PREDICTING THE EFFECTS OF WEIR MANAGEMENT ON DRAINAGE DISCHARGE OF A CONTROLLED DRAINAGE SYSTEM IN A CHANGING CLIMATE By Md Sami Bin Shokrana A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Biosystems Engineering – Doctor of Philosophy 2022 ABSTRACT The widespread adoption of subsurface drainage in the Midwest United States coupled with fertile land and abundant rainfall has made this region the largest producer of corn and soybean in the nation. Although subsurface drainage helps reduce the waterlogging stress on crops by removing nutrient-enriched water from the field, it could contribute to harmful algal blooms in freshwater ecosystems. Controlled drainage (CD) practices can reduce the drainage volume leaving the field and have a positive effect on nutrient load reduction. Although the efficiency of CD under the present climate has been widely studied, it is essential to evaluate its performance in the future to build resilient agricultural systems. In this study, the efficiency of two CD practices was evaluated in reducing drainage discharge for the future based on the height and timing of the weir management. We used the Root Zone Water Quality Model (RZWQM2) to predict the CD management effects on drainage discharge. To obtain reliable simulation results from the RZWQM2 model, it is important to use measured soil-water characteristic parameters and flow data. We used HYPROP to measure the parameters of the soil water characteristic curve which served as input to the model. Additionally, we developed a stage-discharge equation for an AgriDrain metal-edge sharp-crest 45° V-notch weir to accurately estimate the drainage discharge from the field. A reliable estimate of the drainage discharge was necessary to accurately estimate the nutrient loss. The recently developed P module of the RZWQM2, known as RZWQM2-P was used in this study to predict drainage discharge and P loss from a subsurface-drained field with clay loam soil. We used the Nash-Sutcliffe model Efficiency (NSE) and percentage bias (PBIAS) statistics to evaluate model performance. While the model showed “good” and “satisfactory” performance in predicting drainage discharge and total phosphorus (TP) load, respectively, it performed unsatisfactorily in predicting the dissolved reactive phosphorus (DRP) load for both calibration and validation periods. The underperformance of the model in simulating DRP load may be due to the inability of the model to partition fertilizer P into different P pools. We predicted the efficiency of two CD management scenarios (i.e., common and aggressive management) in reducing drainage discharge for future climate using the calibrated RZWQM2 model. The CD management scenarios were performed by maintaining the weir height at a higher (i.e., 15 cm for non-growing season and 40 cm for growing season) or lower (i.e., 30 cm for non-growing season or 50 cm for growing season) level inside a control structure and by altering the timing of the CD management based on the planting and harvesting dates. While both common and aggressive management was efficient in reducing drainage discharge for both historic and future periods, the percent reduction of drainage discharge with aggressive management was about 11% higher than the common management. The projected increase in precipitation and temperature in the future would cause increased drainage discharge during fall and winter. The aggressive management will be able to completely restrict the flow during the non-growing season which would reduce the nutrient loss of the surface water bodies. In the future, farmers should plant early to benefit from the projected increase in spring rainfall and avoid dry summers. In conclusion, it is evident from the results of this study that both common and aggressive CD management will continue to be effective in reducing drainage discharge in a changing climate. Copyright by MD SAMI BIN SHOKRANA 2022 To my parents, who sacrificed everything for my education. I would not have come this far without your love and support. v TABLE OF CONTENTS LIST OF TABLES .............................................................................................................................. viii LIST OF FIGURES .............................................................................................................................. ix LIST OF ABBREVIATIONS ............................................................................................................... xiii CHAPTER 1: LITERATURE REVIEW ....................................................................................................1 REFERENCES .........................................................................................................................6 CHAPTER 2: MEASUREMENT OF SOIL-WATER CHARACTERISTICS INPUT DATA .......................... 10 2.1 Abstract................................................................................................................. 10 2.2 Method Description .............................................................................................. 10 2.3 Soil Sample Collection .......................................................................................... 12 2.4 Preparing HYPROP2 .............................................................................................. 16 2.5 Starting the Measurements .................................................................................. 26 2.6 Stopping the Measurements ................................................................................ 27 2.7 Determining the Dry Weight ................................................................................ 29 2.8 Method Validation ................................................................................................ 30 2.9 Summary ............................................................................................................... 37 REFERENCES ...................................................................................................................... 38 CHAPTER 3: DEVELOPING A STAGE-DISCHARGE EQUATION FOR A V-NOTCH WEIR TO ESTIMATE THE DRAINAGE DISCHARGE ......................................................................................... 41 3.1 Abstract................................................................................................................. 41 3.2 Introduction .......................................................................................................... 41 3.3 Materials and Methods ........................................................................................ 44 3.4 Results and Discussion .......................................................................................... 52 3.5 Conclusions ........................................................................................................... 63 REFERENCES ...................................................................................................................... 64 CHAPTER 4: CALIBRATING THE RZWQM2-P MODEL TO SIMULATE DRAINAGE DISCHARGE AND PHOSPHORUS LOSS IN CLAY LOAM SOIL IN MICHIGAN ....................................................... 66 4.1 Abstract................................................................................................................. 66 4.2 Introduction .......................................................................................................... 67 4.3 Materials and Methods ........................................................................................ 70 4.4 Results................................................................................................................... 85 4.5 Discussions............................................................................................................ 89 4.6 Conclusions ........................................................................................................... 93 REFERENCES ...................................................................................................................... 94 CHAPTER 5: PREDICTING THE EFFECTS OF TWO CONTROLLED DRAINAGE SCENARIOS ON DRAINAGE DISCHARGE FOR FUTURE CLIMATE........................................................................... 100 vi 5.1 Abstract............................................................................................................... 100 5.2 Introduction ........................................................................................................ 101 5.3 Materials and Methods ...................................................................................... 104 5.4 Results and Discussions ...................................................................................... 112 5.5 Conclusions ......................................................................................................... 127 REFERENCES .................................................................................................................... 128 APPENDIX A: CHAPTER 3 ............................................................................................................. 134 APPENDIX B: CHAPTER 4 ............................................................................................................. 141 vii LIST OF TABLES Table 2.1. Saturation time for different types of soils .................................................................. 20 Table 4.1. Annual management practices adopted at the Blissfield site [From Shokrana et al. (2022). Used with permission.] ..................................................................................................... 74 Table 4.2. Input soil hydrologic properties data for Ziegenfuss clay loam soil [From Shokrana et al. (2022). Used with permission.] ............................................................................................ 76 Table 4.3. Input data for P pools to initiate the P model [From Shokrana et al. (2022). Used with permission.] .......................................................................................................................... 78 Table 4.4. Calibrated values of microporosity parameters [From Shokrana et al. (2022). Used with permission.] .......................................................................................................................... 81 Table 4.5. Calibrated soil hydraulic parameters and their values [From Shokrana et al. (2022). Used with permission.] ................................................................................................................. 82 Table 4.6. Other calibrated hydrologic, P loss, soil erosion, and plant parameters and their values. Default values are recorded from Sadhukhan et al. (2019a) [From Shokrana et al. (2022). Used with permission.] ..................................................................................................... 83 Table 4.7. Model performance statistics for drainage discharge, DRP load, and TP load during the calibration and validation period [From Shokrana et al. (2022). Used with permission.] .................................................................................................................................. 89 Table 5.1. Summary of CMIP projections obtained from GCMs used in NASA NEX-GDDP database for RCP 4.5 emission scenario (Rojas-Downing et al., 2018) ...................................... 109 Table 5.2. Management of weir height and timing of the common and aggressive managements used in the simulations during (1975-2004) and future (2030-2059) periods. The reference of the weir height is from the ground surface) ................................................... 111 Table 5.3. The 30-year average monthly temperature and precipitation under historic (1975 -2004) and future (2030-2059) climate scenarios ...................................................................... 116 Table 5.4. Comparison between historic (1975-2004) and future period (2030-2059) for 30 -year average drainage discharge under FD and CD .................................................................. 119 viii LIST OF FIGURES Figure 2.1. Soil sampling tools: (a) sampling ring, (b) sample ring insertion tool, (c) rubber mallet, and (d) trowel ................................................................................................................... 14 Figure 2.2. (a) hammering to collect soil cores, (b) soil core inserted and ready to collect after hammering, (c) removing soil from the sides of the sample ring to collect undisturbed samples, (d) put a white cap on the sample ring before digging it out with a trowel, (e) digging out the sample ring with trowel and (f) flattening the soil surface with the trowel ....... 15 Figure 2.3. Degassing water using a vacuum pump and a vacuum bottle: (a) Assembly of instruments for degassing DI water and (b) Blue tube should be kept in air inside the vacuum bottle while degassing DI water ................................................................................................... 16 Figure 2.4. (a) vacuum pump, (b) beaker mount, (c) vacuum mount and (d) assembly of Hyprop refilling unit ...................................................................................................................... 18 Figure 2.5. (a) cap is removed from the cutting edge of soil core, (b) Soil core is flipped, and a cheesecloth and perforated tray was put on the soil core, (c) Soil core flipped again with the cap removed, (d) soil core was put in a tray, (e) drainage water was poured in the tray, and (f) drainage water was poured until the cutting edge of the tray ................................................ 21 Figure 2.6. HYPROP 2 sensor unit after implementing the tensio shafts ..................................... 23 Figure 2.7. Shiny surface appears on top of soil core ................................................................... 25 Figure 2.8. (a) sensor unit placed on the soil ring, (b) after removing cheesecloth and perforated tray, the user flips the assembled sensor unit and soil core ...................................... 26 Figure 2.9. Assembled sensor unit and soil core is put on a balance to start measurements ..... 27 Figure 2.10. An optimal measuring curve for HYPROP measurements (image source: HYPROP Manual) ......................................................................................................................................... 29 Figure 2.11. Tension curve for a clay loam soil. Tension is in units of hectopascal (1 kPa = 10 hPa) ............................................................................................................................................... 31 Figure 2.12. Volumetric water content vs. pF plot using evaporation method and fitted soil water characteristic curve for a clay loam soil ............................................................................. 