Proximal gamma-ray spectrometry for site-independent in situ prediction of soil texture on ten heterogeneous fields in Germany using support vector machines

详细信息    查看全文

• 作者：Tobias Heggemann^a ; ^{tobias.heggemann@uni-bonn.de} ; ^{to.heggemann@gmail.com} ; Gerhard Welp^a ; Wulf Amelung^a ; ^b ; Gerrit Angst^a ; ^f ; Sven Oliver Franz^c ; Sylvia Koszinski^d ; Karsten Schmidt^e ; Stefan P&auml ; tzold^a
• 关键词：Soil sensing ; Soil mapping ; Precision agriculture ; Machine learning ; Field survey
• 刊名：Soil and Tillage Research
• 出版年：2017
• 出版时间：May 2017
• 年：2017
• 卷：168
• 期：Complete
• 页码：99-109
• 全文大小：1344 K
• 卷排序：168

文摘

Proximal gamma spectrometry as a tool soil texture prediction requires calibration. Linear calibration models failed when transferred between different sites. Mineralogy partly led to adverse relationships between gamma features and texture. Non-linear SVM based models yielded site-independent reliable texture predictions. Universal application of gamma spectrometry for precision farming is possible.