递函数模型(PTFs)
优化PTFs Optimized PTFs函数方程Functional equationMPESDPERMSPER2
Shiri et al(2017)*0.0000.0990.0090.427
韩光中等(2016)-a* (黑色石灰土)0.0000.0960.0090.934
韩光中等(2016)-c* (黄色石灰土)0.0000.1280.0160.414
注 Note:韩光中等(2016)-a和韩光中等(2016)-c两个模型在原作者文章中分别是均腐土和雏形土的容重传递函数模型,改进后分别对应本研究区的黑色石灰土和黄色石灰土。其余注释同表2。Han et al (2016) -a and Han et al (2016) -c two models in the original author's article are the pedo-transfer function models of soil bulk density of isohumosols and cambosols respectively. After improvement, they correspond to the black limestone and the yellow limestone in this study area respectively. The rest of the comments are the same as in Table 2.


图3   优化前后土壤容重实测值与预测值的比较
Fig.3 Comparison of measured value and predicted value of soil bulk density before and after optimization
2.4.2   石灰土亚类土壤容重传递函数(PTFs)的新建
由于成土条件的差异,同为石灰土类但有不同的土壤发育阶段和成土过程,导致土壤属性出现分异,不同亚类的石灰土也具有质的差异(庞纯焘和宋铭荷,1990)。有研究表明,基于土壤分类的数据分组可以明显提高土壤PTFs的预测精度(韩光中等,2016)。所以,本文将基于石灰土亚类进行新函数模型的建立。为进一步提高喀斯特地区石灰土容重预测模型的精度,需要根据本研究区的土壤属性特征重新拟合得到最佳PTFs(表5和图4)。与已发表PTFs和优化后的PTFs相比,基于石灰土亚类分类的PTFs的精确度更高,对于南方喀斯特地区的石灰土BD的预测具有更高的适用性。各石灰土亚类分别留取预测精度较高的两个模型,从模型的方程式来看,不同石灰土亚类土壤BD的影响因素存在差异。黑色石灰土BD主要受土壤有机质含量(OM)因素控制;棕色石灰土BD主要受土壤砂粒含量(Sand)、土壤粉砂粒含量(Silt)和土壤全氮含量(TN)等影响因素的控制;黄色石灰土BD主要受OM、土壤全钾含量(TK)以及土壤深度(Depth)的影响控制;而红色石灰土BD则主要受OM、Depth等的影响。总之,OM是影响石灰土容重预测的关键因素,这一点与大多数学者的结论一致。不同石灰土亚类的PTFs的预测精确度也存在差异,顺序为:黑色石灰土 > 棕色石灰土 > 红色石灰土 > 黄色石灰土。研究认为其原因有二:一是可能与各亚类土壤的成土阶段有关,黑色石灰土是初期阶段形成物,继续发育顺序为棕色石灰土— 黄色石灰土— 红色石灰土(袁红等, 2016),随着土壤发育阶段的不同,土壤的淋溶作用和淋溶时间加强,土壤元素迁移流失,元素的迁移流失可能会改变土壤属性间的相关性;二是因为土壤OM含量对石灰土BD的影响,土壤OM含量越高对BD值的影响越大(Howard et al, 1995),本文中OM平均含量为:黑色石灰土(4.81%)> 棕色石灰土(2.67%)> 红色石灰土(2.64%)> 黄色石灰土(2.09%)。
表5   适宜南方喀斯特地区石灰土亚类土壤容重传递函数模型(PTFs)
模型Models方程Equations模型精度验证指标
Model accuracy verification index
MPESDPERMSPER2
黑色石灰土 Black limestone soil
Black L-aLnBD=0.791508(1/OM)+0.0224830.0000.0690.0050.925
Black L-bLnBD=0.3951–0.03283OM0.0000.1000.0100.938
棕色石灰土 Brown limestone soil
Brown L-aLnBD= –0.01077 – 0.00328sand+0.01837silt – 1.65809TN0.0000.0980.0090.938
Brown L-bLnBD= –0.15259+0.020988silt – 1.95159TN0.0000.1040.0110.937
黄色石灰土 Yellow limestone soil
Yellow L-aBD=1.467034+0.001305depth – 0.18159OM0.50.0000.1430.0210.361
Yellow L-bBD=1.383977 – 0.18525OM0.5+0.86638TK0.0000.1220.0150.464
红色石灰土 Red limestone soil