32 Figure 2.13. Tension curve for a sandy loam soil. Tension is in units of hectopascal (1 kPa = 10 hPa) ............................................................................................................................................... 35 ix Figure 2.14. Volumetric water content vs. pF plot using evaporation method and fitted soil water characteristic curve for a sandy loam soil .................................................................... 36 Figure 3.1. Top: Diagram of the experimental setup showing water flow through a V-notch weir. Bottom: Sideview photo of the control structure with flowing water [From Shokrana and Ghane (2021). Used with permission.] .................................................................................. 48 Figure 3.2. Left: Schematic diagram of a metal-edge sharp-crest 45° V-notch weir. The apex is the point at which both inclined sides of the metal crest meet. The dimensions of the V-notch weir do not include the thickness of the rubber gaskets attached to the board. Right: Photo of the metal-edge sharp-crest 45° V-notch weir [From Shokrana and Ghane (2021). Used with permission.] ..................................................................................................... 49 Figure 3.3. Diagram of the side view of a control structure and the distances needed to calculate the head (H) flowing through a V-notch weir using equation (3.5). A Solinst water- depth sensor was used to measure the distance "a" from the top of the structure to the upstream water surface and “f” from the top of the structure to the downstream water surface. A tape measure was used to measure the distance “b” inside the structure, and a meter-stick was used to measure the distance “c” from the top of the structure to the apex of the V-notch weir. Distance “d” was calculated by subtracting “a” from “b” [From Shokrana and Ghane (2021). Used with permission.] .................................................................................. 50 Figure 3.4. Measurement of the distance “c” from the top of the control structure to the apex of the V-notch weir. Multiple meter-sticks were glued and riveted together to make a long one. To measure the distance “c”: (a) We lowered the meter-stick inside the apex of the V-notch weir, and (b) next, we measured the distance from the apex to the top of the structure by placing another ruler on the top and reading the value on the vertical meter stick [From Shokrana and Ghane (2021). Used with permission.] ............................................... 51 Figure 3.5. Calibration equation for a metal-edge sharp-crest V-notch weir. We developed this equation for a maximum flow rate of 409.42 L/min with a head of 16.3 cm. We used the weighing method for the low flow rates (16-59 L/min) and the flow meter for the higher flow rates (105-409 L/min). This equation is valid for an H less than the height of the V-notch (i.e., flow through the V-notch). In our experiment, the height of the V-notch was 17.0 cm [From Shokrana and Ghane (2021). Used with permission.] .................................................................. 53 Figure 3.6. Comparison of the V-notch equation developed in this study to those in previous studies [From Shokrana and Ghane (2021). Used with permission.] ............................ 55 Figure 3.7. Measurement of water level with water-finding paste. A small amount of paste is applied to a meter stick, which is inserted into water. The dry paste is white but turns red upon contact with water [From Shokrana and Ghane (2021). Used with permission.] .............. 58 x Figure 3.8. Measurement of water level using a water-depth sensor: (a) Solinst water-depth sensor; the sensor is located at the middle of the probe. Once the sensor comes in contact with water, sound and light signals will indicate when to take the reading. (b) The sensor probe is lowered inside the control structure through a PVC pipe to reduce the effect of flow turbulence. Once in place, the length of the flat tape along the wall of the control structure indicates the water depth [From Shokrana and Ghane (2021). Used with permission.] ............. 59 Figure 3.9. A diagram (not drawn to scale) to represent the water flow inside a 25-cm control structure in agricultural fields. Top diagram: If there is heavy rainfall, the water level in the ditch will rise and the water level in the downstream chamber of the control structure will rise above the apex of the V-notch, causing submerged flow. Bottom diagram: Adding a 17.8 cm (7-inch) bottom board below the V-notch weir would give extra height to the apex of the V-notch weir, so there would be a greater chance of achieving freely flowing water through the V-notch [From Shokrana and Ghane (2021). Used with permission.] ................................... 62 Figure 4.1. (a) geographic location of the Blissfield site. This site is part of the River Raisin watershed which directly discharges into the western basin of Lake Erie at Monroe Harbor (b) drainage layout of the study site [From Shokrana et al. (2022). Used with permission.] ...... 73 Figure 4.2. Comparison between observed and simulated drainage discharge during the calibration period (October 1, 2018 – September 30, 2020) [From Shokrana et al. (2022). Used with permission.] ................................................................................................................. 86 Figure 4.3. Comparison between observed and simulated drainage discharge during the validation period (October 1, 2020 – June 30, 2022) [From Shokrana et al. (2022). Used with permission.] .................................................................................................................................. 86 Figure 4.4. Comparison between observed and simulated DRP loss through drainage discharge during the calibration period (October 1, 2018 – September 30, 2020) [From Shokrana et al. (2022). Used with permission.]............................................................................ 87 Figure 4.5. Comparison between observed and simulated DRP loss through drainage discharge during the validation period (October 1, 2020 – June 30, 2022) [From Shokrana et al. (2022). Used with permission.] ................................................................................................ 87 Figure 4.6. Comparison between observed and simulated TP loss through drainage discharge during the calibration period (October 1, 2018 – September 30, 2020) [From Shokrana et al. (2022). Used with permission.] ............................................................................ 88 Figure 4.7. Comparison between observed and simulated TP loss through drainage discharge during the validation period (October 1, 2020 – June 30, 2022) [From Shokrana et al. (2022). Used with permission.] ................................................................................................ 88 xi Figure 5.1. Hydrologic boundary of the river raisin watershed. The Blissfield is part of this river raisin watershed which directly discharges into the wester Lake Erie basin at Monroe Harbor ......................................................................................................................................... 105 Figure 5.2. Change in the 30-year average monthly maximum, minimum, and Mean temperature ................................................................................................................................ 113 Figure 5.3. Changes in precipitation amount under historical (1975-2004) and future climate scenario (2030-2059) ..................................................................................................... 115 Figure 5.4. 30-year average monthly drainage discharge for historic and future period .......... 119 Figure 5.5. The 30-year average monthly drainage discharge under FD, common management, and aggressive management for the historic period (1975-2004) ..................... 122 Figure 5.6. The 30-year average monthly drainage discharge under FD, common management, and aggressive management for the future period (2030-2059) ....................... 123 Figure 5.7. The variability in simulated future drainage discharge using 21 GCMs under FD management ............................................................................................................................... 125 Figure 5.8. The variability in simulated future drainage discharge using 21 GCMs under common management................................................................................................................ 125 Figure 5.9. The variability in simulated future drainage discharge using 21 GCMs under aggressive management ............................................................................................................. 126 xii LIST OF ABBREVIATIONS CD Controlled Drainage DRP Dissolved Reactive Phosphorus FD Free Drainage gSSURGO Gridded Soil Survey Geographic Database HYPROP Hydraulic Property Analyzer NSE Nash-Sutcliffe Efficiency PBIAS Percentage Bias RZWQM2 Root Zone Water Quality Model SWCC Soil Water Characteristics Curve TP Total Phosphorus WLEB Western Lake Erie Basin xiii CHAPTER 1: LITERATURE REVIEW The increasing population around the world demands an increase in food production. To meet this demand, agricultural lands have not been increased at the same rate. In many cases, forested and agricultural lands have been transformed into urban areas, thus increasing surface runoff and inundation of low-land areas. Agricultural lands are under heavy stress to increase crop production. Irrigation increases the crop yield but if water does not drain properly from the fields waterlogging might decrease the production of certain crops such as corn and soybean. Therefore, agricultural fields need to be properly drained. Most of the drainage occurred in the Midwest United States during the 1900s as federal and local government imposed laws regarding drainage and provided the necessary support for farmers to implement it (University of Illinois Extension, 2022). Since then, subsurface drainage practice is widely used in water-abundant areas of the US to improve crop production (King et al., 2014). Subsurface drainage helps reduce crop waterlogging stress by lowering the groundwater table and allowing farmers to perform timely field operations (Blann et al., 2009; Fausey et al., 1995). While the agronomic benefits of subsurface drainage are well documented (Fraser et al., 2001; Gardner et al., 1994; Hill, 1976), downstream environmental problems have been associated with the practice (Jaynes et al., 2001; Sims et al., 1998; Wright & Sands, 2001). A large number of studies have linked subsurface drainage to water quality problems in the receiving surface water bodies, leading to downstream eutrophication and hypoxia problems (Ahiablame et al., 2011; Rabalais et al., 2002). Farmers in the Midwest United States apply Phosphorus (P) and Nitrogen (N) based fertilizers in their fields which are transported to surface 1 water through surface runoff and subsurface drainage pathways. As a result, the presence of excessive nutrients causes harmful algal blooms (HABs) in freshwater ecosystems. The golden rule of drainage states: “Drain only the amount of water that is necessary for crop production and not a drop more” – all drainage systems that do not meet this rule most likely transport nutrients and water that is no longer available for crop uptake. Therefore, it is evident that restricting drainage discharge from leaving the field is important to reduce nutrient load. Controlled drainage (CD) is a conservation practice of managing water in the field by changing the outlet level of the drainage system(Evans et al., 1995; Lalonde et al., 1996; Nash et al., 2015; Saadat et al., 2018; Wesstrom & Messing, 2007; Williams et al., 2015b). This practice reduces the drainage volume which in turn reduces the nutrient load coming from the fields. The level is controlled with weir boards or stoplogs inside a control structure. The weir boards are usually lowered a few weeks before planting and harvesting, and then raised after the field operations are completed (Thorp et al., 2008; Youssef et al., 2018). The goal is to keep water inside the field for increased crop uptake during dry periods of the year and also to reduce drainage discharge during the non-growing season. Several on-farm and on-station replicated plot experiments were conducted to determine the efficiency of CD in reducing drainage discharge and nutrient loss (especially P). Since P is the limiting nutrient for the HABs in freshwater systems, we focused on the studies that worked with the P loss and drainage discharge reduction performance of the CD systems. Williams et al., (2015a) performed an on-farm experiment in Ohio, USA and they found a 7.5 to 33.6% reduction and drainage discharge and a 40 to 68% reduction in dissolved reactive phosphorus (DRP) load under CD management. These findings were in agreement with the findings observed in 2 Denmark, where the authors found a 37 to 54% reduction in drainage discharge and a 41 to 51% reduction in DRP load (Carstensen et al., 2019). A 60% reduction in drainage discharge and a 66% reduction in total phosphorus (TP) load were obtained from a study in Canada (Sunohara et al., 2016). On the other hand, Saadat et al. (2018) reported no reduction in P loading from Indiana, USA, while a 25 to 39% reduction in drainage discharge was obtained under CD. King et al., (2022) found an insignificant reduction in drainage discharge, but the TP load reduction was significant. The above discussion states that the performance of CD in reducing drainage discharge and P loading varies in different parts of the world and more studies are needed. While long-term field experiments with CD management can provide the best estimates of drainage discharge and nutrient loss from subsurface-drained fields, they are expensive and time-consuming. On the contrary, field-scale hydrologic models are robust, inexpensive, and can provide results in a cost-effective manner compared to field studies. Therefore, several field- scale models (DRAINMOD, RZWQM2, ICECREAM, etc.) were developed by researchers to predict drainage discharge and nutrient loss from agricultural fields. While a lot of these models can satisfactorily predict the drainage discharge from subsurface-drained fields, their performance in predicting P loss with discharge is still questionable and largely uninvestigated. Therefore, there is a need to assess the performance of these P models in simulating drainage discharge and P loss from fields. A comprehensive literature review on the P model suggested that ICECREAM is the best model that can predict P loss from surface runoff and subsurface drainage discharge (Qi & Qi, 2016; Radcliffe et al., 2015). However, since ICECREAM lacks a water-table based tile drainage component and uses a simple storage routing concept to mimic tile drainage, there is room to 3 improve this model (Pferdmenges et al., 2020). To improve the limitations of ICECREAM, Sadhukhan et al. (2019) developed the P module for the Root Zone Water Quality Model (RZWQM2) known as RZWQM2-P. The RZWQM2 is a process-based model that has been widely used in North America (Ahmed et al., 2007; Jiang et al., 2018; Ma et al., 2007; Malone et al., 2014; Thorp et al., 2008), but the P component of this model has been only tested twice by the developers for the same on-station plot experiment under a free drainage scenario. It is important to test the performance of the P model under CD management coupled with different climates, soil types, fertilization, and cropping practices. The performance of CD in reducing drainage discharge under the present climate has been widely tested around the world (Nash et al., 2015; Saadat et al., 2018; Tolomio & Borin, 2018; Wahba et al., 2001; Wesstrom & Messing, 2007; Williams et al., 2015a), but the efficiency of CD for the future climate is still limited. Pease et al. (2017) conducted a study in Ohio, USA, and reported a reduction in drainage discharge in the future although the climate projections indicated a significant increase in precipitation and temperature for the future. The authors concluded that increased evapotranspiration (ET) due to increased temperature likely has caused reduced drainage discharge in the future. Another study in Poland reported that the most efficient way of reducing drainage discharge with CD management will be to implement it in the early spring (Sojka et al., 2020). Another study investigated the effects of the timing of weir management under CD for future climate (Salla et al., 2022). They found that drainage discharge under differently timed CD management was decreased by 11% to 23% under future climate conditions. 