吸附
化学
土壤水分
表土
草甘膦
环境化学
土壤碳
溶解有机碳
偏最小二乘回归
总有机碳
矿物学
土壤科学
吸附
农学
环境科学
数学
有机化学
生物
统计
作者
Cecilie Hermansen,Trine Nørgaard,Lis Wollesen de Jonge,Per Møldrup,Karin Müller,Maria Knadel
出处
期刊:Geoderma
[Elsevier]
日期:2019-11-12
卷期号:360: 114009-114009
被引量:23
标识
DOI:10.1016/j.geoderma.2019.114009
摘要
Glyphosate [N-(phosphonomethyl) glycine] is the active ingredient in Roundup, which is the most used herbicide around the world. It is a non-selective herbicide with carboxyl, amino, and phosphonate functional groups, and it has a strong affinity to the soil mineral fraction. Sorption plays a major role for the fate and transport of glyphosate in the environment. The sorption coefficient (Kd) of glyphosate, and hence its mobility, varies greatly among different soil types. Determining Kd is laborious and requires the use of wet chemistry. In this study, we aimed to estimate Kd using basic soil properties, and visible near-infrared spectroscopy (vis–NIRS). The latter method is fast, requires no chemicals, and several soil properties can be estimated from the same spectrum. The data set included 68 topsoil samples collected across the South Island of New Zealand, with clay and organic carbon (OC) contents ranging from 0.001 to 0.520 kg kg−1 and 0.021 to 0.217 kg kg−1, respectively. The Kd was determined with batch equilibration sorption experiments and ranged from 13 to 3810 L kg−1. The visible near-infrared spectra were obtained from 400 to 2500 nm. Multiple linear regression was used to correlate Kd to oxalate extractable aluminium and phosphorous and pH, which resulted in an R2 of 0.89 and an RMSE of 259.59 L kg−1. Further, interval partial least squares regression with ten-fold cross-validation was used to predict Kd by vis–NIRS, and an R2 of 0.93 and an RMSECV of 207.58 L kg−1 were obtained. Thus, these results show that both basic soil properties and vis–NIRS can predict the variation in Kd across these samples with high accuracy and hence, that glyphosate sorption to a soil can be determined with vis–NIRS.
科研通智能强力驱动
Strongly Powered by AbleSci AI