1 Factors Controlling Riffle-Scale Hyporheic Exchange Flows and Their Seasonal Changes in a Gaining Stream: A Three- Dimensional Groundwater Flow Model R.G. Story, K.W.F. Howard, and D.D. Williams 2003 Geology 230 Kent E. Parrish, P.G., C.Hg February 14, 2013

2 Lecture Outline Introduction Site Field Methods Model Description Results Discussion Conclusions Limitations

3 Introduction lifeinfreshwater.org.uk Riffle-pool: Basic unit of exchange area Hyporheic Flow

4 Introduction Harvey and Bencala, 1993 Used a model to predict that, in mountain streams, exchange can occur even during aquifer discharge – Discontinuities in stream gradient (>20% slope) But they did not describe other properties of streambed or aquifer that are required to allow exchange Did not attempt to find relationships between hydrological or geological parameters of the system and vertical or lateral extent of exchange flows

5 Introduction Tracer Studies versus Modeling Tracers have been main tool to compare storage sizes and exchange rates But too many factors to simply describe system Tracers cannot distinguish between surface and subsurface storage Tracers operate from surface water perspective but hyporheic flow is through porous medium Color represents a take-home point

6 Introduction Past Modeling Efforts Hadn’t attempted to evaluate range of controlling factors for exchange Had simulated streams as only one cell wide in models Had been two-dimensional models

7 Introduction This Article’s Objectives 1.Identify hydrological and geological conditions that are required for hyporheic exchange to occur during aquifer discharge 2.Identify key factors that are sufficient to explain seasonal changes in exchange flows 3.Describe differences in vertical versus lateral exchange flows and paths in different parts of streambed

8 Site Speed River, southern Ontario Gravel bed Flows across undulating glacial terrain Low topo relief (2 - 5 m/km) Dolomite aquifer bedrock 20 m below ground surface Domomite overlain by low-K till, kame, and outwash deposits (K = m/s (0.028 ft/d) to m/s [ ft/d)

9 Site Speed River, southern Ontario Stream lies in recent alluvium m deep and 5-10 m wide on each side of the stream ( K = 2 x m/s [57 ft/d] ) At Site, stream is 6 m wide; 0.15 – 0.35 cm deep in summer Summer baseflow = 0.1 m 3 /s Winter baseflow = 2 – 3 times summer baseflow

10 Field Methods Field Studies Used nested minipiezometers (dia. 1.3 cm) Each piezo had single 5 mm opening Nest consisted of piezos 0, 20, 40,60, 80, or 100 cm below stream bottom Nests installed about 1 m apart in two transects – Across stream at upstream end of riffle – Along axis of stream between upstream and downstream end of riffle

11 Field Methods Field Studies Measured hydraulic head distributions in 3-D in one 13- m-long riffle site

12 Field Methods Field Studies Data collected over four seasons (Aug 1996 to July 1997 Additional measurements over high and low base flow periods until November 1998

13 Field Methods Field Studies NaCl tracers used to confirm flow directions interpreted from piezo heads Injected in up-welling and down-welling zones, at center and near sides of stream channel Measured EC Time to peak EC was used to calculate K – K = - ( vn e )/ i ; where, v = flow path length/tracer travel time » n e = effective porosity of sediments » i = hydraulic gradient between release and detection points

14 Field Methods Field Studies Measured temperature variations over 24-hour period in stream channel and across upper transect Measured every 3 hours Criterion for when surface water reached measurement depth: when piezo temp cycle had amplitude > 10% of stream channel temp cycle Criterion based on Silliman et al. (1995)

15 Field Methods Field Studies Time delay between temp peaks in stream channel and in each piezo was used to calculate a first-order travel time estimate for surface water down-welling NaCl not conservative tracer so only first-order estimate possible

16 Model Description Model Domain 1,000m x 500 m E and W boundaries were Speed River catchment edges N and S boundaries parallel to groundwater flow Grid cell sizes – 8 m x 8 m across domain – Refined to 1 m x 1 m at the riffle 12 model layers Dolomite aquifer bottom, K x,y = m/s (0.28 ft/d); K z = m/s (0.028 ft.d)