Red L-aLnBD=0.474332 – 0.11598OM0.50.0000.0690.0050.853
Red L-bLnBD= –0.07315 – 0.56102OM0.5 – 0.09365depth0.0000.0670.0040.848
注 Note:BD:容重 Bulk density(g·cm-3);OM:有机质含量 Organic matter content(%);sand:砂粒含量Sand content(%);silt:粉砂粒含量Silt content(%);TN:全氮含量Total nitrogen content(%) ;TK:全钾含量Total potassium content(%); depth:土壤深度 Soil depth(cm)。


图4   南方喀斯特地区石灰土亚类容重实测值与预测值的比较
Fig. 4 Comparison of measured value and predicted value of limestone soil bulk density in South Karst region
3   结论
本文首次探讨了喀斯特石灰土容重预测模型及影响因素 ,对于喀斯特地区的石灰土容重预测工作有了更进一步的认识。(1) 国内外已发表的土壤容重传递函数模型(PTFs)不能直接用来对喀斯特石灰土容重进行预测。通过对其中三个预测精度相对较高的模型进行优化发现,优化后的模型预测精度明显提高。(2)土壤容重预测函数的适用性不仅与研究区域有关,同时也与研究的土壤类型有关。为进一步提高喀斯特石灰土容重预测的准确度,本文基于石灰土亚类进一步进行了PTFs的建立,结果表明基于石灰土亚类拟合的PTFs更适合本研究区的BD预测研究。(3)不同石灰土亚类的容重(BD)预测的影响因子存在差异。总体上,土壤有机质含量(OM)是影响石灰土BD预测的关键因素;此外,粉粒含量(silt)、砂粒含量(sand)和全氮含量(TN)是棕色石灰土BD预测的影响因素,全钾含量(TK)和土壤深度(depth)是黄色石灰土BD预测的影响因素,土壤深度(depth)是红色石灰土BD预测的影响因素。因此,在今后喀斯特地区的土壤BD预测研究中,应同时考虑地域差异性和土壤类型差异。本文也存在着一定的不足之处,由于目前关于南方喀斯特石灰土土壤属性的完整数据存在严重不足,导致本文章中各类石灰土的实验数据较少(但各组数据也均超过20组数据),同时由于模型适用的石灰土类型及区域已经限定的较为详细,故本文中不再对模型进行适用性的验证。
致谢
Abdelbaki A M. 2016. Evaluation of pedotransfer functions for predicting soil bulk density for U.S. soils[J]. Ain Shams Engineering Journal, 12(02):19.
Alexander E B. 1980. Bulk densities of California soils in relation to other soil properties[J]. Soil Science Society of America Journal, 44: 689 692.
Al-Qinna M I, Jaber S M. 2013. Predicting soil bulk density using advanced Pedotransfer functions in an arid environment[J]. Trans ASABE, 56, (3): 96376.
Behzad N, Olivier F, Dino T. 2015. Coupling published pedotransfer functions for the estimation of bulk density and saturated hydraulic conductivity in stony soils[J]. Catena, 131: 99‒108.
Benites VM, Machado PLOA, Fidalgo ECC, et al. 2007.Pedotransfer functions for estimating soil bulk density from existing soil survey reports in Brazil[J]. Geoderma, 139(1–2):90 ‒ 97.
Brahim N, Bernoux M, Gallali T. 2012. Pedotransfer functions to estimate soil bulk density for northern Africa: Tunisia case[J]. J. Arid Environ, 81: 77 – 83.
Calhoun F G, Smeck N E, Slater B L, et al. 2001.Predicting bulk density of Ohio soils from morphology, genetic principles, and laboratory characterization data[J]. Soil Science Society of America Journal, 65(3): 811819.
