基于RBF组合模型的山地红壤有机质含量光谱估测
|
摘要
【目的】探讨组合模型在山地红壤有机质含量高光谱估算中应用的可行性,以期为土壤有机质含量估测提供基础数据和科学依据。【方法】基于山地红壤光谱的全波段(400~2 450 nm)研究范围,选择偏最小二乘回归(PLSR)、BP神经网络(BP)和支持向量机回归分析(SVMR)3种单一高光谱估测模型,分别获得预测结果,并重构预测结果数据,以绝对误差和最小为目标,计算固定权重与不固定权重两种组合模型的权重值,并基于径向基函数(RBF)神经网络法建立组合模型,探讨不同赋权方法与是否重构数据条件下的最优组合模型。通过均方根误差(RMSE)、预测偏差比(RPD)和决定系数(R2)评价山地红壤有机质含量的预测精度。【结果】单一预测模型中的SVMR估测精度最高,验证决定系数(R2)为0.64,均方根误差为9.76 g·kg-1,测定值标准差与标准预测误差的比值为1.67;在组合模型数据不重构的条件下,不定权组合模型要优于定权组合模型;在组合模型数据重构的条件下,定权组合模型要略优于不定权组合模型,估测精度相差不大;最优模型是数据重构定权组合模型,模型验证决定系数(R2)为0.87,均方根误差为7.91 g·kg-1,测定值标准差与标准预测误差的比值为2.06;组合模型验证精度优于单一模型,说明利用RBF组合模型估算山地红壤有机质含量是可行的。【结论】对山地红壤有机质含量的快速估测而言,单一模型具有操作简单、运算速度快等特点,因而具有较大应用价值,但组合模型能较大限度地利用各种预测样本信息,从而能有效减少应用单一模型时所受随机因素的影响,从而提高山地红壤有机质含量的估测精度。
【Objective】It is of great significance for rapid acquisition of soil fertility parameters using spectral technology to explore the quantitative estimation of the contents of soil organic matters.【Method】Two hundred and forty eight soil samples at 0-30 cm depth were collected from mountain red soil region of northern Jiangxi Province. The spectral reflectance of mountain red soil samples was measured by an ASD Field Spec3 instrument under laboratory conditions.Meanwhile the content of organic matters of each soil sample was analyzed using potassium dichromate external heating method in laboratory. Correlation analyses between raw mountain red soil spectral reflectance and organic matter content mountain red soil were conducted. Kennard-Stone algorithm was used to divide mountain red soil samples into calibration sets with 186 samples and prediction sets with 62 samples. Based on the full band( 400-2 450 nm) of mountain red soil spectra in this study,the partial least squares regression( PLSR),BP neural network( BP) and support vector machine regression( SVMR) were selected to obtain and rebuild the prediction result,and establish fixed weight and variable weights in combination model with the minimum absolute error sum as the goal. Then,the combination model can be built based on radial basis function( RBF) neural network. The optimal combination model with different weighting and with or without reconstructed data was investigated. Accuracies of the content predictions of organic matters of the mountain redsoil were evaluated by root mean squared error( RMSE),ratio of partial deviation( RPD) and determination coefficients( R2).【Result】The results showed the best single model is SVMR,its determination coefficient( R2),the root mean square error( RMSE) and the ratio of standard error of performance to standard deviation( RPD) value of validation set were 0. 64,9. 76 g·kg-1 and 1. 67. Under the condition of prediction data,the combination model of variable weights is better than the fixed weight combination model, while under the condition of prediction data reconstruction, the combination model of fixed weight is slightly better than the variable weight combination model,but the estimation accuracy of the two models is very close. The optimal model of the combination model is the reconstruction of fixed weight combination model,its determination coefficient( R2),the root mean square error( RMSE) and ratio of standard error of performance to standard deviation( RPD) value of validation set were respectively 0. 87,7. 91 g·kg-1 and 2. 06.Therefore,the accuracy of the RBF combination model is better than that of the single model,indicating that it is feasible to estimate the content of organic matters in mountain red soil by RBF combination model. 【Conclusion】The rapid content estimation of organic matters in mountain red soil showed that the single model has the characteristics of simple operation and fast calculation. Therefore,it has a greater value of application. However,the combination model was able to make full use of various sample information,therefore,effectively reducing the impacts of random factors on the use of single model,and hence to improve the precision of estimating the content of organic matters in mountain red soil.
