- Email: [email protected]
Impact of tile drainage on water budget and spatial distribution of sediment generating areas in an agricultural watershed
Agricultural Water Management 184 (2017) 124–134
Contents lists available at ScienceDirect
Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat
Impact of tile drainage on water budget and spatial distribution of sediment generating areas in an agricultural watershed Golmar Golmohammadi a,∗ , Ramesh Rudra a , Shiv Prasher b , Ali Madani b , Mohamed Youssef c , Pradeep Goel d , Kourosh Mohammadi a a
School of Engineering, University of Guelph, Guelph, Ontario, N1G 2W1, Canada Department of Bioresource Engineering, McGill University, Macdonald Campus, Ste-Anne-de-Bellevue, Quebec, H9X 3V9, Canada c Department of Biological and Agricultural Engineering, North Carolina State University, Raleigh, USA d Ontario Ministry of the Environment and Climate Change, Etobicoke, Ontario, M9P 3V6, Canada b
a r t i c l e
i n f o
Article history: Received 1 April 2016 Received in revised form 8 December 2016 Accepted 1 February 2017 Keywords: Hydrological modeling SWATDRAIN Watershed scale Sediment load Streamflow
a b s t r a c t A recently developed model, SWATDRAIN, was used to assess the alterations in water balance components, discharge, and sediment loads due to tile drainage practices in a heavily tile drained watershed in Ontario, Canada. Furthermore, the model was implemented to determine the spatial variability of sediment loads which can be explained by a combination of spatially distributed variables within a watershed, including those controlling the hydrology, geology, soil and land use. Three scenarios were examined across the watershed, including conventional drainage (existing condition), controlled drainage, and no tile drainage. The model predicted that streamflow was not significantly impacted due to tile drainage, while the total runoff and sediment loads from the basin due to controlled drainage were increased by 27.1% and 22.2%, respectively, while removing tile drain infrastructures resulted in 37.1% increase in surface runoff and 55% increase in sediment load from the watershed The areas with high sediment load generation were identified by the model and the impact of tile drainage in producing sediment in those areas was assessed. The results showed that the sediment load generation rate in the areas with the highest load (class V) increased by 8% only due to controlled drainage, while in the second ranked sediment generating areas (class IV), the sediment load generation rate was increased by 32%. © 2017 Elsevier B.V. All rights reserved.
1. Introduction The watersheds of Southern Ontario which provide food, feed, and fiber for millions of people and livestock are facing increasing pressure as a result of water pollution from agricultural activities. The alteration of settlements and development along the Grand River basin in Ontario has led to deterioration in water quality (Boyd et al., 2009). The Canagagigue Creek watershed, located in the Grand River Basin of south western Ontario, is one of the fastest developing areas in Ontario; this region has some of the most intensive agricultural production in the watershed. In the Canagagigue watershed, the vast majority of land (80%) is used for agricultural purposes and a high percentage of the agricultural lands (27–49% in the upper or more northern region and 0.2–6% in lower region) is enhanced with subsurface drainage infrastructure (Cooke, 2006). Subsurface drainage has the potential to reduce pollutants associ-
∗ Corresponding author. E-mail address: [email protected] (G. Golmohammadi). http://dx.doi.org/10.1016/j.agwat.2017.02.001 0378-3774/© 2017 Elsevier B.V. All rights reserved.
ated with surface runoff and sediment (Bengston and Xu, 1995). On the other hand, tile drains are the main pathways to increase the export of contaminants from fields to surface water bodies; this can result in substantial agricultural contaminants by translocation of sediments, nutrients and pesticides from the field to streams and lakes in those landscapes with intensive tile drainage, especially during the non-growing season and after heavy summer rains (Tan et al., 1993; Tan et al., 2007; Drury et al., 2007). Subsurface drainage systems have been identified as a major source of nutrients and other pollutants such as (NO3 -N), herbicides, insecticides and fungicides exported to water bodies (Moriasi et al., 2012; Thomas et al., 1992; Zucker and Brown, 1998). Over the past few decades, controlled drainage practices have been applied at both the field and the watershed scale to conserve water, regulate tile drainage flow, improve water quality, and increase crop yield (Tan et al., 2007; Drury et al., 2014; Busscher et al., 1992; Evans et al., 1992; Evans et al., 1995; Lalonde et al., 1996; Breve et al., 1998; Fausey. 2004; Singh et al., 2007; Ale et al., 2009). In order to quantify the effect of subsurface drainage on flow, pollutant losses and sediment transport, observed data must be collected and ana-
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
lyzed. Historically, the monitoring data were collected manually and conclusions were mostly empirically based. This process was very intensive, both in terms of time and human resources. Over the past few decades, great strides have been made in technology and in modeling techniques that help users to make relatively accurate predictions of ungagged watersheds which previously would have been impractical (Frana, 2012). Nowadays, with proper calibration, the hydrologic models can be implemented to widely varying landscapes with very useful results (Frana, 2012). The main objective of this study is to assess the impacts of tile drainage on discharge, sediment load and water budget of the Canagagigue West watershed and also to identify the potential sediment source areas under tile drainage practices; therefore, the current research has the objectives of i) predicting and analyzing the water balance components of the watershed; ii) quantifying the total sediment yield of the watershed and determining the spatial variability of sediment loads across the watershed; iii) comparing the water budget and sediment yield and assessing the susceptibility of agricultural lands under different scenarios of tile drainage. The recently developed model, SWATDRAIN (Golmohammadi et al., 2016), which is capable of simulating different water management scenarios, such as controlled drainage was selected. The SWATDRAIN model was calibrated and validated for an agricultural watershed, called Green Belt in Ontario, Canada (Golmohammadi et al., 2016) and the results showed that by using the SWATDRAIN model, the flow and water table dynamics were significantly improved compared to SWAT model. This research makes an important contribution to watershed sedimentation control practices and to planning the Best Management practices (BMPs).
125
2. Materials and methods 2.1. Watershed description The Canagagigue Creek has a total drainage area of 143 km2 and is a tributary of the Grand River. It lies between latitudes 43◦ 36 N and 43◦ 42 N and longitudes 80◦ 33 W and 80◦ 38 W, and is about 25 kilometers northwest of the city of Guelph, Ontario. The climate of the area, according to the Koppen-Geiger climatic classification system, can be characterized as humid continental with warm summers and moderate winters. Based on the availability of observed flow and sediment data and also the presence of tile drainage in agricultural regions mostly located in the western portion of the Canagagigue Watershed, this study targeted the upstream portion of the Canagagigue Creek west. The major land use is agriculture including winter wheat, corn, soybean, and other row crops. Table 2 provides the landuse distribution on that watershed. The observed flow data were obtained from the available hydrometric station called “Canagagigue Creek near Floradale” by the Atmospheric Environment Services of Environment Canada. Sediment loads measurements at this station for period 1974–1984 are provided as archived data at the web site of Environment Canada (www.wsc.ec.gc.ca). It has available historical observation data for the daily flow rate and sediment load for the period 1975–1984. The topography of the watershed is flat to gently undulating with a slight slope towards the outlet in the south. The average elevation is 417 m. Fig. 1 shows the location of the Canagagigue Creek and the sub-watershed used in this study.
Fig. 1. Location of the study area, Canagagigue West watershed, in Grand River Basin.
126
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
2.2. General overview of SWATDRAIN model SWATDRAIN is a watershed model which was recently developed by incorporating DRAINMOD into SWAT (Golmohammadi et al., 2016). The hydrologic cycle that is simulated by SWATDRAIN computes the soil water balance, on a daily basis, for each hydrological response unit (HRU) in every subbasin of the watershed. The daily values of tile drainage are calculated using the DRAINMOD model (Skaggs, 1980), which computes the tile drainage flux based on the Kirkham and/or Hooghoudt’s equations and the hydraulic capacity as a function of the daily water table elevation midway between the drains. The water table depth is calculated based on drainage volume versus the water table depth relationship which is calculated from the soil water characteristics curve. SWATDRAIN uses either the SCS curve number procedure (SCS, 1972), or the Green and Ampt infiltration method (Green and Ampt, 1911), to estimate infiltration and surface runoff. Different methods can be used in SWATDRAIN for potential evapotranspiration (PET) estimation: the Penman-Monteith method (Monteith, 1965), the Priestly-Taylor method (Priestley and Taylor, 1972), the Hargreaves method (Hargreaves and Samani, 1985), the Thornthwaite method (Thornthwaite, 1984) or PET can also be entered as daily or monthly PET input by the user. Detailed information on the SWATDRAIN
modeling approach can be found in Golmohammadi et al. (2014, 2016). 2.3. Input data and model setup Daily climate data, including maximum and minimum temperatures, precipitation, wind speed and humidity for the Fergus station were obtained from Environment Canada. In addition to the weather data inputs, the model requires Digital Elevation Model (DEM), soil and land use and agricultural management data, if applicable. A DEM with a 10 m × 10 m spatial resolution was obtained from the Grand River Conservation Authority (Fig. 2). The digital elevation model of the study area was processed to define the subbasin parameters, such as slope and slope length, and to define the stream network characteristics, such as channel slope, length and width. The resulting stream network was implemented to define a layout of 13 subbasins. The soil and land use classification across the watershed was defined by polygon shape files, provided by the Ontario Ministry of Agriculture and Agri-Food (Fig. 2). The combination of land use and soil type resulted in 67 HRUs. The soil surveys of Waterloo County presented by Presant and Wicklund (1971) and Wellington County presented by Hoffman et al. (1963) indicated that the major portion of the watershed has 20–60 centimeters of
Fig. 2. Digital Elevation model (DEM), soil, landuse and tile drainage maps.
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
127
Table 1 Soil distribution in Canagagigue Watershed. Soil type
Soil texture
Area (%)
BUF: Burford BAY: Brady BKN: Brookstone CTG: Conestogo BRR: Berrin BSB: Birsbane CAD: Caledone Total
loam/loam gravel/sand clay gravel/sand gravel sandy loam/sandy loam/sand loam/clay loam/clay loam silt loam/loam/silt loam/loam sandy loam/loamy sand/sandy loam/clay/clay loam/loam/loamy gravel/gravel sandy loam/sandy loam/sandy loam/sandy loam/sand/loamy sand
37.2 16.2 23.2 6.5 10.7 0.5 5.7 100
Table 2 Land use distribution on Canagagigue Watershed. Landuse
Area (%)
Winter wheat Corn Soybean Agricultural land-row crops Forest Hay Residential Medium density Total
30.56 26.92 22.87 7.53 6.89 5.14 0.09 100.00
loam or silty loam of Huron and Harriston series overlying a loam till. In the northern part of the watershed, clay loam is predominant. Loam is the main soil type in the southern portion of the watershed. Different soil type’s distribution across the watershed is presented in Table 1. Approximately 93.02% of the total area was under agricultural production, with the majority of land use under winter wheat, followed by corn, soybean, agricultural land row crops, and hay. The remaining 6.98% of the area was occupied by forest and residential lands. Land use distribution across the watershed is presented in Table 2. The tile drained area across the watershed was determined based on the tile drainage map obtained from the Land Information Ontario (Fig. 2). 2.4. Model evaluation The SWATDRAIN model was manually calibrated (1975–1979) and validated (1980–1984) for streamflow discharged from the Canagagigue Creek Watershed. These periods were selected in order to include dry, average, and wet years in each calibration and validation phase. The calibration and validation were performed based on the results of sensitivity analysis and on findings of previous studies (Rong, 2009; Oogathoo, 2006; Golmohammadi et al., 2014). The watershed has both systematic and non-uniform tile drainage system. The SWATDRAIN model is capable of evaluating subsurface drainage system whether they are uniform or non-uniform. Therefore, the model allows for various drainage layouts. However, in this study, due to lack of information on drainage layouts of different fields across the watershed, tile drainage systems was assumed uniform for entire watershed based on the literature and previous studies in Ontario condition A subsurface drainage system with drain spacing of 17 m at a depth of 1.00 m, operated as an approach of free drainage at the drain outlet, were recommended (GRCA, 2008; OMAFRA, 2009; Oogathoo, 2006; Golmohammadi et al., 2016). The initial Soil Conservation Service runoff curve number to moisture condition II (CN2) values calibrated using the SWAT model ranged from 66 to 78 (Rong, 2009). Some other calibration parameters adjusted in this study included the curve number coefficient (CNCOEF), surface runoff lag coefficient (SURLAG), and initial soil water storage expressed as a fraction of field capacity water content (FFCB).
