Carl Mendoza
Carl Mendoza
Carl Mendoza
PII: S0022-1694(15)00377-7
DOI: http://dx.doi.org/10.1016/j.jhydrol.2015.05.027
Reference: HYDROL 20462
Please cite this article as: Thompson, C., Mendoza, C.A., Devito, K.J., Petrone, R.M., Climatic controls on
groundwater-surface water interactions within the Boreal Plains of Alberta: Field observations and numerical
simulations, Journal of Hydrology (2015), doi: http://dx.doi.org/10.1016/j.jhydrol.2015.05.027
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1 Climatic controls on groundwater-surface water interactions within the Boreal Plains of
2 Alberta: Field observations and numerical simulations
a
5 Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton, AB, T6G
6 2E3, Canada
b
7 Department of Biological Sciences, University of Alberta, Edmonton, AB, T6G 2R3, Canada
c
8 Cold Regions Research Centre, Wilfrid Laurier University, Waterloo, ON, N2L 3C5, Canada
*
9 Correspondence to: Craig Thompson, Department of Earth and Atmospheric Sciences,
10 University of Alberta, 1-26 Earth Sciences Building, Edmonton, AB, T6G 2E3, Canada.
12
15
18
1
1 Abstract
2 Hydrologic data were collected over eleven years at a catchment situated in low permeability
3 glacial terrain within the Boreal Plains (Alberta, Canada) to evaluate the hydrologic interactions
4 occurring between landscape units (i.e., shallow pond, extensive peatlands, aspen forested
6 groundwater-surface water model to evaluate key landscape features that allow these
7 ecosystems to persist within the sub-humid climate. Study results show that the dynamic
8 interactions between the pond and peatlands are driven by precipitation and
9 evapotranspiration. As a result, pond and peatland water levels reflect recent climatic trends.
10 Limited hillslope contributions to the peatlands occur, indicating they are not required within
11 this climatic setting for long-term maintenance. Instead, the peatlands conserve water within
12 the landscape and supply it to adjacent landscape units. By contrast, the pond and the aspen
13 forested hillslopes are dominated by high rates of evapotranspiration, and represent net water
14 sinks within the landscape. Further simulations indicate these hydrologic systems are sensitive
15 to pond and peatland evapotranspiration rates, and the hydraulic conductivity of the underlying
17
19 hydrology; hydrogeology
20
2
1 1.0 Introduction
2 The Boreal Plains of northern Alberta, Canada, are host to numerous wetlands, small
3 lakes, and ponds interspersed within forested uplands. The wetlands provide important
4 seasonal habitat for migratory birds and also represent a significant carbon pool (Smith and
5 Reid, 2013; Gorham, 1991). However, they are under significant development pressures as a
6 result of expanding petroleum developments and timber harvesting (Devito et al., 2012).
7 Furthermore, large areas have been disturbed by open-pit mining of oil sands which will require
8 reclamation on an unprecedented scale over the next 30 to 50 years (Kelln et al., 2008).
9 Consequently, characterization of the processes governing the movement of water within these
10 ecosystems is of pressing importance for both managing existing ecosystems and restoring
12 Ecosystems within the Boreal Plains are sustained by sub-humid climatic conditions,
13 where annual precipitation (P) is commonly less than potential evapotranspiration (PET;
14 Marshal et al., 1999). Water deficit conditions occur frequently, making them highly vulnerable
15 to developments that may alter their water budget. A large portion of annual P falls during the
16 summer months (Marshal et al., 1999) when the PET is greatest (Petrone et al., 2007; Brown et
17 al., 2014). Therefore, in contrast to more humid regions such as the Boreal Shield, limited
18 water is available to replenish the subsurface during the growing season in most years. Instead,
19 the majority of groundwater recharge is derived from spring snowmelt and rain events
20 occurring outside the growing season (Smerdon et al., 2008; Redding and Devito, 2011),
21 particularly in the forested uplands where evapotranspiration (ET) by species such as aspen
3
1 The elevated upland ET, combined with the deep glacial soils within the region (Vogwill,
2 1978), result in deep upland water tables that do not follow topography and often decline away
3 from adjacent ponds and peatlands (Ferone and Devito, 2004; Smerdon et al., 2005).
4 Consequently, following snowmelt and rain events the upland hydrologic response is
5 dominated by fluctuations in storage, with little potential for generation of overland flow
6 (Devito et al., 2005; Redding and Devito, 2008). Thus, pond and peatland water levels are
7 dominated by the atmospheric fluxes of P and ET (Ferone and Devito, 2004; Smerdon et al.,
9 settings.
10 Despite their primary reliance on P for maintenance within the dry sub-humid climate,
11 pond and peatland ecosystems comprise a large and essential portion of the landscape within
12 the Boreal Plains. Restoration of these ecosystems will represent an important challenge in the
13 coming years, particularly in the oil sands mining region where complete reconstruction of the
14 landscape will be required (Devito et al., 2012). The re-establishment of peatland terrain has
15 not been previously attempted (Price et al., 2010), thus questions remain regarding the key
16 features that may ultimately lead to successful reclamation, particularly within the context of
17 the relatively dry sub-humid climate. Research aimed at addressing these questions has been
19 landscape units (i.e., pond, peatland, and hillslope; Devito et al., 2012) for relatively short
21 This study evaluated the hydrologic linkages occurring between landscape units at a
4
1 hydrologic dataset was collected over eleven years at a heterogeneous low permeability
2 catchment within the region to assess the interactions occurring between landscape units and
3 to evaluate how they vary over a range of climatic variability. The empirical dataset was used
4 to develop numerical models capable of representing the dominant hydrological processes and
5 to evaluate the key features of the landscape that allow these ecosystems to persist within the
6 sub-humid climate. The goals of the study were to assess how the interactions vary between
7 landscape units within the observed climatic cycles, to evaluate the sensitivity of the system to
8 a range of soil characteristics and catchment configurations commonly found within the Boreal
9 Plains, and to address how catchment hydrology within the region is influenced by the
13 56.07 N, Long: 115.5 W; Figure 1), situated within the Utikuma Region Study Area (URSA). The
14 URSA is located within the plains region of the western Boreal Forest approximately 350 km
16 discontinuous permafrost region (Woo and Winter, 1993). Previous studies at the URSA have
17 investigated pond water budgets situated in a range of glacial landforms (Ferone and Devito,
18 2004; Smerdon et al., 2005); groundwater-surface water interactions in glacial outwash terrain
19 (Smerdon et al., 2007); groundwater recharge and runoff dynamics (Smerdon et al., 2008;
20 Redding and Devito, 2008, 2010, 2011); the variability in ecosystem ET (Petrone et al., 2007;
21 Brown et al., 2010, 2014), and the influence of subsurface ice formation (Petrone et al., 2008;
5
1 The URSA experiences cold winters and warm summers, with average January and July
2 temperatures of -14.5°C and 15.6°C. The climate is sub-humid, with annual PET (517 mm)
3 exceeding annual P (485 mm) in an average year (Marshal et al., 1999). However, wetter years
4 where P exceeds PET occur every 10 to 25 years (Mwale et al., 2009). A large fraction of P (i.e.,
5 50-60%) falls during the summer months when the PET is greatest (Petrone et al., 2007; Brown
6 et al., 2014); therefore, the majority of groundwater recharge occurs outside of the growing
10 (Ferone and Devito, 2004). The catchment is characterized by deep, low permeability glacial
11 disintegration moraine deposits with subtle topographic relief. The till is heterogeneous with
12 sand and silt lenses interspersed throughout the soil column (Figure 2), and ranges in depth
13 from 40 m to 50 m (Pawlowicz and Fenton, 2005). The till is underlain by marine shales of the
15 Mature forests consisting of predominately aspen cover the hillslopes (62% of area). In
16 these areas, near-surface soils are classified as gray luvisols and overlay oxidised clay till 5 m to
17 8 m in depth with unoxidised clay till beneath (Ferone and Devito, 2004). In flatter areas and
18 depressions, peatlands with low density stunted black spruce (32%) are the dominant surface
19 cover and directly overly unoxidised clay till. The organic materials within the peatlands range
20 in depth from 2 to 5 m. Groundwater and surface water flow within the peatlands are
21 influenced by seasonally frozen lenses, which can persist well into the summer months
22 (Petrone et al., 2008; Brown et al., 2010). In the vicinity of the pond (6%), the peatlands
6
1 transition to gyttja that extends to greater than 3 m depth. Surface drainage within the
2 peatlands is generally poorly developed (Ferone and Devito, 2004) and frequently modified by
3 beaver activity.