4 While previous studies concentrated on the timing of weir management (Salla et al., 2022), they ignored the effects on drainage discharge caused by the height of weir management. Those that investigated the height of weir management on CD performance, used the same weir height under CD for both non-growing and growing seasons (Williams et al., 2015), which is not a common practice in the Midwest USA. Only Saadat et al. (2018) performed an on-farm experiment to evaluate the height of weir management on CD performance for existing climate and drainage design. Therefore, it is important to predict the efficiency of CD in reducing drainage discharge under both varying heights and time management for future climate. The objective of this study is to predict the effects of height and timing of weir management on the drainage discharge of a CD system for future climate conditions in Michigan. The outcome of this study will help farmers make informed decisions on weir management in the future to reduce the waterlogging stress on crops, and to reduce the drainage discharge and nutrient loss from the fields. The following tasks were performed to achieve this objective: Task 1: Measurement of soil-water characteristics input data Task 2: Developing a stage-discharge equation for a V-notch weir to estimate the drainage discharge Task 3: Calibrating the RZWQM2-P model to simulate drainage discharge and phosphorus loss in clay loam soil in Michigan Task 4: Predicting the effects of two controlled drainage scenarios on drainage discharge for future climate 5 REFERENCES Ahiablame, L. M., Chaubey, I., Smith, D. R., & Engel, B. A. (2011). Effect of tile effluent on nutrient concentration and retention efficiency in agricultural drainage ditches. Agricultural Water Management, 98(8), 1271–1279. Ahmed, I., Rudra, R., McKague, K., Gharabaghi, B., & Ogilvie, J. (2007). Evaluation of the Root Zone Water Quality Model (RZWQM) for Southern Ontario: Part II. Simulating Long-Term Effects of Nitrogen Management Practices on Crop Yield and Subsurface Drainage Water Quality. WATER QUALITY RESEARCH JOURNAL OF CANADA, 42(3), 219–230. https://doi.org/10.2166/wqrj.2007.025 Blann, K. L., Anderson, J. L., Sands, G. R., & Vondracek, B. (2009). Effects of agricultural drainage on aquatic ecosystems: a review. Critical Reviews in Environmental Science and Technology, 39(11), 909–1001. Carstensen, M. v, Borgesen, C. D., Ovesen, N. B., Poulsen, J. R., Hvid, S. K., & Kronvang, B. (2019). Controlled Drainage as a Targeted Mitigation Measure for Nitrogen and Phosphorus. JOURNAL OF ENVIRONMENTAL QUALITY, 48(3), 677–685. https://doi.org/10.2134/jeq2018.11.0393 Evans, R. O., Skaggs, R. W., & Gilliam, J. W. (1995). CONTROLLED VERSUS CONVENTIONAL DRAINAGE EFFECTS ON WATER-QUALITY. JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING-ASCE, 121(4), 271–276. https://doi.org/10.1061/(ASCE)0733- 9437(1995)121:4(271) Fausey, N. R., Brown, L. C., Belcher, H. W., & Kanwar, R. S. (1995). Drainage and Water-quality in Great-Lakes and Corn-belt States. Jounral of Irrigation and Drainage Monitoring-ASCE, 121(4), 283–288. https://doi.org/10.1061/(ASCE)0733-9437(1995)121:4(283) Fraser, H., Fleming, R., & Eng, P. (2001). Environmental benefits of tile drainage. Prepared for: LICO. Land Improvement Contractors of Ontario. Ridgetown College, University of Guelph. Gardner, W. K., Drendel, M. F., & McDonald, G. K. (1994). Effects of subsurface drainage, cultivation, and stubble retention on soil porosity and crop growth in a high rainfall area. Australian Journal of Experimental Agriculture, 34(3), 411–418. Hill, A. R. (1976). The environmental impacts of agricultural land drainage. Jaynes, D. B., Colvin, T. S., Karlen, D. L., Cambardella, C. A., & Meek, D. W. (2001). Nitrate loss in subsurface drainage as affected by nitrogen fertilizer rate. Journal of Environmental Quality, 30(4), 1305–1314. 6 Jiang, Q. J., Qi, Z. M., Madramootoo, C. A., & Singh, A. K. (2018). Simulating hydrologic cycle and crop production in a subsurface drained and sub-irrigated field in Southern Quebec using RZWQM2. COMPUTERS AND ELECTRONICS IN AGRICULTURE, 146, 31–42. https://doi.org/10.1016/j.compag.2018.01.021 WE - Science Citation Index Expanded (SCI- EXPANDED) King, K. W., Fausey, N. R., & Williams, M. R. (2014). Effect of subsurface drainage on streamflow in an agricultural headwater watershed. Journal of Hydrology, 519(A), 438–445. https://doi.org/10.1016/j.jhydrol.2014.07.035 King, K. W., Hanrahan, B. R., Stinner, J., & Shedekar, V. S. (2022). Field scale discharge and water quality response, to drainage water management. Agricultural Water Management, 264, 107421. Lalonde, V., Madramootoo, C. A., Trenholm, L., & Broughton, R. S. (1996). Effects of controlled drainage on nitrate concentrations in subsurface drain discharge. Agricultural Water Management, 29(2), 187–199. Ma, L., Malone, R. W., Heilman, P., Jaynes, D. B., Ahuja, L. R., Saseendran, S. A., Kanwar, R. S., & Ascough II, J. C. (2007). RZWQM simulated effects of crop rotation, tillage, and controlled drainage on crop yield and nitrate-N loss in drain flow. Geoderma, 140(3), 260–271. Malone, R. W., Nolan, B. T., Ma, L., Kanwar, R. S., Pederson, C., & Heilman, P. (2014). Effects of tillage and application rate on atrazine transport to subsurface drainage: Evaluation of RZWQM using a six-year field study. Agricultural Water Management, 132, 10–22. Nash, P. R., Nelson, K. A., Motavalli, P. P., Nathan, M., & Dudenhoeffer, C. (2015). Reducing Phosphorus Loss in Tile Water with Managed Drainage in a Claypan Soil. JOURNAL OF ENVIRONMENTAL QUALITY, 44(2), 585–593. https://doi.org/10.2134/jeq2014.04.0146 Pease, L. A., Fausey, N. R., Martin, J. F., & Brown, L. C. (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. https://doi.org/10.2489/jswc.72.3.240 Pferdmenges, J., Breuer, L., Julich, S., & Kraft, P. (2020). Review of soil phosphorus routines in ecosystem models. Environmental Modelling & Software, 126, 104639. Qi, H., & Qi, Z. (2016). Simulating phosphorus loss to subsurface tile drainage flow: a review. Environmental Reviews, 25(2), 150–162. Rabalais, N. N., Turner, R. E., & Wiseman Jr, W. J. (2002). Gulf of Mexico hypoxia, aka" The dead zone". Annual Review of Ecology and Systematics, 235–263. 7 Radcliffe, D. E., Reid, D. K., Blombäck, K., Bolster, C. H., Collick, A. S., Easton, Z. M., Francesconi, W., Fuka, D. R., Johnsson, H., & King, K. (2015). Applicability of models to predict phosphorus losses in drained fields: A review. Journal of Environmental Quality, 44(2), 614–628. Saadat, S., Bowling, L., Frankenberger, J., & Kladivko, E. (2018). Nitrate and phosphorus transport through subsurface drains under free and controlled drainage. Water Research, 142, 196– 207. https://doi.org/10.1016/j.watres.2018.05.040 Sadhukhan, D., Qi, Z., Zhang, T., Tan, C. S., Ma, L., & Andales, A. A. (2019). Development and evaluation of a phosphorus (P) module in RZWQM2 for phosphorus management in agricultural fields. Environmental Modelling and Software, 113, 48–58. https://doi.org/10.1016/j.envsoft.2018.12.007 Salla, A., Salo, H., & Koivusalo, H. (2022). Controlled drainage under two climate change scenarios in a flat high-latitude field. HYDROLOGY RESEARCH, 53(1), 14–28. https://doi.org/10.2166/nh.2021.058 Sims, J. T., Simard, R. R., & Joern, B. C. (1998). Phosphorus loss in agricultural drainage: Historical perspective and current research. Journal of Environmental Quality, 27(2), 277–293. Sojka, M., Kozlowski, M., Kesicka, B., Wrozynski, R., Stasik, R., Napierala, M., Jaskula, 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-BASEL, 10(5). https://doi.org/10.3390/agronomy10050625 Sunohara, M. D., Gottschall, N., Craiovan, E., Wilkes, G., Topp, E., Frey, S. K., & Lapen, D. R. (2016). Controlling tile drainage during the growing season in Eastern Canada to reduce nitrogen, phosphorus, and bacteria loading to surface water. AGRICULTURAL WATER MANAGEMENT, 178, 159–170. https://doi.org/10.1016/j.agwat.2016.08.030 Thorp, K. R., Jaynes, D. B., & Malone, R. W. (2008). Simulating the long-term performance of drainage water management across the Midwestern United States. Transactions of the ASABE, 51(3), 961–976. Tolomio, M., & Borin, M. (2018). Water table management to save water and reduce nutrient losses from agricultural fields: 6 years of experience in North-Eastern Italy. Agricultural Water Management, 201(January), 1–10. https://doi.org/10.1016/j.agwat.2018.01.009 University of Illinois Extension. (2022). Drainage Tile History in the U.S. Wahba, M. A. S., El-Ganainy, M., Abdel-Dayem, M. S., Gobran, A., & Kandil, H. (2001). Controlled drainage effects on water quality under semi-arid conditions in the Western Delta of Egypt. IRRIGATION AND DRAINAGE, 50(4), 295–308. https://doi.org/10.1002/ird.29 8 Wesstrom, I., & Messing, I. (2007). Effects of controlled drainage on N and P losses and N dynamics in a loamy sand with spring crops. AGRICULTURAL WATER MANAGEMENT, 87(3), 229–240. https://doi.org/10.1016/j.agwat.2006.07.005 Williams, M. R., King, K. W., & Fausey, N. R. (2015a). Drainage water management effects on tile discharge and water quality. AGRICULTURAL WATER MANAGEMENT, 148, 43–51. https://doi.org/10.1016/j.agwat.2014.09.017 Williams, M. R., King, K. W., & Fausey, N. R. (2015b). Drainage water management effects on tile discharge and water quality. AGRICULTURAL WATER MANAGEMENT, 148, 43–51. https://doi.org/10.1016/j.agwat.2014.09.017 Wright, J., & Sands, G. (2001). Planning an agricultural subsurface drainage system. College of Agricultural, Food and Environmental Sciences, University of Minnesota. BU-07685, 1–10. Youssef, M. A., Abdelbaki, A. M., Negm, L. M., Skaggs, R. W., Thorp, K. R., & Jaynes, D. B. (2018). DRAINMOD-simulated performance of controlled drainage across the U.S. Midwest. Agricultural Water Management, 197, 54–66. https://doi.org/10.1016/j.agwat.2017.11.012 9 CHAPTER 2: MEASUREMENT OF SOIL-WATER CHARACTERISTICS INPUT DATA 2.1 Abstract Soil water characteristic curve (SWCC) has an important relationship with application in drainage, irrigation, soil physical behavior, and modeling hydrology and nutrient transport. However, measurement of the SWCC is often very time consuming, inaccurate and requires a lot of effort. In order to determine an accurate SWCC, we used HYPROP2. This method article extensively describes the topics which were not covered well by the instrument’s manual such as collecting soil samples, use of the HYPROP refill unit, degassing water prior to degassing the tensio shafts and other procedures. Advice is provided in terms of better handling of the equipment to receive all four phases of an optimal measuring curve. Following the step-by-step procedure mentioned in this article would provide a high-quality SWCC. Our measurements were performed on both clay loam and sandy loam soils to show differences in the SWCC. We found that the upper tensio shaft took longer to cavitate for sandy loam soil compared to the clay loam soil. 2.2 Method Description This method explains the use of HYPROP (Hydraulic Property Analyzer) as an alternative technique to the conventional methods of measuring Soil water characteristic curve (SWCC) and unsaturated hydraulic conductivity of soil (METER 2015b; Schindler and Mueller 2006). Measuring SWCCs have always been difficult due to the lack of sufficient data points using conventional methods and the amount of time these methods require to generate these curves. HYPROP2 follows an evaporation method where two tensiometers measure the tension implied by water to the soil column. Also, how the water content of the soil column changes over time at 10 different tension values is observed (Peters and Durner 2006; Peters et al. 2015; Schindler 1980). However, the limitation of those tensiometers was that they would cavitate at a much smaller pressure typically ranging between 70 to 90 kPa. A new design of tensiometers in the late 2000s allowed withstanding cavitation to a much higher tension values as high as 435 kPa (Peters 2013; Schelle et al. 2013; Schelle, Iden, and Durner 2011; Schindler et al. 2010; Schindler and Müller 2017). HYPROP2 was developed based on those new designs of tensiometers. HYPROP2 measures water potential at different soil saturation levels with the help of two tensiometers (i.e., one long and one short). This device is a more advanced version of HYPROP that has more robust and precise pressure transducers. It also introduces a faster optical monitoring of the measurements through a LED ring for better visualization of the current status of the device (METER 2015b). In HYPROP2, the soil sample stays on a laboratory balance during the experiment. Water evaporates from the soil over time, and HYPROP2 records the change of water potential during this process. This instrument also records the changing weight of the soil sample. This change of weight helps determine the moisture content of the soil. Finally, HYPROP- Fit program plots water potential against the changes in moisture contents to create a SWCC. The user should know that HYPROP2 can only measure the matric potential of the soil. It can measure water potential at the wet range of the soil, so the curve created with HYPROP2 will not be a complete curve. WP4C instrument can be combined with HYPROP2 to measure water potential in the dry range of the soil sample. WP4C is capable of measuring both matric and osmotic potential that completes the SWCC. However, HYPROP2 has its merits on producing high resolution data at the wet range of a soil sample. 