Spatio-temporal relevance and controls of preferential flow at the landscape scale

Dominic Demand,Markus Weiler,Theresa Blume

doi:10.5194/hess-23-4869-2019

Abstract

Abstract. The spatial and temporal controls of preferential flow (PF) during infiltration are still not fully understood. As soil moisture sensor networks allow us to capture infiltration responses in high temporal and spatial resolution, our study is based on a large-scale sensor network with 135 soil moisture profiles distributed across a complex catchment. The experimental design covers three major geological regions (slate, marl, sandstone) and two land covers (forest, grassland) in Luxembourg. We analyzed the responses of up to 353 rainfall events for each of the 135 soil moisture profiles. Non-sequential responses (NSRs) within the soil moisture depth profiles were taken as one indication of bypass flow. For sequential responses maximum porewater velocities (vmax⁡) were determined from the observations and compared with velocity estimates of capillary flow. A measured vmax⁡ higher than the capillary prediction was taken as a further indication of PF. While PF was identified as a common process during infiltration, it was also temporally and spatially highly variable. We found a strong dependence of PF on the initial soil water content and the maximum rainfall intensity. Whereas a high rainfall intensity increased PF (NSR, vmax⁡) as expected, most geologies and land covers showed the highest PF under dry initial conditions. Hence, we identified a strong seasonality of both NSR and vmax⁡ dependent on land cover, revealing a lower occurrence of PF during spring and increased occurrence during summer and early autumn, probably due to water repellency. We observed the highest fraction of NSR in forests on clay-rich soils (slate, marl). vmax⁡ ranged from 6 to 80 640 cm d−1 with a median of 120 cm d−1 across all events and soil moisture profiles. The soils in the marl geology had the highest flow velocities, independent of land cover, especially between 30 and 50 cm depth, where the clay content increased. This demonstrates the danger of treating especially clay soils in the vadose zone as a low-conductive substrate, as the development of soil structure can dominate over the matrix property of the texture alone. This confirms that clay content and land cover strongly influence infiltration and reinforce PF, but seasonal dynamics and flow initiation also have an important impact on PF.

Highlights

Preferential flow (PF) in soils describes different flow processes with higher flow velocities than soil matrix flow and heterogeneous flow patterns (Hendrickx and Flury, 2001)
Between 63.2 % and 79.5 % of the infiltration events per landscape units were not classifiable (NC) in their infiltration response, with the marl forest sites having the lowest amount of NC. 49.6 % of all NC events resulted from events with a Psum of 3 mm or less
The importance of preferential flow (PF) during infiltration was highlighted by the fact that observed sequential response (SR) flow velocity was most of the time faster than pure soil matrix flow and depended on the landscape unit Nonsequential responses (NSRs) accounted for 18 %–44 % of the responses deeper than 10 cm

Summary

Introduction

Preferential flow (PF) in soils describes different flow processes with higher flow velocities than soil matrix flow and heterogeneous flow patterns (Hendrickx and Flury, 2001). PF can affect water distribution in soil (Ritsema et al, 1996), groundwater recharge (Ireson and Butler, 2011), root water uptake (Schwärzel et al, 2009) and solute transport (Larsbo et al, 2014). Most studies focusing on different PF processes, such as fingered flow (Selker et al, 1992), macropore flow (Weiler and Naef, 2003) or funnel flow (Kung, 1990), were carried out at the point or plot scale (spatial scale smaller than a few meters). Since PF increases the range of flow velocities in the vadose zone by orders of magnitudes (Nimmo, 2007), it is essential to include this process when modeling water and solute transport in soil.

Objectives

Methods

Results

Discussion

Conclusion