This simulation study focused on the hydrological effects of climate change and controlled drainage operated with subsurface drains and an open collector ditch in an agricultural field. The objective was to understand the potential of controlled drainage and open ditch schemes for managing groundwater levels and field water balance in climate conditions projected to take place in Finland during the 21st century with representative concentration pathways 8.5 and 2.6. A methodological aim was to find ways to condense hourly hydrological results to understand future changes in field hydrology. During the historical reference interval (1970–2005), controlled drainage caused 17–36 cm higher mean groundwater levels and decreased the mean annual drain discharge by 11–23% compared to conventional subsurface drainage. Controlled drainage was projected to increase groundwater levels by additional 1–4 cm in the future compared to its effect on drainage during the reference interval. The effect on annual drain discharge did not change significantly. The open collector ditch lowered groundwater tables and diminished the effect of controlled drainage on groundwater levels in the vicinity of the ditch. Controlled drainage was shown to remain an effective method for countering early summer drought and reducing drain discharge.

  • Controlled drainage provides an option to mitigate early summer droughts.

  • Controlled drainage can be used to decrease annual drain discharge.

  • The potential effects of controlled drainage will remain similar or increase slightly under the projected future climate.

  • An open collector ditch affected the performance of controlled drainage.

The main limitations for crop yields in high-latitude areas such as Finland are the short growing season and the early summer droughts (Peltonen-Sainio et al. 2016). While the long-term mean rainfall is sufficient for cropping, summer rainfall tends to take place contrary to crop requirements (Peltonen-Sainio 2012). Environmental impacts are another problem in agriculture, as it is responsible for most of the nutrient loads to surface waters (Vuorenmaa et al. 2002). Climate change is expected to exacerbate these challenges. The annual mean air temperature and annual precipitation are projected to increase in Finland (Ruosteenoja et al. 2016), but the projected increase in early summer precipitation is only moderate, especially in the productive southwestern crop areas (Ylhäisi et al. 2010), and future precipitation is likely to be distributed more unevenly (Lehtonen 2011). Increased runoff is likely to occur outside the growing season increasing the environmental impact of agriculture (Barnett et al. 2005; Huttunen et al. 2015). All these challenges can be mitigated with proper field water management. Currently, field water management in Finland is mostly based on conventional subsurface drainage, but the climate change-induced hydrological variability makes more flexible methods attractive.

Controlled drainage is a method where drainage efficiency can be reduced by elevating the drain outlet. Its effects on field hydrology have not been comprehensively studied in Nordic countries. The research consists mainly of experimental studies, most of which have been done in acid sulfate soils and peatlands, where the environmental impacts of agriculture, such as runoff acidity and greenhouse gas emissions, can be reduced by controlling the groundwater table. Joukainen & Yli-Halla (2003), Österholm et al. (2015) and Åström et al. (2007) showed in comparisons of controlled drainage and conventional subsurface drainage in acidic sulfate soils that controlled drainage alone was not able to sufficiently affect the groundwater table, and the targeted environmental impact reductions were not reached. However, Myllys (2019) noted that controlled drainage was an effective method to raise the groundwater table and reduce the environmental impacts in peatlands. In a comparison study conducted in mineral soil, where the environmental impact is mostly caused by nutrient leaching via runoff or discharge, Wesström & Messing (2007) found that controlled drainage had a large effect on both drain discharge and nutrient loads.

Unlike experimental studies, the use of a model application enables the research of field hydrology and water management methods under selected future climate scenarios. The future performance of controlled drainage has been previously simulated in southern Sweden (Abdelbaki 2015), central-western Poland (Sojka et al. 2020) and Ohio, USA (Pease et al. 2017). According to Abdelbaki (2015), the effects of controlled drainage on drain discharge reduction will remain similar in the future (emission scenario B2). Sojka et al. (2020), who simulated the effects of controlled drainage from March to September with representative concentration pathway (RCP) 4.5 as the emission scenario, found that the groundwater table spent less time above the drainage depth in the future resulting in a lower effect of controlled drainage on groundwater levels and drain discharge. Pease et al. (2017) used emission scenarios RCP 4.5 and RCP 8.5 and found that the drain discharge decreased in the future, but the effect of controlled drainage on annual drain discharge remained similar in the 21st century, whereas the effect on summer drain discharge increased. These studies did not consider groundwater outflow, whose importance in the field water balance has been noted, e.g. by Turunen et al. (2013) and Rozemeijer et al. (2016). Especially in Nordic countries, little to no research exists on controlled drainage operated with subsurface drains and an open collector ditch under future climate conditions with water balance including groundwater outflow. What is also needed is better understanding of how different control schedules affect groundwater table at different times of the year and how this will change in future conditions.

In simulations of future hydrology, climate projections extending to many decades in the future contain large uncertainties. Systematic biases are found in regional climate models (RCMs) (Jacob et al. 2007; Lind & Kjellström 2009), and while many of the biases originate from the general circulation models (GCMs) used as boundary conditions, RCMs can amplify the biases in some cases (Kjellström & Lind 2009). Projected changes in precipitation are more uncertain than those of temperature, which are more consistent among different climate models (Räisänen & Ruokolainen 2006). To deal with the uncertainties related to climate projections, projection ensembles including different models and/or scenarios have been used (Benestad et al. 2017), although combining projections and interpreting the ensembles have their own challenges as well (Tebaldi & Knutti 2007). One approach to reduce the overall uncertainty of model results is to tie the model setup to the existing field site and use data tested model setup. This way the uncertainties can be more confidently limited to the input data.

An important step in hydrological simulation is the selection of a model that fits the purpose. All three simulation studies cited above were conducted with DRAINMOD, which is a lumped, field scale, process-based hydrologic model developed to simulate water balance in artificially drained soils. Lumped models are parameterized with the average characteristics of the modeled field. They are relatively easy to use and require fewer input parameters than spatially distributed models, but they can be insufficient when the modeled area has high spatial heterogeneity in hydro-geological characteristics and results are studied at high temporal resolution (e.g. Dai et al. 2010) or during high flow periods (e.g. Muma et al. 2016).