11 There are studies where results obtained from HYPROP have also been compared with the results from conventional methods or to different models. Öztürk et al. (2013) compared HYPROP outputs with the outputs from sand box (or pressure plate), which validated the use of HYPROP as a potential method for creating SWCC. Another study compared HYPROP with the BEST model (Beerkan Estimation of Soil Transfer Parameters) (Leitinger, Obojes, and Lassabatère 2015), which showed that the retention curves (SWCC) from HYPROP followed a faster and continuous dehydration process compared to the retention curves from the BEST model. Also, their comparison revealed that the results of soil water characteristic may vary based on the methodological approach used between different soil types. A different study evaluated the accuracy of HYPROP Measurement Systems (HMS) with HYDRUS-1D software package (Bezerra- Coelho et al. 2018). HYDRUS-1D can generate virtual pressure head in soil columns as a function of time, and independent tests were performed by this package on HYPROP system using the Van-Genuchten-Mualem model for a wide range of soil textures. The results from HYDRUS-1D showed that accurate estimates of SWCC and other parameters were obtained by HYPROP within the range of available retention measurements. Since HMS failed to operate at a very high pressure range, Wang et al. (2012) evaluated the data points at a high-pressure range using centrifugal method. He added one control point at the high-pressure range of the SWCC already developed by HYPROP. The results showed that adding a control point to the high-pressure range makes the extrapolation of SWCC more reliable. 2.3 Soil Sample Collection Soil samples were collected from two privately-owned farms. The first one is near Blissfield, Michigan. The soil type is predominantly Ziegenfuss clay loam, which is classified as a 12 poorly drained soil (reference of NRCS web soil survey). The farmer uses a corn-soybean rotation in this field and applies commercial fertilizer. The second farm is near Palmyra, Michigan. The soil type is Brady and Macomb sandy loam at this farm, which is classified as a somewhat poorly drained soil (reference of NRCS web soil survey). The cropping system is corn-soybean rotation with commercial fertilizer application. For both farms, soil samples were collected at two different locations with three replicates at each location. It is important to note that soil samples must remain undisturbed during collection. Four tools (Figure 2.1) are needed to collect the soil samples: (i) a soil sampling ring (Meter Group), (ii) a sample ring insertion tool (Meter Group), (iii) a rubber mallet, and (iv) a trowel. Figure 2.2 explains the whole soil sample collection process. Firstly, the soil sampling ring was attached to the sample ring insertion tool in such a way that the cutting edge of the ring was facing the soil surface. Then, the apparatus was hammered using the rubber mallet. The hammering was continued until the sampling ring had completely penetrated the soil. It is important to remember that the sampling ring just holds onto the insertion tool, and it cannot be attached to this without holding the bottom of the ring. Thus, whenever the ring had penetrated the soil, the insertion tool came off leaving the sampling ring inside the soil. Subsequently, a trowel was used to dig around the ring to loosen it up. Each ring comes with two white plastic caps to cover both ends of it. One of the caps were placed on the top surface of the ring. The trowel was put under the cutting edge of the ring, one hand was placed on the top of the ring, and the soil sampling ring with soil sample was taken out and was flipped. The excess soil was removed and levelled along the cutting edge using the trowel. Another cap was placed 13 to cover the cutting edge of the ring. After collection, soil samples were placed in a box and transported to the lab for further analysis. (a) (b) (c) (d) Figure 2.1. Soil sampling tools: (a) sampling ring, (b) sample ring insertion tool, (c) rubber mallet, and (d) trowel 14 (a) (b) (d) (c) (e) (f) Figure 2.2. (a) hammering to collect soil cores, (b) soil core inserted and ready to collect after hammering, (c) removing soil from the sides of the sample ring to collect undisturbed samples, (d) put a white cap on the sample ring before digging it out with a trowel, (e) digging out the sample ring with trowel and (f) flattening the soil surface with the trowel 15 2.4 Preparing HYPROP2 In this section, we will explain the steps required to prepare the HYPROP2 for determining the soil water characteristic curve. 2.4.1 Degassing water The first step in operating HYPROP2 is degassing water. A vacuum bottle was filled with deionized (DI) water. The tube coming out of the vacuum bottle was connected to the vacuum mount of the HYPROP2 system, and the vacuum mount was connected to the vacuum pump ( Figure 2.3a). Then, the pump was turned on to create vacuum in order to evacuate all the bubbles or gas from the DI water. It is important to remember that the tube inside the vacuum bottle needs to stay in the air (not submerged in DI water), so that it can evacuate as much air from the bottle without removing water ( Figure 2.3b). The water was degassed for a couple of hours. (a) (b) Figure 2.3. Degassing water using a vacuum pump and a vacuum bottle: (a) Assembly of instruments for degassing DI water and (b) Blue tube should be kept in air inside the vacuum bottle while degassing DI water 2.4.2 Degassing the HYPROP sensor unit For running a HYPROP experiment, two of the devices need to be degassed completely: the HYPROP sensor unit and the two tensio shafts. There are two ways to perform degassing: (i) 16 degassing the device using the HYPROP refill unit and (ii) degassing using syringes. Meter Group recommended degassing using the HYPROP refill unit because it is more accurate. Alternative to the HYPROP refill unit is manual degassing using syringes, which creates challenges to degas the water completely. This manual method requires a lot of labor, and chances of error is more than using the automated HYPROP refill unit. The HYPROP manual gives a fair instruction about how to degas the device manually using syringes but does poorly on explaining the degassing process using HYPROP refill unit. The scope of this article is to give a better understanding of how the HYPROP refill unit works in degassing the tensio shafts and HYPROP sensor unit. 2.4.3 Degassing the sensor unit using the HYPROP refill unit 2.4.3.1 Assembly This method involves the use of a high-performance vacuum pump that can generate a vacuum pressure of around 0.85 to 0.90 bars, which can degas the sensor unit and both of the tensio shafts. The total arrangement consisted of four instruments connected to each other (Figure 2.4). The first instrument was the vacuum pump, which was connected to a vacuum mount. The vacuum mount consisted of a pressure gauge and a vacuum flask. Whenever the pump was running, the vacuum pressure could be monitored looking at the pressure gauge. Also, when the degassing process took place, air bubbles from both tensio shafts and the sensor unit were collected in the vacuum flask to prevent water entering the pump. The vacuum mount was connected to a beaker mount where four tensio shafts can be degassed at the same time. The beaker mount has four ports. Each port is connected to a tube and each tube connects to an adapter. Finally, the beaker mount was connected to the HYPROP sensor unit. The top part of the 17 HYPROP sensor unit was the acrylic adapter, which was attached to the HYPROP sensor unit base. The beaker mount was connected to the sensor unit with a tube. (a) (b) (c) (d) Figure 2.4. (a) vacuum pump, (b) beaker mount, (c) vacuum mount and (d) assembly of Hyprop refilling unit 2.4.3.2 Procedure The degassing process using the HYPROP refill unit should be continued for about 12 to 24 hours. In our case, it was about 20 hours. All the tubes were connected to their respective connections or ports (Figure 2.4d). The color of the tubes should match the color of the connections, so that wrong tubes are not connected to wrong ports. Blind plugs were put in the connections not being used during the experiment. In the beaker mount, the beakers were filled with water from the vacuum bottle which had already been degassed. The tubes in the beaker mount were connected to the glass adapters and then each of the tensio shafts were screwed in 18 the adapters. Then, the tensio shafts were placed in the beakers. Each beaker had a long and a short tensio shaft submerged in degassed water. After placing the tensio shafts in the beakers, the black pressure valve attached to the beaker mount was closed by rotating it counterclockwise towards a vertical position of the knob. The pump was turned on for 10 minutes until the system reached full vacuum. Once full vacuum was achieved, the pump was turned off and the system retained the vacuum. The system then kept on degassing for about 2 hours. After 2 hours, the system started losing full vacuum, so the pump needed to be started again for 10 minutes to take the system back to full vacuum. If the user wants to perform this degassing process overnight, purchasing a programmable timer would be very helpful. This timer can run the pump for 10 minutes in every couple of hours and shut it off, thus saving time by running the degassing process overnight and performing the experiments during the daytime. Getting full vacuum is extremely important for running experiments using HYPROP2. If a vacuum pressure value of around 0.85 to 0.90 bars (85 to 90 kPa) cannot be achieved, proper degassing will not be accomplished. Also, even if the system reaches full vacuum, the user needs to check if it can still hold the vacuum after the pump is turned off. The purpose of using the pump is to allow the system to reach full vacuum and then retain it for a couple of hours even if the pump is not in action. If the system is not able to hold the vacuum, there is definitely a leak in any of the refill unit components. A leak can happen in different ways such as a damaged vacuum bottle, the tubes may not be pushed all the way in through the ports of the vacuum mount and beaker mount, or the pressure valve attached to the beaker mount may be open. Thus, proper attention needs to be directed to these details. 19 2.4.3.3 Saturating the soil sample After degassing the sensor unit using the HYPROP2 refill unit, the soil sample was saturated with water. The HYPROP manual suggests saturating the soil sample in degassed water. But in reality, the water for saturating soil sample solely depends on the purpose of the experiment. For agricultural applications, the soil core needs to be saturated in subsurface drainage water to simulate water movement in the soil matrix. Thus, we started saturating the soil cores in subsurface drainage water (collected from the on-farm site). The required time of saturating different kinds of soils may vary (Table 2.1). Table 2.1. Saturation time for different types of soils Soil Texture Time for saturation Clay loam About 1 hour1 Sandy loam About 45 minutes1 Coarse sands About 10 minutes2 Fine sands About 45 minutes2 Silt About 6 hours2 1- Based on our experience. 2- Based on HYPROP manual A step-by-step saturation process of the soil sample is shown in (Figure 2.5). The white cap from the blunt end of the soil core was removed. The soil core was covered with a cheesecloth, and the perforated tray was placed on top of the soil core. Then, the soil core was flipped, and the white cap was removed from the cutting-edge end of the soil core. Later, the soil core was placed in an empty tray. Drainage water was filled in the tray up to the cutting edge of the soil core. The user should be careful not to pour water on the top surface of the soil core to avoid trapping air. Also, pouring water from the top may immediately create a shiny surface indicating the soil sample is saturated although it may not be saturated in the middle. A white 20 cap can be placed on top of the cutting edge to prevent evaporation and to protect the soil from solar radiation. (a) (b) (c) (d) (e) (f) Figure 2.5. (a) cap is removed from the cutting edge of soil core, (b) Soil core is flipped, and a cheesecloth and perforated tray was put on the soil core, (c) Soil core flipped again with the cap removed, (d) soil core was put in a tray, (e) drainage water was poured in the tray, and (f) drainage water was poured until the cutting edge of the tray 2.4.3.4 Preparing the programs Once the soil was saturated, these two programs were used to determine the SWCC: (i) HYPROP-View and (ii) HYPROP-Fit. HYPROP-View was used during the beginning of the experiment to select the saving directories, selection of the HYPROP device, and to determine 21 the data collection frequency from the sensor unit. HYPROP-Fit was used at a later stage to find the stop point and air entry point of the sample and further evaluation of the results. HYPROP- Fit allowed to select different computational methods for evaluation of results, and to export the results to the computer. Both programs were fairly easy to use, and we recommend using the HYPROP2 manual for more explanation. 