17 Model Description K x,y of Layer 12 dolomite estimated from specific capacity tests Layers 1-3 till and outwash with K = 3 x m/s (0.008 ft/d) estimated from slug tests Layers 4-11 till and outwash with K = m/s (0.28 ft/d) and m/s (2.8 ft/d) “ calibrated” layers 1-3 to match vertical gradients across piezos and historical stream discharge data (Water Survey of Canada, 1992) 3-9 horizontal 0.25 m thick near stream High K zone (est. via salt tracers) along and 1.5 m beneath stream K = 2 x m/s (57 ft/d)

18 Model Description

19 Model Description Note thin, constant thickness cells. Allowed finer-scale modeling

20 Model Description Stream as constant heads

21 Model Description

22 Model Description Aquifer Recharge Estimated from stream discharge records Recharge = summer and winter base flow / catchment area Applied as constant flux to top model layer Model Application Steady state runs (36) Varied parameters to simulate winter/summer conditions – Stream heads winter and summer (factor of 2) – Groundwater discharge doubled (field-based) [raised heads by 2 m and doubled aerial recharge. Then doubled groundwater discharge again K varied over 2 orders of magnitude (field-based)

23 Field Results

24 Field Results Temp expressed as % of variations in stream temperature

25 Model Results Key Model Factors (model most sensitive to these) Hydraulic conductivity Boundaries of hyporheic zone – Head difference between upstream and downstream ends of riffle – Flux of groundwater entering the alluvial zone from the sides and beneath – Steeper Summer stream gradient causes increased exchange flux – Hyporheic flow travel times related to both flow velocity and distance

26 Model Results Vertical versus lateral exchange flows Vertical exchange in the channel occurred more consistently than later flows into the stream banks Downwelling extended to the bottom layer of the alluvial deposits in majority of simulations

27 Model Results Note flow direction change when K exceeded threshold

28 Model Results

29 Model Results Summer Heads (2 x Steeper stream gradient than Winter) Hyporheic Flux vs K

30 Model Results Winter Heads Hyporheic Flux vs K

31 Model Results Summer Heads (Steeper stream gradient) Hyporheic Zone Depth vs K

32 Model Results Winter Heads Hyporheic Zone Depth vs K

33 Model Results Hyporheic Travel Time vs K Summer Heads (Steeper stream gradient)

34 Model Results Winter Heads Hyporheic Travel Time vs K

35 Model Results Little exchange

36 Model Results High exchange

37 Model Results Summer Heads (Steeper stream gradient) Hyporheic Flux vs K

38 Model Results Winter Heads Hyporheic Flux vs K

39 Model Results All Seasons Upstream Transect

40 Conclusions 1.Low-gradient streams, riffle-scale exchange flows are possible only when high-permeability materials ( K x,y =10 -5 m/s [2.8 ft/d) ]) 2.Moderate- to low-permeability catchment K x,y = m/s (0.28 ft/d) to K = 3 x m/s (0.008 ft/d) with alluvial sediments surrounding the stream 3.Amount of exchange flux, lateral and vertical extent of surface water penetration, and travel times through hyporheic zone determined by three parameters: – K of the alluvium – GW flux to the alluvium – Hydraulic gradient between riffle ends

41 Conclusions 4.Exchange flows tend to be stronger but more variable at the sides than at the center of the stream channel 5.Hydraulic conductivity of the streambed can vary by up to 40% with season due to changes in water temperature

42 Model Limitations 1.Model not calibrated (fatal flaw) 2.Homogenous K in streambed 3.No bank storage (not transient model) 4.Isotropic conditions in high K zone around stream (unrealistic) 5.Use of constant head cells allows unrestricted flow into/out of model. Can lead to unrealistic water balance. 6.Drastic changes in model descretization can lead to numerical dispersion (unrealistic and instable results)

43 Suggested Improvements 1.Calibrate the model using field data 2.Perform more rigorous sensitivity analysis 3.Produce table(s) and figure(s) of the calibration and sensitivity analysis 4.Check and present internal water balance of calibrated model 5.Improve model descretization 6.Simplify figures

44 References Harvey, J.W. and K.E. Bencala, 1993, The effect of streambed topography on surface- subsurface water exchange in mountain catchments, Water Resources Research, v. 29, p Silliman, S.E., J. Ramirez, and M.G. Scafe, 1997, The hydrogeology of southern Ontario, Ontario Ministry of Environment and Energy, Toronto, Canada. Water Survey of Canada, 1992, Historical streamflow summary-Ontario, Inland Waters Directorate, Water Resources Branch, Department of the Environment, Ottawa, Ontario, Canada.