Based on the literature, some parameters important to the snow-water mass balance were also investigated for their sensitivity to surface runoff, baseflow, evapotranspiration and streamflow within the watershed under study. These parameters were SMTMP (snow melt base temperature), SFTMP (snowfall temperature), TIMP (snow pack temperature lag factor), SMFMN (melt factor for snow on December 21) and SMFMX (melt factor for snow on June 21). Based on several tile drainage studies in southern Ontario, drainage coefficient (DC), the lateral hydraulic conductivity multiplication factor (LATKSATF) and effective radius of drains (RE) were set at 20.0 mm d−1 , 1.1 mm h−1 , and 15 mm, respectively (Thooko et al., 1990). The values of other streamflow calibration parameters were obtained from previous Canagagigue Creek Watershed studies (Golmohammadi et al., 2014; Rong, 2009). Additional calibration parameters were required for the SWATDRAIN model, given the new input parameters in the model, including soil retention curve parameters, saturated hydraulic conductivities of different soil layers, ET monthly factors, rooting depths and all parameters regarding freeze/thaw conditions. The calibrated parameters for the SWATDRAIN model used in this study are presented in Table 3. The model also requires inputs for the initial soil temperature profile, upper boundary condition and the lower boundary condition. Since a long record of measured soil surface temperatures is not usually available for most applications, air temperature was used (Luo et al., 2000, 2001). The lower boundary was assumed to be a constant soil temperature, which can be approximated from the long-term average air temperature (Penrod et al., 1958). In SWATDRAIN, lateral saturated hydraulic conductivity is soil-specific and it is read into the model by the soil file used for each HRU. The effective rooting depth is a difficult parameter to determine (Skaggs, 1980); and range of 45–55 cm was suggested by Rudra et al., 1985; Golmohammadi et al., 2016. No field measurements were made in this research; an initial rooting depths were obtained from the previous studies in Ontario (Golmohammadi et al., 2016; Wang et al., 2006; Singh et al., 2007; Dayyani et al., 2009; Thooko et al., 1990). It was assumed that, a restrictive layer set at the bottom of the soil profile. An initial estimated depth of 200 cm from the surface to the restrictive layer was used for the soil in this study. Although parameter input values can be determined by model calibration within the recommended range of values, known parameters values from previous studies can be used as well (Golmohammadi et al., 2016, 2014). Several trial and error runs were performed by varying these input parameters. The calibrated values for the sediment load are also listed in Table 3. In order to enter the land use and cover factor (USLE C), and soil erodibility factor (USLE K), properly reflecting the local characteristics, a soil erosion screening tool, Guelph Model for Evaluating the Effects of Agricultural Management Systems (GEAMS) on Erosion and Sedimentation (University of Guelph, Ontario, Canada, (1985) was used. Furthermore in order to reflect the real values of the soil properties of the studied watershed, two soil survey reports in Ontario by Hoffman et al., 1963 and Presant and Wicklund (1971) were used. In addition to graphical methods such as hydrographs, three performance criteria were used to evaluate the performance of the
128
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
Table 3 Adjusted values of parameters of the SWATDRAIN model for Canagagigue Creek Watershed. Parameter
Description
Value
ESCO EPCO FFCB
Soil evaporation compensation factor Plant uptake compensation factor Initial soil water storage expressed as a fraction of field capacity water content Surface runoff lag coefficient (day) Curve number method flag Curve number coefficient Depth to impervious layer (mm) Multiplication factor to determine lateral hydraulic conductivity (mmh−1 ) Lateral hydraulic conductivity (mmh−1 ) Effective radius of drains (mm) Drainage coefficient (mmday−1 ) Distance between two drain or tile tube (mm) Depth from soil surface to drain (mm)
0.80 0.65 0.95
SURLAG ICN CNCOEFF DEP IMP LATKSATF Ks Re Dc Drain spacing (L) Drain Depth ET monthly factors Rooting depth TKA, TKB TLAG TB Tsnow Tmelt CDEG CICE USLE K USLE C SPCON ADJ PKR PRF CH N2 CH K2 CH EROD CH COV
— Thermal conductivity functions W m−1 ◦ C Diurnal phase lag of air temperature (h) Soil temperature at the bottom of the profile (◦ C) Rain/snow-dividing temperature (◦ C) Snowmelt base temperature (◦ C) Snow melt coefficient (mm day−1 ◦ C−1 ) Critical ice content (cm3 cm−3 ) USLE soil erodibility factor USLE minimum crop factor Linear factor for channel sediment routing Subbasin peak sediment adjustment factor Main channel peak sediment adjustment factor Main channel manning’s roughness coefficient Main channel hydraulic conductivity(mmh−1 ) Channel erodibility factor Channel cover factor
model: Nash-Sutcliffe efficiency (NSE); (Nash and Sutcliffe, 1970), the RMSE-observations standard deviation ratio (RSR), and percent bias (PBIAS) (Gupta et al., 1999).
2.0 1 1.40 2200 1.1 Varies 15 20 15000 1000 Jan = 1, Feb = 1, March = 1, April = 0.8, May = 0.6, June = 0.8, July = 0.8, Aug = 0.8, Sept = 1, Oct = 1, Nov = 1, Dec = 0.6 Varies a = 0.553,b = 1.963 9 7 0 1 5 0.3 Varies Varies 0.01 1.0 2.0 0.04 10 0.23 0.85
Table 4 Statistical comparisons of observed and simulated streamflow at Canagagigue West watershed. Monthly
NSE = 1 − [
ni=1 (Yi obs − Yi sim )
ni=1 (Yi obs − Yobs mean )
RSR =
[
2
(1)
]
Index R2 NSE RSR PBIAS
Calibration 0.92 88.93 0.27 −6.07
Daily Validation 0.82 74.0 0.37 3.67
Calibration 0.85 78.70 0.36 −6.07
Validation 0.75 60.01 0.47 3.67
2
ni=1 (Yobs − Ysim ) ] i i
[
2
(2)
2
ni=1 (Yobs − Ymean ) ] i
3. Results and discussion 3.1. Streamflow and sediment loads
PBIAS = (
ni=1 (Yi obs − Yi sim )*(100) ni=1 (Yobs i )
)
(3)
where Yi obs is the observed value, Yobs Mean is the mean observed value, is the mean simulated value, and Yi sim is the predicted value,Ysim mean n is the total number of observations. A Multiple regression method was also used to find a relationship between the two main parameters impacting the sediment yield (Restrepo et al., 2006). The purpose of multiple regression in this research was to learn about the relationship between independent or predictor variables in the watershed and a dependent or criterion variable such as sediment yield. For this method, six different parameters including runoff, peak flow, mean water discharge rate at the outlet, catchment length, river length and mean slope were considered. Using step-wise regression, two control variables were selected which were mean annual runoff and peak flow. These two parameters were used to predict sediment yield for whole watershed and evaluate the impact factor of each parameter on sediment yield.
Time series plots of monthly streamflow during calibration and validation period is illustrated in Fig. 3. The results showed that the observed and simulated streamflows were in a good agreement. In general, during the calibration period, the model closely simulated streamflow values, except for two months of March 1976 and April 1979, which the streamflow was underestimated by the model (Fig. 4). During the validation period, observed and simulated sediment loads matched very well, except for the month of March 1984. The calibration and validation model performance results for the daily and monthly streamflow are presented in Table 4. During the calibration period, the monthly NSE, RSR, and PBIAS values were 0.89, 0.27 and −6.07, respectively (Table 4). According to Moriasi et al. (2007) a model is considered well calibrated for flow if monthly NSE ≥ 0.65, RSR ≤ 0.60 and PBIAS ≤ ± 10%. Therefore, the SWATDRAIN model was well calibrated as presented by the statistics in Table 4 and also supported by the monthly hydrographs (Fig. 3). The monthly NSE, RSR, and PBIAS values for the validation phase were calculated to be 0.74, 0.37 and 3.67 (Table 4). Watershed
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
129
Fig. 3. Observed and simulated monthly streamflows during calibration and validation periods.
Fig. 4. Observed and simulated monthly sediment loads during calibration and validation period.
model evaluation guidelines, developed by Moriasi et al. (2007), state that a model simulation performance for flow is considered good if NSE ≥ 0.65, RSR ≤ 0.60 and PBIAS ≤ ± 15. Therefore, in this study, the streamflow trends simulation performance rating was acceptable according to Moriasi et al., 2007. Observed and simulated monthly sediment loads during calibration and validation periods are presented in Fig. 4. During the calibration phase, the observed and simulated sediment load matched well, although in the month of March 1977 sediment is overestimated and in the months of April 1978 and April 1979, the sediment is underestimated by the model (Fig. 4).
During the validation period, measured and simulated sediment loads matched very well, except for the snowmelt periods of the years 1980 and 1984. The values of NSE, RSR and PBIAS for sediment load are presented in Table 5. Watershed model evaluation guidelines, developed by Moriasi et al. (2007), state that a model simulation performance for sediment is considered good if monthly values of NSE ≥ 0.65, RSR ≤ 0.60 and PBIAS ≤ ± 15.Therefore, the simulation results are very good due to the high statistical results during the validation period (NSE = 0.80; RSR = 0.44; PBIAS = 25.88).
