Controls on groundwater response timing of a subalpine catchment

Abstract:

Introduction

Lateral subsurface storm flow within the soil profile is one of the main process that efficiently transfers soil water between runoff source areas in a catchment and the channel network (Weiler et al. 2005). Groundwater dynamics are therefore important in understanding states of hydrologic connectivity and runoff response during rainfall events. In order to better understand how catchments are organized we need to study spatial patterns of rainfall-runoff processes and the governing mechanisms and controls that determine them. (McDonnell et al. 2007). In mountain headwater catchments topography, soil and vegetation properties are expected to play an important role in groundwater response dynamics but only a few studies have tried to identify the dominant catchment characteristics that allow prediction of spatial groundwater dynamics. Groundwater responses were shown to be related to topography on steep slopes and under wet environmental conditions, where groundwater levels are predominantly shallow (Anderson & Burt 1978; Burt & Butcher 1985; Lana-Renault et al. 2013). In flatter and drier conditions (Detty & McGuire 2010), and especially in permeable soils (Seibert et al. 1997; Dhakal & Sullivan 2014; Anderson et al. 2010), it has been difficult to demonstrate the impact of the topography of groundwater levels and dynamics.

Some topographically based conceptual models (like TOPMODEL) assume groundwater levels to respond in unison across a catchment and approximate groundwater variations by a succession of steady-state situations. This implies a spatially persistent pattern of groundwater levels in a catchment, which can be modeled by the Topographic Wetness Index (TWI) (Beven & Kirkby 1979). However from existing studies we know that groundwater levels do not always respond in unison across a hillslope or an entire catchment (Penna et al. 2014; Bachmair & Weiler 2012; Fannin et al. 2000) but differ between landforms (Detty & McGuire 2010), shallow and deep soils (Tromp-van Meerveld & McDonnell 2006) and the upper or lower part of hillslopes (Haught & van Meerveld 2011). Locations in the riparian zone and in subsurface depressions react earlier to rainfall events than upslope locations (Detty & McGuire 2010) Haught & van Meerveld (2011) found that groundwater lagtimes got shorter with increasing antecedent moisture content.

Despite the knowledge gained in previous hillslope studies on sites with transmissive soils, we know little about catchment-scale groundwater dynamics in steep mountain environments with less permeable soils. Therefore, we set up a comprehensive monitoring and measured groundwater fluctuations in a subalpine catchment with the aim to answer the following questions:

1. To what extent does topography govern the groundwater response timing?
2. How do rainfall intensity and antecedent conditions influence groundwater response timing?
3. Do observations support the concept of steady-state successions, as assumed in the TOPMODEL?

Methods

The study site is a 20 ha headwater catchment located in the Alptal, a pre-alpine valley about 40 km southeast of Zurich, Switzerland extending from 1270 m asl. to 1650 m asl.. Due to the high mean annual precipitation of 2300 mm/year (Feyen et al. 1999) and low permeable bedrock (Flysch) and soils (Gleysols) (Schleppi et al. 1998), moor landscapes have formed with wet grassland growing in flat or concave parts of the catchment and open coniferous forest stands on steeper slopes and ridge-sites.

Fifty-one groundwater monitoring sites in seven nested sub-catchments were installed based on a stratified sampling procedure to cover the range of topographic positions, soil types and vegetation in the experimental catchments. The well depth varied between 0.5 to 2 m. Water levels were measured in the wells between September 2010 and the end of November 2012. The stage at the outlet of each sub-catchment was measured continuously and climatologic variables were recorded at a long-term weather station 1 km from the experimental catchment. For the analyses, we selected 133 rainfall events and classified them according to rainfall intensity and antecedent wetness.

For the timing analyses we considered time lags of groundwater response relative to the start of the rainfall event, namely the time to rise, (t_rise.) and the time to peak, (t_peakP). The groundwater peak duration (t_dur) was calculated as the time lag between the time that the water level had increased to 95% of the maximum water level rise on the rising limb of the groundwater hydrograph and the corresponding point of time on the falling limb (called 95% recession). The duration of the recession (t_rec) was defined as the time between 95% of the rise and 20% of the rise. In order to investigate synchronicity between groundwater and streamflow peaks we calculated t_peakQ as the time lag between the peak groundwater level and runoff peak at the catchment outlet.