7 checked using manual measurements from bulk rain gauges distributed across the site. Snow
8 surveys were used to measure the snowpack each winter and spring. The surveys indicated
9 similar snow depths accumulated within each landscape unit, with little drifting observed.
10 Snow water equivalent (SWE) was determined from composite samples obtained during each
11 survey.
12 Rainfall interception within aspen and spruce stands was estimated using integrated
13 throughfall measurements. Throughfall was collected in 10 m long troughs placed below the
14 shrub understory within both forest types. Bulk samples were collected at roughly two week
15 intervals following the snowmelt period before aspen leaf on through to mid-September.
16 Pond evaporation was measured using a Class A evaporation pan which was partially
17 submerged to maintain thermal equilibrium between the pond and pan. Daily to continuous
18 water level measurements within the pan were obtained throughout the pond’s ice-free
19 season. Additional evaporation data from Pond 43, peatlands, and the aspen forests have
20 previously been reported (Petrone et al., 2007; Brown et al., 2010, 2014).
7
1 3.2. Surface Water
2 Continuous pond stage measurements were obtained during the ice-free season using
3 pressure transducers and augmented with manual measurements at established staff gauges.
4 The bathymetry of Pond 43 was determined using depth to gyttja measurements from the
5 water surface along two perpendicular transects. The bathymetry measurements were
8 Surface water flows and water levels in the pond and peatland channels were measured
9 daily to weekly at V-notch weirs throughout the study period. The frequency of measurements
10 was higher during the spring melt and following summer rain events to better capture
11 increased flow rates during these periods. The resulting dataset was interpolated between
12 measurements to estimate daily surface water inflows and outflows from the pond.
13 3.3. Groundwater
14 Groundwater levels were monitored throughout the duration of the study period in
15 monitoring well and piezometer nests installed within each material. A total of 180 monitoring
16 points were installed to a maximum depth of 23 m (Figures 1 and 2). Groundwater levels were
17 typically measured weekly to biweekly from spring through autumn with less frequent
20 The saturated hydraulic conductivity (Ks) of the glacial substrate, peat, and gyttja has
21 been previously reported (Ferone and Devito, 2004; Petrone et al., 2008; Redding and Devito,
22 2008). The dataset was augmented with the results of slug tests conducted at 51 additional
8
1 monitoring points distributed within each material. The dataset indicates that the Ks within the
2 peat and gyttja displays a distinct decreasing trend with depth. Similar results have been
3 reported by other researchers (e.g., Beckwith et al., 2003; Quinton et al., 2008). Within the
4 glacial till, results of infiltration tests indicate a similar decline in Ks with depth in the near-
5 surface materials, ranging from 10-3 m/s in the upper 0.1 m to 10-8 m/s at 0.75 m depth
6 (Redding and Devito, 2010). However, no distinct depth trend was observed from slug tests
7 conducted at greater depth. Instead, significant variation in Ks (i.e., 10-5 m/s to 10-10 m/s) is
11 occurring between the different landscape units. The goal of the simulations was to reproduce
12 observed hydrologic responses to climatic forcing and evaluate the sensitivity of the system to
13 changes to specified parameters and boundary conditions, rather than to simulate hydraulic
14 heads exactly at a specific well within the heterogeneous subsurface. Therefore, the models
15 were simplified representations of the hydrologic system that retained sufficient complexity to
16 simulate the dominant behavior, but may not capture finer-scale intricacies present at a given
17 location.
18 Simulations were performed from January 1, 2000 to March 15, 2011. Models were
19 developed for two transects to bracket the range in hydrologic conditions present at the site
20 (Figure 1 and 2). The location of each transect was selected to follow interpreted surface
21 water/groundwater flow paths while incorporating the largest number of data points available.
22 Each transect features a pond near the center of the domain along with peatlands of varying
9
1 depth and length on either side. Section A includes an aspen forested hillslope at its northern
2 end that rises in elevation approximately 9 m above the pond. Section B differs by not having a
3 hillslope in the vicinity of the pond and by terminating at an adjacent pond (i.e., Pond 48).
6 the diffusion-wave approximation of the Saint Venant equations in the surface water domain
7 and the variably-saturated Richard’s equation in the subsurface domain. HGS was selected as
8 the numerical simulator due to the integrated simulation capabilities of the code which allow
9 the user to specify boundary conditions in the form of atmospheric fluxes, rather than
10 specifying groundwater recharge rates and surface water levels that necessitate additional a
11 priori assumptions. HGS has been applied to a wide range of problems, ranging in scale from
13 The finite-element mesh for each 2D model was discretized using uniform 1 m nodal
14 spacing horizontally. Vertically, nodal spacing varied from 0.05 m to 0.1 m in the upper 0.5 m.