2.4.3.5 Implementing tensio shafts in sensor unit Once degassing was done, the valve connected to the beaker mount was opened very slowly. If the valve was opened too quickly, a sudden pressure shock could damage the pressure sensors in the sensor unit. Then, the tensio shafts were removed from the adapters. Once the tension shafts have been removed from the adapters, it is very important to keep the tensio shafts hydrated, so putting the silicone caps on them is a good practice. The user can also add degassed water on ceramic tips of the tensio shafts from time to time to prevent them from drying. The respective connection ports for each shaft is already drawn on the sensor unit. When the tensio shafts are completely degassed and their ceramic tips are fully wet, they are ready to be screwed in the sensor unit. The HYPROP manual suggests that it takes about 9 turns for the tensio shafts to get sealed in the ports. From our experience, that was not always the case, or it was hard to measure exact 9 turns (Figure 2.6). Thus, a better practice is to keep an eye on the HYPROP-View program where the refilling wizard shows how much pressure is being exerted by each tensio shaft. At the initial stages of screwing in the shaft, the pressure values remained small or negative, and the pressure started increasing with time. It is advisable to monitor the pressure values while screwing in the tensio shafts. Once it feels that the tensio shafts are close to sealing, they should be screwed in very slowly. Also, monitoring the pressure 22 values in the HYPROP-View program will help avoid exceeding the maximum limit of 200 kPa (2000 hPa). The pressure values should always remain below this range. Once the shafts were fully screwed in the sensor unit, the silicone caps were removed. Then, a silicone gasket was put on the sensor unit to protect it from dust. The tensio shafts are very sensitive and can get damaged by the slightest negligence while operating, so it is good to occasionally check whether the tensio shafts are working properly. An easy step to check whether the tensio shafts are functioning is to add small drops of water on the ceramic tip of the tensio shafts. The refilling wizard window in the HYPROP-View program will show the change in pressure due to addition of water droplets. If the pressure for both tensio shafts go back to zero in a little while, then the tensio shafts are functioning properly. This is to check the zero potential of the tensio shafts. Another way to check is to dry the ceramic tip with paper towel and to keep an eye to the pressure values. In this case, the pressure values should readily increase to the atmospheric air pressure. Figure 2.6. HYPROP 2 sensor unit after implementing the tensio shafts 23 2.4.3.6 Precautions One important precaution before starting the experiment with HYPROP2 is to check whether offset recalibration is necessary. METER Group sets a narrow range of values for HYPROP sensors to function efficiently. If HYPROP has not been used for some time, the offset values of the electronic sensors may drift from the values set by METER Group. In these circumstances, the user needs to check for offset recalibration. From our experience, it was seen that during consecutive experiments the offset recalibration was unnecessary. However, the user should always check whether calibration is needed. Also, the HYPROP manual does a good job in explaining this calibration process. After degassing the sensor unit, the user needs to connect it with the provided USB adapter. Then, HYPROP-View program needs to be started. After clicking on the “refilling wizard”, the HYPROP sensor should be selected. Then, the software will check whether calibration is required. If calibration is needed, clicking the button “set zero value” will recalibrate the HYPROP sensors. 2.4.4 Assembling sensor unit to the soil sampling ring When the tensiometers were fully functioning and the soil sample was saturated, i.e., a shiny surface appeared on top of the soil core (Figure 2.7), it was time to attach the tensiometers to the soil core. The USB adapter was removed from the sensor unit, and the sensor unit was ready to be connected to the soil sampling ring. One important advice is to avoid removing the saturated soil sample from the tray filled with drainage water until the user is ready to start the measurements. For sandy soils, water drains through the soil core quickly, so the soil still might not be saturated while implementing the sensor unit to the soil core. A small auger and tensio shaft adapter were used for drilling holes in the soil core. The adapter should be set in such a way 24 so that the user can remember which holes in the soil core are for long and short tensio shafts. HYPROP manual states that the adapter should be adjusted to the sample ring in such a way, so that the small hole of the adapter aligns with the sample ring number. Once holes were made with the auger, there was a chance that air would enter those holes, so it was imperative to fill the holes again with water using the droplet syringe. At this stage, the soil was fully saturated and ready to be assembled to the sensor unit. Before assembling the soil core to the sensor unit, the silicone caps from the tensio shafts were removed. Then, the sensor unit was inverted and slowly placed close to the holes in the soil core. Once both tensio shafts were aligned with their respective holes in the soil core, the tensio shafts along with the sensor unit was slowly penetrated to the soil core (Figure 2.8a). Next, the assembled sensor unit and sample ring was flipped (Figure 2.8b) and the perforated bowl along with the cheesecloth was removed. Now, the ring is ready for the experiment. Figure 2.7. Shiny surface appears on top of soil core 25 (a) (b) Figure 2.8. (a) sensor unit placed on the soil ring, (b) after removing cheesecloth and perforated tray, the user flips the assembled sensor unit and soil core 2.5 Starting the Measurements After the soil sampling ring was safely attached to the sensor unit, the combined piece was ready to be placed on a balance (Figure 2.9). But before that, the balance was placed on a flat top and made level using the vertical screws and bubble. Next step was to calibrate the balance using the instructions in the HYPROP manual. The sensor unit was then placed on the balance. One end of a magnet clamp (i.e., comes with the HYPROP unit) was connected to the balance and another end to the sensor unit. A white USB cable (i.e., comes with the HYPROP unit) connected the balance to the computer. Once both the sensor unit and the balance were connected to the computer, HYPROP- View program started recording change of weight of the soil core and pressure potentials at the long and short tensio shafts (METER 2015b). HYPROP-View allows selecting the frequency of the pressure potential data. The default value is 10 minutes, and it is possible to collect higher frequency data at different intervals during the experiment. The weight of the sampling ring (i.e., written on each ring) needs to be entered into the HYPROP-View program for correct evaluation of the results. These weight values are slightly different for each sampling ring. HYPROP-View also allows the user to select the units of pressure potential for example, Hectopascal (hPa) or 26 Kilopascal (kPa) (1 kPa = 10 hPa). Usually, the experiment keeps on running for 2 to 7 days (depending on the type of soil) and data is collected continuously by the HYPROP-View program. The user can see the data in the HYPROP-Fit program without stopping the measurements. HYPROP-Fit also allows to look for the stopping point or air-entry point in the plot. Once this point is found, the measurements in the HYPROP-View program can be stopped. Figure 2.9. Assembled sensor unit and soil core is put on a balance to start measurements 2.6 Stopping the Measurements HYPROP measurements should be run for several days to create an optimal measuring curve with time. An ideal measuring curve (Figure 2.10) is composed of 4 phases: (phase 1) regular measurement range, (phase 2) boiling delay phase, (phase 3) cavitation phase and (phase 4) air entry phase (METER 2015b). Phase 1 is the regular measurements where the tension values keep on increasing over time without showing any sign of decrease. In phase 2, tension is increased above the ambient air pressure, and the increase keeps on continuing until the tension value decreases. The boiling delay phase is often very difficult to achieve, but the results are still 27 valid without this phase. Phase 2 can only be achieved if the user is able to completely degas the tensio shafts and the sensor unit, which would mean no air bubbles should be present in the system. In phase 3, some water vapor starts to form in the tensio shafts, and the tension values will have a sudden drop to the ambient air pressure, which is the point where the curve becomes flat. There will be very small decreases in the tension values after this phase. Air enters the ceramic tube at phase 4, and tension values instantly drop to zero. After the tension values in the tensio shafts start dropping, a measurement can be manually stopped in three ways. First, using the stop point where the long tensio shaft reaches the cavitation phase, so the tension values do not increase anymore. The curve will become almost flat or reduce very little after this point. Second, using the air-entry point when the tension values at the long tensio shaft reduce to zero as air enters the ceramic tip. Third, using the air entry point when the tension values at both the long and short tensio shafts reaches zero due to air entrapment in the ceramic tip. To automatically stop the measurements, the user needs to click on the “search stop point” button in the HYPROP-Fit measurements tab. This program will detect a stop point in the tension curve of the top tensio shaft whenever it starts cavitating. It is important to remember that one should only use this “search stop point” option if the tensio shafts have been fully degassed and boiling delay phase have been achieved. For our experiments, we were not able to achieve a boiling delay phase for both soil types due to presence of water bubbles in the tensio shafts even after degassing them for more than 20 hours. When it is not possible to get a boiling delay phase, suboptimal curves develop, and the tension values will not go through a rapid drop. Instead, the tension values will become flat 28 after increasing for some time. In this case, the automatic stop point detection using the HYPROP- Fit program may not be reliable. The user should manually find the stop point using the stop point cursor by moving right or left in the measurement tab of the HYPROP-Fit program. It is advisable to select the stop point at such a point where the slope of the tension curve is positive. In this paper, the same concept has been used to identify the stop point of tension curves for both clay and sandy loam soils. Figure 2.10. An optimal measuring curve for HYPROP measurements (image source: HYPROP Manual) 2.7 Determining the Dry Weight After the experiment was ended, we determined the dry weight of the soil sample (METER 2015b). This dry weight helped to measure the volumetric water content based on how much weight was lost from the soil core due to evaporation. First, the soil core was detached from the 29 HYPROP sensor unit, which required care because the tensio shafts were strongly attached to the soil core and a little extra force in the wrong direction could break them. Once the soil core was detached from the sensor unit, it was placed in an aluminum tray. The weight of the empty tray was measured beforehand. It is normal that after several days of drying within ambient temperature, the soil becomes very crumbly (especially sandy soils), and the soil will easily fall off the sampling ring. To address this, the sampling ring was cleaned properly with a brush so that the tray could catch all the grains of the soil sample. The tray was placed in an oven for 24 hours at 105°C. After drying, the soil sample were weighed again. The dry soil weight was calculated by subtracting the empty aluminum tray weight from the combined weight of soil and tray after drying. This dry weight was recorded in the HYPROP-Fit program to measure the volumetric water content (METER 2015a). There are several soil hydraulic models available in the HYPROP-Fit program including Brooks-Corey, Fredlund-Xing, Kosugi, and van Genuchten models. The user can use any of these models, but the van Genuchten model is the most popular and the default model in the HYPROP-Fit program (Brooks and Corey 1966; Fredlund and Xing 1994; Van Genuchten 1980; Kosugi 1996). 2.8 Method Validation HYPROP measurements were conducted for both clay loam and sandy loam soil. The results for both soils are shown in figures 2.11 to 2.14 as part of our method validation. After finishing measurements, HYPROP produced one spreadsheet in Excel Worksheet (.xlsx) and five plots in image format (.png). The spreadsheet kept record of all the data measured during the experiment. The plots featured (i) change in tension values obtained from tensio shafts over time, (ii) retention or change in volumetric water content with change in pressure head (pF), (iii) change 30 in unsaturated hydraulic conductivity with pressure head (pF), (iv) change in unsaturated hydraulic conductivity with volumetric water content, and (v) weight loss of undisturbed soil core over time. 2.8.1 Clay loam soil Figure 2.11. Tension curve for a clay loam soil. Tension is in units of hectopascal (1 kPa = 10 hPa) The water tension did not rise immediately at the beginning of the measurements, but rather both tensio shafts showed a very gradual increase and it was difficult to distinguish their pressure difference until about first 3 hours of the measurements (Figure 2.11). At 2 kPa (20 hPa), the tensio shafts were far enough away to determine the hydraulic conductivities. After 6 hours of measurements, the tension values increased more rapidly than the beginning of the measurements and formed two gradually increasing curves. At this stage of gradual increment of 31 curves, tension at the top shaft was higher than the tension at the bottom shaft. Both tension curves kept on increasing and after a day of measurements, the top tensio shaft reached its cavitation phase and air bubbles started entering the tensio shaft. This is the stage where the experiment should have been stopped, but measurements were not stopped, and the experiment was continued. Since the measurements were not stopped, the tension at the bottom shaft kept on rising and it started cavitating after 12 hours. At this point, the soil sample had lost 9% of water over the drying process (Figure 2.12). Since experiment was not stopped automatically using the “search stop point” option of the HYPROP-Fit program, the experiment was stopped by manually finding a stop point before the cavitation phase until where the slope of the tension curve of top tensio shaft kept on increasing without showing any sign of decrease (method 1 explained in section 1.4). Figure 2.12. Volumetric water content vs. pF plot using evaporation method and fitted soil water characteristic curve for a clay loam soil 32 HYPROP-Fit program lets the user choose different models to analyze the HYPROP2 measurements. The default is Van Genuchten model (1980) where volumetric water content of the soil is calculated by the following equations Θ−Θr Se = (2.1) Θs −Θr 1 m and Se = [1+(αh)n ] (2.2) where Se is effective saturation (cm3/cm3), h is pressure head (cm), ϴ is volumetric water content (cm3/cm3; multiply by 100 to get in percentage), ϴr is residual water content (cm3/cm3) and ϴs is saturated water content (cm3/cm3). Additionally, α is the shape parameter related to the inverse of the air entry pressure (cm -1), n is the shape parameters that controls both the bending of the retention curve at the air-entry region and the asymptotic curvature towards the residual water content and m is the additional shape parameter which equals to 1 – (1/n). It is also possible to select 4 other models (Brooks-Corey, Freudland-Xing, Kosugi and Van Genuchten mnvar) in HYPROP2 for analysis, but we used the Van Genuchten (1980) model also known as the traditional constrained Van Genuchten-Mualem model for our study. After running this model by HYPROP-Fit program for a clay loam soil, values for the following parameters were generated as outputs: α = 0.0730, n = 1.661, m = 1- (1/n) = 0.398, Θs = 0.371 and Θr = 0.280 Thus, the equation of volumetric water content for clay loam soil is calculated as 1 0.398 Θ = 0.091 [1+(0.0730∗h)1.661 ] + 0.280 (2.3) HYPROP-Fit program also provides estimates of field capacity and permanent wilting point. Field capacity is estimated at 33 kPa matric potential. The water content at permanent 33 wilting point is estimated at 1500 kPa. The above model provides the water content at 6 kPa (h = 63.1 cm, and pF=1.8), 33 kPa (h = 316.2 cm, and pF=2.5) and 1500 kPa (h= 15848.9 kPa, and pF=4.2) to be 31.3%, 29.1% and 28.1%, respectively. These values are important for irrigation timing, and can be used as an input for hydrological models such as DRAINMOD (Skaggs et al. 2012). 