The temporal resolution of hydrological simulations is tightly linked to the simulated processes and the modeling objectives. On a field scale, the hydrological processes can have a strong sub-daily variation. The effects of intense but short precipitation events on runoff generation or fast preferential flow in macropores on field scale hydrology cannot necessarily be captured with daily resolution (Warsta et al. 2013).

This study focused on the effects of controlled drainage, an open collector ditch and climate change on the hydrology of an agricultural field located at latitude 64° N in Northern Ostrobothnia, Finland. The field represents a typical flat subsurface-drained field in the coastal agricultural area along the Bothnian Bay. The objective was to computationally estimate (1) the theoretical potential of controlled drainage in regulating groundwater levels, especially with respect to mitigating early summer droughts, (2) the potential of controlled drainage in reducing drain discharge, (3) how the effects of controlled drainage change in the 21st century and (4) how the groundwater table profile of the field and controlled drainage are affected by an open collector ditch. As previous studies in Finland have focused on the environmental aspects of controlled drainage, the effects on groundwater levels and runoffs both currently and in future decades are yet to be examined. A hydrological model FLUSH was applied in this study due to its spatial description and its previous applications in Nordic conditions (e.g. Turunen et al. 2013, 2015; Warsta et al. 2013; Nousiainen et al. 2015). The model was calibrated and validated by Salo et al. (2021) using drain discharge and groundwater table measurements from the field when it was under conventional subsurface drainage. The FLUSH model, which has been benchmarked to several field sites in Finland, and a field parameterization under conventional subsurface drainage were used as a basis for the control scenario simulations. To study hydrological impacts of various control scenarios at the selected site, computational drain control and open collector ditch were added to the model application. The study provides a novel computational methodology to study the holistic effects of field water management solutions in future climate conditions which can be divided into three parts: (a) the use of hourly climate model data sets in a process-based state-of-the-art spatial hydrological model, (b) the description of potential future field hydrology, including groundwater outflow, of a Nordic field under controlled drainage operated with subsurface drainage and an open collector ditch and (c) the visualization of the simulation results to understand the future trends of the effects of controlled drainage on field drainage potential.

Study site description

The study site is a field of arable land located in Sievi, Northern Ostrobothnia, Finland (N 63° 55,978′ and E 24° 20,646′). In 2015, a field experiment was established at the site to study subsurface drainage implemented with two different drainage installation machines with 15 m drain spacing and drainage depth of 1 m (Äijö et al. 2017). The field was modified for a controlled drainage experiment in June 2019 when half of the field was equipped with controlled drainage and the other half was left as conventional subsurface drainage (Äijö et al. 2021). Data from a controlled drainage period were not available for this study, as the first year (2019) was dry and no discharge data were recorded.

The field has a mean slope of less than 0.2% and is surrounded by similar flat arable land. It has been used for cultivating oat, barley, grass and field mustard. At the drainage depth (∼1 m), the soil varies between loam, sandy loam and loamy sand. The topsoil consists of loamy sand with high organic matter content (Äijö et al. 2017).

Climate data

The meteorological input in this study was bias-corrected time series from regional climate projections by EURO-CORDEX for the years 1970–2100 (Table 1). For the years 2006–2100, future climate projections RCP 8.5 and RCP 2.6 were used, whereas a historical projection was used for the years 1970–2005. To form the time series used in this study, average values were taken over nine cells of the RCM (3 × 3 grid) located over the Sievi field, and the values for air temperature, precipitation, shortwave radiation and longwave radiation were bias-corrected by comparing simulated values of the historical period (1970–2005) to observations from the Finnish Meteorological Institute (FMI). The temporal and spatial resolutions of the regional climate simulations were 1 h and 12.5 km, respectively. In addition to the meteorological input from the climate model, hourly potential evapotranspiration was calculated with the Penman–Monteith equation described in Allen et al. (1998). Each year in the time series consisted of 365 days, meaning leap days were not considered.

Table 1

GCM and the RCM that produced the EURO-CORDEX time series

GCM 
 Institute Norwegian Climate Center (NCC) 
 Model NorESM1-M 
 Run r1i1p1 
RCM 
 Institute Swedish Meteorological and Hydrological Institute (SMHI) 
 Model RCA4 
 Resolution Eur-11 (∼12.5 km) 
GCM 
 Institute Norwegian Climate Center (NCC) 
 Model NorESM1-M 
 Run r1i1p1 
RCM 
 Institute Swedish Meteorological and Hydrological Institute (SMHI) 
 Model RCA4 
 Resolution Eur-11 (∼12.5 km) 

Figure 1 shows the temporal development of variables describing the climate conditions used in the simulations during the years 1970–2100. Precipitation, air temperature and longwave radiation increase in both future scenarios, the increase being larger in RCP 8.5. Despite the rising air temperature, the long-term development of potential evapotranspiration is moderate, which leads to a larger ratio of precipitation to potential evapotranspiration in the future. Potential evapotranspiration is affected by the decreasing shortwave radiation, which in turn is linked to increased cloudiness.

Figure 1

Annual mean air temperature (a), precipitation (b), shortwave radiation (c), longwave radiation (d) and potential evapotranspiration (e) of the used meteorological input for years 1970–2100 (historical emission scenario + RCP 8.5 and RCP 2.6).

Figure 1

Annual mean air temperature (a), precipitation (b), shortwave radiation (c), longwave radiation (d) and potential evapotranspiration (e) of the used meteorological input for years 1970–2100 (historical emission scenario + RCP 8.5 and RCP 2.6).

Close modal

Bias correction of meteorological input

A bias correction was performed for air temperature, precipitation, shortwave radiation and longwave radiation of the original EURO-CORDEX time series. The corrections were done by comparing available reference observations from the nearest FMI stations with the values by EURO-CORDEX from the historical period 1970–2005 and forming correction values that were applied to the whole length of the time series. The used stations were in the range of 44–195 km from Sievi. For longwave radiation, observations were not used, but the values corresponding to observations were determined using the air temperature observations and Equation (1):
formula
(1)
where R0 is a longwave radiation value used for the bias correction, T0 is a temperature observation as Kelvin, TM is a temperature value by EURO-CORDEX as Kelvin and RM is a longwave radiation value by EURO-CORDEX.
Equations (2) and (3) were used for correcting temperature and the other variables, respectively.
formula
(2)
formula
(3)
where ci is the corrected value of the time series to be corrected at timestep i, ci,0 is the original value of the series to be corrected at timestep i, is the average of the correcting time series over timesteps jA and is the average of the time series to be corrected over timesteps jA. The correction values are and . Seasonal correction values were used for winter (December–February), spring (March–May), summer (June–August) and autumn (September–November). For these seasons, the correction values for temperature were 1.15, 3.04, 4.43 and 1.80, respectively, and for precipitation 0.73, 0.65, 0.77 and 0.69.