16 0.25 m. The surface of the mesh was set to ground surface using the 1 m resolution LiDAR
17 surface with the pond’s bathymetry incorporated. Glacial sediments within the modeled area
18 reach depths of 50 m, thus no natural near-surface hydrologic boundary was present to define
19 the base of the model. Instead, the base of the mesh was set to an arbitrary elevation of
10
1 4.1. Material Properties
2 Three hydrogeologic units were specified within the models including peat, gyttja, and
3 glacial till. Parameter zones within each unit were defined based on observed depth trends in
4 Ks. Within the peat and gyttja, the Ks was specified to decrease with depth (Table 1). An
5 anisotropy ratio of 10:1 was assumed for both materials (Beckwith et al., 2003; Nagare et al.;
6 2013). The remainder of the model was specified to be composed of glacial till. Highly
7 decomposed organic materials at the edge of the peatland were also included within this unit,
8 as their hydraulic properties were found to be similar to the till (i.e., low permeability and
9 porosity, high bulk density). The glacial till was subdivided into three zones with Ks decreasing
10 with depth. The depth of the upper two layers was specified to include near-surface sand and
12 permeability resulting from fractures in the till, decayed roots, and animal burrows (Hayashi
13 and van der Kamp, 2009). The remainder of the material was assigned a lower Ks of
14 1 x 10-8 m/s. An anisotropy ratio of 100:1 was assumed for each till layer and evaluated during
15 model calibration. An additional 0.1 m thick forest floor was specified above the glacial till. Soil
16 water characteristic and hydraulic conductivity curves were derived from the literature for the
17 peat and gyttja (Silins and Rothwell, 1998; Price et al.; 2010) and glacial till using the properties
18 of a clay loam (Carsel and Parrish, 1988). Limited data was available to characterize the
19 porosity and specific storage of each material; therefore, measurements were augmented with
20 literature values (Redding and Devito, 2006; Petrone et al., 2008; Quinton et al., 2008; Smith
21 and Wheatcraft, 1993). Within the peat and gyttja, the porosity and specific storage were
11
1 specified to follow a similar decreasing with depth trend as Ks; within the heterogeneous glacial
5 fluxes and hydraulic heads. Specified fluxes consisting of daily rainfall and snowmelt were
6 applied to the surface of each model (Table 2). Rainfall and snowmelt rates differed between
7 landscape units to account for varying P interception, snow sublimation, and frozen season
8 length. Within the pond, all rainfall was assumed to reach the surface during the non-ice
9 covered period. The rainfall rate within the forested areas was calculated based on a simple
10 relationship used to relate measured interception, which was found to be similar in aspen and
11 black spruce areas, to rainfall intensity and season. For the summer months when the ET is
12 highest, the daily canopy interception was assumed to be 2 mm, with the remaining quantity of
13 rainfall reaching the ground. During the cooler month of May and the remaining winter
14 months, the interception was reduced to 0.5 mm and 0 mm, respectively. The resultant
15 calculated throughfall was found to provide a reasonable match to measured values (Figure 3)
17 Spring snowmelt was applied based on snow surveys and measured changes in water
18 levels within the pond and observation wells. The depth of the winter snow pack was manually
19 calculated in each landscape unit and applied as snowmelt over a 1 to 3 week period during the
20 spring. Rain falling during the snow-covered period was assumed to be incorporated into the
21 snowpack. Snow surveys indicated similar SWE volumes were present within each landscape
22 unit in the spring, suggesting that varying snow interception was largely offset by higher mid-
12
1 winter melt and evaporation rates in more open areas (Koivusalo and Kokkonen, 2002).
2 Therefore, similar combined snowmelt evaporation and sublimation equal to 25% and 30% of
3 the annual snowfall were assumed within the respective forested and pond areas.
4 Daily ET was also applied to the models’ surface from the spring through the fall. Within
5 the pond, ET was limited to ice-free periods, which were estimated from measured air
6 temperatures and observed pond water levels. In the aspen-dominated areas, ET was applied
7 from May 15 to September 15 (Brown et al., 2014); whereas, in the peatlands ET was specified
9 Measurements of ET were not available for the entire study period; therefore, the
11 Calculated values were found to be similar to measured pond ET (Petrone et al., 2007). The
12 generated dataset was then scaled to account for differing ET within the peatlands (Petrone et
13 al., 2007; Brown et al., 2010) and aspen areas (Brown et al., 2014). The resultant applied ET
14 ratios between the peatlands and pond and between the aspen and pond were 0.5 and 1.3,
17 different characteristics of the predominant vegetation type. Within the aspen hillslopes and
18 black spruce peatlands, the ET was distributed within the upper 3 m and 0.5 m, respectively,
19 based on typical rooting depths (Debyle and Winokur, 1985; Lieffers and Rothwell, 1987). In
20 the pond, the ET was limited to the upper 0.1 m. In all zones, ET was focused at near-surface
21 nodes through use of a cubic function available within HGS. The actual ET (AET) was limited by
13
1 the available water, with the maximum rate occurring at saturations greater than field capacity
2 (i.e., -33 kPa) and decreasing to zero at the permanent wilting point (i.e., -1500 kPa).
3 Within the pond, the specified P and ET fluxes also included measured stream flow into
4 and out of the pond. The continuous stream flow dataset was converted to an areal flux using
6 Additional boundary conditions applied to the models included specified heads at the
7 base of the domain, a specified head at the eastern edge of Section B where it terminates at
8 Pond 48, and no flow conditions along the remaining lateral edges of the models. At the base,
9 a uniform hydraulic head was applied to simulate connection with the regional groundwater
10 flow system. Measured hydraulic heads from deeper wells (i.e., >15 m depth) showed little
11 fluctuation throughout the year; therefore, a constant head was assumed to provide a
12 reasonable representation. At the eastern edge of Section B, daily specified hydraulic heads
13 were applied within the peatland to simulate the adjacent Pond 48. Along the remaining lateral
14 edges of the models, no flow boundaries were specified as groundwater flow has been
16 Initial conditions were generated by specifying the pressure head across the surface of
17 the models and running them to steady state. Pressure heads were determined from measured
18 hydraulic heads nearest to the start of the simulation period and interpolated between
19 measurement points. Boundary conditions applied to the base (i.e., constant head) and lateral
20 edges (i.e., no flow and specified head at Pond 48) of the models were unchanged from the
21 transient simulations.
14
1 5.0 Results and Discussion
4 annual P was less than average in all but 2005 (Table 3, Figure 5). The driest years occurred in
5 2001 and 2002, with both years receiving less than 60% of average P. Wetter climatic
6 conditions occurred in consecutive years in 2007 and 2008 before dropping again in the
7 following years. Despite the variability in annual climatic conditions, rainfall patterns were
8 relatively consistent with the majority of daily rainfall occurring in small events. On average,
9 90% of days with rain received less than 10 mm, with 51% of total annual rainfall occurring on
10 days receiving less than 10 mm. Larger storms occurred infrequently, with daily rainfall totals
11 exceeding 30 mm occurring on a total of 5 days over the duration of the study period, resulting
12 in little potential for overland flow generation from the hillslopes (Redding and Devito, 2008).