2.8.2 Sandy loam soil At the beginning of the experiment, the water tension at both tensio shafts followed a gradual increase for the first 3 days (Figure 2.13), and both the tensio shafts showed the same tension values during this time. At about 140 hPa (14 kPa), the tensio shafts were far enough away to determine the hydraulic conductivities. After about 4 days of measurements, the tension values increased at a greater slope. Top shaft showed higher tension value than the bottom shaft. The measurements were completed by the failure of upper tensio shaft after more than four days. At this point, the soil sample had lost about 25% of water over the drying process (Figure 2.14). The experiment was stopped by manually finding a stop point on the top tension curve as explained in section 1.4. 34 Figure 2.13. Tension curve for a sandy loam soil. Tension is in units of hectopascal (1 kPa = 10 hPa) 35 Figure 2.14. Volumetric water content vs. pF plot using evaporation method and fitted soil water characteristic curve for a sandy loam soil After collecting the data for the sandy loam soil, HYPROP-Fit program gave parameter values as follows: α = 0.0302, n = 1.975, m = 1- (1/n) = 0.494, Θs = 0.386 and Θr = 0.138 Using the Van Genuchten-Mualem (1980) model, the equation for this soil is calculated as: 1 0.494 Θ = 0.248 [1+(0.0302∗h)1.975] + 0.138 (2.4) This model gives the water content at 6 kPa (h = 63.1 cm, and pF=1.8), 33 kPa (h = 316.2 cm, and pF=2.5) and 1500 kPa (h= 15848.9 kPa, and pF=4.2) to be 31.3%, 29.1% and 28.1%, respectively. 36 2.9 Summary This chapter provides an extensive guideline for using HYPROP2 including collecting an undisturbed soil sample and using the instrument to determine and accurate SWCC. The HYPROP2 manual does not give good instructions on how the samples should be collected and how to saturate those samples in water based on one’s research needs. Also, this article talks about the degassing process using the HYPROP refill unit. Even though the optimal curve is expected from HYPROP2 based on the instrument manual, a suboptimal curve can also be used to determine the SWCC. This method is highly efficient in degassing water and to generate good results. Anyone correctly following this method article should be able to replicate the results obtained in this study. 37 REFERENCES Bezerra-Coelho, Camila R., Luwen Zhuang, Maria C. Barbosa, Miguel Alfaro Soto, and Martinus Th Van Genuchten. 2018. “Further Tests of the HYPROP Evaporation Method for Estimating the Unsaturated Soil Hydraulic Properties.” Journal of Hydrology and Hydromechanics 66(2):161–69. Brooks, Royal Harvard and Arthur T. Corey. 1966. “Properties of Porous Media Affecting Fluid Flow.” Journal of the Irrigation and Drainage Division 92(2):61–90. Fredlund, Delwyn G. and Anqing Xing. 1994. “Equations for the Soil-Water Characteristic Curve.” Canadian Geotechnical Journal 31(4):521–32. Van Genuchten, M. Th. 1980. “A Closed-Form Equation for Predicting the Hydraulic Conductivity of Unsaturated Soils 1.” Soil Science Society of America Journal 44(5):892–98. Kosugi, Ken’ichirou. 1996. “Lognormal Distribution Model for Unsaturated Soil Hydraulic Properties.” Water Resources Research 32(9):2697–2703. Leitinger, Georg, Nikolaus Obojes, and Laurent Lassabatère. 2015. “Assessing Soil Hydraulic Characteristics Using HYPROP and BEST: A Comparison.” in EGU General Assembly Conference Abstracts. Vol. 17. METER. 2015a. HYPROP-Fit User User Manual. 2015-01, 96 pp, UMS GmbH, Gmunder Strasse 8, 81379 Munich, Germany.URL: http://www.meter-group.com/hyprop-2/#support. METER. 2015b. Manual HYPROP. Version 2015-01, 96 pp, UMS GmbH, Gmunder Strasse 37, Munich, Germany. URL: http://www.meter-group.com/hyprop-2/#support. Öztürk, Hasan S., Wolfgang Durner, Amir Haghverdi, and Birgit Walter. 2013. “Comparison of Soil Moisture Retention Characteristics Obtained by the Extended Evaporation Method and the Pressure Plate/Sand Box Apparatus.” P. 3911 in EGU General Assembly Conference Abstracts. Vol. 15. Peters, A. and W. Durner. 2006. “Improved Estimation of Soil Water Retention Characteristics from Hydrostatic Column Experiments.” Water Resources Research 42(11). Peters, Andre. 2013. “Simple Consistent Models for Water Retention and Hydraulic Conductivity in the Complete Moisture Range.” Water Resources Research 49(10):6765–80. Peters, Andre, Sascha C. Iden, and Wolfgang Durner. 2015. “Revisiting the Simplified Evaporation Method: Identification of Hydraulic Functions Considering Vapor, Film and Corner Flow.” Journal of Hydrology 527:531–42. 38 Bezerra-Coelho, Camila R., Luwen Zhuang, Maria C. Barbosa, Miguel Alfaro Soto, and Martinus Th Van Genuchten. 2018. “Further Tests of the HYPROP Evaporation Method for Estimating the Unsaturated Soil Hydraulic Properties.” Journal of Hydrology and Hydromechanics 66(2):161–69. Brooks, Royal Harvard and Arthur T. Corey. 1966. “Properties of Porous Media Affecting Fluid Flow.” Journal of the Irrigation and Drainage Division 92(2):61–90. Fredlund, Delwyn G. and Anqing Xing. 1994. “Equations for the Soil-Water Characteristic Curve.” Canadian Geotechnical Journal 31(4):521–32. Van Genuchten, M. Th. 1980. “A Closed-Form Equation for Predicting the Hydraulic Conductivity of Unsaturated Soils 1.” Soil Science Society of America Journal 44(5):892–98. Kosugi, Ken’ichirou. 1996. “Lognormal Distribution Model for Unsaturated Soil Hydraulic Properties.” Water Resources Research 32(9):2697–2703. Leitinger, Georg, Nikolaus Obojes, and Laurent Lassabatère. 2015. “Assessing Soil Hydraulic Characteristics Using HYPROP and BEST: A Comparison.” in EGU General Assembly Conference Abstracts. Vol. 17. METER. 2015a. HYPROP-Fit User User Manual. 2015-01, 96 pp, UMS GmbH, Gmunder Strasse 8, 81379 Munich, Germany.URL: http://www.meter-group.com/hyprop-2/#support. METER. 2015b. Manual HYPROP. Version 2015-01, 96 pp, UMS GmbH, Gmunder Strasse 37, Munich, Germany. URL: http://www.meter-group.com/hyprop-2/#support. Öztürk, Hasan S., Wolfgang Durner, Amir Haghverdi, and Birgit Walter. 2013. “Comparison of Soil Moisture Retention Characteristics Obtained by the Extended Evaporation Method and the Pressure Plate/Sand Box Apparatus.” P. 3911 in EGU General Assembly Conference Abstracts. Vol. 15. Peters, A. and W. Durner. 2006. “Improved Estimation of Soil Water Retention Characteristics from Hydrostatic Column Experiments.” Water Resources Research 42(11). Peters, Andre. 2013. “Simple Consistent Models for Water Retention and Hydraulic Conductivity in the Complete Moisture Range.” Water Resources Research 49(10):6765–80. Peters, Andre, Sascha C. Iden, and Wolfgang Durner. 2015. “Revisiting the Simplified Evaporation Method: Identification of Hydraulic Functions Considering Vapor, Film and Corner Flow.” Journal of Hydrology 527:531–42. 39 Schelle, H., L. Heise, K. Jänicke, and W. Durner. 2013. “Water Retention Characteristics of Soils over the Whole Moisture Range: A Comparison of Laboratory Methods.” European Journal of Soil Science 64(6):814–21. Schelle, H., S. C. Iden, and W. Durner. 2011. “Combined Transient Method for Determining Soil Hydraulic Properties in a Wide Pressure Head Range.” Soil Science Society of America Journal 75(5):1681–93. Schindler, U. 1980. “Ein Schnellverfahren Zur Messung Der Wasserleitfahigkeit Im Teilgesattigten Boden and Stechzylinderproben.” Archiv Fur Acker-Und Pflanzenbau Und Bodenkunde. Schindler, U., W. Durner, G. von Unold, and L. Mueller. 2010. “Evaporation Method for Measuring Unsaturated Hydraulic Properties of Soils: Extending the Measurement Range.” SOIL SCIENCE SOCIETY OF AMERICA JOURNAL 74(4):1071–83. Schindler, Uwe Georg and Lothar Müller. 2017. “Soil Hydraulic Functions of International Soils Measured with the Extended Evaporation Method (EEM) and the HYPROP Device.” Open Data Journal for Agricultural Research 3. Schindler, Uwe and Lothar Mueller. 2006. “Simplifying the Evaporation Method for Quantifying Soil Hydraulic Properties.” JOURNAL OF PLANT NUTRITION AND SOIL SCIENCE 169(5):623– 29. Skaggs, R. Wayne, M. A. Youssef, and G. M. Chescheir. 2012. “DRAINMOD: Model Use, Calibration, and Validation.” Transactions of the ASABE 55(4):1509–22. Wang, Xiao-Lei, Shuang Huang, J. S. Huang, and Zi-yun Peng. 2012. “Using HYPROP System and centrifugal method to measure soil Water characteristic curve.” China Rural Water and Hydropower 6(4):7. 40 CHAPTER 3: DEVELOPING A STAGE-DISCHARGE EQUATION FOR A V-NOTCH WEIR TO ESTIMATE THE DRAINAGE DISCHARGE 3.1 Abstract A reliable empirical flow equation for V-notch weirs will provide flow estimates that can be used to calculate nutrient loads leaving fields with subsurface drainage. The objective of this study was to develop such an equation for an AgriDrain metal-edge sharp-crest 45° V-notch weir. In this undertaking, we measured flow rate with a combination of the weighing method for low flow and a turbine flow meter for high flow. The head of water (H) was measured inside a 25-cm AgriDrain control structure with a three-step method. First, we measured the water level (a) and height of the control structure (b). Second, we measured the height of the V-notch apex (c). Third, we calculated head using this equation: H=(b-a) – (b-c). Based on the flow meter readings (Q) and H measurements, we developed the following stage-discharge equation: Q = 0.749H 2.25, with Q in liters per minute and H in centimeters. This equation is valid for an H less than the height of the V-notch (i.e., flow through the V-notch) with unsubmerged flow. Based on field experience, we provide a standard procedure for accurate estimation of drainage discharge. In conclusion, the stage-discharge equation developed in this study can provide reliable flow estimates for subsurface drainage studies. 3.2 Introduction Subsurface drainage is a widely used practice that removes excess water from poorly drained soils (King et al., 2014; Konyha et al., 1992; Mourtzinis et al., 2021; Schilling and Helmers, 2008) and is implemented by installing perforated plastic drainage pipes below the soil surface. In addition to removing water, these pipes allow nitrogen and phosphorus to drain out of the 41 soil. These can accumulate in lakes and oceans and cause eutrophication (Fausey et al., 1995; Ghane et al., 2016; King et al., 2015; Pease et al., 2018). Several drainage best management practices (BMPs: controlled drainage, saturated buffer, denitrifying bioreactor, etc.) have the potential to mitigate this issue of nutrient transport. Some of these BMPs require installation of an in-line water-level control structure, which manages drainage discharge at the edge of the field (Christianson et al., 2012; Ghane et al., 2012; Jaynes and Isenhart, 2014; Williams et al., 2015). To assess the effectiveness of the BMPs in reducing nutrient loads, it is necessary to estimate drainage discharge, and an accurate method for doing so is needed. Researchers have used various devices and methods, including weirs, sump pumps, tipping buckets, and ultrasonic flow measurements to estimate flow rate (Chun and Cooke, 2008; Kanwar et al., 1999). Calculating flow rate with a weir is one of the most inexpensive and reliable methods available to assess subsurface drainage systems. Weirs are placed inside a control structure to measure drainage discharge from the field. Researchers use different types of weirs (e.g., rectangular, triangular or V-notch, or trapezoidal) to develop flow equations (Walkowiak, 2006). These equations are mostly empirical and vary based on the design and dimension of the weir. In 2018, AgriDrain introduced a new type of metal-edge sharp-crest V-notch weir. The reason for the new design was to manufacture a standard sharp-crest V-notch weir as opposed to a broad-crest V-notch weir (Huffman et al., 2013). The advantage of a sharp-crest V-notch weir is to prevent the jet of water leaving the weir apex from adhering to any part of the weir's downstream side (Walkowiak, 2006). Instead, the jet of water would spring free of the apex, with 42 minimal contact with the downstream side of the weir. As a result, the new design improves the accuracy of flow measurements. Christianson et al. (2019) calibrated these sharp-crest V-notch weirs for 15- and 25-cm AgriDrain control structures. These dimensions “15- and 25-cm” refer to the diameter of the drainage pipe connected to the control structure. The equation developed from their experiment for flow rate as related to the depth of water from the apex of the V-notch, often called the head of water, is: Q = 0.66H2.28 (3.1) where Q = flow rate (L/min), and H = head of water (cm) Christianson et al. (2019) also calibrated the flow rate measured with these different- sized (15- and 25-cm) V-notch weirs at three different placements inside the control structure: on the bottom of the structure, 48 cm above it, and 97 cm above it. They found no significant differences in the equations they developed to calculate flow rate, regardless of placement or the size of the weir. The authors developed their flow equation with a weighing method of flow measurement. Due to the importance of V-notch equations in subsurface drainage studies, there is a need to verify the V-notch equation in another setting. Also, Christianson et al. (2019) developed their equation for a head of about 14 cm, which does not cover the entire height of a standard metal-edge sharp-crest 45° V-notch weir. The objective of this study was to develop a stage-discharge relationship for an AgriDrain