130
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
Table 5 Statistical comparisons of observed and simulated streamflow at Canagagigue West watershed. Monthly Index R2 NSE RSR PBIAS
Calibration 0.88 0.83 0.41 −1.71
Daily Validation 0.89 0.80 0.44 25.88
Calibration 0.82 0.74 0.48 −1.71
Validation 0.73 0.71 0.49 25.88
Table 6 Average annual water balance over the simulation period for the Canagagigue West watershed. Components
Precipitation; PCP Water yield; WYLD Tile flow; TILEQ Surface runoff; SURQ Lateral flow; LATQ Groundwater contribution to streamflow; GQ Actual Evapotranspiration; ET Percolation; PERC
Average Annual [1975–1984] (mm)
(%)
950.3 444.4 113.1 243.9 50.1 37.3 442.5 41.3
100 46.8 11.9 25.6 5.3 3.9 46.6 4.3
4. Water balance analysis After the model evaluation, the water budget of Canagagigue West Watershed was analyzed based on the amount of precipitation received, and simulated discharges and evapotranspiration. The average annual values for water balance components are presented in Table 6. The relative percentages of the watershed balance components were calculated by dividing each component into the total precipitation received. The results of the simulated water budget show that the model simulated water processes much like those that would occur in nature. The simulated water balance results under existing condition (65% of the watershed tile drained) showed that the major portion of the precipitation (46.6%) received by the watershed contributed to streamflow at the outlet of the watershed during the 10 years of the simulation. This corresponds to the observed average annual streamflow of 439 mm (45%) of the precipitation. The simulated evapotranspiration accounted for 46.6% of the total precipitation which was similar to the historical average ET of –444 mm obtained by the various empirical methods that used canopy cover, latitude, elevation and precipitation, and PET for this site (Rong, 2009; Oogathoo, 2006). While the simulated average values of surface runoff and tile drainage were 25.6% and 11.9% of precipitation, lateral flow, calculated as a percentage of average annual precipitation, was 5.3%. During the simulation period, 46.6% of annual precipitation was simulated as being lost in evapotranspiration across the watershed. The high evapotranspiration rate could be attributed to the land cover of the study area. The average annual contribution of groundwater as a relative percentage to precipitation is only 3.9% for the study period. Deep aquifer recharge is very low with an average of 0.1% which is negligible. However, the difference between the amount of precipitation and the sum of the discharges and the evapotranspiration was obvious. The water balance error was obtained by balancing the major hydrologic components simulated in the model. The model performed well with a relative error of 2.3% for the simulation period. 5. Impact of tile drainge The impact of tile drainage on water balance of the Canagagigue watershed was assessed by implementing different tile drainage
practices across the watershed. Three scenarios of i) existing condition (65% of the watershed under tile drainage), ii) controlled drainage (65% of the watershed under tile drainage), and iii) no tile drain infrastructure installed in the watershed were implemented to illuminate the effects of drainage. Even though a shift to non-drained agriculture in Ontario is not expected, it is a useful exercise to provide evidence for the benefits of tile drainage systems to ensure ease of planting, growth, and harvesting. Based on the simulated average annual water balance (Fig. 5), under existing condition, 46.7% of precipitation (442.5 mm) was simulated as evapotranspiration; this was increased to 49.4% (469.5 mm) for the situation in which controlled drainage was applied instead of conventional drainage. While, for the watershed with no tile drainage, the ET value was increased to 50.8% of the annual precipitation (482.5 mm). Therefore, based on the simulated average annual water balance [1974–1984], ET was significantly decreased by implementing tile drainage. The model simulated that 46.8% of precipitation (444.4 mm) contributed to streamflow at the outlet of the watershed during the 10 years of simulation (Table 6). This value was increased to 48.0% (455.9 mm) under the scenario of controlled drainage. While, if there were no drainage infrastructure installed across the watershed, the streamflow decreased to 48.5% (461.0 mm). These simulated results of average annual water budget provide evidence for the hydrological impact on the watershed by the scenarios of controlled drainage and the absence of tile drainage. Evapotranspiration increased as there was more water available on the soil surface for evapotranspiration. Although the streamflow was not decreased significantly by controlled tile drainage, but surface runoff increased significantly. In addition, without subsurface drainage, water is allowed to saturate the upper profile of the unsaturated zone, which in turn may increase the deep recharge in the groundwater aquifer (Fig. 5).
6. Streamflow impact analysis In order to assess the significant changes in streamflow due to potential changes in subsurface drainage in the Canagagigue watershed, certain hydrologic analyses, including flood frequency analysis and normalized hydrography analysis, were conducted (Rahman et al., 2014). Flood frequency analysis was conducted to identify the changes in the magnitude and frequency of annual peak flows in the watershed for the controlled drainage and no tile drainage scenarios. Fig. 6 compares the flood frequency curves under the existing, controlled and no tile drainage scenarios. The flood frequency results showed that the subsurface drainage under the controlled drainage scenario will increase the frequency of peak flows while decreasing the frequency of minor flows. The results also showed that the no tile drainage scenario will increase the frequency of peak flows and decrease the frequency of minor flows more than the scenario of controlled drainage. Nevertheless, Fig. 6 shows that the changes in magnitude and frequency of peak flows are not statistically significant, since all three peak flow frequency curves lie completely within the 95% confidence interval curves. In order to evaluate the possible changes in the shape of hydrographs, particularly during the snowmelt period, a normalized hydrograph analysis was conducted using the criteria adopted from Rahman et al. (2014). The hydrographs were selected from the snowmelt events, which included only one main peak with complete daily records of 41 days (15 days before and 25 days after the peak), and no other complications in the shape of the hydrograph such as minor flows (Miller and Frink, 1984; Rahman et al., 2014). Therefore, those hydrographs which did not provide a useful characterization of a simple runoff hydrograph shape were removed from the analysis.
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
131
Fig. 5. Average annual water balance for Canagagigue Watershed under i) existing condition, ii) Entire Watershed tile drained, and iii) no tile draianage.
The selected hydrographs were normalized by dividing each discharge value by the peak discharge value, which resulted in normalized hydrographs varying between 0 and 1. Fig. 7 compares the averaged normalized hydrographs under the three tile drainage scenarios. It shows that the normalized hydrographs of the existing condition has a steeper rising limb which translates into the shorter time to peak of the hydrograph. Fig. 7 also shows that the alteration in shape of the hydrograph is not statistically significant, since all three hydrographs lie completely within the 95% confidence intervals. Therefore, it can be concluded that the increase of peak flow rates at smaller recurrence probabilities under the controlled
drainage is mainly due to the flow volumes rather than increasing the time to peak of the hydrograph. 7. Impact of tile drainage on sediment yield Watershed runoff and sediment load simulation results are presented in Table 7. The sum of the surface runoff simulated for the 1975–1984 years was 2539.2 mm under the existing condition compared with 3227.1 mm and 3481.1 mm under controlled drainage and no tile drainage scenarios. Controlled drainage lead to an increase of 27.1% in total annual surface runoff. The scenario of no tile drainage across the watershed resulted in an increase of 37.1% in total annual surface runoff. Total sediment loads estimated
132
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
Table 7 Simulated surface runoff and sediment load results for conventional, controlled and no tile scenarios. Year
1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 Total
Runoff
Sediment loads
Existing Condition
Controlled Drainage
No Drainage
Existing Condition
Controlled Drainage
No Drainage
317.7 301.9 258.3 192.8 267.9 227.3 198.9 294.7 173.4 306.3 2539.2
397.97 387.89 300.12 243.96 365.04 275.2 247.21 390.82 230.1 388.76 3227.1
412.9 409.2 319.9 266.7 382.1 284.1 286.7 404.1 287.6 428.4 3481.1
1.0 0.7 0.8 0.5 0.6 0.6 0.4 0.6 0.2 0.9 6.3
1.3 0.9 1.0 0.6 0.7 0.6 0.5 0.8 0.3 1.0 7.7
1.2 1.1 1.0 1.0 1.0 0.8 0.6 1.1 0.7 1.4 9.8
Fig. 6. Peak flow frequency analysis for the Canagagigue Watershed.
Fig. 7. Average Normalized-Hydrograph (NH) based on the simulated streamflow in the Canagagigue Watershed.
for the study period were 6.3 tons/ha for the exiting condition while it was predicted 7.7 tons/ha for the scenario of controlled drainage and 9.8 tons/ha for the condition in which the entire tile drains were
removed. These values translate to 22.2% and 55.0% increases associated with the controlled drainage and no tile drainage scenarios. The simulated average seasonal and annual sediment loads are presented in Fig. 8. The figure shows that most of sediment load increase happened during the spring season. Comparing water balance simulation results, under three different scenarios which was presented in Fig. 5, indicates that reduction in subsurface drainage due to controlled drainage was reflected in an increase in surface runoff at the watershed scale. The increase of surface runoff may lead to increase soil erosion and sedimentation in the watershed. Analysis of sediment loads and different hydrologic variables in the Canagagigue Watershed indicated that the surface runoff is the main factor explaining the variation in observed sediment loads. The sediment load was estimated using a multiple regression model that includes runoff and peak flow. The results showed that these estimators explain 61% of variance in sediment yield. It can be concluded that the spatially distributed variables, such as topography, soil and land use are also important factors in explaining the sediment load spatial variability. Fig. 9 shows the spatial distribution of sediment yield for the existing condition, controlled drainage and no tile drainage scenarios. The sediment yield varied in the range of 0.0 t ha −1 to 4.0 t ha −1 across the watershed which was divided into 5 different classes (Fig. 9) according to a previous study in Ontario by Dickinson and Rudra (1986). It was noted that the great majority of sediment loads were generated from a small percentage of the watershed. By combining the sediment spatial distribution and subbasin maps, it was determined that under the existing condition, the highest sediment load generating areas, Class V, are mainly located in subbasins number 5 and 6 and 7, followed by the areas with the load class IV which are mostly located in subbasins 8, 9, 12, 13, and 3. The results of this study showed that sediment yield rates and its distribution vary with land cover and soil. Analyzing the spatial distribution of the sediment loads and the soils and land use across the watershed indicate that the areas with sediment loads of type V are covered by agricultural row crops and winter wheat. The soil types in this area are mostly Conestoga (silt loam), Brookston (clay loam), and Burford (loam) which have a high erodibility factor (30 and more) compared to the other soils (ranges from 10 to 14) in the study area. Therefore, it can be concluded that the interaction of highly erodible soils and agricultural land use, such as agricultural row crops and winter wheat, resulted in high sediment loads (Class V). The results also indicate that the areas with low amounts of sediment are mostly under hay and forest while the soil types of Bradly (sandy loam) and Caledone (sandy loam) with low erodibility factor (10–12) are dominant in these areas. The impact of tile drainage on spatial distribution of sediment loads was assessed by comparing the sediment load maps of the three scenarios presented in Fig. 9. By comparing the results of sediment load distribution under
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
133
Fig. 8. Simulated average seasonal and annual sediment loads.