15 (Figure 4). Water levels in both landscape units reflect recent trends in climate variability
16 (Figure 5), although the correlation is stronger in the pond (i.e., correlation coefficient (R) =
17 0.65) than the peatlands (R = 0.40). Seasonally, peatland water levels were generally higher
18 than the pond following spring snowmelt. The resulting groundwater flow direction was from
19 peatland to pond, with the annual discharge period varying in duration in response to the
20 frequency and magnitude of early spring rains. Pond discharge from the peatlands continued
21 through the growing season following drier periods when pond water levels were low (e.g.,
22 2002 to 2003) as the deeper, lower Ks peat layers maintained peatland water levels above the
15
1 pond. In contrast, lateral groundwater flow from the pond to the peatlands was observed from
2 late summer through spring following the relatively dry 1999 growing season (Ferone and
3 Devito; 2004); however, pond water levels were elevated during this period by comparison.
4 During wetter periods when the pond was full (e.g., 2007 to 2008), pond-peatland interactions
5 were much more dynamic, with hydraulic gradients reversing frequently in response to rain
6 events. However, as the growing season progressed, peatland water levels generally declined
7 below the pond as previously observed (Ferone and Devito; 2004), resulting in lateral
8 groundwater flow from the pond to the peatlands throughout a large portion of the year. In
9 contrast to horizontal hydraulic gradients, water level data collected within the peat, gyttja, and
11 occurred regardless of climatic conditions, consistent with the interpretation that the area
13 Within the hillslope, water level elevations were generally well below those observed in
14 both the peatland and pond as previously observed (Ferone and Devito), indicating that the
15 hillslope acted as a net water sink to the peatland. All wells remained dry throughout the study
16 period except at the hillslope toe. At the toe, wells were dry prior to the spring of 2003 (Figure
17 4), indicating that the water table declined rapidly at the peatland edge to a depth of at least 2
18 to 3 m below the base of the pond. Annual snowfall during this period was well below average,
19 suggesting limited water was available to recharge the glacial materials. As a result, the
20 direction of groundwater flow was always from the peatland to the hillslope regardless of
21 season. In 2003, spring water levels at the toe of the hillslope increased to within about 1 m of
22 those in the peatland following an average snow fall year. However, a rapid water level decline
16
1 was observed throughout the growing season despite near average rainfall, suggesting that
2 most of the water was either removed through ET or incorporated into storage. Consequently,
3 by mid-July the water table dropped below the base of the well located at the hillslope toe.
4 Water levels at the hillslope toe remained low until the spring of 2005, when they rose
5 above those in the adjacent peatland following snowmelt. The increase in water level followed
6 another average snow fall year that was preceded by a relatively wet autumn, suggesting that
7 the majority of groundwater recharge in the aspen hillslopes occurs outside of the growing
8 season in the spring and late fall, consistent with the findings of other studies in the URSA
9 (Smerdon et al., 2008; Redding and Devito, 2011). Comparable responses were observed in the
10 relatively wet years of 2006 to 2009, with water levels at the toe of the hillslope exceeding
11 those in the peatland each spring. Thus some groundwater discharge from the hillslope to the
12 peatland likely occurred, although the combination of generally low Ks glacial till and seasonally
16 Section A; 43 in Section B) along with the computed stage at Pond 43. Calculated calibration
17 statistics (i.e., R, residual mean (rm), root mean square error (RMS), and RMS normalized by the
18 range in observed water levels (NRMS)) indicate the models were able to provide a reasonable
19 representation of the hydrologic system (i.e., R = 0.65; RMS = 0.29 m; Figure 6d). Water levels
20 within the pond were particularly well-represented (i.e., R = 0.97; RMS = 0.06 m; Figure 6a),
21 although some bias is present at shallower pond depths where the predicted stage is somewhat
17
1 0.17 m; Figure 6b), with the RMS equal to about half the range of annual peatland water level
2 fluctuations. The largest residuals were generally from deeper monitoring wells within the
3 glacial till (i.e., R = 0.47; RMS = 0.50 m; Figure 6c), where considerable heterogeneity is present.
4 Temporally, simulated trends in water levels were found to be in good agreement with
5 measured values (Figure 7). Computed residuals typically decreased from spring to fall when P
6 inputs were lowest, with similar error values obtained in both dry and wetter years. At the
7 hillslope toe in Section A, the general trend of a rapidly rising water level following snowmelt
8 and a continuous decline throughout the growing season is well-represented by the model
9 (Figure 7e). Water levels at this location were over-predicted early in the simulated period,
10 which likely occurred as a result of the initial conditions in the hillslope being too wet. In later
11 years, the predicted peak of the spring water level rise differed from measured values in some
12 years. This may be due to a number of factors not incorporated into the model. Such factors
13 might include drifting of snow, higher Ks lenses or fractures that lead to more rapid
14 transmission of water to this location, the presence of frozen peat that may act as a dam
15 preventing water exchange, or flow contributions from outside of the assumed 2D flow system.
16 Simulated hydraulic heads during the winter and early spring are not as well represented, as
17 flow of water from the peatlands towards the hillslope leads to an increase in water level not
18 observed in the field. This discrepancy occurs as a result of not including freezing and thawing
19 within the simulations, as the peatlands freeze solid throughout the winter preventing water
21 Above the hillslope toe, all wells within the hillslope remained dry throughout the
22 duration of the simulation, with the simulated water table elevation declining with distance
18
1 upslope. The high aspen ET and thick unsaturated zone prevented significant recharge from
2 reaching the water table, resulting in minimal fluctuations in groundwater levels and no
3 overland flow generation. Thus, results of the simulations suggest that significantly wetter
4 climate cycles would be required before noticeable rises in groundwater levels could occur.
5 The simulated AET integrated over the entire hillslope was limited by soil moisture availability,
6 ranging from 56-76% of the potential rate (Table 3), and was found to be similar to measured
7 rates at an adjacent catchment (Brown et al., 2014). In contrast, simulated AET within the pond
8 and peatlands was predicted to equal the potential rate throughout the study period.
9 Despite neglecting the effects of peatland freezing and thawing, observed seasonal
10 trends in hydraulic heads at peatland wells are well replicated (Figure 7a, b, d). The model is
11 able to capture the early spring rise in water levels following snowmelt and the general decline
12 throughout the growing season. Periodic smaller scale increases in water levels following larger
13 rain events are reasonably well captured, although some deviations are apparent which may be
15 The stage of Pond 43 was also well replicated (Figure 7c), and was found to mimic
16 measured water levels throughout most of the simulation. Some discrepancy is apparent
17 during the summers of 2002 to 2004 when the measured water level decline is under-
18 predicted. This may be due to the relatively simple air temperature based method that was
19 used to calculate the daily pond ET. Incorporation of a more rigorous approach that includes
20 the effects of water depth and temperature may have provided a better representation of the
19
1 5.4. Sensitivity Simulations
2 Additional simulations were conducted to evaluate the sensitivity of the hydrologic
4 configurations. Each scenario did not necessarily provide a realistic representation of the
5 system, as simulated water levels at some locations were significantly different than what has
6 been observed. However, the simulated cases do provide insight into the controlling aspects of
8 Results of the sensitivity simulations are described below and summarized in Table 4.
9 The first simulations investigated the sensitivity of the model results to the specified set
10 of boundary conditions. Simulated water levels within the pond were found to be particularly
11 sensitive to the timing and duration of snowmelt in both the pond and peatlands. Modification
12 of the timing of the snowmelt period, particularly within the peatlands, led to simulated peaks
13 in pond stage being out of phase with measured values. Annual water levels within the
14 peatlands were not as sensitive to this boundary input provided that the snowmelt was added
15 in a similar time frame as the observed increase in peatland water level. The overall response
16 of the system was less sensitive to the timing of snowmelt in the hillslope, suggesting limited
17 water was transferred to the adjacent peatland. Modification to the onset of the melt period
18 primarily resulted in changes to the duration of the peak water level at the toe, which was
19 predicted to persist until removal of water from aspen ET commenced each spring.