FLUSH

FLUSH is a three-dimensional (3D) process-based hydrological model developed for simulating water flow in subsurface drained fields (Warsta et al. 2013). In the model, water flow is divided into a two-dimensional (2D) overland flow and a 3D subsurface flow. The overland flow is described with the diffuse wave simplification of the Saint-Venant equations, and the subsurface flow is represented with the Richards (1931) equation. The soil porosity is divided into soil matrix and macropores. This dual pore system allows simulation of slow flow in the soil matrix, fast preferential flow in the macropores and an exchange of water between the pore systems. In both pore systems, water retention properties are described with the van Genuchten (1980) model.

The water cycle in the model is initiated by precipitation falling on the soil surface or snowpack. Water from rainfall or snowmelt can infiltrate into both pore systems. When the infiltration capacity is exceeded or the groundwater table rises to the soil surface, overland flow is initiated. Water is lost from the field via evapotranspiration, open ditches (overland flow and groundwater seepage), groundwater outflow, overland flow and subsurface drains. Ditches and drains are represented as sinks. The water flow from the soil to the drains is defined with an equation based on Darcy's law:
formula
(4)
where q (L3 T−1) is the volumetric drain flux, K (LT−1) is the arithmetic mean of horizontal and vertical unsaturated hydraulic conductivities, A (L2) is the surface area of the cylindrically shaped drain in the cell, H (L) is the hydraulic head in the cell, HS (L) is the hydraulic head in the drain, Hcontrol (L) is the control term to describe the effect of drain outlet blocking and λ (L) is the entrance resistance term. If Hs+Hcontrol > H, q is set to zero. The pressure head in the drain is assumed to be equal to atmospheric pressure (0 m). A soil heat transport model (Warsta et al. 2012) and a process-based snow model (Koivusalo et al. 2001) have been integrated into FLUSH, which enables the simulation of snow accumulation and soil freezing.

Model application

The simulations were computed in a 2D grid oriented in the xz-plane, where the z-axis is orthogonal to the soil surface. In the xy-plane, the grid consisted of a row of 192 cells with dimensions of 2 m × 2 m. In the z-direction, the grid was divided into 32 layers with varying thicknesses the total depth being 3.4 m. The hydrological parameters of the soil at depths 0–1.15 m were based on samples taken from the Sievi field (Äijö et al. 2021). The bottom layer (1.15–3.4 m), including the lower boundary, was parameterized following Turunen et al. (2013). The parameterization was done in Salo et al. (2021) using groundwater table depth and drain discharge measurements taken when the field was under conventional subsurface drainage. The parameter values and calibration and validation results are listed in Supplementary material, Tables 1 and 2. The boundaries at the bottom and at the sides of the grid were impermeable. Groundwater outflow was possible only to a hypothetical open collector ditch with a depth of 1.5 m at the left end of the computation grid. The grid also had a hypothetical lengthwise runoff collector ditch with a depth of 0.1 m to generate topsoil layer runoff. Both ditches had fixed water levels at the elevation of the ditch bottom. In the x-direction, the total length of the grid was 384 m. Based on the slope of the Sievi field, the hydrological gradient was set to be towards the open collector ditch.

In this study, the time series describing the root depth was based on a constant cropping cycle, where the crop is sowed on June 1 and harvested 3 months later on August 31. The root depth function was defined following Turunen et al. (2015): before sowing the depth is 0.05 m. Starting from the sowing day, the depth increases linearly for 67 days after which a constant depth of 0.75 m is reached. After the harvest day, the root depth returns to 0.05 m. This enabled evapotranspiration to occur outside the main growing period. In the model, evapotranspiration is dependent on the rooting depth, soil moisture and potential evapotranspiration calculated according to Allen et al. (1998).

Subsurface drainpipes with a diameter of 5 cm were placed into the drain depth (0.98–1.18 m) with 15 m drain spacing following the actual drainage system in the Sievi field. The subsurface drainpipes were oriented orthogonally to the computation grid, forming local sinks along the grid transect. The simulated groundwater levels were recorded at the midpoints of every two adjacent drains. Controlled drainage was implemented in the model according to Equation (1). Each drain was controlled separately with a regulation depth of 0.6 m from the soil surface. Four different drainage schemes were simulated to study the potential hydrological impacts of controlled drainage: conventional subsurface drainage without regulation (ND), two scenarios with regulation alternating on and off (CD1 and CD2) and constant regulation (CC). Both CD1 and CD2 have a summer regulation period from June 1 (sowing) to August 24 (one week before harvest). In addition, CD1 has a long winter regulation period from October 1 to March 31, and CD2 has a shorter autumn regulation period from October 1 to November 15. The regulation schedules of CD1 and CD2 are visualized in Figure 2(h). In high-latitude areas such as Finland, both CD1 and CD2 can be used depending on the risk of damage caused by water freezing in drainage pipes and control wells in winter.

Figure 2

Daily mean groundwater table depths with different drainage schemes and emission scenarios for time intervals S0 (a), S1 (b–c), S2 (d–e) and S3 (f–g). The regulation schedules of CD1 (blue line) and CD2 (green line), i.e. the variation between free drainage depth and regulation depth are described in (h). The light green area shows the growing season. Please refer to the online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/nh.2021.058.

Figure 2

Daily mean groundwater table depths with different drainage schemes and emission scenarios for time intervals S0 (a), S1 (b–c), S2 (d–e) and S3 (f–g). The regulation schedules of CD1 (blue line) and CD2 (green line), i.e. the variation between free drainage depth and regulation depth are described in (h). The light green area shows the growing season. Please refer to the online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/nh.2021.058.