Fig. 9. Spatial distribution of sediment loads categories across the watershed.
three different scenarios, it was noted that the implementation of controlled drainage will not affect the sediment load produced in critical sediment load areas of class V (> 2 t ha−1 ) under existing condition (Subbasins 5, 6 and 7). While in the areas with sediment load of Class IV (1.0–2 t ha−1 ) under existing condition, (Subbasins 8, 9, 12, 13, and 3), implementing controlled drainage increased the sediment loads significantly. This can be attributed to different soil permeability in these areas. In the areas with high sediment production, the Conestoga and Brookston soils are imperfectly drained and poorly-drained soils, respectively, while the Burford soil is a well-drained soil. The results of this study indicate that the areas with the soil type, Burford, which is a well-drained soil, were most impacted by tile drainage. The impact of tile drainage in the areas with Brookston soil was not significant. The results showed that removing the controlled drainage, in general increased the sediment loads across the watershed. This may be due to an increase
in surface runoff which may lead to increase soil erosion and sedimentation in the watershed. Spatial distribution of the sediment loads under no drainage scenario. 8. Conclusions A recently developed SWATDRAIN model was used to assess the impact of tile drainage on water budget and sediment loads from the Canagagigue watershed in southern Ontario. The results showed that the controlled drainage settings lead to surface runoff and a sediment load increase of 27.1% and 22.2% respectively. While removing tile drain infrastructures resulted in 37.1% increase in surface runoff and 55% increase in sediment load in the watershed. The results demonstrated that tile drainage has a significant impact on watershed hydrology and sediment yields and SWATDRAIN is an appropriate model to simulate these impacts.
134
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
The areas contributing to sediment loads and the extent to which the sediment loads increased in those areas due to tile drainage was also estimated by the model. The spatial variability of sediment loads can be explained by a combination of spatially distributed variables within the watershed, including those controlling hydrology, geology, soil and land use. The SWATDRAIN model predicted that in the Canagagigue watershed, the areas with the highest sediment load values (> 2 tones/ha) were not affected by controlled drainage significantly while the areas with high sediment load values (1.5–2 tones/ha) were affected significantly. The investigation of sediment load results can help in identifying the critical areas of sediment generation in order to design appropriate management practices for sedimentation control. Acknowledgement The funding received for this project from the Natural Sciences and Engineering Research Council of Canada is gratefully acknowledged. References Ale, S., Bowling, L.C., Brouder, S.M., Frankenberger, J.R., Youssef, M.A., 2009. Simulated effect of drainage water management operational strategy on hydrology and crop yield for Drummer soil in the Midwestern United States. Agric. Water Manage. 96, 653–665. Bengston David N., Xu Z., 1995. Changing National Forest Values: a content analysis. Research Paper NC-323. St. Paul, MN: U.S. Dept. of Agriculture, Forest Service, North Central Forest Experiment Station. Boyd, D., Cooke, S., Yerex, W., 2009. Exploring Grand-Erie connections: flow, quality and ecology. 9th Annual Grand River Watershed Water Forum. 18 September 2009. Breve, M.A., Skaggs, R.W., Parsons, J.E., Gilliam, J.W., 1998. Using the DRAINMOD-N model to study the effects of drainage system design and management on crop productivity, profitability and NO3-N losses in drainage water. Agric. Water Manage. 35 (3), 227–243. Busscher, W.J., Sadler, E.J., Wright, F.S., 1992. Soil and crop managementaspects of water table control practices. J. Soil Water Conserv. 47, 71–74. Cooke, S., 2006. Water quality in the grand river: A summary of current conditions (2000–2004) long term trends. Grand River Conservation Authority. Dayyani, S., Madramootoo, C.A., Enright, P., Simard, G., Gullamudi, A., Prasher, S.O., Madani, A., 2009. Field evaluation of DRAINMOD 5. 1 under a cold climate: simulation of daily midspan water table depths and drain outflows. J. Am. Water Resour. Assoc. 45 (3), 779–792. Dickinson, W.T., Rudra, R.P., 1986. Identification of soil erosion and fluvial sediment problems. J. Hydrol. Proc. 1, 111–124. Drury, C.F., Yang, J.Y., De Jong, R., Yang, X.M., Huffman, E.C., Kirkwood, V., Reid, K., 2007. Residual soil nitrogen indicator for agricultural land in Canada. Can. J. Soil Sci. 87, 167–177. Drury, C.F., Tan, C.S., Welacky, T.W., Reynolds, W.D., Zhang, T.Q., Oloya, T.O., McLaughlin, N.B., 2014. Reducing nitrogen loss in tile drainage water with cover crops and water-table management systems. J. Environ. Qual. 43, 578–598. Evans, R.O., Parsons, J.E., Stone, K., Wclls, W.B., 1992. Water tahle management on a watershed scale. J. Soil Water Conserv. 47 (1), 58–64. Evans, R.O., Skaggs, R.W., Gilliam, J.W., 1995. Controlled versus conventional drainage effects on water quality. J. Irrig. Drain. Eng. ASCE 121 (4), 271–276. Fausey, N., 2004. Comparison of free drainage, controlled drainage, and subirrigation water management practices in an Ohio Lakebed soil. In: ASAE Meeting Paper No. 042-237, ASAE, St. Joseph, MI. Frana, A.S., 2012. Applicability of MIKE SHE to Simulate Hydrology in Heavily Tile Drained Agricultural Land and Effects of Drainage Characteristics on Hydrology M.Sc. Thesis. Iowa State University. GRCA, 2008. Grand River Watershed Characterization Report. Grand River Conservation Authority. Golmohammadi, G., Prasher, S., Madani, A., Rudra, R., 2014. Evaluating three hydrological distributed watershed models: MIKE-SHE, APEX, SWAT. Hydrology 1 (1), 20–39. Golmohammadi, G., Prasher, S., Madani, A., Rudra, R., Youssef, M.A., 2016. SWATDRAIN, a new model to simulate the hydrology of agricultural lands, model development and evaluation. J. Biosyst. Eng. 141, 31–47.
Green, W.H., Ampt, G.A., 1911. Studies on soil physics: part I. The flow of air and water through soils. J. Agric. Sci. 4, 1–24. Gupta, H.V., Sorooshian, S., Yapo, P.O., 1999. Status of automatic calibration for hydrologic models: comparison with multilevel expert calibration. J. Hydrol. Eng. 4 (2), 135–143. Hargreaves, G.H., Samani, Z.A., 1985. Reference crop evapotranspiration from temperature. Appl. Eng. Agric. 1, 96–99. Hoffman, D.W., Matthews, B.C., Wicklund, R.E., 1963. Soil survey of wellington county Ontario, report No.35, Ontario soil survey Canada. Dep. Agric. Ontario Dep. Agric. Lalonde, V., Madramootoo, C.A., Trenholm, L., Broughton, R.S., 1996. Effects of controlled drainage on nitrate concentrations in subsurface drain discharge. Agric. Water Manage. 29, 187–199. Luo, W., Skaggs, R.W., Chescheir, G.M., 2000. DRAINMOD modifications for cold conditions. Trans. ASAE 43 (6), 1569–1582. Luo, W., Skaggs, R.W., Madani, A., Cizikci, S., Mavi, A., 2001. Predicting field hydrology in cold conditions with DRAINMOD. Trans. ASAE 44 (4), 825–834. Miller, J.E., Frink, D.L., 1984. Changes in Flood Response of the Red River of the North Basin, North Dakota-Minnesota. Water-Supply Paper 2243. U.S. Geological Survey, Alexandria, Virginia. Monteith, J.L., 1965. Evaporation and environment. Symp. Soc. Exp. Biol. 19, 205–234. Moriasi, D.N., Rossi, C.G., Arnold, J.G., Tommer, M.D., 2012. Evaluating hydrology of the Soil and Water Assessment Tool (SWAT) with new tile drain equations. J. Soil Water Conserv. 67 (6), 513–524. Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel, R.D., Veith, T.L., 2007. Model evaluation guidelines for systematic quantification of accuaracy in watershed simulations. Trans. ASABE 50 (3), 885–900. Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models: part 1. A discussion of principles. J. Hydrol. 10 (3), 282–290. OMAFRA, 2009. Agronomy Guide for Field Crops. Ministry of Agriculture, Food and Rural Affairs (Pub. 811, PP306). Oogathoo, Sh., 2006. Runoff Simulation in the Canagagigue Creek Watershed Using the MIKE SHE Model. Master’s Thesis. Sainte Anne-de-Bellevue, QC; McGill University, Department of Bioresource Engineering. Penrod, E.B., Walton, W.W., Terrel, D.V., 1958. A method to describe soil temperature variation. J. Soil Mech. Found ASCE 1537, 1–21. Presant, E.W., Wicklund, R.E., 1971. Soil survey of Waterloo county, report No. 44 of the Ontario soil survey Canada. Dep. Agric. Ontario Dep. Agric. Priestley, C.H.B., Taylor, R.J., 1972. On the assessment of surface heat flux and evaporation using large-scale parameters. Mon. Weather Rev. 100, 81–92. Rahman, M.M., Lin, Z.H., Jia, X., Steele, D.D., DeSutter, T.M., 2014. Impact of subsurface drainage on streamflows in the Red River of the North basin. J. Hydrol. 511, 474–483. Restrepo, J.D., Kjerfve, B., Hermelin, M., Restrepo, J.C., 2006. Factors controlling sediment yield in a major South American drainage basin: the Magdalena River, Colombia. J. Hydrol. 316, 213–232. Rong, H., 2009. Water Quality Evaluation Under Climate Change Impacts for Canagagigue Creek Watershed in Southern Ontario Master’s Thesis. Guelph University, Department of Bioresource Engineering. Rudra, R.P., Dickinson, W.T., Wall, G.J., 1985. Application of the CREAMS model in southern Ontario conditions. Trans. ASAE. 28 (4), 1233–1240. Soil Conservation Service (SCS), 1972. Section 4: Hydrology. In: National Engineering Handbook. USDA Soil Conservation Service, Washington, D.C. Singh, R., Helmers, M.J., Crumpton, W.G., Lemke, D.W., 2007. Predicting effects of drainage water management in Iowa’s subsurface drained landscapes. Agric. Water Manage. 92, 162–170. Skaggs R.W., 1980. DRAINMOD Reference Report. Methods for Design and Evalua-tion of Drainage-Water Management Systems for Soils with High Water Tables. USDA-SCS, South National Technical Center, Fort Worth, Texas 329. Tan, C.S., Drury, C.F., Gaynor, J.D., Welacky, T.W., 1993. Integrated soil, crop and water management system to abate herbicide and nitrate contamination of the Great Lakes. Water Sci. Technol. 28, 497–507. Tan, C.S., Zhang, T.Q., Drury, C.F., Reynolds, W.D., Oloya, T., Gaynor, J.D., 2007. Water quality and crop production improvement using a wetland reservoir and drainage/subsurface irrigation system. Can. Water Resour. J. 32, 129–136. Thomas, D.L., Hunt, P.G., Gilliam, J.W., 1992. Water table management for water quality improvement. J. Soil Water Conserv. 47 (1), 65–70. Thooko, L.W., Rudra, R.P., Dickinson, W.T., 1990. DRAINMOD in Ontario Conditions. ASAE., Paper No. 902064, St. Joseph, Mich. Thornthwaite, C.W., 1984. An Approach toward a Rational Classification of Climate. Geogr. Rev. 38 (1), 55–94. Wang, Sh., Prasher, S.O., Patel, R.M., Yang, C.C., Kim, S.H., Madani, A., Macdonald, P.M., Robertson, S.D., 2006. Fate and transport of nitrogen compounds in a cold region soil using DRAINMOD. J. Comput. Electron. Agric. 53 (2), 113–121.