【Objective】It is of great significance for rapid acquisition of soil fertility parameters using spectral technology to explore the quantitative estimation of the contents of soil organic matters.【Method】Two hundred and forty eight soil samples at 0-30 cm depth were collected from mountain red soil region of northern Jiangxi Province. The spectral reflectance of mountain red soil samples was measured by an ASD Field Spec3 instrument under laboratory conditions.Meanwhile the content of organic matters of each soil sample was analyzed using potassium dichromate external heating method in laboratory. Correlation analyses between raw mountain red soil spectral reflectance and organic matter content mountain red soil were conducted. Kennard-Stone algorithm was used to divide mountain red soil samples into calibration sets with 186 samples and prediction sets with 62 samples. Based on the full band( 400-2 450 nm) of mountain red soil spectra in this study,the partial least squares regression( PLSR),BP neural network( BP) and support vector machine regression( SVMR) were selected to obtain and rebuild the prediction result,and establish fixed weight and variable weights in combination model with the minimum absolute error sum as the goal. Then,the combination model can be built based on radial basis function( RBF) neural network. The optimal combination model with different weighting and with or without reconstructed data was investigated. Accuracies of the content predictions of organic matters of the mountain redsoil were evaluated by root mean squared error( RMSE),ratio of partial deviation( RPD) and determination coefficients( R2).【Result】The results showed the best single model is SVMR,its determination coefficient( R2),the root mean square error( RMSE) and the ratio of standard error of performance to standard deviation( RPD) value of validation set were 0. 64,9. 76 g·kg-1 and 1. 67. Under the condition of prediction data,the combination model of variable weights is better than the fixed weight combination model, while under the condition of prediction data reconstruction, the combination model of fixed weight is slightly better than the variable weight combination model,but the estimation accuracy of the two models is very close. The optimal model of the combination model is the reconstruction of fixed weight combination model,its determination coefficient( R2),the root mean square error( RMSE) and ratio of standard error of performance to standard deviation( RPD) value of validation set were respectively 0. 87,7. 91 g·kg-1 and 2. 06.Therefore,the accuracy of the RBF combination model is better than that of the single model,indicating that it is feasible to estimate the content of organic matters in mountain red soil by RBF combination model. 【Conclusion】The rapid content estimation of organic matters in mountain red soil showed that the single model has the characteristics of simple operation and fast calculation. Therefore,it has a greater value of application. However,the combination model was able to make full use of various sample information,therefore,effectively reducing the impacts of random factors on the use of single model,and hence to improve the precision of estimating the content of organic matters in mountain red soil.