Close modal

Processing output data

FLUSH produced hourly values for the output variables. The results were examined by dividing the years 1970–2100 into four consecutive time intervals: 1970–2005 (S0), 2006–2037 (S1), 2038–2069 (S2) and 2070–2100 (S3). S0 corresponds to the historical period of the EURO-CORDEX time series whereas S1–S3 represent the future periods of emission scenarios RCP 8.5 and RCP 2.6. These time intervals were used to aggregate data to describe the climatic and hydrologic conditions of each time interval. The statistical significance between time intervals was analyzed with ANOVA and Tukey's honest significance test with α = 0.05.

Effect of controlled drainage on groundwater levels

The changes in groundwater levels due to projected climate change were more distinct in the high-emission scenario (RCP 8.5) than in the low-emission scenario (RCP 2.6). In RCP 8.5, the groundwater level peaks in March–April gradually flattened, and groundwater levels rose at other times throughout most of the year (Figure 2(b), 2(d) and 2(f)). Similar development took place in RCP 2.6, but the changes were much more subtle (Figure 2(c), 2(e) and 2(g)). In RCP 2.6, a clear rise in the springtime groundwater levels was still visible at the end of the 21st century.

In the simulations, controlled drainage caused elevated groundwater levels when compared to ND, both during the regulation periods and outside of them (Figure 2). The reverting of the mean groundwater levels after the regulation periods was slow, as they never reached the levels of ND. At the beginning of the summer regulation period (sowing time), CD1 and CD2 had approximately 20 and 5 cm higher groundwater levels, respectively, than ND. Compared to CD2, the higher summertime groundwater levels in CD1 remained until July–August. At the start of regulation periods, the mean groundwater levels in CD1 and CD2 immediately began to recede from the levels of ND and quickly began to rise upwards.

The mean effects of controlled drainage on groundwater levels during the regulation periods are shown in Figure 3. The effects were larger in winter and autumn than in summer and larger in CD1 than in CD2. The controlled drainage effects were larger in the future intervals (S1–S3) than in the historical interval (S0) in both emission scenarios, but there were both increases and decreases between S1, S2 and S3. The effects were on average 1–4 cm (RCP 8.5) and 1–2 cm (RCP 2.6) larger during the intervals S1–S3 when compared to the interval S0, depending on the season. Except for the summer period in CD2 during S2, differences between the emission scenarios were small.

Figure 3

The mean effects of controlled drainage on the groundwater table during the regulation periods. The vertical axis describes how much higher the groundwater table was when compared to normal drainage. All differences between time intervals were statistically significant, except between intervals S0 and S2 in CD2 autumn RCP 2.6.

Figure 3

The mean effects of controlled drainage on the groundwater table during the regulation periods. The vertical axis describes how much higher the groundwater table was when compared to normal drainage. All differences between time intervals were statistically significant, except between intervals S0 and S2 in CD2 autumn RCP 2.6.

Close modal

In both emission scenarios, CD1 and CD2 notably increased the frequency of groundwater tables at the depth of 0.3–0.6 m and reduced the occurrence of deep groundwater tables during summer regulation periods in every time interval S0–S3 (Figure 4). CD1 (Figure 4(a) and 4(b)), in which drainage regulation was on over winter, decreased the occurrence of deep (>1 m) groundwater tables 69–71%. This is considerably more than the 44–47% reduction in CD2, in which drainage was regulated for a shorter period in autumn.

Figure 4

Distributions of groundwater table depths during summer regulation periods for different drainage scenarios (ND, CD1 and CD2) and time intervals (S0–S3). Emission scenarios of RCP 8.5 are in (a) and (c), and RCP 2.6 in (b) and (d). The vertical axis describes relative frequency.

Figure 4

Distributions of groundwater table depths during summer regulation periods for different drainage scenarios (ND, CD1 and CD2) and time intervals (S0–S3). Emission scenarios of RCP 8.5 are in (a) and (c), and RCP 2.6 in (b) and (d). The vertical axis describes relative frequency.

Close modal

Effect of controlled drainage on water balance

Mean annual water balance output components (evapotranspiration, drain discharge and groundwater outflow) for all drainage schemes and mean annual precipitations are shown in Figure 5. All these components had an increasing trend towards the future, but the effects of controlled drainage remained similar to the historical time interval (S0). A continuous rise in evapotranspiration can be seen between the time intervals S0–S3, which reflects the rising air temperature in both emission scenarios. The increase in drain discharge and groundwater outflow was continuous in RCP 2.6, but in RCP 8.5, the highest values were reached during the time interval S2 after which there was a small drop in S3. As the precipitation and air temperature were higher in RCP 8.5 than RCP in 2.6, the water balance components were also greater. Between the intervals S0 and S3, the mean of the annual maximum snow water equivalent changed from 67 mm to 23 and 52 mm in RCP 8.5 and RCP 2.6, respectively (data not presented). The average annual topsoil layer runoff, which includes overland flow, was less than 1 mm/a in each time interval S0–S3 and could be neglected.

Figure 5

Mean annual evapotranspiration, drain discharge and groundwater outflow with different drainage schemes for RCP 8.5 (a) and RCP 2.6 (b). Mean annual precipitations are in (c). The horizontal dashed line in (a) and (b) shows the drain discharge level of ND in the historical interval S0. The average topsoil layer runoff during the time intervals S0–S3 was always less than 1 mm/a.

Figure 5

Mean annual evapotranspiration, drain discharge and groundwater outflow with different drainage schemes for RCP 8.5 (a) and RCP 2.6 (b). Mean annual precipitations are in (c). The horizontal dashed line in (a) and (b) shows the drain discharge level of ND in the historical interval S0. The average topsoil layer runoff during the time intervals S0–S3 was always less than 1 mm/a.