20 To assess the uncertainty in the tabulated ET boundary fluxes for each landscape unit,
21 simulations were conducted with the specified ET modified by 30%. Pond and peatland water
22 levels were sensitive to this boundary flux. Increasing the ET led to significantly reduced water
20
1 levels (Figure 8). Conversely, where the ET decreased, pond and peatland water levels were
2 predicted to be elevated relative to the base case simulation. Similar responses in water levels
3 at the toe of the hillslope were predicted for these scenarios, although the magnitude of the
4 change was generally less than within the peatlands. Simulation results were less sensitive to
5 increased aspen ET, as the simulated AET was already limited by the available soil moisture
6 (Figure 8). Reducing the aspen ET led to increased water levels at the toe of the hillslope,
7 although this had little effect on water levels within the pond and peatland. Similar to the base
8 case simulation, all wells located further up the hillslope remained dry in both scenarios.
9 The next sensitivity simulations evaluated the influence of the Ks distribution. Results
10 were relatively insensitive to the Ks of the organic materials. Modification of the rate of decline
11 in peatland Ks with depth resulted in minor deviations from the base case. Uniform order of
12 magnitude increases or decreases to the Ks to either the peat or gyttja also led to relatively
13 small changes to pond and peatland water levels. Similarly small variations in water levels at
14 the toe of the hillslope were predicted for changes in peat Ks.
15 Water levels in all materials were most sensitive to the Ks of the glacial till. Increasing
16 the Ks of each till unit by an order of magnitude led to water levels in the pond and peatlands
17 dropping by up to 0.5 m due to increased vertical groundwater flow. Consequently, the pond
18 was predicted to dry out in 2010 for this scenario (Figure 9). Conversely, decreasing the Ks of
19 each till unit by an order of magnitude led to increased pond and peatland water levels of
20 about 0.3 m. Representation of the glacial materials as a uniform material of high (i.e.,
21 1x10-7 m/s) or low (i.e., 1x10-8 m/s) Ks yielded similar results as the respective increased and
22 decreased Ks simulations.
21
1 Groundwater levels at the toe of the hillslope were also sensitive to the Ks of the till. For
2 the increased Ks scenario, water level peaks at the toe of the hillslope following snowmelt were
3 generally lower and the magnitude of annual fluctuations was greater (Figure 9). In contrast,
4 for the decreased Ks scenario, the simulated water level did not fluctuate in response to
5 snowmelt inputs. Instead, the annual influx of water was held in storage prior to being
6 removed by aspen ET during the growing season. A similar response was predicted for
7 simulations that represented the glacial materials as a uniform high and low permeability
8 material.
9 The final sensitivity simulations investigated the individual influence that the pond and
10 peatlands exerted on the hydrological behavior of the system. In the first simulation, the pond
11 was removed from the modeled area to represent catchments that occur within the URSA that
12 have extensive peatlands that do not terminate at a surface water body. For this scenario, the
13 topography within the footprint of the pond was set to a similar elevation as the adjacent
14 peatlands. The base of the organic materials was unchanged from the base case. Results
15 indicated that the pond acts as a net water sink within the landscape, as water levels increased
16 throughout the peatland (Figure 10). Similar conclusions were reached by Smerdon et al.
17 (2005, 2007) at a lake situated within more permeable glacial outwash within the region. For
18 this simulation, peatland water levels rose above ground surface by the spring of 2007, with
19 ponded water predicted to remain to the end of the simulation. Thus, in the absence of a pond,
20 increased overland flow would have been generated for the simulated climatic conditions.
21 The final sensitivity simulation was used to investigate the influence the peatlands exert
22 within the landscape. For this scenario, the peatlands were replaced with glacial till and aspen
22
1 was assumed to be the vegetative cover. Although this scenario is unlikely to occur in an
2 undisturbed setting, it may ensue following disturbance by timber harvesting, as roads and skid
3 trails are commonly developed on flatter areas occupied by peatlands. Furthermore, it could
4 be representative of a potential reclamation scenario within the oil sands mining region if
5 difficulties are encountered in reestablishing peatlands within the landscape. Results indicated
6 that the peatlands play a major role in maintaining the pond, as the predicted pond water level
7 decreased significantly relative to the base case (Figure 10). Extension of the aspen stand led to
8 increased removal of water through ET. As a result, wells that were formerly located within the
9 peatland were predicted to dry out seasonally, leading to increased seepage losses from the
10 pond. Thus, the peatlands act to conserve water within the landscape, providing a counter-
11 balance to the higher ET demands associated with both the pond and hillslope.
14 system to a range of climate conditions. Results indicate dynamic interactions occur between
15 Pond 43 and the peatlands that are driven by the fluxes of P and ET, with little to no additional
16 input from the surrounding uplands. As a result, pond and peatland water levels reflect recent
17 climatic variability (Figure 5), generally following trends in the cumulative difference between P
18 and ET. Comparable responses to climate forcing have been observed in prairie pothole
19 wetlands such as those found in the Cottonwood Lake area of North Dakota. Situated in a
20 regional recharge zone within low permeability glacial terrain, water levels measured over 2
21 decades were found to mirror trends in P (Winter et al., 2001). In contrast to the pond and
22 peatlands, water levels within the hillslope are more indicative of longer-term trends in climatic
23
1 variability. Periods of groundwater recharge within the hillslope are generally limited to the
2 late fall and to the spring following snowmelt as most P inputs are lost to ET over the growing
3 season or taken up into storage. Therefore, a series of wetter years are required before soil
4 moisture thresholds are exceeded and an increase in water levels is exhibited (Redding and
5 Devito, 2010).
6 The rapid response to climate fluctuations at the pond and peatlands is a direct
7 consequence of their position within the context of regional groundwater flow systems. Due to
8 their location at elevated topographic positions, the pond and peatlands and their associated
9 catchments function as regional recharge areas with little potential for discharge of
10 groundwater to the system from outside sources. Thus, they are heavily reliant on P inputs and
11 have limited buffering capacity available to sustain water levels during periods of drought.
12 Wetlands falling into this category are highly vulnerable to changes in climate at both short-
14 The long-term viability of wetlands within the Boreal Plains of Alberta may therefore be
15 in question when considering the potential impact of climate change. Projections for the region
16 indicate a gradual warming trend, with mean annual temperature increases of approximately 2
17 to 5oC by the 2050’s (Barrow and Yu, 2005). It is probable that the length of the growing season
18 will also increase as the climate warms, potentially leading to increased water stress resulting
19 from greater ET losses. This stress may be offset by predicted general increases in annual P.
21 predicting reduced P, particularly during the summer period (Barrow and Yu, 2005). However,
22 the ability of peatlands to restrict water losses during periods of drought (Kettridge and
24
1 Waddington, 2014; Waddington et al., 2015) along with the resilience of the deeper peat to
2 decomposition (Waddington et al., 2015) may render them more resistant to climate change
3 than otherwise expected. Future research will involve developing simulations of predicted
4 climate change scenarios that may prove insightful to the future sustainability of these
5 ecosystems.
8 within the peatlands (Petrone et al., 2008; Smerdon and Mendoza, 2010; Brown et al., 2010).