引文
鲍士旦.1999.土壤农化分析.北京:中国农业出版社.(Bao S D.1999.Soil agrichemical analysis.Beijing:China AgriculturePress.[in Chinese])
代海波,单锐,刘文.2012.组合预测方法中权系数的应用研究.科学技术与工程,12(32):8483-8497.(Dai H B,Shan R,Liu W.2012.Application research on weight coefficients of the combination forecasting method.Science Technology and Engineering,12(32):8483-8497.[in Chinese])
丁美青,肖红光,陈松岭,等.2012.基于BP神经网络的土地开发整理区土壤有机质含量遥感定量反演.湘潭大学自然科学学报,34(2):103-106.(Ding M Q,Xiao H G,Chen S L,et al.2012.Remote sensing quantitative of soil organic matter content in the land development and consolidation region based on BP neural network.Natural Science Journal of Xiangtan University,34(2):103-106.[in Chinese])
方少文,杨梅花,赵小敏,等.2014.红壤区土壤有机质光谱特征与定量估算---以江西省吉安县为例.土壤学报,51(5):1003-1010.(Fang S W,Yang M H,Zhao X M,et al.2014.Spectral characteristics and quantitative estimation of SOM in red soil typical of Ji’an county,Jiangxi Province.Acta Pedologica Sinica,51(5):1003-1010.[in Chinese])
付华,许凡,徐耀松,等.2013.基于变权重RBF组合模型的煤与瓦斯突出强度预测.中国安全科学学报,23(8):65-70.(Fu H,Xu F,Xu Y S,et al.2013.Prediction of goal and gas outburst intensity based on variable weight RBF combination model.China Safety Science Journal,23(8):65-70.[in Chinese])
高志海,白黎娜,王琫瑜,等.2011.荒漠化土地土壤有机质含量的实测光谱估测.林业科学,47(6):9-16.(Gao Z H,Bai L N,Wang B Y,et al.2011.Estimation of soil organic matter content in desertified lands using measured soil spectral data.Scientia Silvae Sinicae,47(6):9-16.[in Chinese])
黄仁东,张海彬,杨志辉,等.2015.基于多准则的组合预测模型权重研究及其应用.中南大学学报:自然科学版,46(5):1778-1785.(Huang R D,Zhang H B,Yang Z H,et al.2015.Research and application of multi-criteria combination forecast mode.Journal of Central South University:Science and Technology,46(5):1778-1785.[in Chinese])
纪文君,李曦,李成学,等.2012.基于全谱数据挖掘技术的土壤有机质高光谱预测建模研究.光谱学与光谱分析,32(9):2393-2398.(Ji W J,Li X,Li C X,et al.2012.Using different data mining algorithms to predict soil organic matter based on visible near infrared spectroscopy.Spectroscopy and Spectral Analysis,32(9):2393-2398.[in Chinese])
刘磊,沈润平,丁国香.2011.基于高光谱的土壤有机质含量估算研究.光谱学与光谱分析,31(3):762-766.(Liu L,Shen R P,Ding G X.2011.Studies on the estimation of soil organic matter content based on hyper-spectrum.Spectroscopy and Spectral Analysis,31(3):762-766.[in Chinese])
沈润平,丁国香,魏国栓,等.2009.基于人工网络的土壤有机质含量高光谱反演.土壤学报,46(3):391-397.(Shen R P,Ding G X,Wei G S,et al.2009.Retrieval of soil organic matter content from hyper-spectrum based on ANN.Acta Pedologica Sinica,46(3):391-397.[in Chinese])
史舟,王乾龙,彭杰,等.2014.中国主要土壤高光谱反射特性分类与有机质光谱预测模型.中国科学:地球科学,44(5):978-988.(Shi Z,Wang Q L,Peng J,et al.2014.Debelopment of a national VNIR soil-spectral library for soil classification and prediction of organic matter concentrations.Scienc China:Earth Sciences,44(5):978-988.[in Chinese])
田永超,张娟娟,姚霞,等.2012.基于近红外光声光谱的土壤有机质含量定量建模方法.农业工程学报,28(1):145-152.(Tian Y C,Zhang J J,Yao X,et al.2012.Quantitative modeling method of soil organic matter content based on near-infrared photoacoustic spectroscopy.Transactions of the Chinese Society of Agricultural Engineering,28(1):145-152.[in Chinese])
卫敏,余乐安.2012.具有最优学习率的RBF神经网络及其应用.管理科学学报,15(4):50-57.(Wei M,Yu L A.2012.A RBF neural network with optimum learning rates and its appliction.Journal of Management Sciences in China,15(4):50-57.[in Chinese])