We expected the groundwater response to be correlated to topography and therefore determined several topographic indices based on a 6 by 6 meter Digital Terrain Model (DTM). These indices were used to assess local controls that characterize the monitoring site itself and upslope controls that represent the properties of the upslope contributing area. Site characteristics selected for this study were: local slope gradient (Tarboton 1997), local curvature (Evans 1980; Travis et al. 1975), TWI (Beven & Kirkby 1979), upslope contributing area, mean slope, mean curvature and mean TWI of the upslope contributing area. To quantify the relation between topographic characteristics and median event timing characteristics the Spearman rank correlation coefficient (r_s) was used.

Results

The groundwater response, duration (width of the peak) and recession and differed for the monitoring sites and were closely tied to topography. Half of the groundwater monitoring sites responded on average within 1.5 h after the onset of rainfall. Moderate-intense rainfall events had the shortest response times, which is in agreement with the perception that soil water deficits are satisfied faster when the the rainfall intensity is higher. In fact, we found t_rise to be highly correlated to the median sum of rainfall till response (r_s= 0.98). The inter quartile range (IQR) of the mean and median response times of all sites was twice as high for the low-intense rainfall events than for the moderate-intense rainfall events.

The response time (t_rise) was correlated to topographic indices, namely the mean curvature of the upslope contributing area (r_s = 0.82), TWI (r_s = -0.81), upslope contributing area (r_s = -0.74), mean TWI of the upslope contributing area (r_s = -0.66), and local slope (r_s = 0.64). The r_s was lower for the mean slope of the upslope contributing area (r_s = 0.29) and the local curvature (r_s = 0.28). The median t_rise decreased with TWI for sites with a TWI < 5; the variability decreased with increasing TWI. The decrease in median t_rise with TWI for sites with a TWI < 5 was steeper for the low-intensity rainfall events than for the high-intensity events. The influence of the antecedent condition on t_risewas small.

Groundwater peaks lagged rainfall centroids by ca. 1 hour for half of the sites. Differences in t_peakP between the four rainfall event types were small, except for the low intensity events during dry antecedent conditions for which t_peakP was more than twice as long. Streamflow peaks were delayed relative to groundwater peaks by a 10s of minutes and in ca. 25% of all rainfall events groundwater levels lagged the runoff peak at the catchment outlet at most sites. The timing of the groundwater peak (median t_peakP and t_peakQ) was not correlated to the topographic indices, except TWI.

We speculated that the peak duration (t_dur) was governed by subsurface inputs from upslope and therefore correlated to the upslope contributing area. Instead we found the median t_dur only to be correlated to local slope (r_s = -0.32) and mean TWI of the upslope contributing area (r_s= 0.29) but not to any of the other indices.

We expected the groundwater recessions to be slower for sites with a larger or longer water input from the upslope contributing area and a low hydraulic gradient. Indeed we found the median t_rec was correlated to local slope (r_s = -0.39) and TWI (r_s = 0.38), as well as to mean curvature of the upslope contributing area (r_s = -0.37) and upslope contributing area (r_s = 0.32) but not to any of the other indices. The median t_recincreased with increasing TWI for sites with a TWI < 6 but the variability was high. Differences between the four rainfall event types were small.

We also wanted to know, if the correlation between absolute groundwater levels and TWI was constant over time, as assumed by TOPMODEL. However we found the correlation to decrease strongly at the beginning of rainfall events, reaching the lowest values shortly after peak streamflow. After the peak, r_s increased quickly and reached the highest values twelve hours to two days after the event. During dry periods, r_sgradually decreased until the beginning of the next event.

Discussion

Our study revealed that timing of groundwater rise and recession, but not time to peak, was strongly controlled to topography in a catchment with low permeable soils and shallow groundwater tables. The way subsurface water flow is generated and concentrated in the upslope contributing area (as described by our predictor variables: mean curvature of the upslope contributing area and upslope contributing area) and the local drainage conditions at that site (as described by the local slope) seems to determine the timing of rise and recession of the groundwater levels during a rainfall event. One can speculate that the storage capacity of wet sites with a TWI > 5 was filled to a larger degree for most of the time, which is why they responded quickly. Sites with TWI < 5 showed an increasing time to rise with decreasing TWI, implying that the soil water deficit needed to be satisfied before the groundwater levels started to rise. The slope of this relation appers to be dominated by rainfall intensity and to a lesser extent by antecedent wetness (see Fig. 1).