metal-edge sharp-crest 45° V-notch weir. In our calibration method, we used a combination of the weighing method and a flow meter to measure flow rates. The stage-discharge equation that 43 we developed should accurately estimate drainage discharge when water is flowing through the V-notch weir. This equation will aid in evaluating the effectiveness of BMPs in reducing nutrient load. 3.3 Materials and Methods 3.3.1 Commonly used V-notch equations The triangular or V-notch weirs are popular in drainage studies because of their high accuracy at low flow rates (Chanson and Wang, 2013; Haan et al., 1994; Troskolanski, 1960; USGS, 1982). The estimation of flow through V-notch weir follows a stage-discharge relationship. The discharge is directly proportional to the head. The discharge equation for a sharp-crest V-notch weir with an apex angle ϴ is written as (World Meteorological Organization, 1971): 8 Θ Q= Cd (2g)0.5 tan ( 2 ) H2.5 (3.2) 15 where Q = flow rate (L/min), H = head of water (cm), g = acceleration due to gravity, and Cd = discharge coefficient. For a triangular sharp-crest weir with an angle of 90°, Cd is assumed to be 0.61 (Bijankhan and Ferro, 2017). A simplified equation was developed (USGS, 1982) to calculate flow rate through thin- plate V-notch weirs when the angle of V-notch was less than or equal to 90°. For a metal-edge sharp-crest 45° V-notch weir, the United States Geological Survey (USGS) equation is: Q = 0.343H2.5 (3.3) 44 where Q = flow rate (L/min), and H = head of water (cm) Since the area of the V-notch weir is quite small compared to the cross-sectional area of the channel, it is safe to neglect the approach velocity for V-notch angles of 90° or less (USGS, 1982). While this may be true for open channel flow but may not apply to the subsurface drainage control structures. Depending on the diameter of the pipe and the slope leading into the control structure, the approach velocity can be enough to create turbulence inside the structure. Another study suggested that if two criteria are satisfied, it is acceptable to ignore the approach velocity in a V-notch weir (World Meteorological Organization, 1971). Those criteria are met if H/P < 0.4 (where H is the head of water, P is the distance from the bottom of the structure to the apex of the V-notch weir) and H/B < 0.2 (where B is the width of the V-notch board). These criteria may be invalid in subsurface drainage control structures, so we cannot use the USGS equation. Therefore, there is a need to calibrate the V-notch weir inside a control structure. In our experiment, we fitted flow and head data to a simplified version of equation 3.2. This equation is written as: Q = aHb (3.4) where a and b are fitted parameters from a regression analysis. 3.3.2 Experimental setup and flow measurement Our experiment to develop a flow equation was conducted in September 2021 near a pond located south campus of Michigan State University. The setup is shown in figure 3.1 (Supplementary information: figure S1). We used a Flomec GPI turbine flow meter (model 45 number: TM20NQ9GMB) with a flow measurement range of 75-757 L/min (20-200 gallons/min, with a ±3% accuracy). This flow meter was factory calibrated by the manufacturer. It is important to note that the experimental flow range (16-409 L/min) from this study was determined based on experience and literature review. For a 25-cm control structure, 17 cm is the maximum head that can be obtained when water is flowing through an Agri Drain metal- edge sharp-crest 45° V-notch weir. Therefore, this experimental flow range mentioned above approximately covers the entire flow range through a V-notch weir. For the lower flow rates (16-59 L/min), we used a weighing method. In this method, we held a 5-gallon bucket at the outlet of the control structure, recorded the time required to collect a specific volume of water in the bucket, weighed the bucket on a portable scale (Brecknell PS- USB), then calculated the flow rate. We took three weight readings (triplicate readings) for each flow rate and used their mean value as our final flow reading. It is important to note that the weighing scale was calibrated with known weights before and after the experiment. For the higher flow rates (105-409 L/min), we used the Flomec GPI turbine flow meter mentioned above. The upstream chamber of the control structure was connected to a 25-cm (10 inches) SDR-35 pipe. We reduced the pipe size to 5 cm (2 inches) to accommodate the flow meter. We pumped (Pacer pumps, model number: SEB2ULE51C) water out of a pond with a flexible PVC pipe. The benefit of using a pond as the water source is that the water level in the pond does not drop as compared to a water tank, which reduces the fluctuation of the pumped water. Nevertheless, we still saw minor fluctuation in the pumped water, as is expected with the pump. To reduce the effect of flow fluctuation, we took numerous (at least 4) flow meter readings for each measurement and took an average of those readings as our final flow. 46 The discharge side of the pump was connected to a 100-ft long blue lay flat discharge hose, which supplied water to the flow meter. The V-notch weir (figure 3.2) was placed on top of a 17.8-cm tall (7-in; height does not include the thickness of compressed rubber gasket) bottom board inside the control structure. It is important to note that there was a minor difference (17.0 vs. 16.5 cm) in the height of the V-notch in our experiment compared to the one used by Christianson et al. (2019), even though both are standard V-notch weirs from the same manufacturer. The total height and angle of the V-notch board were the same for both studies. Therefore, any calibrated stage-discharge equation is still valid when the angle of the sharp-crest V-notch weir is the same. 47 Figure 3.1. Top: Diagram of the experimental setup showing water flow through a V-notch weir. Bottom: Sideview photo of the control structure with flowing water [From Shokrana and Ghane (2021). Used with permission.] 48 Apex Figure 3.2. Left: Schematic diagram of a metal-edge sharp-crest 45° V-notch weir. The apex is the point at which both inclined sides of the metal crest meet. The dimensions of the V- notch weir do not include the thickness of the rubber gaskets attached to the board. Right: Photo of the metal-edge sharp-crest 45° V-notch weir [From Shokrana and Ghane (2021). Used with permission.] 3.3.3 Head measurement procedure Phase 1. Water-level and measurement of the height of the structure To determine the head of water inside the V-notch weir, we put a PVC pipe along the wall of the control structure and made holes in it so that water can seep through that hole and maintain the same head as the control structure inside that pipe, but with less instability. Afterwards, we lowered a water-depth sensor (model 101 P2, Solinst Canada Ltd.) inside the PVC pipe to measure the head for each flow reading obtained from the flow meter and the weighing method (figure 3.3). This water-depth sensor provided the distance from the top of the control structure to the upstream water surface (distance “a”). We also measured the distance from the top to the bottom of the control structure (distance “b”) with a tape measure. 49 Figure 3.3. Diagram of the side view of a control structure and the distances needed to calculate the head (H) flowing through a V-notch weir using equation (3.5). A Solinst water-depth sensor was used to measure the distance "a" from the top of the structure to the upstream water surface and “f” from the top of the structure to the downstream water surface. A tape measure was used to measure the distance “b” inside the structure, and a meter-stick was used to measure the distance “c” from the top of the structure to the apex of the V-notch weir. Distance “d” was calculated by subtracting “a” from “b” [From Shokrana and Ghane (2021). Used with permission.] Phase 2. Measurement of the height of the apex of the V-notch and the procedure for its verification We measured the height of the apex of the V-notch using a top-down approach. In this approach, we measured the distance from the top of the control structure to the apex of the V- notch (distance “c” in figure 3.3) with a meter-stick (figure 3.4). Then, the distance from the top of the structure to the apex of the V-notch was measured by placing another ruler on the top and reading the value on the vertical meter stick (figure 3.4 (b)). The height of the V-notch apex was calculated by estimating the distance “b-c”, shown in figure 3.3. 50 Figure 3.4. Measurement of the distance “c” from the top of the control structure to the apex of the V-notch weir. Multiple meter-sticks were glued and riveted together to make a long one. To measure the distance “c”: (a) We lowered the meter-stick inside the apex of the V-notch weir, and (b) next, we measured the distance from the apex to the top of the structure by placing another ruler on the top and reading the value on the vertical meter stick [From Shokrana and Ghane (2021). Used with permission.] Phase 3. Equation for head calculation Based on the measured distances in phases 1 and 2 (figure 3.3), head was calculated as: H = (b − a) − (b − c) (3.5) where a = distance from the top of the control structure to the upstream water surface (cm), b = distance from the top of the control structure to the bottom (the height of the structure, cm), c = distance from the top of the control structure to the apex of the V-notch weir (cm), and H = head of water (cm) 3.3.4 Quality control measures We ensured high-quality data collection by taking cautionary steps throughout the experiment. These are listed below. 1. According to the manufacturer’s user manual, we maintained the optimum pipe length for the turbine flow meter. Based on the turbine flow meter manual, the upstream and 51 downstream pipe lengths of the flow meter should be at least ten (10) times and five (5) times the diameter of the turbine (i.e., 5 cm), respectively. In our experiment, the upstream pipe length of the flow meter was 101.6 cm (20 times the turbine’s diameter) and the downstream pipe length was 50.8 cm (10 times the turbine’s diameter). 2. We checked the level of the whole setup before and during the experiment, including the control structure and the pipes, to ensure that the setup was horizontal. 3. We made sure that the turbine of the flow meter was not clogged with algae. We disconnected the flow meter from the pipes before, during, and after the experiment. Then, we visually checked for algae, and we did not find any algae in the turbine of the flow meter. 3.3.5 Developing the empirical flow equation with regression analysis We used a least squares regression to develop the empirical flow equation for this experiment. Measured flow rates were plotted against the corresponding head measurements, and the coefficient of determination (R-squared) was used to evaluate the goodness of fit. 3.4 Results and Discussion 3.4.1 V-notch apex height measurement using top-down approach We measured the distances “b” and “c” (figure 3.3) with the top-down approach. Distance “b” was 149.8 cm and distance “c” was 118.1 cm, so that the distance “b-c” was 31.7 cm, which is the height of the V-notch apex, that is, the distance from the structure bottom to the V-notch apex, as measured with the top-down approach. The distance “b-c” was constant for all head measurements. It is important to mention that the distance “b-c” represented the field condition 52 where the V-notch board was placed on top of a 17.8-cm (7-in) bottom board, and the rubber gaskets attached to those boards were compressed. 3.4.2 Calibrated V-notch weir equation The least squares regression showed a strong relationship (R-square= 0.999) between head and flow rate (figure 3.5). The empirical stage-discharge relationship equation is written as: Q = 0.749H2.25 (3.6) where Q = flow rate (L/min), and H = head of water (cm) Figure 3.5. Calibration equation for a metal-edge sharp-crest V-notch weir. We developed this equation for a maximum flow rate of 409.42 L/min with a head of 16.3 cm. We used the weighing method for the low flow rates (16-59 L/min) and the flow meter for the higher flow rates (105-409 L/min). This equation is valid for an H less than the height of the V-notch (i.e., flow through the V-notch). In our experiment, the height of the V-notch was 17.0 cm [From Shokrana and Ghane (2021). Used with permission.] 53 Christianson et al. (2019) tested their V-notch flow equation for two different control structure sizes (15 cm and 25 cm). The authors concluded that the V-notch flow equation could be used for other control structures, regardless of their size. They also tested their equation by placing the V-notch weir at three different heights inside each of the different sized structures. They found no statistically significant difference in the flow equations among three different V- notch weir placements. Consequently, we expect that our empirical equation will be valid for structures of different sizes when the same AgriDrain metal-edge sharp-crest 45° V-notch weir is used. 3.4.3 Comparison with previously reported V-notch weir equations We compared the V-notch equation of this study to that of previous studies (figure 3.6). The USGS (1982) equation (eq. 3.3) underestimated the flow rates for all the head measurements. This may be explained by the differences in the scope for application of these two equations. The equation developed in this study can estimate flow rates for 45° V-notch weirs inside a 25-cm AgriDrain control structure, whereas the USGS equation is used in different applications in an open-channel flow. 54 Figure 3.6. Comparison of the V-notch equation developed in this study to those in previous studies [From Shokrana and Ghane (2021). Used with permission.] We developed our equation for a maximum flow rate of 409.42 L/min with a head of 16.3 cm. Our equation is valid for an H less than the height of the V-notch (i.e., flow through the V- notch). In our experiment, the height of the V-notch was 17.0 cm. The flow rates obtained from the equation developed in this study (eq. 6) are consistently higher than the equation developed by Christianson et al. (2019) (figure 3.6). Christianson et al. (2019) used one method (weighing) of flow rate reading to develop their equation, but they did not calibrate and verify their weighing scale as reported in Christianson et al. (2021). Figure 3.6 shows that the equation from Christianson et al. (2019) underestimates the higher flow rates compared to the equation obtained in our study. This could be because their weighing scale was not calibrated and verified. 