杨扬,高小红,贾伟,等.2015.三江源不同土壤类型有机质含量高光谱反演.遥感技术与应用,30(1):186-198.(Yang Y,Gao X H,Jia W,et al.2015.Hyperspectral retrieval of soil organic matter for different soil types in the Three-River Headwaters region.Remote Sensing Technology and Application,30(1):186-198.[in Chinese])
于雷,洪永胜,耿雷,等.2015.基于偏最小二乘回归的土壤有机质含量高光谱估算.农业工程学报,31(14):103-109.(Yu L,Hong Y S,Geng L,et al.2015.Hyperspectral estimation of soil organic matter content based on partial least squares regression.Transactions of the Chinese Society of Agricultural Engineering,31(14):103-109.[in Chinese])
于雷,洪永胜,周勇,等.2016a.高光谱估算土壤有机质含量的波长变量筛选方法.农业工程学报,32(13):95-102.(Yu L,Hong Y S,Zhou Y,et al.2016a.Wavelength variable selection methods for estimation of soil organic matter content using hyperspectral technique.Transactions of the Chinese Society of Agricultural Engineering,32(13):95-102.[in Chinese])
于雷,洪永胜,周勇,等.2016b.连续小波变换高光谱数据的土壤有机质含量反演模型构建.光谱学与光谱分析,36(5):1428-1433.(Yu L,Hong Y S,Zhou Y,et al.2016b.Inversion of soil organic matter content using hyperspectral data based on continuous wavelet transformation.Spectroscopy and Spectral Analysis,36(5):1428-1433.[in Chinese])
袁征,李希灿,于涛,等.2014.高光谱土壤有机质估测模型对比研究.测绘科学,39(5):117-120.(Yuan Z,Li X C,Yu T,et al.2014.Contrast research on soil organic matter estimation model using hyper-spectral data.Science of Surveying and Mapping,39(5):117-120.[in Chinese])
曾胤,陆宇振,杜昌文,等.2014.应用红外光声光谱技术及支持向量机模型测定土壤有机质含量.土壤学报,51(6):1262-1269.(Zeng Y,Lu Y Z,Du C W,et al.2014.Applying infrared photoacoustic spectroscopy and support vector machine model to quantify soil organic matter content.Acta Pedologica Sinica,51(6):1262-1269.[in Chinese])
张娟娟,余华,乔红波,等.2012.基于高光谱特征的土壤有机质含量估测研究.中国生态农业学报,20(5):566-572.(Zhang J J,Yu H,Qiao H B,et al.2012.Soil organic matter content estimation based on hyperspectral properties.Chinese Journal of Eco-Agriculture,20(5):566-572.[in Chinese])
Al-Abbas A H,Swain P H,Baumgarder M F.1972.Relating organic matter and clay content to the multi-spectral radiance of soils.Soil Science,114(6):477-485.
Bates J M,Granger C W J.1969.The combiantion of forecasts.Operational Research Society,20(4):451-468.
Baumgardner M F,Kristof S J,Johannsen C J,et al.1970.Effects of organic matter on the multispectral properties of soils.Proceedings of the Indiana Academy of Science,(79):413-422.
Galvao L S,Pizarro M A.2001.Variation in reflectance of tropical soil spectral chemical composition relationships from AVIRISD data.Remote Sensing Environment,75(2):245-255.
Gunsaulis F R,Kocher M F,Griffis C L.1991.Surface structure effects on close-range reflectance as a function of soil organic matter content.American Society of Agricultural Engineer,34(2):641-649.
Karnieli A,Verchovsky I,Hall J K,et al.1998.Geographic information system for semi-detailed mapping of soils in semi-arid region.Geocarto International,13(3):29-43.
Krishnan P,Alexander J D,Butler B J,et al.1980.Reflectance technique for predicting soil organic matter.Soil Society of American Journal,44(6):1280-1285.
Robin G,Viacheslav I A.2010.Precision agriculture and food security.Science,327(5967):828-831.