Close modal

Figure 5 shows that the drainage schemes with more time under the reduced drainage efficiency resulted in larger reductions in the drain discharge and larger increases in the groundwater outflow. In RCP 8.5, the mean annual drain discharge reductions caused by CD1 and CD2 at different time intervals were 41.3–46.2 mm and 22.4–26.1 mm, respectively. In RCP 2.6, these values were 41.3–43.2 mm and 21.0–24.8 mm, respectively. The differences between time intervals were not statistically significant. Approximately 80 and 70% of the drain discharge reductions turned into groundwater outflow in CD1 and CD2, respectively, the rest of the reductions turning mostly into evapotranspiration. In RCP 8.5, the drain discharges of CD1 during intervals S2 and S3 exceed that of ND during interval S0. In RCP 2.6, CD1 causes lower drain discharges in all intervals than ND in interval S0.

The differences in evapotranspiration between drainage schemes were small when compared to the other components, but the effect of controlled drainage was visible. The highest evapotranspiration resulted from CD1. The second highest value was from CC in the historical interval S0 and from CD2 in the future intervals S1–S3.

Effect of an open collector ditch on the groundwater table profile

The open ditch had a clear effect on groundwater levels in its close surroundings (Figure 6(a)). In a range of 115 m, the ditch caused groundwater tables below the drainage depth (∼1 m) to be much more common while decreasing the frequency of groundwater tables near the soil surface. This caused the effect of controlled drainage on groundwater levels to be clearly weaker as seen in Figure 6(b) and 6(c).

Figure 6

Distribution of simulated groundwater tables in the soil profile (a) and mean groundwater table depths near (b) and far (c) from the collector ditch during the last future interval S3 in RCP 8.5. The distribution is based on an 80 × 111 grid. Each row depicts a 5 cm layer and each column a horizontal location every 3 m. To the columns located at the observation tubes (blue marks), the frequency of the groundwater table being at each layer was computed. To the columns between the observation tubes, values were computed with second-degree polynomial interpolation. Please refer to the online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/nh.2021.058.

Figure 6

Distribution of simulated groundwater tables in the soil profile (a) and mean groundwater table depths near (b) and far (c) from the collector ditch during the last future interval S3 in RCP 8.5. The distribution is based on an 80 × 111 grid. Each row depicts a 5 cm layer and each column a horizontal location every 3 m. To the columns located at the observation tubes (blue marks), the frequency of the groundwater table being at each layer was computed. To the columns between the observation tubes, values were computed with second-degree polynomial interpolation. Please refer to the online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/nh.2021.058.

Close modal

The model results showed two ways in which controlled drainage can theoretically reduce early summer drought, which is one of the main limitations for crop yield in Finland (Peltonen-Sainio et al. 2016). First, the groundwater levels in control schemes CD1 and CD2 were higher compared to ND at the very beginning of the growing season due to the slow reversion of groundwater levels after regulation periods preceding the growing season. Second, the summer regulation periods in CD1 and CD2 caused a quick rising of groundwater levels compared to the levels in ND. The difference between CD1 and CD2 during summer highlights the role of storing water in the soil over winter in preventing summer droughts. The control drainage impacts extending from winter and spring to early summer is a finding not addressed in earlier experimental studies. While the magnitude and extent of the impact are likely affected by various site-specific factors, recognizing the opportunities of winter regulation is potentially beneficial regarding adaptation to future climate conditions.

The simulation results with climate model data showed similar control capacity on groundwater levels that has been seen in field experiments in Finland. During the regulation periods in the historical interval S0, controlled drainage caused 17–36 cm higher groundwater levels compared to ND, which resembles the observations of 31 cm by Joukainen & Yli-Halla (2003) and 10–20 cm by Österholm et al. (2015). Model simulations revealed that the effects on groundwater levels were largest during autumn and winter regulation periods.

Based on the simulation results, winters in both emission scenarios became milder with less snow accumulation, which most likely caused higher drain discharge potential. In Finland, most of the nutrient load takes place during autumn and spring, but the role of winter regulation will likely become more important in the future since winter temperatures are projected to increase in Northern Europe (Puustinen et al. 2007; Olesen et al. 2011). Regulating the groundwater level at these times can have a positive impact on reducing the environmental loads if drain discharge can be retained at the field and the discharge via groundwater outflow can be increased and discharge via surface runoff can be avoided. The simulated water balance components showed that annual drain discharge can be reduced with controlled drainage and most of that reduction turns into groundwater outflow. The differences between CD1 and CD2 also indicate that winter drain discharges can be clearly reduced with regulation: CD1, in which the regulation period starting in October 1 continues over winter, resulted on average 19.4 mm smaller annual drain discharge than CD2, which has no winter regulation, in both emission scenarios. Topsoil layer runoff, which includes overland flow, was found to be negligible for the field with sandy soil layers (on average less than 1 mm/a), which can be attributed to the efficient drainage and relatively high conductivity of the soil. Also in the experiments of Frey et al. (2016) and Wesström & Messing (2007), the overland flow was not observed to take place.

Instead of using a root distribution that depends on meteorological conditions such as temperature, the annual root depth pattern was kept constant to reduce the number of variables explaining the changes in hydrology between the time intervals. As root growth varies between crops and no data of actual root depths from the Sievi field were available, the root distribution was defined according to a previous study (Turunen et al. 2015). Different root depth patterns might result in larger differences in evapotranspiration between the control schemes, which were relatively small in this study. A possible increase in biomass production enabled by warmer climate was not considered. Larger crop growth could lead to higher evapotranspiration during the growing season, which would lead to lower groundwater levels.

When comparing the effect of controlled drainage on groundwater levels during the regulation periods in different time intervals, the effects were larger in the future intervals S1–S3 than in the historical interval S0, but the difference was not large and there were both increases and decreases between S1, S2 and S3. The results in this study contrast with Sojka et al. (2020), who found the controlled drainage effects on groundwater levels to be weaker in the future, due to the differences in projected future hydrology. In their simulations, the groundwater table spent less time above the drainage depth in the future, while the mean groundwater levels mostly increased in this study. The performance of controlled drainage is strongly dependent on hydro-meteorological conditions.