Curtis R O, Post B W. 1964. Estimating bulk density from organic matter content in some Vermont forest soils[J]. Soil. Sci. Soc. Am. Pro., 28: 285286.
De Vos B, Van M M, Quataert P, et al. 2005.Predictive quality of pedotransfer functions for estimating bulk density of forest soils[J]. Soil Science Society of America Journal, 69: 500 – 510.
Federer C A. 1983. Nitrogen mineralization and nitrification: depth variation in four New England forest soils[J]. Soil Science Society of America Journal, 47: 10081014.
Federer C, Turcotte D, Smith C. 1993. The organic fraction-bulk density relationship and the expression of nutrient content in forest soils[J]. Canadian Journal of Forest Research, 23: 10261032.
Han G Z, Zhang G L, Gong Z T, et al. 2012.Pedotransfer functions for estimating soil bulk density in China[J]. Soil Science, 177(3): 158 – 164.
Hobley E, Willgoose G R, Frisia S, et al. 2013. Environmental and site factors controlling the vertical distribution and radiocarbon ages of organic carbon in a sandy soil[J]. Biology and Fertility of Soils, 49( 8): 10151026.
Hong S Y, Minasny B, Han K H, et al. 2013.Predicting and mapping soil available water capacity in Korea[J]. Peer J, 1: e71.
Howard P J A, Loveland P J, Bradley R I. 1995.The carbon content of soil and its geographical distribution in Great Britain [J]. Soil Use Manage, 11: 915.
Huntington T G, Johnson C E, Johnson A H, et al. 1989. Carbon, organic matter, and bulk density relationships in a forested spodosol[J]. Soil Science, 148: 380 – 386.
Jeffrey D W. 1970. A note on the use of ignition loss as a means for the approximate estimation of soil bulk density[J]. Journal of Ecology, 58: 297299.
Jobbágy E G, Jackson R B. 2000. The vertical distribution of soil organic carbon and its relation to climate and vegetation[J]. Ecological Applications, 10(2): 423436.
Kaur R, Kumar S, Gurung H P. 2002. A pedo-transfer function (PTF) for estimating soil bulk density from basic soil data and its comparison with existing PTFs[J]. Aust J Soil Res, 40: 847858.
Leonavičiute N. 2000. Predicting soil bulk and particle densities by pedotransfer functions from existing soil data in Lithuania[J]. Geoandgrafijos Metraštis, 33: 317330.
Manrique L, Jones C. 1991. Bulk density of soils in relation to soil physical and chemical properties[J]. Soil Science Society of America Journal, 55(2): 476481.
Martín M A, Miguel R F, Javier T. 2017. Estimating soil bulk density with information metrics of soil texture[J]. Geoderma, 287: 66 – 70.
Milne E, Al Adamat R, Batjes N H, et al. 2007. National and sub-national assessments of soil organic carbon stocks and changes: the GEFSOC modelling system[J]. Agric Ecosyst Environ, 122: 3 – 12.
Moreels E, De Neve S, Hofman G, et al. 2003. Simulating nitrate leaching in bare fallow soils: a model comparison[J]. Nutrient Cycling in Agroecosystems, 67:137–144.
Nanko K, Ugawa S, Hashimoto S, et al. 2014. A pedotransfer function for estimating bulk density of forest soil in Japan affected by volcanic ash[J]. Geoderma, 213: 36 – 45.
Perie C, Ouimet R. 2008. Organic carbon, organic matter and bulk density relationships in boreal forest soils[J]. Canadian Journal of Soil Science, 88: 315325.
Prévost, M. 2004. Predicting soil properties from organic matter content following mechanical site preparation of forest soils[J]. Soil Science Society of America Journal, 68: 943949.
Shiri J, Ali K, Ozgur K, et al. 2017. Modeling soil bulk density through a complete data scanning procedure: Heuristic alternatives[J]. Journal of Hydrology, 549: 592602.