代海波,单锐,刘文.2012.组合预测方法中权系数的应用研究.科学技术与工程,12(32):8483-8497.(Dai H B,Shan R,Liu W.2012.Application research on weight coefficients of the combination forecasting method.Science Technology and Engineering,12(32):8483-8497.[in Chinese])
丁美青,肖红光,陈松岭,等.2012.基于BP神经网络的土地开发整理区土壤有机质含量遥感定量反演.湘潭大学自然科学学报,34(2):103-106.(Ding M Q,Xiao H G,Chen S L,et al.2012.Remote sensing quantitative of soil organic matter content in the land development and consolidation region based on BP neural network.Natural Science Journal of Xiangtan University,34(2):103-106.[in Chinese])
方少文,杨梅花,赵小敏,等.2014.红壤区土壤有机质光谱特征与定量估算---以江西省吉安县为例.土壤学报,51(5):1003-1010.(Fang S W,Yang M H,Zhao X M,et al.2014.Spectral characteristics and quantitative estimation of SOM in red soil typical of Ji’an county,Jiangxi Province.Acta Pedologica Sinica,51(5):1003-1010.[in Chinese])
付华,许凡,徐耀松,等.2013.基于变权重RBF组合模型的煤与瓦斯突出强度预测.中国安全科学学报,23(8):65-70.(Fu H,Xu F,Xu Y S,et al.2013.Prediction of goal and gas outburst intensity based on variable weight RBF combination model.China Safety Science Journal,23(8):65-70.[in Chinese])
高志海,白黎娜,王琫瑜,等.2011.荒漠化土地土壤有机质含量的实测光谱估测.林业科学,47(6):9-16.(Gao Z H,Bai L N,Wang B Y,et al.2011.Estimation of soil organic matter content in desertified lands using measured soil spectral data.Scientia Silvae Sinicae,47(6):9-16.[in Chinese])
黄仁东,张海彬,杨志辉,等.2015.基于多准则的组合预测模型权重研究及其应用.中南大学学报:自然科学版,46(5):1778-1785.(Huang R D,Zhang H B,Yang Z H,et al.2015.Research and application of multi-criteria combination forecast mode.Journal of Central South University:Science and Technology,46(5):1778-1785.[in Chinese])
纪文君,李曦,李成学,等.2012.基于全谱数据挖掘技术的土壤有机质高光谱预测建模研究.光谱学与光谱分析,32(9):2393-2398.(Ji W J,Li X,Li C X,et al.2012.Using different data mining algorithms to predict soil organic matter based on visible near infrared spectroscopy.Spectroscopy and Spectral Analysis,32(9):2393-2398.[in Chinese])
刘磊,沈润平,丁国香.2011.基于高光谱的土壤有机质含量估算研究.光谱学与光谱分析,31(3):762-766.(Liu L,Shen R P,Ding G X.2011.Studies on the estimation of soil organic matter content based on hyper-spectrum.Spectroscopy and Spectral Analysis,31(3):762-766.[in Chinese])
沈润平,丁国香,魏国栓,等.2009.基于人工网络的土壤有机质含量高光谱反演.土壤学报,46(3):391-397.(Shen R P,Ding G X,Wei G S,et al.2009.Retrieval of soil organic matter content from hyper-spectrum based on ANN.Acta Pedologica Sinica,46(3):391-397.[in Chinese])
史舟,王乾龙,彭杰,等.2014.中国主要土壤高光谱反射特性分类与有机质光谱预测模型.中国科学:地球科学,44(5):978-988.(Shi Z,Wang Q L,Peng J,et al.2014.Debelopment of a national VNIR soil-spectral library for soil classification and prediction of organic matter concentrations.Scienc China:Earth Sciences,44(5):978-988.[in Chinese])
田永超,张娟娟,姚霞,等.2012.基于近红外光声光谱的土壤有机质含量定量建模方法.农业工程学报,28(1):145-152.(Tian Y C,Zhang J J,Yao X,et al.2012.Quantitative modeling method of soil organic matter content based on near-infrared photoacoustic spectroscopy.Transactions of the Chinese Society of Agricultural Engineering,28(1):145-152.[in Chinese])
卫敏,余乐安.2012.具有最优学习率的RBF神经网络及其应用.管理科学学报,15(4):50-57.(Wei M,Yu L A.2012.A RBF neural network with optimum learning rates and its appliction.Journal of Management Sciences in China,15(4):50-57.[in Chinese])