The absolute reductions in the mean annual drain discharge did not change significantly between the time intervals, but the relative reduction decreased due to a clear increase in the amount of drain discharge. The drainage scheme CD1, in which drainage efficiency was reduced longer than in CD2, was able to keep the drain discharge levels lower than the historical ND levels during the future interval S1 in RCP 8.5 and during all future intervals in RCP 2.6. This was due to the larger increase in drain discharge in RCP 8.5. Sojka et al. (2020) projected the controlled drainage effect on drain discharge during the growing season to decrease in the future, while the results by Pease et al. (2017) align more with this study, as they projected the effect on the annual drain discharge to remain similar during the 21st century, even though the annual drain discharge decreased in their simulations.

The modeled open collector ditch was used for simulating the groundwater outflow from the field, and the water level at the ditch was assumed to be at the bottom of the 1.5-m-deep ditch. The ditch had only a drainage effect and was not used for maintaining the groundwater level higher in the field area. Rozemeijer et al. (2016) noticed that the water level at an open ditch next to the field was affecting the efficiency of controlled drainage in maintaining the groundwater levels closer to the control level. The same effect took place in this study, as the effect of controlled drainage on the groundwater levels was clearly diminished near the open ditch. The results are calling for investigation on how water level management in collector ditches can increase controlled drainage efficiency in the fields.

A major source of uncertainty in this study was the meteorological input. Projections (RCP 8.5 and RCP 2.6) from a single climate model were used instead of projection ensembles. Ensembles can be used in attempts to tackle uncertainties related to initial values, parameters and structure of climate models, and unweighted multi-model means have been used as ‘best guess projections’ by, e.g., IPCC (2001), but serious challenges exist in assembling and interpreting projection ensembles and the extent to which ensembles can reduce uncertainties is unclear (Tebaldi & Knutti 2007). Another uncertainty arises from the controlled drainage description and the fact that the control impact was not tested against observations. In this study, each drainpipe was separately controlled. In actual controlled drainage, fewer control structures are typically used, and field slope diminishes their effect, whereas the description in this study can be seen as producing the maximum effect for the regulation depth that was used (0.6 m). While testing the model performance against observations would increase the model reliability, a benchmarked process-based model combined with in-situ field parameterization offers a transparent method for describing field hydrology under different drainage settings and climate conditions. It should also be noted that a calibrated model setup using historical climate forcing and field data typically contains uncertainties when applied to changed conditions (e.g. Seifert et al. 2012). This study demonstrates the potential hydrological impacts of controlled drainage in a northern field site under varying meteorological forcing using two out of many possible future climate scenarios, and the results should not be interpreted as a strict future forecast.

The hydrological model FLUSH enabled spatially distributed simulations including sub-daily dynamics. The selected visualizations aggregated large quantities of data to depict both short-term (sub-annual) and long-term (between multi-decade intervals) variations. Most of the visualizations were based on average values, which are useful especially in describing changes between time intervals but can leave out information about distributions and extreme values.

The simulated controlled drainage scenarios caused higher groundwater levels compared to conventional subsurface drainage and were useful in reducing early summer droughts, which are a major limiting factor in Finnish crop production. Comparison of different drainage schemes showed that the probability of harmful droughts in early summer could be further reduced by reducing drainage efficiency over winter (CD1), as the winter regulation period resulted in higher groundwater levels during the growing season compared to a shorter reduction in drainage efficiency during autumn (CD2). Given the simulated future climate scenarios, the effect of controlled drainage on groundwater levels will be on average 1–4 cm larger in the future intervals S1–S3 compared to the historical interval S0.

The simulated controlled drainage scenarios changed the annual water balance by reducing drain discharge and increasing groundwater outflow. During the whole simulation period, CD1 and CD2 reduced annual drain discharge 41.3 to 46.2 mm and 22.4 to 26.1 mm, respectively, in RCP 8.5, and 41.3 to 43.2 mm and 21.0 to 24.8 mm, respectively, in RCP 2.6. The differences between intervals S0–S3 were not statistically significant. Thus, given the simulated future climate scenarios, controlled drainage will remain a relevant method for decreasing drain discharge. Controlled drainage will have a strong impact in reducing wintertime drain discharge in future climate conditions with milder winters.

The open collector ditch caused clearly lower groundwater levels in the vicinity of the ditch. This decreased the effect of controlled drainage on groundwater levels, as the groundwater table spent less time above the drainage depth, which is the functional soil domain of controlled drainage. Close to the open ditch, the effect of controlled drainage on groundwater levels was undertaken by groundwater outflow to the open ditch.

This study was done in VesiHave, a research project led by Field Drainage Research Association. Funding was received from the Environmental Ministry of Finland, Drainage Foundation sr, Maa-ja vesitekniikan tuki ry and Sven Hallin research foundation sr. CSC – IT Center for Science Ltd provided computation resources for running the simulations. We acknowledge the World Climate Research Programme's Working Group on Regional Climate, and the Working Group on Coupled Modeling, former coordinating body of CORDEX and responsible panel for CMIP5. We also thank the climate modeling groups (listed in Table 1 of this paper) for producing and making available their model output. We also acknowledge the Earth System Grid Federation infrastructure an international effort led by the U.S. Department of Energy's Program for Climate Model Diagnosis and Intercomparison, the European Network for Earth System Modeling and other partners in the Global Organisation for Earth System Science Portals (GO-ESSP). We acknowledge Dr Mika Tähtikarhu for his advice on the manuscript.

Data cannot be made publicly available; readers should contact the corresponding author for details.

