Abstract
The global population is growing rapidly, however, one-third of the world’s population is
suffering from water scarcity. One reason is that agricultural water withdrawal accounts for
approximately 69% of freshwater use. In order to improve water use efficiency, knowledge of
the spatio-temporal variation in the soil volumetric water content (Ɵ) is essential. Apparent soil
electrical conductivity (ECa) measured by non-invasive electromagnetic (EM) induction
instruments is increasingly being used but mostly limited to map average Ɵ. In this thesis,
electromagnetic conductivity images (EMCI) generated by inverting ECa data have been used to
map the spatial and temporal variations in across two study fields situated in San Jacinto,
California, USA and Cobbitty, New South Wales, Australia. The thesis introduces the quasi-2d
inversion of ECa and its application for mapping Ɵ by horizontal slices and vertical crosssections.
To account for the temporal continuity of the time-lapse ECa data and include physical
soil-water models, a novel spatial-temporal quasi-2d inversion algorithm and the ensemble
Kalman filter have been used, respectively. Furthermore, the spatial and temporal scale-specific
variations in EMCI and the scale-specific correlation between EMCI and different soil
properties were studied using wavelet analysis. In addition, the diurnal drifts of the DUALEM
measurements were tested. It was concluded that non-invasive EM induction techniques
combined with inversion algorithm and the ensemble Kalman filter can be applied to accelerate
the mapping and monitoring depth-specific Ɵ and improve irrigation efficiency and water
management.