杨扬,高小红,贾伟,等.2015.三江源不同土壤类型有机质含量高光谱反演.遥感技术与应用,30(1):186-198.(Yang Y,Gao X H,Jia W,et al.2015.Hyperspectral retrieval of soil organic matter for different soil types in the Three-River Headwaters region.Remote Sensing Technology and Application,30(1):186-198.[in Chinese])
于雷,洪永胜,耿雷,等.2015.基于偏最小二乘回归的土壤有机质含量高光谱估算.农业工程学报,31(14):103-109.(Yu L,Hong Y S,Geng L,et al.2015.Hyperspectral estimation of soil organic matter content based on partial least squares regression.Transactions of the Chinese Society of Agricultural Engineering,31(14):103-109.[in Chinese])
于雷,洪永胜,周勇,等.2016a.高光谱估算土壤有机质含量的波长变量筛选方法.农业工程学报,32(13):95-102.(Yu L,Hong Y S,Zhou Y,et al.2016a.Wavelength variable selection methods for estimation of soil organic matter content using hyperspectral technique.Transactions of the Chinese Society of Agricultural Engineering,32(13):95-102.[in Chinese])
于雷,洪永胜,周勇,等.2016b.连续小波变换高光谱数据的土壤有机质含量反演模型构建.光谱学与光谱分析,36(5):1428-1433.(Yu L,Hong Y S,Zhou Y,et al.2016b.Inversion of soil organic matter content using hyperspectral data based on continuous wavelet transformation.Spectroscopy and Spectral Analysis,36(5):1428-1433.[in Chinese])
袁征,李希灿,于涛,等.2014.高光谱土壤有机质估测模型对比研究.测绘科学,39(5):117-120.(Yuan Z,Li X C,Yu T,et al.2014.Contrast research on soil organic matter estimation model using hyper-spectral data.Science of Surveying and Mapping,39(5):117-120.[in Chinese])
曾胤,陆宇振,杜昌文,等.2014.应用红外光声光谱技术及支持向量机模型测定土壤有机质含量.土壤学报,51(6):1262-1269.(Zeng Y,Lu Y Z,Du C W,et al.2014.Applying infrared photoacoustic spectroscopy and support vector machine model to quantify soil organic matter content.Acta Pedologica Sinica,51(6):1262-1269.[in Chinese])
张娟娟,余华,乔红波,等.2012.基于高光谱特征的土壤有机质含量估测研究.中国生态农业学报,20(5):566-572.(Zhang J J,Yu H,Qiao H B,et al.2012.Soil organic matter content estimation based on hyperspectral properties.Chinese Journal of Eco-Agriculture,20(5):566-572.[in Chinese])
Al-Abbas A H,Swain P H,Baumgarder M F.1972.Relating organic matter and clay content to the multi-spectral radiance of soils.Soil Science,114(6):477-485.
Bates J M,Granger C W J.1969.The combiantion of forecasts.Operational Research Society,20(4):451-468.
Baumgardner M F,Kristof S J,Johannsen C J,et al.1970.Effects of organic matter on the multispectral properties of soils.Proceedings of the Indiana Academy of Science,(79):413-422.
Galvao L S,Pizarro M A.2001.Variation in reflectance of tropical soil spectral chemical composition relationships from AVIRISD data.Remote Sensing Environment,75(2):245-255.
Gunsaulis F R,Kocher M F,Griffis C L.1991.Surface structure effects on close-range reflectance as a function of soil organic matter content.American Society of Agricultural Engineer,34(2):641-649.
Karnieli A,Verchovsky I,Hall J K,et al.1998.Geographic information system for semi-detailed mapping of soils in semi-arid region.Geocarto International,13(3):29-43.
Krishnan P,Alexander J D,Butler B J,et al.1980.Reflectance technique for predicting soil organic matter.Soil Society of American Journal,44(6):1280-1285.
Robin G,Viacheslav I A.2010.Precision agriculture and food security.Science,327(5967):828-831.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |