Premium
Mapping Clay Content across Boundaries at the Landscape Scale with Electromagnetic Induction
Author(s) -
Weller U.,
Zipprich M.,
Sommer M.,
Castell W. Zu,
Wehrhan M.
Publication year - 2007
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/sssaj2006.0177
Subject(s) - silt , calibration , sampling (signal processing) , soil science , scale (ratio) , field (mathematics) , soil water , environmental science , remote sensing , geology , hydrology (agriculture) , mathematics , geography , statistics , geomorphology , computer science , geotechnical engineering , cartography , computer vision , filter (signal processing) , pure mathematics
Detailed information on soil textural heterogeneity is essential for land management and conservation. It is well known that in individual fields, measurement of the soil's apparent electrical conductivity (EC a ) offers an opportunity to map the clay content of soils with free drainage under a humid climate. At the catchment scale, however, units of different land management and differing sampling dates add variation to EC a and constrain the mapping across field boundaries. We analyzed their influence and compared three approaches for applying electromagnetic induction (EM v ) to clay‐content mapping at the landscape scale across the boundaries of individual fields and different sampling dates. In the study region, a separate calibration of the relation between clay and EC a for each field and sampling date (fieldwise calibration) yielded satisfactory clay‐content predictions only if the costly precondition of sufficient calibration points for each field was fulfilled. We propose a method (nearest‐neighbors EC a correction) for unifying EC a across boundaries based only on the EC a data themselves, and the assumption of continuity of textural properties at field boundaries, which was fulfilled in the landscape studied. Prediction is calibrated once for the entire landscape, which allows a reduced set of calibration points. The coefficient of determination for predicting clay content (here, including silt <4 μm) was improved from R 2 = 0.66 (no correction for land use and sampling date) to R 2 = 0.85 ( n = 46). With the method developed, EC a offers a powerful and cheap method of clay‐content mapping in agricultural landscapes.