Contents lists available at ScienceDirect
Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat
Impact of tile drainage on water budget and spatial distribution of sediment generating areas in an agricultural watershed Golmar Golmohammadi a,∗ , Ramesh Rudra a , Shiv Prasher b , Ali Madani b , Mohamed Youssef c , Pradeep Goel d , Kourosh Mohammadi a a
School of Engineering, University of Guelph, Guelph, Ontario, N1G 2W1, Canada Department of Bioresource Engineering, McGill University, Macdonald Campus, Ste-Anne-de-Bellevue, Quebec, H9X 3V9, Canada c Department of Biological and Agricultural Engineering, North Carolina State University, Raleigh, USA d Ontario Ministry of the Environment and Climate Change, Etobicoke, Ontario, M9P 3V6, Canada b
a r t i c l e
i n f o
Article history: Received 1 April 2016 Received in revised form 8 December 2016 Accepted 1 February 2017 Keywords: Hydrological modeling SWATDRAIN Watershed scale Sediment load Streamflow
a b s t r a c t A recently developed model, SWATDRAIN, was used to assess the alterations in water balance components, discharge, and sediment loads due to tile drainage practices in a heavily tile drained watershed in Ontario, Canada. Furthermore, the model was implemented to determine the spatial variability of sediment loads which can be explained by a combination of spatially distributed variables within a watershed, including those controlling the hydrology, geology, soil and land use. Three scenarios were examined across the watershed, including conventional drainage (existing condition), controlled drainage, and no tile drainage. The model predicted that streamflow was not significantly impacted due to tile drainage, while the total runoff and sediment loads from the basin due to controlled drainage were increased by 27.1% and 22.2%, respectively, while removing tile drain infrastructures resulted in 37.1% increase in surface runoff and 55% increase in sediment load from the watershed The areas with high sediment load generation were identified by the model and the impact of tile drainage in producing sediment in those areas was assessed. The results showed that the sediment load generation rate in the areas with the highest load (class V) increased by 8% only due to controlled drainage, while in the second ranked sediment generating areas (class IV), the sediment load generation rate was increased by 32%. © 2017 Elsevier B.V. All rights reserved.
1. Introduction The watersheds of Southern Ontario which provide food, feed, and fiber for millions of people and livestock are facing increasing pressure as a result of water pollution from agricultural activities. The alteration of settlements and development along the Grand River basin in Ontario has led to deterioration in water quality (Boyd et al., 2009). The Canagagigue Creek watershed, located in the Grand River Basin of south western Ontario, is one of the fastest developing areas in Ontario; this region has some of the most intensive agricultural production in the watershed. In the Canagagigue watershed, the vast majority of land (80%) is used for agricultural purposes and a high percentage of the agricultural lands (27–49% in the upper or more northern region and 0.2–6% in lower region) is enhanced with subsurface drainage infrastructure (Cooke, 2006). Subsurface drainage has the potential to reduce pollutants associ-
∗ Corresponding author. E-mail address: [email protected] (G. Golmohammadi). http://dx.doi.org/10.1016/j.agwat.2017.02.001 0378-3774/© 2017 Elsevier B.V. All rights reserved.
ated with surface runoff and sediment (Bengston and Xu, 1995). On the other hand, tile drains are the main pathways to increase the export of contaminants from fields to surface water bodies; this can result in substantial agricultural contaminants by translocation of sediments, nutrients and pesticides from the field to streams and lakes in those landscapes with intensive tile drainage, especially during the non-growing season and after heavy summer rains (Tan et al., 1993; Tan et al., 2007; Drury et al., 2007). Subsurface drainage systems have been identified as a major source of nutrients and other pollutants such as (NO3 -N), herbicides, insecticides and fungicides exported to water bodies (Moriasi et al., 2012; Thomas et al., 1992; Zucker and Brown, 1998). Over the past few decades, controlled drainage practices have been applied at both the field and the watershed scale to conserve water, regulate tile drainage flow, improve water quality, and increase crop yield (Tan et al., 2007; Drury et al., 2014; Busscher et al., 1992; Evans et al., 1992; Evans et al., 1995; Lalonde et al., 1996; Breve et al., 1998; Fausey. 2004; Singh et al., 2007; Ale et al., 2009). In order to quantify the effect of subsurface drainage on flow, pollutant losses and sediment transport, observed data must be collected and ana-
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
lyzed. Historically, the monitoring data were collected manually and conclusions were mostly empirically based. This process was very intensive, both in terms of time and human resources. Over the past few decades, great strides have been made in technology and in modeling techniques that help users to make relatively accurate predictions of ungagged watersheds which previously would have been impractical (Frana, 2012). Nowadays, with proper calibration, the hydrologic models can be implemented to widely varying landscapes with very useful results (Frana, 2012). The main objective of this study is to assess the impacts of tile drainage on discharge, sediment load and water budget of the Canagagigue West watershed and also to identify the potential sediment source areas under tile drainage practices; therefore, the current research has the objectives of i) predicting and analyzing the water balance components of the watershed; ii) quantifying the total sediment yield of the watershed and determining the spatial variability of sediment loads across the watershed; iii) comparing the water budget and sediment yield and assessing the susceptibility of agricultural lands under different scenarios of tile drainage. The recently developed model, SWATDRAIN (Golmohammadi et al., 2016), which is capable of simulating different water management scenarios, such as controlled drainage was selected. The SWATDRAIN model was calibrated and validated for an agricultural watershed, called Green Belt in Ontario, Canada (Golmohammadi et al., 2016) and the results showed that by using the SWATDRAIN model, the flow and water table dynamics were significantly improved compared to SWAT model. This research makes an important contribution to watershed sedimentation control practices and to planning the Best Management practices (BMPs).
125
2. Materials and methods 2.1. Watershed description The Canagagigue Creek has a total drainage area of 143 km2 and is a tributary of the Grand River. It lies between latitudes 43◦ 36 N and 43◦ 42 N and longitudes 80◦ 33 W and 80◦ 38 W, and is about 25 kilometers northwest of the city of Guelph, Ontario. The climate of the area, according to the Koppen-Geiger climatic classification system, can be characterized as humid continental with warm summers and moderate winters. Based on the availability of observed flow and sediment data and also the presence of tile drainage in agricultural regions mostly located in the western portion of the Canagagigue Watershed, this study targeted the upstream portion of the Canagagigue Creek west. The major land use is agriculture including winter wheat, corn, soybean, and other row crops. Table 2 provides the landuse distribution on that watershed. The observed flow data were obtained from the available hydrometric station called “Canagagigue Creek near Floradale” by the Atmospheric Environment Services of Environment Canada. Sediment loads measurements at this station for period 1974–1984 are provided as archived data at the web site of Environment Canada (www.wsc.ec.gc.ca). It has available historical observation data for the daily flow rate and sediment load for the period 1975–1984. The topography of the watershed is flat to gently undulating with a slight slope towards the outlet in the south. The average elevation is 417 m. Fig. 1 shows the location of the Canagagigue Creek and the sub-watershed used in this study.
Fig. 1. Location of the study area, Canagagigue West watershed, in Grand River Basin.
126
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
2.2. General overview of SWATDRAIN model SWATDRAIN is a watershed model which was recently developed by incorporating DRAINMOD into SWAT (Golmohammadi et al., 2016). The hydrologic cycle that is simulated by SWATDRAIN computes the soil water balance, on a daily basis, for each hydrological response unit (HRU) in every subbasin of the watershed. The daily values of tile drainage are calculated using the DRAINMOD model (Skaggs, 1980), which computes the tile drainage flux based on the Kirkham and/or Hooghoudt’s equations and the hydraulic capacity as a function of the daily water table elevation midway between the drains. The water table depth is calculated based on drainage volume versus the water table depth relationship which is calculated from the soil water characteristics curve. SWATDRAIN uses either the SCS curve number procedure (SCS, 1972), or the Green and Ampt infiltration method (Green and Ampt, 1911), to estimate infiltration and surface runoff. Different methods can be used in SWATDRAIN for potential evapotranspiration (PET) estimation: the Penman-Monteith method (Monteith, 1965), the Priestly-Taylor method (Priestley and Taylor, 1972), the Hargreaves method (Hargreaves and Samani, 1985), the Thornthwaite method (Thornthwaite, 1984) or PET can also be entered as daily or monthly PET input by the user. Detailed information on the SWATDRAIN
modeling approach can be found in Golmohammadi et al. (2014, 2016). 2.3. Input data and model setup Daily climate data, including maximum and minimum temperatures, precipitation, wind speed and humidity for the Fergus station were obtained from Environment Canada. In addition to the weather data inputs, the model requires Digital Elevation Model (DEM), soil and land use and agricultural management data, if applicable. A DEM with a 10 m × 10 m spatial resolution was obtained from the Grand River Conservation Authority (Fig. 2). The digital elevation model of the study area was processed to define the subbasin parameters, such as slope and slope length, and to define the stream network characteristics, such as channel slope, length and width. The resulting stream network was implemented to define a layout of 13 subbasins. The soil and land use classification across the watershed was defined by polygon shape files, provided by the Ontario Ministry of Agriculture and Agri-Food (Fig. 2). The combination of land use and soil type resulted in 67 HRUs. The soil surveys of Waterloo County presented by Presant and Wicklund (1971) and Wellington County presented by Hoffman et al. (1963) indicated that the major portion of the watershed has 20–60 centimeters of
Fig. 2. Digital Elevation model (DEM), soil, landuse and tile drainage maps.
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
127
Table 1 Soil distribution in Canagagigue Watershed. Soil type
Soil texture
Area (%)
BUF: Burford BAY: Brady BKN: Brookstone CTG: Conestogo BRR: Berrin BSB: Birsbane CAD: Caledone Total
loam/loam gravel/sand clay gravel/sand gravel sandy loam/sandy loam/sand loam/clay loam/clay loam silt loam/loam/silt loam/loam sandy loam/loamy sand/sandy loam/clay/clay loam/loam/loamy gravel/gravel sandy loam/sandy loam/sandy loam/sandy loam/sand/loamy sand
37.2 16.2 23.2 6.5 10.7 0.5 5.7 100
Table 2 Land use distribution on Canagagigue Watershed. Landuse
Area (%)
Winter wheat Corn Soybean Agricultural land-row crops Forest Hay Residential Medium density Total
30.56 26.92 22.87 7.53 6.89 5.14 0.09 100.00
loam or silty loam of Huron and Harriston series overlying a loam till. In the northern part of the watershed, clay loam is predominant. Loam is the main soil type in the southern portion of the watershed. Different soil type’s distribution across the watershed is presented in Table 1. Approximately 93.02% of the total area was under agricultural production, with the majority of land use under winter wheat, followed by corn, soybean, agricultural land row crops, and hay. The remaining 6.98% of the area was occupied by forest and residential lands. Land use distribution across the watershed is presented in Table 2. The tile drained area across the watershed was determined based on the tile drainage map obtained from the Land Information Ontario (Fig. 2). 2.4. Model evaluation The SWATDRAIN model was manually calibrated (1975–1979) and validated (1980–1984) for streamflow discharged from the Canagagigue Creek Watershed. These periods were selected in order to include dry, average, and wet years in each calibration and validation phase. The calibration and validation were performed based on the results of sensitivity analysis and on findings of previous studies (Rong, 2009; Oogathoo, 2006; Golmohammadi et al., 2014). The watershed has both systematic and non-uniform tile drainage system. The SWATDRAIN model is capable of evaluating subsurface drainage system whether they are uniform or non-uniform. Therefore, the model allows for various drainage layouts. However, in this study, due to lack of information on drainage layouts of different fields across the watershed, tile drainage systems was assumed uniform for entire watershed based on the literature and previous studies in Ontario condition A subsurface drainage system with drain spacing of 17 m at a depth of 1.00 m, operated as an approach of free drainage at the drain outlet, were recommended (GRCA, 2008; OMAFRA, 2009; Oogathoo, 2006; Golmohammadi et al., 2016). The initial Soil Conservation Service runoff curve number to moisture condition II (CN2) values calibrated using the SWAT model ranged from 66 to 78 (Rong, 2009). Some other calibration parameters adjusted in this study included the curve number coefficient (CNCOEF), surface runoff lag coefficient (SURLAG), and initial soil water storage expressed as a fraction of field capacity water content (FFCB).
Based on the literature, some parameters important to the snow-water mass balance were also investigated for their sensitivity to surface runoff, baseflow, evapotranspiration and streamflow within the watershed under study. These parameters were SMTMP (snow melt base temperature), SFTMP (snowfall temperature), TIMP (snow pack temperature lag factor), SMFMN (melt factor for snow on December 21) and SMFMX (melt factor for snow on June 21). Based on several tile drainage studies in southern Ontario, drainage coefficient (DC), the lateral hydraulic conductivity multiplication factor (LATKSATF) and effective radius of drains (RE) were set at 20.0 mm d−1 , 1.1 mm h−1 , and 15 mm, respectively (Thooko et al., 1990). The values of other streamflow calibration parameters were obtained from previous Canagagigue Creek Watershed studies (Golmohammadi et al., 2014; Rong, 2009). Additional calibration parameters were required for the SWATDRAIN model, given the new input parameters in the model, including soil retention curve parameters, saturated hydraulic conductivities of different soil layers, ET monthly factors, rooting depths and all parameters regarding freeze/thaw conditions. The calibrated parameters for the SWATDRAIN model used in this study are presented in Table 3. The model also requires inputs for the initial soil temperature profile, upper boundary condition and the lower boundary condition. Since a long record of measured soil surface temperatures is not usually available for most applications, air temperature was used (Luo et al., 2000, 2001). The lower boundary was assumed to be a constant soil temperature, which can be approximated from the long-term average air temperature (Penrod et al., 1958). In SWATDRAIN, lateral saturated hydraulic conductivity is soil-specific and it is read into the model by the soil file used for each HRU. The effective rooting depth is a difficult parameter to determine (Skaggs, 1980); and range of 45–55 cm was suggested by Rudra et al., 1985; Golmohammadi et al., 2016. No field measurements were made in this research; an initial rooting depths were obtained from the previous studies in Ontario (Golmohammadi et al., 2016; Wang et al., 2006; Singh et al., 2007; Dayyani et al., 2009; Thooko et al., 1990). It was assumed that, a restrictive layer set at the bottom of the soil profile. An initial estimated depth of 200 cm from the surface to the restrictive layer was used for the soil in this study. Although parameter input values can be determined by model calibration within the recommended range of values, known parameters values from previous studies can be used as well (Golmohammadi et al., 2016, 2014). Several trial and error runs were performed by varying these input parameters. The calibrated values for the sediment load are also listed in Table 3. In order to enter the land use and cover factor (USLE C), and soil erodibility factor (USLE K), properly reflecting the local characteristics, a soil erosion screening tool, Guelph Model for Evaluating the Effects of Agricultural Management Systems (GEAMS) on Erosion and Sedimentation (University of Guelph, Ontario, Canada, (1985) was used. Furthermore in order to reflect the real values of the soil properties of the studied watershed, two soil survey reports in Ontario by Hoffman et al., 1963 and Presant and Wicklund (1971) were used. In addition to graphical methods such as hydrographs, three performance criteria were used to evaluate the performance of the
128
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
Table 3 Adjusted values of parameters of the SWATDRAIN model for Canagagigue Creek Watershed. Parameter
Description
Value
ESCO EPCO FFCB
Soil evaporation compensation factor Plant uptake compensation factor Initial soil water storage expressed as a fraction of field capacity water content Surface runoff lag coefficient (day) Curve number method flag Curve number coefficient Depth to impervious layer (mm) Multiplication factor to determine lateral hydraulic conductivity (mmh−1 ) Lateral hydraulic conductivity (mmh−1 ) Effective radius of drains (mm) Drainage coefficient (mmday−1 ) Distance between two drain or tile tube (mm) Depth from soil surface to drain (mm)
0.80 0.65 0.95
SURLAG ICN CNCOEFF DEP IMP LATKSATF Ks Re Dc Drain spacing (L) Drain Depth ET monthly factors Rooting depth TKA, TKB TLAG TB Tsnow Tmelt CDEG CICE USLE K USLE C SPCON ADJ PKR PRF CH N2 CH K2 CH EROD CH COV
— Thermal conductivity functions W m−1 ◦ C Diurnal phase lag of air temperature (h) Soil temperature at the bottom of the profile (◦ C) Rain/snow-dividing temperature (◦ C) Snowmelt base temperature (◦ C) Snow melt coefficient (mm day−1 ◦ C−1 ) Critical ice content (cm3 cm−3 ) USLE soil erodibility factor USLE minimum crop factor Linear factor for channel sediment routing Subbasin peak sediment adjustment factor Main channel peak sediment adjustment factor Main channel manning’s roughness coefficient Main channel hydraulic conductivity(mmh−1 ) Channel erodibility factor Channel cover factor
model: Nash-Sutcliffe efficiency (NSE); (Nash and Sutcliffe, 1970), the RMSE-observations standard deviation ratio (RSR), and percent bias (PBIAS) (Gupta et al., 1999).
2.0 1 1.40 2200 1.1 Varies 15 20 15000 1000 Jan = 1, Feb = 1, March = 1, April = 0.8, May = 0.6, June = 0.8, July = 0.8, Aug = 0.8, Sept = 1, Oct = 1, Nov = 1, Dec = 0.6 Varies a = 0.553,b = 1.963 9 7 0 1 5 0.3 Varies Varies 0.01 1.0 2.0 0.04 10 0.23 0.85
Table 4 Statistical comparisons of observed and simulated streamflow at Canagagigue West watershed. Monthly
NSE = 1 − [
ni=1 (Yi obs − Yi sim )
ni=1 (Yi obs − Yobs mean )
RSR =
[
2
(1)
]
Index R2 NSE RSR PBIAS
Calibration 0.92 88.93 0.27 −6.07
Daily Validation 0.82 74.0 0.37 3.67
Calibration 0.85 78.70 0.36 −6.07
Validation 0.75 60.01 0.47 3.67
2
ni=1 (Yobs − Ysim ) ] i i
[
2
(2)
2
ni=1 (Yobs − Ymean ) ] i
3. Results and discussion 3.1. Streamflow and sediment loads
PBIAS = (
ni=1 (Yi obs − Yi sim )*(100) ni=1 (Yobs i )
)
(3)
where Yi obs is the observed value, Yobs Mean is the mean observed value, is the mean simulated value, and Yi sim is the predicted value,Ysim mean n is the total number of observations. A Multiple regression method was also used to find a relationship between the two main parameters impacting the sediment yield (Restrepo et al., 2006). The purpose of multiple regression in this research was to learn about the relationship between independent or predictor variables in the watershed and a dependent or criterion variable such as sediment yield. For this method, six different parameters including runoff, peak flow, mean water discharge rate at the outlet, catchment length, river length and mean slope were considered. Using step-wise regression, two control variables were selected which were mean annual runoff and peak flow. These two parameters were used to predict sediment yield for whole watershed and evaluate the impact factor of each parameter on sediment yield.
Time series plots of monthly streamflow during calibration and validation period is illustrated in Fig. 3. The results showed that the observed and simulated streamflows were in a good agreement. In general, during the calibration period, the model closely simulated streamflow values, except for two months of March 1976 and April 1979, which the streamflow was underestimated by the model (Fig. 4). During the validation period, observed and simulated sediment loads matched very well, except for the month of March 1984. The calibration and validation model performance results for the daily and monthly streamflow are presented in Table 4. During the calibration period, the monthly NSE, RSR, and PBIAS values were 0.89, 0.27 and −6.07, respectively (Table 4). According to Moriasi et al. (2007) a model is considered well calibrated for flow if monthly NSE ≥ 0.65, RSR ≤ 0.60 and PBIAS ≤ ± 10%. Therefore, the SWATDRAIN model was well calibrated as presented by the statistics in Table 4 and also supported by the monthly hydrographs (Fig. 3). The monthly NSE, RSR, and PBIAS values for the validation phase were calculated to be 0.74, 0.37 and 3.67 (Table 4). Watershed
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
129
Fig. 3. Observed and simulated monthly streamflows during calibration and validation periods.
Fig. 4. Observed and simulated monthly sediment loads during calibration and validation period.
model evaluation guidelines, developed by Moriasi et al. (2007), state that a model simulation performance for flow is considered good if NSE ≥ 0.65, RSR ≤ 0.60 and PBIAS ≤ ± 15. Therefore, in this study, the streamflow trends simulation performance rating was acceptable according to Moriasi et al., 2007. Observed and simulated monthly sediment loads during calibration and validation periods are presented in Fig. 4. During the calibration phase, the observed and simulated sediment load matched well, although in the month of March 1977 sediment is overestimated and in the months of April 1978 and April 1979, the sediment is underestimated by the model (Fig. 4).
During the validation period, measured and simulated sediment loads matched very well, except for the snowmelt periods of the years 1980 and 1984. The values of NSE, RSR and PBIAS for sediment load are presented in Table 5. Watershed model evaluation guidelines, developed by Moriasi et al. (2007), state that a model simulation performance for sediment is considered good if monthly values of NSE ≥ 0.65, RSR ≤ 0.60 and PBIAS ≤ ± 15.Therefore, the simulation results are very good due to the high statistical results during the validation period (NSE = 0.80; RSR = 0.44; PBIAS = 25.88).