9 However, despite neglecting their influence within the simulations, water levels within Pond 43
10 and the peatlands were generally well replicated. This may be due to a combination of the
11 thermal behavior of the near-surface peat along with the low Ks of the deeper peat layers and
12 underlying glacial substrate. The higher Ks near-surface peat that transmits most of the water
13 has been observed to thaw relatively rapidly in the spring in response to rising air
14 temperatures, thereby reducing the influence of the frozen material on water distribution
15 within the peatlands. Furthermore, the presence of the deeper, lower permeability peat and
16 glacial till restricts vertical groundwater flow in a similar fashion as the development of frozen
17 peat, thereby maintaining peatland water levels during the frozen period.
18 Although the numerical simulations were able to capture the overall trends in the pond
19 and peatlands, simulated peatland water levels were frequently under- and over-predicted
20 during the spring and late autumn, respectively (Figure 6). This discrepancy may be due to
21 temporal changes to the peat Ks resulting from the seasonal formation of ice. During the spring
22 and early in the growing season the Ks of the peat may be lower than specified within the base
25
1 case simulation due to the presence of ice. As a result, increased vertical groundwater flow is
2 predicted, leading to peatland water levels being under-predicted. Likewise, during the latter
3 part of the growing season, once the ice has melted, the specified Ks of the peat may be too low
4 to allow adequate seepage and lowering of peatland water levels. Thus, the specified Ks
5 distribution may be representative of an average thermal state of the peat allowing adequate
7 In the context of a warming climate, the influence of the seasonally frozen peatlands
8 may be diminished as the duration of the frozen season is reduced. As a result, the volume of
9 water retained in ponds and peatlands within the Boreal Plains may be reduced in the early
10 spring due to increased seepage losses. Future studies will focus on the dynamics of how
11 seasonal ice formation influences water distribution within the peatlands, and how this may be
15 Plains region associated with activities such as timber harvesting and petroleum developments
16 (Devito et al., 2012). In particular, expanding open-pit mining of oil sands in northern Alberta
17 has created the need to reclaim large areas of land (Kelln et al., 2008; Price et al., 2010).
19 capability”, with the expectation that a vegetative regime similar to what was present prior to
20 disturbance will be reestablished (OSWWG, 2000; Alberta Environment, 2008). The similarity in
21 hydrologic characteristics (i.e., climate, geology, and vegetation) of the study area to many of
26
1 the impacted areas allow it to be used as a proxy for future landscape reconstruction and
2 reclamation.
3 This study indicates that peatlands within the Boreal Plains do not necessarily require
4 regional groundwater discharge or significant flow contributions from adjacent uplands for
6 is predicated on the assumption that the peatlands be underlain by materials of sufficiently low
8 Study results also suggest the peatlands conserve water within the landscape, supplying water
9 to adjacent ponds and hillslopes within the sub-humid climatic setting. Thus, when considering
10 the hydrologic impact of forestry, the largest effect to neighbouring wetland ecosystems may
11 not be due to tree removal. The deep upland water table, large available water storage, and
12 rapid aspen regeneration may limit the impacts of harvesting (Devito et al., 2005; Macrae et al.,
13 2005, 2006; Carrera-Hernandez et al., 2011). Instead, the primary impacts of forestry may be
14 due to placement of roads within flat-lying peatlands, which may disrupt shallow groundwater
15 flows and potentially lead to isolated pond and peatland areas that are highly susceptible to
17 Although natural established peatlands in the region have continued to persist despite
18 experiencing extended drought conditions with large water level declines, prolonged periods of
19 drought may produce conditions that are hostile to newly placed peat that is less well
20 established. Price et al. (2010) indicate that peatland water pressure should not drop below
21 -1 m to encourage moss establishment and peat development, although the biological limit
22 may be as low as -6 m (McCarter and Price, 2014). The upper bound of this threshold was
27
1 exceeded frequently during the study period. Therefore, an external source of water may be
2 needed to maintain wetland vegetation if initial reclamation is followed by drier climate cycles.
3 However, contrary to some reclamation guidelines (e.g., OSWWG, 2000; Alberta Environment,
4 2008), the placement of small hills and hummocks within the landscape is unlikely to generate
5 significant water. Aspen forested hillslopes in this climate have been shown to act as water
6 sinks to the wetlands, storing and transpiring P inputs and receiving lateral groundwater flow
7 from the peatlands throughout much of the year. Thus, from a purely “health of peat”
8 perspective, inclusion of small hills and hummocks within the reclaimed landscape may prove
9 detrimental to peatland development, removing water from the system rather than supplying it
11
28
1 Acknowledgements
2 This research was supported by the HEAD 2 (Hydrology, Ecology and Disturbance) NSERC-CRD
3 supported by Canadian Oilsands Network for Research and Development (CONRAD), Ducks
4 Unlimited, Tolko Industries, Alberta-Pacific Forest Industries and the Forest Producers
5 Association of Canada and grants from Canadian Foundation for Innovation (CFI), Syncrude
6 Canada (to K.J. Devito and C.A. Mendoza), and NSERC-PGSD to C. Thompson. The authors wish
7 to thank D. Bryant, W. Bell, G. Tondeleir, C. Fraser, T. Jiang, and R. Bucknell for field assistance,
8 and Carolynn Forsyth for camp facilities at Artis Inn. Support in kind and assistance with deep
9 piezometer installation was provided by G. Jean, J. Pawlowicz, and M. Fenton of the Alberta
10 Geological Survey. We thank John Diiwu at the Government of Alberta for the high-resolution
12 References
13 Alberta Environment, 2008. Guideline for wetland establishment on reclaimed oil sands leases
14 (2nd Ed.). Prepared by Harris, M.L., of Lorax Environmental for the Wetlands and
17 Aquanty, 2013. HGS 2013: HydroGeoSphere user manual, release 1.0, Aquanty Inc., Waterloo,
18 ON, 459 p.
19 Barrow, E., and Yu, G., 2005. Climate scenarios for Alberta. A report prepared for the Prairie
21 p.
29
1 Beckwith, C.W., Baird, A.J., and Heathwaite, A.L., 2003. Anisotropy and depth-related
4 Brown, S.M., Petrone, R.M., Mendoza, C., and Devito, K.J., 2010. Surface vegetation controls on
7 Brown, S.M., Petrone, R.M., Chasmer, L., Mendoza, C., Lazerjan, M.S., Landhausser, S.M., Silins,
8 U., Leach, J., and Devito, K.J., 2014. Atmospheric and soil moisture controls on
9 evapotranspiration from above and within a Western Boreal Plain aspen forest.
11 Carrera-Hernandez, J.J., C.A., Mendoza, K.J., Devito, R.M., Petrone, and B.D., Smerdon, 2011.
14 10.1029/2010WR009684.