Abdelbaki
A.
2015
DRAINMOD simulated impact of future climate change on agriculture drainage systems
.
Asian Transactions on Engineering
5
,
13
18
.
Äijö
H.
,
Myllys
M.
,
Sikkilä
M.
,
Salo
H.
,
Nurminen
J.
,
Häggblom
O.
,
Turunen
M.
,
Paasonen-Kivekäs
M.
,
Warsta
L.
,
Koivusalo
H.
,
Alakukku
L.
&
Puustinen
M.
2017
Toimivat salaojitusmenetelmät kasvintuotannossa (TOSKA) – Loppuraportti 2017. [Functioning Subdrainage Methods in Crop Production (TOSKA) – Final Report 2017]. Salaojituksen tutkimusyhdistys ry:n tiedote 32. Finnish Drainage Association, Helsinki, Finland, p. 109
.
Äijö
H.
,
Myllys
M.
,
Sikkilä
M.
,
Salo
H.
,
Salla
A.
,
Nurminen
J.
,
Paasonen-Kivekäs
M.
&
Koivusalo
H.
2021
Vesitalouden hallinta vesiensuojelussa (VesiHave) – Loppuraportti 2021. [Water Management in Water Protection (VesiHave) – Final Report 2021]. Manuscript
.
Allen
R. G.
,
Pereira
L. S.
,
Raes
D.
&
Smith
M.
1998
Crop Evapotranspiration—Guidelines for Computing Crop Water Requirements
.
FAO Irrigation and Drainage Paper 56. Food and Agriculture Organization
,
Rome
.
Åström
M.
,
Österholm
P.
,
Bärlund
I.
&
Tattari
S.
2007
Hydrochemical effects of surface liming, controlled drainage and lime-filter drainage on boreal acid sulfate soils
.
Water, Air, and Soil Pollution
179
,
107
116
.
Benestad
R.
,
Haensler
A.
,
Hennemuth
B.
,
Illy
T.
,
Jacob
D.
,
Keup-Thiel
E.
,
Kotlarski
S.
,
Nikulin
G.
,
Otto
J.
,
Rechid
D.
,
Sieck
K.
,
Sobolowski
S.
,
Szabó
P.
,
Szépszó
G.
,
Teichmann
C.
,
Vautard
R.
,
Weber
T.
&
Zsebeházi
G.
2017
Guidance for EURO-CORDEX Climate Projections Data Use
.
Dai
Z.
,
Amatya
D. M.
,
Sun
G.
,
Trettin
C.
,
Li
C.
&
Li
H.
2010
A comparison of MIKE SHE and DRAINMOD for modeling forested wetland hydrology in Coastal South Carolina, USA. In: ASABE - 9th International Drainage Symposium 2010, Held Jointly with CIGR and CSBE/SCGAB
.
Frey
S.
,
Hwang
H.
,
Park
Y.
,
Hussain
S.
,
Gottschall
N.
,
Edwards
M.
&
Lapen
D.
2016
Dual permeability modeling of tile drain management influences on hydrologic and nutrient transport characteristics in macroporous soil
.
Journal of Hydrology
535
,
392
406
.
Huttunen
I.
,
Lehtonen
H.
,
Huttunen
M.
,
Piirainen
V.
,
Korppoo
M.
,
Veijalainen
N.
,
Viitasalo
M.
&
Vehviläinen
B.
2015
Effects of climate change and agricultural adaptation on nutrient loading from Finnish catchments to the Baltic Sea
.
Science of the Total Environment
529
,
168
181
.
IPCC
2001
Climate Change 2001: the Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change
(
Houghton
J. T.
,
Ding
Y.
,
Griggs
D. J.
,
Noguer
M.
,
van der Linden
P. J.
,
Xiaosu
D.
,
Dai
X.
,
Maskell
K.
&
Johnson
C. A.
, eds.).
Cambridge University Press
,
Cambridge
, pp.
881
.
Jacob
D.
,
Bärring
L.
,
Christensen
O. B.
,
Christensen
H. J.
,
de Castro
M.
,
Déqué
M.
,
Giorgi
F.
,
Hagemann
S.
,
Hirschi
M.
,
Jones
R.
,
Kjellström
E.
,
Lenderink
G.
,
Rockel
B.
,
Sánchez
E.
,
Schär
C.
,
Seneviratne
S. I.
,
Somot
S.
,
van Ulden
A.
&
van den Hurk
B.
2007
An inter-comparison of regional climate models for Europe: model performance in present day climate
.
Climate Change
81
,
31
52
.
Kjellström
E.
&
Lind
P.
2009
Changes in the water budget in the Baltic Sea drainage basin in the future warmer climates as simulated by the regional climate model RCA3
.
Boreal Environment Resarch
14
,
114
124
.
Koivusalo
H.
,
Heikinheimo
M.
&
Karvonen
T.
2001
Test of a simple two-layer parameterisation to simulate the energy balance and temperature of a snowpack
.
Theoretical and Applied Climatology
70
,
65
79
.
Lehtonen
I.
2011
Äärisademäärien muutokset Euroopassa maailmanlaajuisten ilmastomallien perusteella. [Changes in Exreme Precipitation in Europe Based on Global Climate Models]
.
Master's Thesis
,
University of Helsinki, Department of Physics
, p.
70
.
Lind
P.
&
Kjellström
E.
2009
Water budget in the Baltic Sea drainage basin: evaluation of simulated fluxes in a regional climate model
.
Boreal Environment Resarch
14
(
1
),
56
67
.
Muma
M.
,
Rousseau
A. N.
&
Gumiere
S. J.
2016
Modeling of subsurface agricultural drainage using two hydrological models with different conceptual approaches as well as dimensions and spatial scales
.
Canadian Water Resources Journal/Revue Canadienne des Ressources Hydriques
42
(
1
),
38
53
.
Myllys
M.
2019
Säätösalaojituksesta hyötyä turvemailla/Turvemaat hyötyvät säätösalaojituksesta
.
Salaojayhdistys ry:n jäsenjulkaisu [Peatlands Benefit From Controlled Drainage]
1/2019
,
14
18
.
Nousiainen
R.
,
Warsta
L.
,
Turunen
M.
,
Huitu
H.
,
Koivusalo
H.
&
Pesonen
L.
2015
Analyzing subsurface drain network performance in an agricultural monitoring site with a three-dimensional hydrological model
.
Journal of Hydrology
529
,
82
93
.
Olesen
J.
,
Trnka
M.
,
Kersebaum
K.
,
Skjelvåg
A.
,
Seguin
B.
,
Peltonen-Sainio
P.
,
Rossi
F.
,
Jerzy
K.
&
Micale
F.
2011
Impacts and adaptation of European crop production systems to climate change
.
European Journal of Agronomy.
34
,
96
112
.
Österholm
P.
,
Virtanen
S.
,
