Premium
Mapping soil salinity using electromagnetic conductivity imaging—A comparison of regional and location‐specific calibrations
Author(s) -
Farzamian Mohammad,
Paz Maria Catarina,
Paz Ana Marta,
Castanheira Nádia Luísa,
Gonçalves Maria Conceição,
Monteiro Santos Fernando A.,
Triantafilis John
Publication year - 2019
Publication title -
land degradation and development
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.403
H-Index - 81
eISSN - 1099-145X
pISSN - 1085-3278
DOI - 10.1002/ldr.3317
Subject(s) - soil salinity , soil science , salinity , environmental science , calibration , mean squared error , remote sensing , hydrology (agriculture) , soil water , geology , mathematics , statistics , geotechnical engineering , oceanography
Abstract Soil salinization limits agricultural productivity and can ultimately cause desertification and land abandonment. One approach to assess soil salinity over large areas efficiently is to use electromagnetic instruments to measure the soil apparent electrical conductivity (EC a , mS m −1 ). EC a data can be then inverted to generate electromagnetic conductivity images (EMCIs), which provide the vertical distribution of the soil electrical conductivity ( σ , mS m −1 ). In this study, we collected EC a data using an EM38 instrument across four locations with different levels of salinity in an important agricultural area of alluvial origin in Portugal. Using an inversion algorithm, we generated EMCIs and evaluated the potential for prediction of the electrical conductivity of the saturated soil paste extract (EC e ). The main aim of our study is to compare regional and location‐specific calibrations in terms of ability to predict EC e from EMCIs. The results showed that the regional calibration predicted EC e unbiased, precisely (root mean square error [RMSE] = 2.54 dS m −1 ), and with strong concordance (0.93). The location‐specific calibration also predicted EC e unbiased, precisely (RMSE = 1.67 dS m −1 ), and with strong concordance (0.97). We conclude that the location‐specific calibration has slightly better prediction results, but the regional calibration is more practical for mapping soil salinity in the study area because it can be used at any new location without the need for new calibration. The prediction results at locations 3 and 4 show high to severely high soil salinity, which can compromise agricultural productivity. Therefore, monitoring soil salinity is required to conserve and improve agricultural productivity in the south of the study area.