15 Carsel, R., and Parrish, R.S., 1988. Developing joint probability distributions of soil water
17 Debyle, N.V., and Winokur, R.P., 1985. Aspen: ecology and management in the western United
18 States. USDA Forest Service General Technical Report RM-119. Rocky Mountain Forest
20 Devito, K.J., Creed, I.F., and Fraser, C.J.D., 2005. Controls on runoff from a partially harvested
30
1 Devito, K, Mendoza, C, and Qualizza, C., 2012. Conceptualizing water movement in the Boreal
2 Plains. Implications for watershed reconstruction. Synthesis report prepared for the
3 Canadian Oil Sands Network for Research and Development, Environmental and
5 Ferone, J.M., and Devito, K.J., 2004. Shallow groundwater-surface water interactions in pond-
8 Gorham, E., 1991. Northern peatlands: role in the carbon cycle and probable responses to
10 Hamon, R.W., 1963. Computation of direct runoff amounts from storm rainfall, in Symposium
11 on Theoretical Studies and Practical Methods of Forecasting the Yield of Rivers for Both
12 Long and Short Terms (Except Floods), IAHS Publication 63, 25-62.
13 Hayashi, M., and van der Kamp, G., 2009. Groundwater-wetland ecosystem interaction in the
14 semiarid glaciated plains of North America. Hydrogeology Journal, 17: 203-214, doi:
15 10.1007/s10040-008-0367-1.
16 Jones, J.P., Sudicky, E.A., and McLaren, R.G., 2008. Application of a fully-integrated surface-
17 subsurface flow model at the watershed-scale: a case study. Water Resources Research,
19 Kelln, C., Barbour, S.L., and Qualizza, C., 2008. Controls on the spatial distribution of soil
31
1 Kettridge, N., and Waddington, J.M., 2014. Towards quantifying the negative feedback
3 doi: 10.1002/hyp.9898.
4 Koivusalo, H., and Kokkonen, T., 2002. Snow processes in a forest clearing and in a coniferous
6 Lemieux, J.-M., Sudicky, E.A., Peltier, W.R., and Tarasov, L., 2008. Dynamics of groundwater
7 recharge and seepage over the Canadian landscape during the Wisconsinian glaciation.
9 Lieffers, V.J., and Rothwell, R.L., 1987. Rooting of peatland black spruce and tamarack in
11 Lieffers, V.J., and MacDonald, 1990. Growth and foliar nutrient status of black spruce and
14 Macrae, M.L., Redding, T.E., Creed, I.F., Bell, W.R., and Devito, K.J., 2005. Soil, surface water and
17 Macrae, M.L., Devito, K.J., Creed, I.F., and Macdonald, S.E., 2006. Relation of soil-, surface-, and
21 Marshall, I.B., Schut, P., and Ballard, M. (compilers), 1999. Canadian ecodistrict climate normal
22 for Canada 1961-1990. A national ecological framework for Canada: Attribute data.
32
1 Environmental Quality Branch, Ecosystems Science Directorate, Environment Canada
3 McCarter, C.P.R., and Price, J.S., 2014. Ecohydrology of Sphagnum moss hummocks:
6 Mwale, D., Gan, T.Y., Devito, K., Mendoza, C., Silins, U., and Petrone, R., 2009. Precipitation
9 Nagare, R.M., Schincariol, R.A., Mohammed, A.A., Quinton, W.L., and Hayashi, M., 2013.
12 7.
13 Oil Sands Wetlands Working Group (OSWWG), 2000. Guideline for wetland establishment on
14 reclaimed oil sands leases, N. Chymko, ed., Rep. ESD/LM/00-1, Alberta Environment,
16 Pawlowicz, J.G., and Fenton, M.M., 2005. Drift thickness of Peerless Lake Area (NTS 84B), Map
18 Petrone, R.M., Silins, U., and Devito, K.J., 2007. Dynamics of evapotranspiration from a riparian
19 pond complex in the Western Boreal Forests, Alberta, Canada. Hydrological Processes,
33
1 Petrone, R.M., Devito, K.J., Silins, U., Mendoza, C., Brown, S.C., Kaufman, S.C., and Price, J.S.,
4 Price, J.S., McLaren, R.G., and Rudolph, D.L., 2010. Landscape restoration after oil sands mining:
7 10.1080/17480930902955724.
8 Quinton, W.L., Hayashi, M., and Carey, S.K., 2008. Peat hydraulic conductivity in cold regions
9 and its relation to pore size and geometry. Hydrological Processes, 22: 2829-2837, doi:
10 10.1002/hyp.7027.
11 Redding, T.E., and Devito, K.J., 2006. Particle densities of wetlands soils in northern Alberta,
13 Redding, T.E., and Devito, K.J., 2008. Lateral flow thresholds for aspen forested hillslopes on the
14 western boreal plain, Alberta, Canada. Hydrological Processes, 22: 4287-4300, doi:
15 10.1002/hyp.7038.
16 Redding, T., and Devito, K., 2010. Mechanisms and pathways of lateral flow on aspen-forested,
17 Luvisolic soils, Western Boreal Plains, Alberta, Canada. Hydrological Processes, 24:
19 Redding, T., and Devito, K., 2011. Aspect and soil textural controls on snowmelt runoff on
21 10.2166/nh.2011.162.
34
1 Silins, U., and Rothwell, R.L., 1998. Forest and peatland drainage and subsidence affect soil
2 water retention and transport properties in an Alberta peatland. Soil Science Society of
4 Smerdon, B.D., Devito, K.J., and Mendoza, C.A., 2005. Interaction of groundwater and shallow
7 Smerdon, B.D., Mendoza, C.A., and Devito, K.J., 2007. Simulations of fully coupled lake-
10 Smerdon, B.D., Mendoza, C.A., and Devito, K.J., 2008. Influence of subhumid climate and water
13 Smerdon, B.D., Redding, T.E., and Beckers, J., 2009. An overview of the effects of forest
15 10: 22-44.
16 Smerdon, B.D., and Mendoza, C.A., 2010. Hysteretic freezing characteristics of riparian
17 peatlands in the western boreal forest of Canada. Hydrological Processes, 24: 1027-
19 Smith, K., and Reid, F., 2013. Spreading our wings in the Boreal. Ducks Unlimited Canada
21 Smith, L., and Wheatcraft, S.W., 1993. Groundwater flow, in: Maidment, D.R., (Ed.), Handbook
35
1 Vogwill, R.I.J., 1978. Hydrogeology of the Lesser Slave Lake Area, Alberta. Alberta Research
3 Waddington, J.M., Morris, P.J., Kettridge, N., Granath, G., Thompson, D.K., and Moore, P.A.,
5 10.1002/eco.1493.
6 Winter, T.C., 2000. The vulnerability of wetlands to climate change: a hydrologic landscape
8 Winter, T.C., Rosenberry, D.O., Buso, D.C., and Merk, D.A., 2001. Water source to four U.S.
10 Woo, M., and Winter, T.C., 1993. The role of permafrost and seasonal frost in the hydrology of
12
36
1 Figure 1. Location of the Utikuma Region Study Area (URSA). Right: Location of the URSA within
2 the Canadian Boreal Plains relative to the discontinuous permafrost zone. Left: Enlarged map
3 showing the Pond 43 study area including instrumentation, vegetative cover, selected
4 monitoring well locations, surface water catchment boundaries, and the locations of the 2D
5 numerical models.
7 hydrogeologic units, and observation points. Cross section locations are indicated in Figure 1.
8 Note that the base of the models has been truncated at 645 masl for illustration purposes.
10 landscape units.