Rosendahl
R.
,
Uusi-Kämppä
J.
,
Ylivainio
K.
,
Yli-Halla
M.
,
Mäensivu
M.
&
Turtola
E.
2015
Groundwater management of acid sulfate soils using controlled drainage, by-pass flow prevention, and subsurface irrigation on a boreal farmland
.
Acta Agriculturae Scandinavica, Section B – Soil & Plant Science
65
(
1
),
110
120
.
Pease
L.
,
Fausey
N. R.
,
Martin
J. F.
&
Brown
L.
2017
Projected climate change effects on subsurface drainage and the performance of controlled drainage in the Western Lake Erie Basin
.
Journal of Soil and Water Conservation
72
,
240
250
.
Peltonen-Sainio
P.
2012
Crop Production in a Northern Climate
.
FAO/OECD Workshop: Building Resilience for Adaptation to Climate Change in the Agriculture Sector
,
Rome
, pp.
183
216
.
Peltonen-Sainio
P.
,
Venäläinen
A.
,
Mäkelä
H.
,
Pirinen
P.
,
Laapas
M.
,
Jauhiainen
L.
,
Kaseva
J.
,
Ojanen
H.
,
Korhonen
P.
,
Huusela-Veistola
E.
,
Jalli
M.
,
Hakala
K.
,
Kaukoranta
T.
&
Virkajärvi
P.
2016
Harmfulness of weather events and the adaptive capacity of farmers at high latitudes of Europe
.
Climate Research
67
(
3
),
221
240
.
Puustinen
M.
,
Koskiaho
J.
,
Jormola
J.
,
Järvenpää
L.
,
Karhunen
A.
,
Mikkola-Roos
M.
,
Pitkänen
J.
,
Riihimäki
J.
,
Svensberg
M.
&
Vikberg
P.
2007
Maatalouden monivaikutteisten kosteikkojen suunnittelu ja mitoitus. [The Design and Dimensioning of Multifunctioning Wetlands in Agriculture]. Suomen ympäristö 21/2007, The Finnish Environment Institute, p. 77
.
Räisänen
J.
&
Ruokolainen
L.
2006
Probabilistic forecasts of near-term climate change based on a resampling ensemble technique
.
Tellus A: Dynamic Meteorology and Oceanography
58
(
4
),
461
472
.
Rozemeijer
J.
,
Visser
A.
,
Borren
W.
,
Winegram
M.
,
Van der Velde
Y.
,
Klein
J.
&
Broers
H.
2016
High-frequency monitoring of water fluxes and nutrient loads to assess the effects of controlled drainage on water storage and nutrient transport
.
Hydrology and Earth System Sciences
20
,
347
358
.
Ruosteenoja
K.
,
Jylhä
K.
&
Kämäräinen
M.
2016
Climate projections for Finland under the RCP forcing scenarios
.
Geophysica
51
(
1
),
15
50
.
Salo
H.
,
Salla
A.
&
Koivusalo
H.
2021
The seasonal effects of controlled drainage on field water balance and groundwater levels
.
Hydrology Research
52
(
6
),
1633
1647
.
Seifert
D.
,
Sonnenborg
T. O.
,
Refsgaard
J. C.
,
Højberg
A. L.
&
Troldborg
L.
2012
Assessment of hydrological model predictive ability given multiple conceptual models
.
Water Resources Research
48
,
WR011149
.
Sojka
M.
,
Kozłowski
M.
,
Kęsicka
B.
,
Wróżyński
R.
,
Stasik
R.
,
Napierała
M.
,
Jaskuła
J.
&
Liberacki
D.
2020
The effect of climate change on controlled drainage effectiveness in the context of groundwater dynamics, surface, and drainage outflows. Central-western Poland case study
.
Agronomy
10
(
5
),
625
645
.
Tebaldi
C.
&
Knutti
R.
2007
The use of the multi-model ensemble in probabilistic climate projections
.
Philosophical Transactions: Mathematical, Physical, and Engineering Sciences
365
(
1857
),
2053
2075
.
Turunen
M.
,
Warsta
L.
,
Paasonen-Kivekäs
M.
,
Nurminen
J.
,
Myllys
M.
,
Alakukku
L.
,
Äijö
H.
,
Puustinen
M.
&
Koivusalo
H.
2013
Modeling water balance and effects of different subsurface drainage methods on water outflow components in a clayey agricultural field in boreal conditions
.
Agricultural Water Management
121
,
135
148
.
Turunen
M.
,
Warsta
L.
,
Paasonen-Kivekas
M.
,
Nurminen
J.
,
Alakukku
L.
,
Myllys
M.
&
Koivusalo
H.
2015
Effects of terrain slope on long-term and seasonal water balances in clayey, subsurface drained agricultural fields in high latitude conditions
.
Agricultural Water Management
150
,
139
151
.
van Genuchten
M. T.
1980
A closed form equation for predicting the hydraulic conductivity of unsaturated soils
.
Soil Science Society of America Journal
44
,
892
898
.
Vuorenmaa
J.
,
Rekolainen
S.
,
Lepistö
A.
,
Kenttämies
K.
&
Kauppila
P.
2002
Losses of nitrogen and phosphorus from agricultural and forest areas in Finland during the 1980 and 1990s
.
Environmental Monitoring and Assessment
76
,
213
248
.
Warsta
L.
,
Turunen
M.
,
Koivusalo
H.
,
Paasonen-Kivekäs
M.
,
Karvonen
T.
&
Taskinen
A.
2012
Modelling heat transport and freezing and thawing processes in a clayey, subsurface drained agricultural field
. In:
11th ICID International Drainage Workshop on Agricultural Drainage Needs and Future Priorities
, September
23–27, 2012
Cairo, Egypt
.
Proceedings
, p.
10
.
Warsta
L.
,
Karvonen
T.
,
Koivusalo
H.
,
Paasonen-Kivekäs
M.
&
Taskinen
A.
2013
Simulation of water balance in a clayey, subsurface drained agricultural field with three-dimensional FLUSH model
.
Journal of Hydrology
476
,
395
409
.
Ylhäisi
J.
,
Tietäväinen
H.
,
Peltonen-Sainio
P.
,
Venäläinen
A.
,
Eklund
J.
,
Räisänen
J.
&
Jylhä
K.
2010
Growing season precipitation in Finland under recent and projected climate
.
Natural Hazards and Earth System Sciences
10
,
1563
1574
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).

Supplementary data