Predictive performance of mobile vis-near infrared spectroscopy for key soil properties at different geographical scales by using spiking and data mining techniques

详细信息    查看全文

• 作者：Said Nawar^a ; ^{s.m.nawar@cranfield.ac.uk} ; Abdul M. Mouazen^a ; ^b
• 关键词：Diffuse reflectance spectroscopy ; Spectral library ; Soil properties ; Data mining ; Spiking
• 刊名：CATENA
• 出版年：2017
• 出版时间：April 2017
• 年：2017
• 卷：151
• 期：Complete
• 页码：118-129
• 全文大小：1409 K
• 卷排序：151

文摘

The performance of on-line vis-NIR for estimating soil TN, TC and MC were evaluated. Comparing between four datasets single field, two-field, national and European Multivariate analysis PLSR, SVM and MARS and spiking techniques were used. The best results obtained using the MARS calibration method with European dataset