In contrast, t_peakP and t_peakQwere not related to by topographic indices but are more likely controlled by rainfall characteristics. The peak duration seemed to be dominated by the local drainage properties but not by the water input from upslope, which was against our expectations. Groundwater peaks preceded runoff peaks only by a few minutes or even lagged it. This contradicts the common perception that groundwater levels rise first, which induces subsurface flow and leads to a subsequent runoff response in the channel network. The duration of the recession seemed to be controlled by the local drainage conditions (as described by the local slope) and the water input from upstream (as described by the upslope contributing area). Despite large variability, the duration of the recession showed a tendency to increase with TWI.

The correlation between groundwater levels and TWI changed during an event, which does not support a persistent pattern of groundwater levels as assumed in TOPMODEL. Differences in groundwater response and a low degree of hydrological connectivity at the beginning of an event violated the assumption of a series of steady state successions. This concept is more realistic during the groundwater recession, which was indicated by stronger correlations between groundwater levels and TWI. More generally speaking, the assumption of steady-state successions was best met during conditions of small changes in runoff and presumably small changes in groundwater levels.

Weaker correlation between groundwater levels and topography, reported in other studies, suggest that subsurface runoff processes may be different in contrasting catchments. In steep mountain headwater catchments with low permeability soils (e.g., the Gleysols in this study catchment), we expect shallow perched groundwater systems, while saturation and subsequent lateral subsurface flow in transmissive soils occurs at deeper depth. Therefore soil properties, as well as topography and permeability of the bedrock or deep impeding soil layer, are expected to be of less importance in environments with low permeable soils (McDonnell 1990; Uchida et al. 2003; Tromp-van Meerveld et al. 2007).

Our dataset, comprising of more than 50 monitoring sites and 133 rainfall events, allowed us to reveal strong correlations between groundwater response timing and topographic predictors, rainfall characteristics and antecedent wetness, which might not have been clear for a smaller dataset. These results may help us to predict groundwater response patterns in catchments with no monitoring sites. We expect our findings to be applicable in other humid mountain headwater catchments with low permeability soils and shallow groundwater tables as they can be interpreted in physical terms. Our results also show that the TWI assumptions might be useful simplifications for modeling applications in catchments with shallow groundwater levels during periods of the groundwater recession but not at the beginning of an event or during long dry periods. This has implications for using TWI based models to predict the spatial patterns of groundwater levels, their connectivity and the catchment runoff response.

References

Anderson, A.E., Weiler, M., Alila, Y. & Hudson, R.O., 2010. Piezometric response in zones of a watershed with lateral preferential flow as a first-order control on subsurface flow. Hydrological Processes, 24(16), pp.2237-2247.

Anderson, M.G. & Burt, T.P., 1978. The role of topography in controlling throughflow generation. Earth Surface Processes and Landforms, 3, pp.331-334.

Bachmair, S. & Weiler, M., 2012. Hillslope characteristics as controls of subsurface flow variability. Hydrology and Earth System Sciences, 16(10), pp.3699-3715.

Beven, K.J. & Kirkby, M.J., 1979. A physically based, variable contributing area model of basin hydrology. Hydrological Sciences Bulletin, 24(1), pp.43-69.

Burt, T.P. & Butcher, D.P., 1985. Topographic controls of soil moisture distributions. Journal of Soil Science, 36(1978), pp.469-486.

Detty, J.M. & McGuire, K.J., 2010. Topographic controls on shallow groundwater dynamics: implications of hydrologic connectivity between hillslopes and riparian zones in a till mantled catchment. Hydrological Processes, 24(16), pp.2222-2236.

Dhakal, A.S. & Sullivan, K., 2014. Shallow groundwater response to rainfall on a forested headwater catchment in northern coastal California: implications of topography, rainfall, and throughfall intensities on peak pressure head generation. Hydrological Processes, 28, pp.446-463.

