Premium
Mapping Soil Texture by Electromagnetic Induction: A Case for Regional Data Coordination
Author(s) -
Kelley Jason,
Higgins Chad W.,
Pahlow Markus,
Noller Jay
Publication year - 2017
Publication title -
soil science society of america journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.836
H-Index - 168
eISSN - 1435-0661
pISSN - 0361-5995
DOI - 10.2136/sssaj2016.12.0432
Subject(s) - soil texture , precision agriculture , environmental science , soil science , digital soil mapping , image resolution , remote sensing , texture (cosmology) , soil management , pedotransfer function , soil map , computer science , soil water , artificial intelligence , geology , geography , hydraulic conductivity , agriculture , image (mathematics) , archaeology
Core Ideas High‐resolution soil texture maps can improve efficiencies in precision irrigation. Noncontact EMI methods can generate high‐resolution maps quickly and at low cost. Mapping with EMI requires robust correlations between soil EC a and soil texture. Accuracy of high‐resolution maps may be reduced by nontarget soil characteristics. Neural networks can predict soil properties accurately by combining measurements. High‐resolution soil maps are needed to guide precision irrigation, fertilization, and crop management decisions. A mobile, noncontact device can be used to map apparent electromagnetic conductivity (EC a ) with detailed spatial resolution (<10 m). Maps of other characteristics can be generated by determining the correlation between the measured EC a and physical samples. Soil mapping services are now offered by commercial vendors and are used to estimate properties that are relevant to soil management, including nitrate and micronutrients, soil pH, texture, and water holding capacity. In this study, EC a maps were compared with 390 colocated physical samples, distributed over three cultivated fields comprising a total area of 140 ha. Principal component analysis and linear regression were used to identify factors that confounded EC a measurement and reduced map fidelity. When correlation was conducted in each field independently, low variability in soil texture reduced the accuracy of predicting texture from EC a . To compensate, a combination of four properties was used to train a neural network model, which yielded an improved prediction. Using such methods to establish regionally specific models for EC a readings could enable robust mapping of soil characteristics via rapid and low‐cost electromagnetic measurements. The results confirm that EC a can be a used to map texture at field scales and that additional quality controls can augment the predictive power of EC a .