130
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
Table 5 Statistical comparisons of observed and simulated streamflow at Canagagigue West watershed. Monthly Index R2 NSE RSR PBIAS
Calibration 0.88 0.83 0.41 −1.71
Daily Validation 0.89 0.80 0.44 25.88
Calibration 0.82 0.74 0.48 −1.71
Validation 0.73 0.71 0.49 25.88
Table 6 Average annual water balance over the simulation period for the Canagagigue West watershed. Components
Precipitation; PCP Water yield; WYLD Tile flow; TILEQ Surface runoff; SURQ Lateral flow; LATQ Groundwater contribution to streamflow; GQ Actual Evapotranspiration; ET Percolation; PERC
Average Annual [1975–1984] (mm)
(%)
950.3 444.4 113.1 243.9 50.1 37.3 442.5 41.3
100 46.8 11.9 25.6 5.3 3.9 46.6 4.3
4. Water balance analysis After the model evaluation, the water budget of Canagagigue West Watershed was analyzed based on the amount of precipitation received, and simulated discharges and evapotranspiration. The average annual values for water balance components are presented in Table 6. The relative percentages of the watershed balance components were calculated by dividing each component into the total precipitation received. The results of the simulated water budget show that the model simulated water processes much like those that would occur in nature. The simulated water balance results under existing condition (65% of the watershed tile drained) showed that the major portion of the precipitation (46.6%) received by the watershed contributed to streamflow at the outlet of the watershed during the 10 years of the simulation. This corresponds to the observed average annual streamflow of 439 mm (45%) of the precipitation. The simulated evapotranspiration accounted for 46.6% of the total precipitation which was similar to the historical average ET of –444 mm obtained by the various empirical methods that used canopy cover, latitude, elevation and precipitation, and PET for this site (Rong, 2009; Oogathoo, 2006). While the simulated average values of surface runoff and tile drainage were 25.6% and 11.9% of precipitation, lateral flow, calculated as a percentage of average annual precipitation, was 5.3%. During the simulation period, 46.6% of annual precipitation was simulated as being lost in evapotranspiration across the watershed. The high evapotranspiration rate could be attributed to the land cover of the study area. The average annual contribution of groundwater as a relative percentage to precipitation is only 3.9% for the study period. Deep aquifer recharge is very low with an average of 0.1% which is negligible. However, the difference between the amount of precipitation and the sum of the discharges and the evapotranspiration was obvious. The water balance error was obtained by balancing the major hydrologic components simulated in the model. The model performed well with a relative error of 2.3% for the simulation period. 5. Impact of tile drainge The impact of tile drainage on water balance of the Canagagigue watershed was assessed by implementing different tile drainage
practices across the watershed. Three scenarios of i) existing condition (65% of the watershed under tile drainage), ii) controlled drainage (65% of the watershed under tile drainage), and iii) no tile drain infrastructure installed in the watershed were implemented to illuminate the effects of drainage. Even though a shift to non-drained agriculture in Ontario is not expected, it is a useful exercise to provide evidence for the benefits of tile drainage systems to ensure ease of planting, growth, and harvesting. Based on the simulated average annual water balance (Fig. 5), under existing condition, 46.7% of precipitation (442.5 mm) was simulated as evapotranspiration; this was increased to 49.4% (469.5 mm) for the situation in which controlled drainage was applied instead of conventional drainage. While, for the watershed with no tile drainage, the ET value was increased to 50.8% of the annual precipitation (482.5 mm). Therefore, based on the simulated average annual water balance [1974–1984], ET was significantly decreased by implementing tile drainage. The model simulated that 46.8% of precipitation (444.4 mm) contributed to streamflow at the outlet of the watershed during the 10 years of simulation (Table 6). This value was increased to 48.0% (455.9 mm) under the scenario of controlled drainage. While, if there were no drainage infrastructure installed across the watershed, the streamflow decreased to 48.5% (461.0 mm). These simulated results of average annual water budget provide evidence for the hydrological impact on the watershed by the scenarios of controlled drainage and the absence of tile drainage. Evapotranspiration increased as there was more water available on the soil surface for evapotranspiration. Although the streamflow was not decreased significantly by controlled tile drainage, but surface runoff increased significantly. In addition, without subsurface drainage, water is allowed to saturate the upper profile of the unsaturated zone, which in turn may increase the deep recharge in the groundwater aquifer (Fig. 5).
6. Streamflow impact analysis In order to assess the significant changes in streamflow due to potential changes in subsurface drainage in the Canagagigue watershed, certain hydrologic analyses, including flood frequency analysis and normalized hydrography analysis, were conducted (Rahman et al., 2014). Flood frequency analysis was conducted to identify the changes in the magnitude and frequency of annual peak flows in the watershed for the controlled drainage and no tile drainage scenarios. Fig. 6 compares the flood frequency curves under the existing, controlled and no tile drainage scenarios. The flood frequency results showed that the subsurface drainage under the controlled drainage scenario will increase the frequency of peak flows while decreasing the frequency of minor flows. The results also showed that the no tile drainage scenario will increase the frequency of peak flows and decrease the frequency of minor flows more than the scenario of controlled drainage. Nevertheless, Fig. 6 shows that the changes in magnitude and frequency of peak flows are not statistically significant, since all three peak flow frequency curves lie completely within the 95% confidence interval curves. In order to evaluate the possible changes in the shape of hydrographs, particularly during the snowmelt period, a normalized hydrograph analysis was conducted using the criteria adopted from Rahman et al. (2014). The hydrographs were selected from the snowmelt events, which included only one main peak with complete daily records of 41 days (15 days before and 25 days after the peak), and no other complications in the shape of the hydrograph such as minor flows (Miller and Frink, 1984; Rahman et al., 2014). Therefore, those hydrographs which did not provide a useful characterization of a simple runoff hydrograph shape were removed from the analysis.
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
131
Fig. 5. Average annual water balance for Canagagigue Watershed under i) existing condition, ii) Entire Watershed tile drained, and iii) no tile draianage.
The selected hydrographs were normalized by dividing each discharge value by the peak discharge value, which resulted in normalized hydrographs varying between 0 and 1. Fig. 7 compares the averaged normalized hydrographs under the three tile drainage scenarios. It shows that the normalized hydrographs of the existing condition has a steeper rising limb which translates into the shorter time to peak of the hydrograph. Fig. 7 also shows that the alteration in shape of the hydrograph is not statistically significant, since all three hydrographs lie completely within the 95% confidence intervals. Therefore, it can be concluded that the increase of peak flow rates at smaller recurrence probabilities under the controlled
drainage is mainly due to the flow volumes rather than increasing the time to peak of the hydrograph. 7. Impact of tile drainage on sediment yield Watershed runoff and sediment load simulation results are presented in Table 7. The sum of the surface runoff simulated for the 1975–1984 years was 2539.2 mm under the existing condition compared with 3227.1 mm and 3481.1 mm under controlled drainage and no tile drainage scenarios. Controlled drainage lead to an increase of 27.1% in total annual surface runoff. The scenario of no tile drainage across the watershed resulted in an increase of 37.1% in total annual surface runoff. Total sediment loads estimated
132
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
Table 7 Simulated surface runoff and sediment load results for conventional, controlled and no tile scenarios. Year
1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 Total
Runoff
Sediment loads
Existing Condition
Controlled Drainage
No Drainage
Existing Condition
Controlled Drainage
No Drainage
317.7 301.9 258.3 192.8 267.9 227.3 198.9 294.7 173.4 306.3 2539.2
397.97 387.89 300.12 243.96 365.04 275.2 247.21 390.82 230.1 388.76 3227.1
412.9 409.2 319.9 266.7 382.1 284.1 286.7 404.1 287.6 428.4 3481.1
1.0 0.7 0.8 0.5 0.6 0.6 0.4 0.6 0.2 0.9 6.3
1.3 0.9 1.0 0.6 0.7 0.6 0.5 0.8 0.3 1.0 7.7
1.2 1.1 1.0 1.0 1.0 0.8 0.6 1.1 0.7 1.4 9.8
Fig. 6. Peak flow frequency analysis for the Canagagigue Watershed.
Fig. 7. Average Normalized-Hydrograph (NH) based on the simulated streamflow in the Canagagigue Watershed.
for the study period were 6.3 tons/ha for the exiting condition while it was predicted 7.7 tons/ha for the scenario of controlled drainage and 9.8 tons/ha for the condition in which the entire tile drains were
removed. These values translate to 22.2% and 55.0% increases associated with the controlled drainage and no tile drainage scenarios. The simulated average seasonal and annual sediment loads are presented in Fig. 8. The figure shows that most of sediment load increase happened during the spring season. Comparing water balance simulation results, under three different scenarios which was presented in Fig. 5, indicates that reduction in subsurface drainage due to controlled drainage was reflected in an increase in surface runoff at the watershed scale. The increase of surface runoff may lead to increase soil erosion and sedimentation in the watershed. Analysis of sediment loads and different hydrologic variables in the Canagagigue Watershed indicated that the surface runoff is the main factor explaining the variation in observed sediment loads. The sediment load was estimated using a multiple regression model that includes runoff and peak flow. The results showed that these estimators explain 61% of variance in sediment yield. It can be concluded that the spatially distributed variables, such as topography, soil and land use are also important factors in explaining the sediment load spatial variability. Fig. 9 shows the spatial distribution of sediment yield for the existing condition, controlled drainage and no tile drainage scenarios. The sediment yield varied in the range of 0.0 t ha −1 to 4.0 t ha −1 across the watershed which was divided into 5 different classes (Fig. 9) according to a previous study in Ontario by Dickinson and Rudra (1986). It was noted that the great majority of sediment loads were generated from a small percentage of the watershed. By combining the sediment spatial distribution and subbasin maps, it was determined that under the existing condition, the highest sediment load generating areas, Class V, are mainly located in subbasins number 5 and 6 and 7, followed by the areas with the load class IV which are mostly located in subbasins 8, 9, 12, 13, and 3. The results of this study showed that sediment yield rates and its distribution vary with land cover and soil. Analyzing the spatial distribution of the sediment loads and the soils and land use across the watershed indicate that the areas with sediment loads of type V are covered by agricultural row crops and winter wheat. The soil types in this area are mostly Conestoga (silt loam), Brookston (clay loam), and Burford (loam) which have a high erodibility factor (30 and more) compared to the other soils (ranges from 10 to 14) in the study area. Therefore, it can be concluded that the interaction of highly erodible soils and agricultural land use, such as agricultural row crops and winter wheat, resulted in high sediment loads (Class V). The results also indicate that the areas with low amounts of sediment are mostly under hay and forest while the soil types of Bradly (sandy loam) and Caledone (sandy loam) with low erodibility factor (10–12) are dominant in these areas. The impact of tile drainage on spatial distribution of sediment loads was assessed by comparing the sediment load maps of the three scenarios presented in Fig. 9. By comparing the results of sediment load distribution under
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
133
Fig. 8. Simulated average seasonal and annual sediment loads.