Evans, S., 1980. An integrated system of terrain analysis and slope mapping. Zeitschrift für Geomorphologie, Supl. Bd.(36), pp.274-294.

Fannin, R.J., Jaakkola, J., Wilkinson, J.M.T. & Hetherington, E.D., 2000. Hydrologic response of soils to precipitation at Carnation Creek, British Columbia, Canada. Water Resources Research, 36(6), pp.1481-1494.

Feyen, H., Wunderli, H., Wydler, H. & Papritz, A., 1999. A tracer experiment to study flow paths of water in a forest soil. Journal of Hydrology, 225(3-4), pp.155-167.

Haught, D.R.W. & van Meerveld, H.J., 2011. Spatial variation in transient water table responses: differences between an upper and lower hillslope zone. Hydrological Processes, 25(25), pp.3866-3877.

Lana-Renault, N., Regüés, D., Serrano, P. & Latron, J., 2013. Spatial and temporal variability of groundwater dynamics in a sub-Mediterranean mountain catchment. Hydrological Processes, p. n/a.

McDonnell, J.J., 1990. A Rationale for Old Water Discharge through Macropores in a Steep, Humid Catchment. Water Resources Research, 26(11), pp.28212832.

McDonnell, J.J., Sivapalan, M., Vache, K., Dunn, S., Grant, G., Haggerty, R., Hinz, C., Hooper, R., Kirchner, J., Roderick, M.L., Selker, J. & Weiler, M., 2007. Moving beyond heterogeneity and process complexity: A new vision for watershed hydrology. Water Resources Research, 43(7), pp.1-6.

Penna, D., Mantese, N., Hopp, L., Borga, M. & Dalla Fontana, G., 2014. Spatio-temporal variability of piezometric response on two steep alpine hillslopes. Hydrological Processes.

Schleppi, P., Muller, N., Feyen, H., Papritz, A., Bucher, J.B. & Flhler, H., 1998. Nitrogen budgets of two small experimental forested catchments at Alptal , Switzerland. Forest Ecology and Management, 101, pp.177-185.

Seibert, J., Bishop, K. & Nyberg, L., 1997. A test of TOPMODEL's ability to predict spatially distributed groundwater levels. Hydrological Processes, 11, pp.1131-1144.

Tarboton, G.D., 1997. A new method for the determination of flow directions and upslope areas in grid digital elevation models. Water Resources, 33(2), pp.309-319.

Travis, M.R., Iverson, W.D., Gary, H. & Johnson, C.G., 1975. VIEWIT: computation of seenarea, slope and aspect for land-use planning PSW 11, Berkeley, California U.S.A.: USDA - Forest Service.

Tromp-van Meerveld, H.J. & McDonnell, J.J., 2006. Threshold relations in subsurface stormflow: 2. The fill and spill hypothesis. Water Resources Research, 42(2), pp.1-11.

Tromp-van Meerveld, H.J., Peters, N.E. & Mcdonnell, J.J., 2007. Effect of bedrock permeability on subsurface stormflow and the water balance of a trenched hillslope at the Panola Mountain Research Watershed , Georgia , USA. Hydrological Processes, 21, pp.750-769.

Uchida, T., Asano, Y., Ohte, N. & Mizuyama, T., 2003. Seepage area and rate of bedrock groundwater discharge at a granitic unchanneled hillslope. Water Resources Research, 39(1), p.1018.

Weiler, M., McDonnell, J.J., Tromp-Van Meerveld, I. & Uchida, T., 2005. Subsurface Stormflow. In G. M. Anderson & J. J. McDonnell, eds. Encyclopedia of Hydrological Sciences. pp. 1719-1732.

Fig. 1: Time to rise (t_rise) relative to start of the rainfall event as a function of Topographic Wetness Index for S four rainfall event types (type 1a: low-intense/dry, type 1b: low-intense/moist, type 2a: moderate-intense/dry and type 2b: moderate-intense/moist). Grey bars show the inter quartile range (IQR), black line: LOWESS curves fitted to all median points; r_s : Spearman Rank Correlation Coefficient and associated p-value.