Suuster E, Ritz C, Roostalu H, et al .2011. Soil bulk density pedotransfer functions of the humus horizon in arable soils[J]. Geoderma, 163(1–2):74 – 82.
Tomasella J, Hodnett M G. 1998. Estimating soil water retention characteristics from limited data in Brazilian Amazonia[J]. Soil Science, 163: 190202.
Tranter G, Minasny B, Mcbratney A B, et al. 2007. Building and testing conceptual and empirical models for predicting soil bulk density[J]. Soil Use and Management, 23(4): 437443.
Wang Y Q, Shao M A, Liu Z P, et al. 2014.Prediction of Bulk Density of Soils in the Loess Plateau Region of China[J]. Sutv Geophys, 35: 395413.
Wen D, He N P. 2016. Spatial patterns and control mechanisms of carbon storage in forest ecosystem: evidence from the north-south transect of eastern China[J]. Ecol Indic, 61: 960 – 967.
Wiesmeier M, Sporlein P, Geuss U, et al. 2012. Soil organic carbon stocks in southeast Germany (Bavaria) as affected by land use, soil type and sampling depth[J]. Global Change Biol, 18(7): 2233 – 2245.
Xu L, He N P, Yu G R. 2016. Methods of evaluating soil bulk density: Impact on estimating large scale soil organic carbon storage [J]. Catena, 144: 94 – 101.
Yi X S, Li G S, Yin Y Y. 2016. Pedotransfer functions for estimating soil bulk density: A case study in the Three-River Headwater region of Qinghai Province, China[J]. Pedosphere, 26(3): 362 – 373.
陈曦, 彭稳, 曹建华. 2012. 典型峰林平原土壤有机碳储量和分布特征研究[J]. 科技通报, 28(2): 167173. [ Chen X, Peng W, Cao J H. 2012. Study on the distribution and reserves features of soil organic carbon in the typical peak-forest plain, Southwest China[J]. Bulletin of science and technology, 28(2): 167173.]
韩光中, 王德彩, 谢贤健. 2016. 中国主要土壤类型的土壤容重传递函数研究[J]. 土壤学报, 53(1): 93 102. [Han G Z, Wang D C, Xie X J. 2016.Pedotransfer Functions for Prediction of Soil Bulk Density for Major Types of Soils in China[J]. Acta Pedologica Sinica, 53(1): 93102.]
解宪丽, 孙波, 周慧珍, 等. 2004. 中国土壤有机碳密度和储量的估算与空间分布分析[J]. 土壤学报, 41(1): 35 43. [Xie X L, Sun B, Zhou H Z, et al. 2004. Organic Carbon Density And Storage In Soils Of China And Spatial Analysis[J]. Acta Pedologica Sinica, 41(1): 3543.]
刘方, 王世杰, 罗海波, 等. 2008. 喀斯特森林生态系统的小生境及其土壤异质性[J]. 土壤学报, 45(6): 10551062. [ Liu F, Wang S J, Luo H B, et al. 2008. Micro-habitats in Karst forest ecosystem and variability of soils[J]. Acta pedologica sinica, 45(6): 10551062.]
门明新, 彭正萍, 许皞, 等. 2008. 河北省土壤容重的传递函数研究[J]. 土壤通报, 39(1): 33 ‒ 37. [ Men M X, Peng Z P, Xu H, et al. 2008. Investigation on Pedo-transfer function for estimating soil bulk density in Hebei province[J]. Chinese Journal of Soil Science, 39(1): 33 ‒ 37.]
倪九派, 袁道先, 谢德体,等. 2009. 重庆岩溶区土壤有机碳库的估算及其空间分布特征[J]. 生态学报, 29(11): 62926301. [Ni J P, Yuan D X, Xie D T, et al. 2009. Estimation of soil organic carbon storage and the characteristic of carbon spatial distributions in Karst area, Chongqing, China[J]. Acta Ecologica Sinica, 29(11): 62926301.]
