Premium
Mapping Salinity in Three Dimensions using a DUALEM‐421 and Electromagnetic Inversion Software
Author(s) -
Zare E.,
Huang J.,
Santos F.A. Monteiro,
Triantafilis J.
Publication year - 2015
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/sssaj2015.06.0238
Subject(s) - salinity , transect , inversion (geology) , mean squared error , soil science , conductivity , electrical resistivity and conductivity , linear regression , software , mathematics , mineralogy , geology , statistics , computer science , chemistry , physics , geomorphology , oceanography , structural basin , programming language , quantum mechanics
To implement management plans, the salt content across affected fields and with depth needs mapping. In this study, we developed a method to map the distribution of normal, uniformly saline, and inverted salinity profiles. We did this by establishing a linear regression (LR) between calculated true electrical conductivity (σ) and electrical conductivity of the saturated soil‐paste extract (EC e ). We estimated σ by inverting the apparent electrical conductivity (EC a ) collected from a DUALEM‐421. The EC a values were collected along 13 parallel transects spaced 50 m apart. The inversion was performed using a quasi‐three‐dimensional model available in the EM4Soil software, where we chose the full solution and S1 inversion algorithm with a damping factor (λ) of 0.3. Using cross‐validation, the quasi‐three‐dimensional model yielded a high accuracy (RMSE = 5.28 dS m −1 ), small bias (mean error [ME] = −0.03 dS m −1 ), and large R 2 (0.88) and Lin's concordance (0.93). While slightly better results were achieved using individual LRs established at each depth increment overall (i.e., RMSE = 4.35 dS m −1 , ME = −0.17 dS m −1 , R 2 = 0.92, and Lin's concordance = 0.96) and with the DUALEM‐421 EC a , the inversion approach requires the development of a single LR (i.e., EC e = 4.1253 + 0.0167σ), which enables efficiencies in estimating salinity and allows EC e to be estimated at any depth where σ was estimated within a three‐dimensional electromagnetic conductivity image. This can improve understanding of the cause and best management of salinity. Improvements in accuracy and bias can be achieved by collection of EC a on more closely spaced transects.