11 Figure 4. Observed water levels over the study period illustrating the similarity between pond
12 and peatland (228-WS) water levels, as well as the large lateral hydraulic gradients between the
13 peatland and hillslope (223-WD). Note that the hillslope well was dry prior to spring 2003, from
14 mid-July 2003 through to spring 2005, and throughout most of 2010. Well locations are shown
15 in Figures 1 and 2.
16 Figure 5. Variation in annual precipitation (P), pond evapotranspiration (ET), and cumulative
17 pond P - ET over the study period. Observed average annual pond and peatland water levels
18 were found to mimic trends in cumulative P - ET due to limited inputs from adjacent uplands.
19 Figure 6. Comparison of simulated and observed water levels within a) the pond, b) peatlands,
37
1 Figure 7. Comparison of simulated (solid lines) and observed (symbols) water levels for Pond 43
2 and selected monitoring wells. Dry wells are indicated by breaks in lines (simulated) and shaded
5 for simulations with a) the ET of the peat modified by 30% and b) the ET of the aspen modified
7 Figure 9. Predicted sensitivity to glacial till saturated hydraulic conductivity (Ks): comparison of
8 simulated water levels for simulations with the Ks of each glacial till unit increased and
9 decreased by one order of magnitude. Simulated dry wells are indicated by breaks in lines.
10 Figure 10. Predicted sensitivity to pond and peatland landscape units: comparison of simulated
11 water levels for simulations with a) the pond replaced with a continuous peatland and b) the
13
38
1 Table 1. Calibrated subsurface parameters within the numerical models
39
1 Table 2. Summary of annual boundary fluxes applied in the base case models
3 Notes:
a
4 Proportions of rain and snow vary by landscape unit due to different frozen season lengths. Total precipitation
5 (i.e., total P = rain + snow) was assumed to be uniform.
b
6 Maximum evapotranspiration (ET) applied within each landscape unit. The actual ET (AET) removed within the
7 simulations was limited by the available water.
c
8 ET includes rain interception and snow sublimation that was removed externally from the model.
d
9 Stream flow volumes normalized by pond area.
10
11
40
1 Table 3. Comparison of simulated aspen actual evapotranspiration (AET) to maximum
2 evapotranspiration (ET) and precipitation (P)
a,b a,b
P ET AET AET/ET P - AET
Year
(mm) (mm) (mm) (%) (mm)
2000 442 518 395 76 47
2001 257 508 293 58 -35
2002 290 507 284 56 6
2003 443 527 370 70 72
2004 378 512 323 63 55
2005 491 532 392 74 99
2006 432 557 366 66 66
2007 530 578 403 70 127
2008 504 574 397 69 107
2009 391 528 352 67 39
2010 282 539 303 56 -21
3
4 Notes:
a
5 Maximum ET applied within aspen areas. The AET removed within the simulations was limited by the available
6 water.
b
7 Includes rain interception and snow sublimation that was removed externally from the model.
41
1 Table 4. Summary of sensitivity simulation results
2
Influence on Landscape Unit Overall
Scenario
Pond Peat Aspen Sensitivity
Timing of spring
Modified timing and
water level peak
Pond duration of spring Little to no change Moderate to
shifted, minimal
Snowmelt water level peak in the from base case. Low
change throughout
vicinity of the pond.
rest of year.
Timing of spring water
Modified timing and level peak shifted, Modified duration
Peat
duration of spring minimal change of spring water Moderate
Snowmelt
water level peak. throughout rest of level peak.
year.
Modified timing and
Modified duration
Aspen Little to no change duration of spring
of spring water Low
Snowmelt from base case. water level peak at
level peak.
the peatland edge.
Large influence on Large influence on Moderate influence High to
Pond PET
water level. water levels. on water levels. Moderate
Large influence on Large influence on Moderate influence High to
Peat PET
water level. water levels. on water levels. Moderate
Small changes to
water levels,
Aspen Minor change to Moderate influence Low to
decreasing with
PET water level. on water levels. Moderate
distance from the
hillslope.
Low to moderate
Minor change to Minor change to
Peat Ks change to water Low
water level. water levels.
levels.
Glacial Large influence on Large influence on Large influence on
High
Till Ks water level. water levels. water levels.
Large increase in Large increase in
No Pond - High
water levels. water levels.
Small influence on
overall response.
Large decrease in High to
No Peat - No water level
water level. Moderate
recovery during the
winter.
3
42
Pond
43
Pond 48
Figure 1
228-WS
250-WD
223-WD
4-WS 40-WD
Figure 2
Forest: Calculated
160
Cumulative Rain Throughfall (mm)
Aspen: Measured
Spruce: Measured
120
80
40
0
May 15 Jun 15 Jul 15 Aug 15 Sep 15
Date
Figure 3
Pond
656
Peat
Hillslope
Hydraulic Head (masl)
655
654
653
652
Figure 4
600 655.0
P, ET, or Cumulative P-ET (mm)
400 654.6
-200
653.4
-400
2000 2002 2004 2006 2008 2010
Year
Figure 5
Observed Hydraulic Head (masl)
650 652 654 656 650 652 654 656
a) R 0.97 b) R 0.77
656 rm -0.01 m rm 0.002 656
RMS 0.06 m RMS 0.17 m
NRMS 5.93% NRMS 8.44%
654 654
Simulated Hydraulic Head (masl)
Figure 6
654.5 a) R
rm
0.77
0.04 m
RMS 0.16 m
654.0
653.5
4-WS 40-WD
654.5 b) R 0.65
rm 0.03 m
RMS 0.17 m
654.0
653.5
83-WS 250-WD
Hydraulic Head (masl)
654.5 c) R 0.97
rm -0.01 m
RMS 0.06 m
654.0
653.5
Pond 43
654.5 d) R
rm
0.85
0.02 m
RMS 0.10 m
654.0
653.5
228-WS
656.0
e) Dry
R
rm
0.45
0.33 m
655.0 Well RMS 0.90 m
654.0
653.0
652.0 223-WD
Figure 7
654.5
Base Case
High Peat ET
Low Peat ET
654.0
Hydraulic Head (masl)
653.5
653.0
a) 228-WS
654.0
Hydraulic Head (masl)
653.0
652.0
Base Case
High Aspen PET
b) 223-WD
Low Aspen PET
651.0
Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 Jan-10
Date
Figure 8
654.5
Base Case
High Ks
Low Ks
654.0
Hydraulic Head (masl)
653.5
653.0
Pond 43
654.0
Hydraulic Head (masl)
653.0
652.0
Base Case
651.0
High Ks
223-WD Low Ks
650.0
Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 Jan-10
Date
Figure 9
655.5
Base Case
No Pond
655.0
654.0
653.5
a) 228-WS
653.0
Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 Jan-10
Date
654.5
Base Case
No Peat
654.0
Hydraulic Head (masl)
653.5
653.0
b) Pond 43
Figure 10
1 HIGHLIGHTS:
2 • Field data were collected over 11 years from a catchment in Alberta’s Boreal Plains
4 • Results indicate dynamic interactions occur between the pond and peatlands
5 • Field and model results suggest little interaction between peatlands and uplands
43