潘根兴. 1999. 中国土壤有机碳和无机碳库量研究[J]. 科技通报, 15(5): 330332. [Pan G X. 1999. Study on Soil Organic Carbon and Inorganic Carbon in China[J]. Bulletin of Science and Technology, 15 (5): 330 332.]
庞纯焘, 宋铭荷. 1990. 贵州土壤分类依据与土壤分类单位系统[J]. 贵州教育学院学报(自然科学版), 3(4): 4961. [Pang C T, Song M H. 1990. Guizhou Soil Classification Basis and Soil Classification Unit System[J]. Journal of Guizhou Educational College (Natural Science Edition ), 3(4): 4961.]
王益, 王益权, 刘军, 等. 2005.黄土地区影响土壤膨胀因素的研究[J]. 干旱地区农业研究, 23(5) : 9397. [Wang Y, Wang Y Q, Liu J, et al. 2005. Study on soil swelling factors in loessal areas[J]. Agricultural Research in the Arid Areas, 23(5): 9397.]
易小波, 邵明安, 赵春雷, 等. 2017. 黄土高原南北样带不同土层土壤容重变异分析与模拟[J]. 农业机械学报, 48(4): 198 205. [Yi X B, Shao M A, Zhao C L, et al. 2017. Variation Analysis and Simulation of Soil Bulk Density within Different Soil Depths along South to North Transect of Loess Plateau[J]. Transactions of the Chinese Society for Agricultural Machinery, 48(4): 198205.]
袁红,谢红霞,罗兰芳,等. 2016.南方石灰岩土壤发生特性和系统分类研究[J]. 中国农学通报, 32(21):124‒128.[ Yuan H, Xie H X, Luo L F, et al. 2016.Genetic Characteristics and Classification of Limestone Soil in South China[J]. Chinese Agricultural Science Bulletin, 32 (21): 124128.]
张美良, 邓自强. 1994. 我国南方喀斯特地区的土壤及其形成[J]. 贵州工学院学报, 23(1): 6775. [ Zhang M L, Deng Z Q. 1994. The soil and soil-forming processes in Karst area of South China[J]. Journal of Guizhou institute of technology, 23(1): 6775.]
张勇, 史学正, 赵永存, 等. 2008. 滇黔桂地区土壤有机碳储量与影响因素研究[J]. 环境科学, 29(8): 23142319. [ Zhao Y, Shi X Z, Zhao Y C, et al. 2008. Estimates and affecting factors of soil organic carbon storages in Yunnan-Guizhou-Guangxi region of China[J]. Environmental science, 29(8): 23142319.]
张珍明, 周运超, 田潇, 等. 2017. 喀斯特小流域土壤有机碳空间异质性及储量估算方法[J]. 生态学报, 37(22):113. [ Zhang Z M, Zhou Y C, Tian X, et al. 2017. Study on spatial heterogeneity and reserve estimation of soil organic carbon in a small karst catchment[J]. Acta ecologica sinica, 37(22):113.]
郑纪勇, 邵明安, 张兴昌. 2004. 黄土区坡面表层土壤容重和饱和导水率空间变异特征[J]. 水土保持学报, 18(3): 5356. [Zheng J Y, Shao M A, Zhang X C. 2004. Spatial Variation of Surface Soil’s Bulk Density and Saturated Hydraulic Conductivity on Slope in Loess Region[J] . Journal of Soil and Water Conservation, 18(3): 5356.]
稿件与作者信息
李颖1,2,3,刘秀明13*,王世杰1,3,周德全2,罗慧1,3
LI Ying1,2,3, LIU Xiuming1,3*, WANG Shijie1,3, ZHOU Dequan2, LUO Hui1,3
国家重点研发计划项目(2016YFC0502301);贵州省高层次创新型人才培养计划“十”层次人才项目(黔科合平台人才[2016]5648)
National key research and development project of China (2016YFC0502301); Guizhou high-level innovative talent training program “ten” level tatents program (Talents of the cooperation platform of guizhou science and technology[2016]5648)
出版历史
出版时间: 2018年5月18日 (版本3
参考文献列表中查看
地球环境学报
Journal of Earth Environment