Prediction of Sodium Chloride in Soil with Reflectance Spectroscopy
Author(s): TAO Lan-hua, TASHPOLAT·Tiyip, MAMAT·Sawut, JIANG Hong-tao, ZHAO Zhen-liang, College of Resources and Environment Sciences, Xinjiang University, Key Laboratory of Oasis Ecology under Ministry of Education, Xinjiang University
Pages: 46-50
Year: 2014 Issue: 3
Journal: Geography and Territorial Research
Keyword: hyperspectral; NaCl content; PLSR;
Abstract: Based on the advantage of rapid and non-destructive detection of spectral reflectance,compared with conventional methods,this paper takes the reflection of different soil salinity as the information source in the study area,and explores the feasibility of using reflectance spectroscopy estimating soil sodium chloride content.Through the investigation in the region of Yutian in July 2012,data preprocessing and the Stepwise Multiple Linear Regression(SMLR)and the Partial Least Squares Regression(PLSR)were established based on the determination of soil spectra,reed leaf spectral and salt content,then Root Mean Squared Error(RMSE)was introduced to test the predictability and precision of the models,and coefficient of the determination(R2)was used to evaluate stability of the models.Then the applicability of the estimate results is compared using different preprocessing methods and different models in this paper.Results demonstrate:SDR is optimal index for predicting sail content form.The PLSR is the optimal model to establish the relationship between the soil spectrum,reed spectrum with salt content,of which R2 and RMSEis 0.812and 0.105,respectively.Estimating salt content of soil by the reflectance spectra and improving the accuracy of estimation through varieties of pre-processing methods can achieve the purpose on the regional scale for soil salinization monitoring and evaluation.
Pages: 46-50
Year: 2014 Issue: 3
Journal: Geography and Territorial Research
Keyword: hyperspectral; NaCl content; PLSR;
Abstract: Based on the advantage of rapid and non-destructive detection of spectral reflectance,compared with conventional methods,this paper takes the reflection of different soil salinity as the information source in the study area,and explores the feasibility of using reflectance spectroscopy estimating soil sodium chloride content.Through the investigation in the region of Yutian in July 2012,data preprocessing and the Stepwise Multiple Linear Regression(SMLR)and the Partial Least Squares Regression(PLSR)were established based on the determination of soil spectra,reed leaf spectral and salt content,then Root Mean Squared Error(RMSE)was introduced to test the predictability and precision of the models,and coefficient of the determination(R2)was used to evaluate stability of the models.Then the applicability of the estimate results is compared using different preprocessing methods and different models in this paper.Results demonstrate:SDR is optimal index for predicting sail content form.The PLSR is the optimal model to establish the relationship between the soil spectrum,reed spectrum with salt content,of which R2 and RMSEis 0.812and 0.105,respectively.Estimating salt content of soil by the reflectance spectra and improving the accuracy of estimation through varieties of pre-processing methods can achieve the purpose on the regional scale for soil salinization monitoring and evaluation.
Related Articles