Fig. 9. Spatial distribution of sediment loads categories across the watershed.
three different scenarios, it was noted that the implementation of controlled drainage will not affect the sediment load produced in critical sediment load areas of class V (> 2 t ha−1 ) under existing condition (Subbasins 5, 6 and 7). While in the areas with sediment load of Class IV (1.0–2 t ha−1 ) under existing condition, (Subbasins 8, 9, 12, 13, and 3), implementing controlled drainage increased the sediment loads significantly. This can be attributed to different soil permeability in these areas. In the areas with high sediment production, the Conestoga and Brookston soils are imperfectly drained and poorly-drained soils, respectively, while the Burford soil is a well-drained soil. The results of this study indicate that the areas with the soil type, Burford, which is a well-drained soil, were most impacted by tile drainage. The impact of tile drainage in the areas with Brookston soil was not significant. The results showed that removing the controlled drainage, in general increased the sediment loads across the watershed. This may be due to an increase
in surface runoff which may lead to increase soil erosion and sedimentation in the watershed. Spatial distribution of the sediment loads under no drainage scenario. 8. Conclusions A recently developed SWATDRAIN model was used to assess the impact of tile drainage on water budget and sediment loads from the Canagagigue watershed in southern Ontario. The results showed that the controlled drainage settings lead to surface runoff and a sediment load increase of 27.1% and 22.2% respectively. While removing tile drain infrastructures resulted in 37.1% increase in surface runoff and 55% increase in sediment load in the watershed. The results demonstrated that tile drainage has a significant impact on watershed hydrology and sediment yields and SWATDRAIN is an appropriate model to simulate these impacts.
134
G. Golmohammadi et al. / Agricultural Water Management 184 (2017) 124–134
The areas contributing to sediment loads and the extent to which the sediment loads increased in those areas due to tile drainage was also estimated by the model. The spatial variability of sediment loads can be explained by a combination of spatially distributed variables within the watershed, including those controlling hydrology, geology, soil and land use. The SWATDRAIN model predicted that in the Canagagigue watershed, the areas with the highest sediment load values (> 2 tones/ha) were not affected by controlled drainage significantly while the areas with high sediment load values (1.5–2 tones/ha) were affected significantly. The investigation of sediment load results can help in identifying the critical areas of sediment generation in order to design appropriate management practices for sedimentation control. Acknowledgement The funding received for this project from the Natural Sciences and Engineering Research Council of Canada is gratefully acknowledged. References Ale, S., Bowling, L.C., Brouder, S.M., Frankenberger, J.R., Youssef, M.A., 2009. Simulated effect of drainage water management operational strategy on hydrology and crop yield for Drummer soil in the Midwestern United States. Agric. Water Manage. 96, 653–665. Bengston David N., Xu Z., 1995. Changing National Forest Values: a content analysis. Research Paper NC-323. St. Paul, MN: U.S. Dept. of Agriculture, Forest Service, North Central Forest Experiment Station. Boyd, D., Cooke, S., Yerex, W., 2009. Exploring Grand-Erie connections: flow, quality and ecology. 9th Annual Grand River Watershed Water Forum. 18 September 2009. Breve, M.A., Skaggs, R.W., Parsons, J.E., Gilliam, J.W., 1998. Using the DRAINMOD-N model to study the effects of drainage system design and management on crop productivity, profitability and NO3-N losses in drainage water. Agric. Water Manage. 35 (3), 227–243. Busscher, W.J., Sadler, E.J., Wright, F.S., 1992. Soil and crop managementaspects of water table control practices. J. Soil Water Conserv. 47, 71–74. Cooke, S., 2006. Water quality in the grand river: A summary of current conditions (2000–2004) long term trends. Grand River Conservation Authority. Dayyani, S., Madramootoo, C.A., Enright, P., Simard, G., Gullamudi, A., Prasher, S.O., Madani, A., 2009. Field evaluation of DRAINMOD 5. 1 under a cold climate: simulation of daily midspan water table depths and drain outflows. J. Am. Water Resour. Assoc. 45 (3), 779–792. Dickinson, W.T., Rudra, R.P., 1986. Identification of soil erosion and fluvial sediment problems. J. Hydrol. Proc. 1, 111–124. Drury, C.F., Yang, J.Y., De Jong, R., Yang, X.M., Huffman, E.C., Kirkwood, V., Reid, K., 2007. Residual soil nitrogen indicator for agricultural land in Canada. Can. J. Soil Sci. 87, 167–177. Drury, C.F., Tan, C.S., Welacky, T.W., Reynolds, W.D., Zhang, T.Q., Oloya, T.O., McLaughlin, N.B., 2014. Reducing nitrogen loss in tile drainage water with cover crops and water-table management systems. J. Environ. Qual. 43, 578–598. Evans, R.O., Parsons, J.E., Stone, K., Wclls, W.B., 1992. Water tahle management on a watershed scale. J. Soil Water Conserv. 47 (1), 58–64. Evans, R.O., Skaggs, R.W., Gilliam, J.W., 1995. Controlled versus conventional drainage effects on water quality. J. Irrig. Drain. Eng. ASCE 121 (4), 271–276. Fausey, N., 2004. Comparison of free drainage, controlled drainage, and subirrigation water management practices in an Ohio Lakebed soil. In: ASAE Meeting Paper No. 042-237, ASAE, St. Joseph, MI. Frana, A.S., 2012. Applicability of MIKE SHE to Simulate Hydrology in Heavily Tile Drained Agricultural Land and Effects of Drainage Characteristics on Hydrology M.Sc. Thesis. Iowa State University. GRCA, 2008. Grand River Watershed Characterization Report. Grand River Conservation Authority. Golmohammadi, G., Prasher, S., Madani, A., Rudra, R., 2014. Evaluating three hydrological distributed watershed models: MIKE-SHE, APEX, SWAT. Hydrology 1 (1), 20–39. Golmohammadi, G., Prasher, S., Madani, A., Rudra, R., Youssef, M.A., 2016. SWATDRAIN, a new model to simulate the hydrology of agricultural lands, model development and evaluation. J. Biosyst. Eng. 141, 31–47.
Green, W.H., Ampt, G.A., 1911. Studies on soil physics: part I. The flow of air and water through soils. J. Agric. Sci. 4, 1–24. Gupta, H.V., Sorooshian, S., Yapo, P.O., 1999. Status of automatic calibration for hydrologic models: comparison with multilevel expert calibration. J. Hydrol. Eng. 4 (2), 135–143. Hargreaves, G.H., Samani, Z.A., 1985. Reference crop evapotranspiration from temperature. Appl. Eng. Agric. 1, 96–99. Hoffman, D.W., Matthews, B.C., Wicklund, R.E., 1963. Soil survey of wellington county Ontario, report No.35, Ontario soil survey Canada. Dep. Agric. Ontario Dep. Agric. Lalonde, V., Madramootoo, C.A., Trenholm, L., Broughton, R.S., 1996. Effects of controlled drainage on nitrate concentrations in subsurface drain discharge. Agric. Water Manage. 29, 187–199. Luo, W., Skaggs, R.W., Chescheir, G.M., 2000. DRAINMOD modifications for cold conditions. Trans. ASAE 43 (6), 1569–1582. Luo, W., Skaggs, R.W., Madani, A., Cizikci, S., Mavi, A., 2001. Predicting field hydrology in cold conditions with DRAINMOD. Trans. ASAE 44 (4), 825–834. Miller, J.E., Frink, D.L., 1984. Changes in Flood Response of the Red River of the North Basin, North Dakota-Minnesota. Water-Supply Paper 2243. U.S. Geological Survey, Alexandria, Virginia. Monteith, J.L., 1965. Evaporation and environment. Symp. Soc. Exp. Biol. 19, 205–234. Moriasi, D.N., Rossi, C.G., Arnold, J.G., Tommer, M.D., 2012. Evaluating hydrology of the Soil and Water Assessment Tool (SWAT) with new tile drain equations. J. Soil Water Conserv. 67 (6), 513–524. Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel, R.D., Veith, T.L., 2007. Model evaluation guidelines for systematic quantification of accuaracy in watershed simulations. Trans. ASABE 50 (3), 885–900. Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models: part 1. A discussion of principles. J. Hydrol. 10 (3), 282–290. OMAFRA, 2009. Agronomy Guide for Field Crops. Ministry of Agriculture, Food and Rural Affairs (Pub. 811, PP306). Oogathoo, Sh., 2006. Runoff Simulation in the Canagagigue Creek Watershed Using the MIKE SHE Model. Master’s Thesis. Sainte Anne-de-Bellevue, QC; McGill University, Department of Bioresource Engineering. Penrod, E.B., Walton, W.W., Terrel, D.V., 1958. A method to describe soil temperature variation. J. Soil Mech. Found ASCE 1537, 1–21. Presant, E.W., Wicklund, R.E., 1971. Soil survey of Waterloo county, report No. 44 of the Ontario soil survey Canada. Dep. Agric. Ontario Dep. Agric. Priestley, C.H.B., Taylor, R.J., 1972. On the assessment of surface heat flux and evaporation using large-scale parameters. Mon. Weather Rev. 100, 81–92. Rahman, M.M., Lin, Z.H., Jia, X., Steele, D.D., DeSutter, T.M., 2014. Impact of subsurface drainage on streamflows in the Red River of the North basin. J. Hydrol. 511, 474–483. Restrepo, J.D., Kjerfve, B., Hermelin, M., Restrepo, J.C., 2006. Factors controlling sediment yield in a major South American drainage basin: the Magdalena River, Colombia. J. Hydrol. 316, 213–232. Rong, H., 2009. Water Quality Evaluation Under Climate Change Impacts for Canagagigue Creek Watershed in Southern Ontario Master’s Thesis. Guelph University, Department of Bioresource Engineering. Rudra, R.P., Dickinson, W.T., Wall, G.J., 1985. Application of the CREAMS model in southern Ontario conditions. Trans. ASAE. 28 (4), 1233–1240. Soil Conservation Service (SCS), 1972. Section 4: Hydrology. In: National Engineering Handbook. USDA Soil Conservation Service, Washington, D.C. Singh, R., Helmers, M.J., Crumpton, W.G., Lemke, D.W., 2007. Predicting effects of drainage water management in Iowa’s subsurface drained landscapes. Agric. Water Manage. 92, 162–170. Skaggs R.W., 1980. DRAINMOD Reference Report. Methods for Design and Evalua-tion of Drainage-Water Management Systems for Soils with High Water Tables. USDA-SCS, South National Technical Center, Fort Worth, Texas 329. Tan, C.S., Drury, C.F., Gaynor, J.D., Welacky, T.W., 1993. Integrated soil, crop and water management system to abate herbicide and nitrate contamination of the Great Lakes. Water Sci. Technol. 28, 497–507. Tan, C.S., Zhang, T.Q., Drury, C.F., Reynolds, W.D., Oloya, T., Gaynor, J.D., 2007. Water quality and crop production improvement using a wetland reservoir and drainage/subsurface irrigation system. Can. Water Resour. J. 32, 129–136. Thomas, D.L., Hunt, P.G., Gilliam, J.W., 1992. Water table management for water quality improvement. J. Soil Water Conserv. 47 (1), 65–70. Thooko, L.W., Rudra, R.P., Dickinson, W.T., 1990. DRAINMOD in Ontario Conditions. ASAE., Paper No. 902064, St. Joseph, Mich. Thornthwaite, C.W., 1984. An Approach toward a Rational Classification of Climate. Geogr. Rev. 38 (1), 55–94. Wang, Sh., Prasher, S.O., Patel, R.M., Yang, C.C., Kim, S.H., Madani, A., Macdonald, P.M., Robertson, S.D., 2006. Fate and transport of nitrogen compounds in a cold region soil using DRAINMOD. J. Comput. Electron. Agric. 53 (2), 113–121.