Spatial and Temporal Occurence of Preferential Flow at the Catchment Scale

Abstract:

1. Introduction

There is abundant evidence that preferential flow is a common phenomenon in the vadose zone, which stresses the importance of considering preferential flow processes at different hydrological scales. However, predicting the occurrence of preferential flow becomes troublesome when moving from the plot towards the catchment scale, as data coverage diminishes. Despite all previous research, information on factors and properties that promote preferential flow still remain limited as monitoring the occurrence of preferential flow pathways through time and in space remains a challenging task [Beven and Germann, 2013; Lin and Zhou, 2008]. Moreover, high spatial and temporal variability in preferential flow make it difficult to qualify and quantify the occurrence of preferential flow. Currently, no general set of rules exists that explains spatial patterns of preferential flow at larger scales.

A promising approach to identify the occurrence of temporal and spatial variability in preferential flow at the catchment scale is the use of soil moisture sensor response times [Graham and Lin, 2011; Hardie et al., 2013; Lin and Zhou, 2008]. After determining the sequence of soil moisture response times for different depths, the spatial occurrence of preferential flow and other flow regimes can be identified for single rainfall events. The aim of this study is to investigate the dominant controls on preferential flow at the catchment scale by applying this method on data from the wireless soil moisture sensor network SoilNet, which is installed in the Wuestebach catchment within the framework of TERENO (Terrestrial Environmental Observatories). To understand the factors and processes that cause spatial variability in preferential flow, the results of this classification were related to site and event characteristics (e.g. precipitation characteristics, initial soil moisture content, soil type, bulk density, organic matter content).

2. Methodology

In the Wuestebach catchment, soil moisture has been monitored since August 2009 at three depths and more than 100 locations using the wireless soil moisture sensor network SoilNet with a temporal resolution of 15 minutes [Rosenbaum et al., 2012]. The measurement locations of the soil moisture network were selected using a 60 * 60 meter grid (74 points) combined with random locations (76 points) in order to assure sufficient spatial coverage while also allowing the investigation of small-scale variability. At each individual measurement location, one 5TE and one 5-EC sensor were installed at 0.05 and 0.5 m depth and two 5-EC sensors were installed at 0.2 m (Decagon, Devices Inc., Pullman, USA).

Our approach to identify preferential flow is based on the principle that non-uniform flow can be identified by non-sequential sequences of sensor response times within the soil [Graham and Lin, 2011; Hardie et al., 2013; Lin and Zhou, 2008; Van Schaik, 2010]. To understand the reaction of soil moisture to precipitation, the soil moisture and precipitation datasets were delineated into events using the measured rainfall. The start of an event was marked by a minimum hourly precipitation of 1 mm. Once started, an event continued until there was at least a period of three hours with rainfall intensities of less than 1 mm/h.

For all delineated events, the six soil moisture time series at each sensor network location were examined and classified into several categories, depending on their response sequence. The response time for a particular sensor was determined if the soil moisture increased by more than 1 vol. % within the delineated event. In such a case, the response time was defined as the time where the soil moisture content increased above the instrumental noise (0.4 vol. %). Subsequently, the order of the response times for the six sensors was used to assign one of the following classes for the particular location and event: (1) preferential non-sequential flow, (2) preferential flow based on flow velocity, (3) sequential flow, (4) no response, (5) no data.

Almost 400 events have been classified at 100 locations, covering a time period of 3 years. Results of this classification were used to understand the controlling factors of preferential flow in time and space. Here, we present the results of four events to analyze the effect of different events characteristics (specific precipitation sums, rainfall intensities and initial soil moisture conditions) as given in the table within Figure 1.

3. Results and conclusions

The results of our analysis show that preferential flow is a common phenomenon within the Wuestebach catchment, but shows a high variability in space and time. At some locations, preferential flow occurred in more than 92% of all events, whereas other locations only showed preferential flow in 7% of all events. Our results indicate that both rainfall characteristics (total precipitation, maximum precipitation intensity) and soil moisture play an important role in determining the occurrence of preferential flow at the catchment scale (Figure 1).

For events with a small amount of precipitation (event 1 and 2 in Figure 1), the percentage of preferential response within the catchment was much lower than for events with a high precipitation sums (events 3 and 4). Although event 4 had similar initial soil moisture and rainfall intensity conditions as event 1, it showed a remarkably larger overall response related to the larger total sum of precipitation. During events with little precipitation and low precipitation intensities (1 and 2), a large part of the sensors within the catchment (39.2 and 71.6 %) did not respond. Precipitation event 3 was exceptional both in terms of intensity and amount, causing preferential flow response at almost all sensor locations.

If all 400 events are considered, no large-scale patterns in preferential flow occurrence could be detected. Although there are certain locations that react quite predictable over time, there are also many locations that react very variable in time. This indicates that it is no easy task to upscale preferential flow processes to landscape units in the Wuestebach catchment. Currently, we are analyzing the responses of each location in order to inspect whether the observed behavior is related to local and spatially variable thresholds that complicate interpretations at the catchment scale.

References

Beven, K., and P. Germann (2013), Macropores and water flow in soils revisited, Water Resources Research, 49(6), 3071-3092.

Graham, C. B., and H. S. Lin (2011), Controls and frequency of preferential flow Occurrence: A 175-Event Analysis, Vadose Zone Journal, 10(3), 816-831.

Hardie, M., S. Lisson, R. Doyle, and W. Cotching (2013), Determining the frequency, depth and velocity of preferential flow by high frequency soil moisture monitoring, Journal of contaminant hydrology, 144(1), 66-77.

Lin, H., and X. Zhou (2008), Evidence of subsurface preferential flow using soil hydrologic monitoring in the Shale Hills catchment, European Journal of Soil Science, 59(1), 34-49.

Rosenbaum, U., H. R. Bogena, M. Herbst, J. A. Huisman, T. J. Peterson, A. Weuthen, A. W. Western, and H. Vereecken (2012), Seasonal and event dynamics of spatial soil moisture patterns at the small catchment scale, Water Resources Research, 48(10), W10544.

Van Schaik, L. (2010), The role of macropore flow from plot to catchment scale : a study in a semi-arid area, 174 p. pp., Koninklijk Nederlands Aardrijkskundig Genootschap : Faculteit Geowetenschappen Universiteit Utrecht, Utrecht.

Figure 1: Four characteristic rainfall events and the corresponding occurrences of preferential flow in the Wuestebach catchment. Total P. = total precipitation; Max. P. Intensity = maximum precipitation intensity; Mean Theta [Vol. %] = mean water content in volume percentage over all sensors.: (1) % preferential non-sequential flow, (2) % preferential flow based on flow velocity, (3) % sequential flow, (4) % no response, (5) % no data.