Th e soil column was 89 mm in diameter and 84 mm high. Th e dimensions of the soil core were limited by the inner diameter of the MRI probe available at the time of imaging. Aft er the sample was extracted, the top and bottom surfaces were gently fl attened. A sample cylinder with a steel cutting edge attached to the lower end was then gradually pushed into the soil while the soil was carefully excavated at the outer perimeter. At the experimental site, the upper layer of soil was excavated to a depth of 40 cm and the bottom of the pit was fl attened.
For the MRI study, an undisturbed soil sample was taken in a Perspex cylinder. For MRI, this soil represented a “diffi cult” medium (Hall et al., 1997) the signal intensity distribution obtained in the MR image depicted only water present in large pores. Th e soil exhibited preferential fl ow, as well as fl ow rate instability in the form of a dependence of the steady-state infi ltration rate on the initial soil moisture content (Císlerová et al., 1988, 1990). An undisturbed sample of a coarse sandy loam soil (a Dystric Cambisol), taken from the experimental site Korkusova Hut, Czech Republic, was examined. Th e presence of the encapsulated air was indicated by a decrease in the MRI signal intensity during the second infi ltration run. The standard flow and pressure head measurements were performed simultaneously with the MRI monitoring.
While at the top of the sample a constant pressure head was maintained, the bottom was supported by a perforated plate to induce free outfl ow (a seepage face boundary condition). We present the results of a fully controlled recurrent ponded infi ltration–outfl ow (RPI) experiment, where the fi rst and second ponded infi ltration runs were started into soil with diff erent initial water contents. Th e experimental setup was described by Sněhota et al. Better magnetic fi eld homogeneity of the whole- body magnet and an advanced control system allowed imaging in a multiple-slice scheme that covered almost the entire volume of the sample. (2003) were extended and adapted for three-dimensional imaging of a larger sample in a “whole body” unit. The procedures developed for MRI of soils by Votrubová et al. Th e aim of the present study was to assess air trapping in an undisturbed sample of natural heterogeneous soil. Magnetic resonance imaging was used for spatial measurement of local fl ow velocities (Seymour and Callaghan, 1997 Baumann et al., 2000). Recently, MRI was used to investigate the fate and transport of nonaqueous-phase liquids (Chu et al., 2004 Zhang et al., 2008 Nestle et al., 2008), to study fi nger fl ow in sand (Posadas et al., 2009), to trace the transport of D 2 O (Pohlmeier et al., 2009) and the transport of Ni(NO 3 ) 2 (Jelínková et al., 2010), and to monitor fl ow in a root zone (Pohlmeier et al., 2008). large pores produces a signifi cant magnetic resonance (MR) signal, MRI studies of preferential fl ow in undisturbed samples of heterogeneous soil were performed (Císlerová et al., 1999, 2002 Votrubová, 2002 Císlerová and Votrubová, 2002 Votrubová et al., 2003 Sněhota et al., 2003 Sněhota, 2003).