55 To assure the reliability of our equation (eq. 3.6), we obtained stage-discharge relationship data from a separate experiment conducted in Farrall Hall at Michigan State University. In that experiment, we used two turbine flow meters (one for low flow and one for high flow), different from the one used to develop equation 3.6. Comparison between that experiment (two flow meters) and the one conducted herein shows that stage-discharge data closely agreed with each other (Supplementary information: figure S2). Therefore, the two- turbine experiment verifies the stage-discharge equation 3.6. For more information, refer to the supplementary information. 3.4.4 Standard procedure for accurately measuring flow rate Based on our combined results and experiences, we provided a standard procedure for accurately estimating drainage discharge. These steps can be used for rectangular or cylindrical structures. Step 1: Measure the height of the control structure (distance “b” in figure 3.3) Measure the distance from the top to the bottom of the control structure with a tape measure, making sure that the tape measure is in contact with the structure wall so that it doesn’t bend. Step 2: Measure the depth of the V-notch apex (distance “c” in figure 3.3) Measure the distance from the top of the control structure to the V-notch apex, distance “c”, by first placing a long, thin measuring stick inside the apex. Then, carefully measure distance “c” by placing another ruler on the top of the control structure and reading the value on the vertical meter stick (figure 3.4). 56 Step 3: Measure the upstream water-level from top or bottom of the structure (distance “a” or “d” in figure 3.3, respectively) There are at least two methods for measuring the water level from the bottom of the control structure (distance “d” in figure 3.3): a pressure transducer or a water-finding paste. The pressure transducer method is useful for measuring the water level when continuous flow rates will be measured. The pressure transducer is placed inside a polyvinyl chloride (PVC) pipe and lowered to the bottom of the upstream chamber of the structure. The PVC pipe reduces the effects of flow turbulence on the pressure transducer. To ensure accurate water-level measurements, the pressure transducer reading is verified with manual water-level measurements regularly (e.g., with water-finding paste) and sediments are removed from the pressure transducer as they accumulate. The second method is useful for measuring water level when the instantaneous flow rate will be measured. This method requires a meter stick and water-finding paste. The water-finding paste is applied on a meter stick and then slowly lowered to the bottom of the control structure. The meter stick is lowered along the wall of the upstream chamber of the control structure to ensure that it does not bend. The paste will turn red upon contact with water (figure 3.7). 57 Figure 3.7. Measurement of water level with water-finding paste. A small amount of paste is applied to a meter stick, which is inserted into water. The dry paste is white but turns red upon contact with water [From Shokrana and Ghane (2021). Used with permission.] Various water-depth sensors are available to measure the water depth from the top of the structure (distance “a” in figure 3.3) as performed in this study. Some of them need to contact water to measure the depth (figure 3.8 (a)). These types of sensors can be acquired from Solinst, Geotech, and Heron, Inc. Others are contactless, for example, an ultrasonic level sensor. Both methods are used when measuring instantaneous flow rate. The sensors that need to contact water for depth measurements (e.g., the one from Solinst) usually consist of a probe connected to a flat tape measure. When using this type of sensor, the probe is lowered into the upstream chamber of the control structure near the wall (figure 3.8 (b)). Once the probe meets the water surface, the unit will emit both sound and light signals. At that time, the value on the flat tape, read by looking at the marked numbers vertically, is an estimate of the water depth inside the control structure, distance “a” in figure 3.3. Another sensor for measuring water depth from the top of the structure is an ultrasonic sensor, which is currently being investigated by researchers to determine if it works with the AgriDrain control structures. 58 Figure 3.8. Measurement of water level using a water-depth sensor: (a) Solinst water-depth sensor; the sensor is located at the middle of the probe. Once the sensor comes in contact with water, sound and light signals will indicate when to take the reading. (b) The sensor probe is lowered inside the control structure through a PVC pipe to reduce the effect of flow turbulence. Once in place, the length of the flat tape along the wall of the control structure indicates the water depth [From Shokrana and Ghane (2021). Used with permission.] Step 4: Measure the downstream water-level from structure top or bottom (distance “f” or “e” in figure 3.3) It is important to measure the water level in the downstream chamber of the control structure to ensure flow impedance (i.e., submerged flow) does not occur. This means that the water level in the downstream chamber must not rise above the V-notch apex. The downstream water depth can be measured from the structure top (distance “b-f”) using a water-depth meter. It is also possible to measure the downstream water level from the bottom (distance “e”) with a pressure transducer or water-finding paste. The details about these measurement procedures are explained in step 3. Our calibrated equation will not be valid under submerged flow conditions. Step 5: Calculate head inside the V-notch weir To calculate head inside the structure when the height of the water level in the upstream chamber (distance “d” in figure 3.3) can be identified from a pressure transducer or from the water-finding-paste method, use the following equation: 59 H = d − (b − c) (3.7) To calculate head when the depth of water in the upstream chamber (distance “a” in figure 3.3) can be determined from a water-depth meter, use equation 3.5. Step 6: Calculate flow rate Use equation 3.6 to calculate flow rate based on a measured head when the following two conditions are met: (1) water must be flowing through the V-notch weir and not flowing over the weir (that is, H < 17.0 m for this study) and (2) the height of the water level in the downstream chamber (distance “e” in figure 3.3) must not exceed the V-notch apex (distance “b-c” in figure 3.3), so that submerged flow is avoided. 3.4.5 Things to avoid when following the standard procedure 3.4.5.1 Avoid adding nominal board sizes to measure height of V-notch apex A common method of estimating the height of the V-notch apex (distance “b-c” in figure 3.3) is to add the nominal sizes of boards, which could introduce errors. When adding nominal board sizes, the thickness of the rubber gaskets will be neglected. To demonstrate this point, we estimated the height of the V-notch apex by adding two nominal 5-in boards, one nominal 7-in board, one nominal 7-in bottom-board, and a standard metal-edge sharp-crest 45° V-notch board. Based on adding nominal board sizes, the height of the V-notch apex was 74.5 cm. Based on the top-down approach, we determined the height of the V-notch apex to be 74.9 cm; there is a difference of 0.4 cm between the two measurements, perhaps due to the thickness of the rubber gaskets, or variations between the nominal and actual board sizes. If not considered, these factors could result in an error in the measurement of the V-notch apex height. At the end of 2013, AgriDrain replaced the traditional boards with foam gaskets (or foam O-rings) with the 60 injection-molded boards with rubber gaskets. The compressed thickness of the new rubber gaskets may not be the same as the foam ones. This error would be even greater if more boards were used inside the control structure. Therefore, nominal board sizes should not be used to estimate the height of the V-notch apex (distance “b-c” in figure 3.3), unless the height of each board is carefully measured, and the thickness of compressed rubber gaskets is included. 3.4.5.2 Avoid using a V-notch weir board without a bottom board During heavy rainfall, the water level in a ditch may rise and create outlet submergence (figure 3.9). In this situation, the water level in the downstream chamber of the structure can rise above the V-notch apex, thus preventing the use of equation 3.6 because water is no longer freely flowing through the V-notch. However, if a bottom board is used, the equation 3.6 can be used to calculate flow as long as the water level in the downstream chamber remains below the V- notch apex. If the downstream water level rises higher, equation 3.6 cannot be used. Overall, the benefit of using a bottom board is that it gives extra height to the apex of the V-notch weir, so there is a greater chance of achieving freely flowing water through the V-notch. 61 Figure 3.9. A diagram (not drawn to scale) to represent the water flow inside a 25-cm control structure in agricultural fields. Top diagram: If there is heavy rainfall, the water level in the ditch will rise and the water level in the downstream chamber of the control structure will rise above the apex of the V-notch, causing submerged flow. Bottom diagram: Adding a 17.8 cm (7- inch) bottom board below the V-notch weir would give extra height to the apex of the V-notch weir, so there would be a greater chance of achieving freely flowing water through the V-notch [From Shokrana and Ghane (2021). Used with permission.] 62 3.5 Conclusions We measured head and flow rate for a metal-edge sharp-crest 45° V-notch weir inside a 25- cm AgriDrain control structure. Based on those measurements, we developed an empirical stage- discharge equation while following quality-control procedures. In our calibration method, we used a weighing method for the lower flow rates and a turbine flow meter for the higher flow rates. The empirical metal-edge sharp-crest 45° V-notch weir equation can be used for any AgriDrain control structure, as long as the water is flowing through the V-notch weir and there is no flow submergence. If using a cylindrical control structure, we recommend developing a stage- discharge equation before drainage monitoring. Based on field experience, we provide a standard procedure to accurately estimate drainage discharge. In conclusion, the V-notch weir equation developed in this study can provide reliable flow estimates that can be used for nutrient load reductions in subsurface drainage studies. 63 REFERENCES Bijankhan, M., Ferro, V. (2017). Dimensional analysis and stage-discharge relationship for weirs: a review. Journal of Agricultural Engineering, 48(1), 1–11. https://doi.org/10.4081/jae.2017.575 Chanson, H., Wang, H. (2013). Unsteady discharge calibration of a large V-notch weir. Flow Measurement and Instrumentation, 29, 19–24. https://doi.org/10.1016/j.flowmeasinst.2012.10.010 Christianson, L., Bhandari, A., Helmers, M., Kult, K., Sutphin, T., Wolf, R. (2012). Performance evaluation of four field-scale agricultural drainage denitrification bioreactors in Iowa. Transactions of the ASABE, 55(6), 2163–2174. https://doi.org/10.13031/2013.42508 Christianson, L. E., Christianson, R. D., Lipka, A. E., Bailey, S. B., Chandrasoma, J., McCoy, C., Fontes, G. P., Roh, J., Bailon, A. P. S. B., Wickramarathne, N. M., Cooke, R. A. (2019). Calibration of stainless steel-edged v-notch weir stop logs for water level control structures. Applied Engineering in Agriculture, 35(5), 745–749. https://doi.org/10.13031/aea.13350 Chun, J. A., Cooke, R. A. (2008). Calibrating agridrain water level control structures using generalized weir and orifice equations. Applied Engineering in Agriculture, 24(5), 595–602. https://doi.org/10.13031/2013.25274 Fausey, N. R., Brown, L. C., Belcher, H. W., Kanwar, R. S. (1995). Drainage and Water-quality in Great- Lakes and Corn-belt States. Journal of Irrigation and Drainage Monitoring-ASCE, 121(4), 283– 288. https://doi.org/10.1061/(ASCE)0733-9437(1995)121:4(283) Ghane, E., Fausey, N. R., Shedekar, V. S., Piepho, H. P., Shang, Y., Brown, L. C. (2012). Crop yield evaluation under controlled drainage in Ohio, United States. Journal of Soil and Water Conservation, 67(6), 465–473. https://doi.org/10.2489/jswc.67.6.465 Ghane, E., Ranaivoson, A. Z., Feyereisen, G. W., Rosen, C. J., Moncrief, J. F. (2016). Comparison of Contaminant Transport in Agricultural Drainage Water and Urban Stormwater Runoff. Plos One, 11(12). https://doi.org/10.1371/journal.pone.0167834 Haan, C. T., Barfield, B. J., Hayes, J. C. (1994). Design hydrology and sedimentology for small catchments. Elsevier. Jaynes, D. B., Isenhart, T. M. (2014). Reconnecting Tile Drainage to Riparian Buffer Hydrology for Enhanced Nitrate Removal. Journal of Environment Quality, 43(2), 631–638. https://doi.org/10.2134/jeq2013.08.0331 Kanwar, R. S., Bjorneberg, D., Baker, D. (1999). An automated system for monitoring the quality and quantity of subsurface drain flow. Journal of Agricultural Engineering Research, 73(2), 123– 129. https://doi.org/10.1006/jaer.1998.0398 64 King, K. W., Fausey, N. R., Williams, M. R. (2014). Effect of subsurface drainage on streamflow in an agricultural headwater watershed. Journal of Hydrology, 519(A), 438–445. https://doi.org/10.1016/j.jhydrol.2014.07.035 King, K. W., Williams, M. R., Fausey, N. R. (2015). Contributions of Systematic Tile Drainage to Watershed-Scale Phosphorus Transport. Journal of Environmental Quality, 44(2), 486–494. https://doi.org/10.2134/jeq2014.04.0149 Konyha, K. D., Skaggs, R. W., Gilliam, J. W. (1992). Effects of Drainage and Water-management Practices on Hydrology. Journal of Irrigation and Drainage Engineering, 118(5), 807–819. https://doi.org/10.1061/(ASCE)0733-9437(1992)118:5(807) Mourtzinis, S., Andrade, J. F., Grassini, P., Edreira, J. I. R., Kandel, H., Naeve, S., Nelson, K. A., Helmers, M., Conley, S. P. (2021). Assessing benefits of artificial drainage on soybean yield in the North Central US region. Agricultural Water Management, 243. https://doi.org/10.1016/j.agwat.2020.106425 Pease, L. A., King, K. W., Williams, M. R., LaBarge, G. A., Duncan, E. W., Fausey, N. R. (2018). Phosphorus export from artificially drained fields across the Eastern Corn Belt. Journal of Great Lakes Research, 44(1), 43–53. https://doi.org/10.1016/j.jglr.2017.11.009 Schilling, K. E., Helmers, M. (2008). Effects of subsurface drainage tiles on streamflow in Iowa agricultural watersheds: Exploratory hydrograph analysis. Hydrological Processes, 22(23), 4497–4506. https://doi.org/10.1002/hyp.7052 Troskolanski, A. T. (1960). Hydrometry: Theory and practice of hydraulic measurements. Pergamon Press. USGS. (1982). Measurement and computation of streamflow, vol. 2, Computation of discharge. US Geological Survey Water Supply Paper 2175, Page 32. Walkowiak, D. K. (2006). ISCO open channel flow measurement handbook. Williams, M. R., King, K. W., Fausey, N. R. (2015). Drainage water management effects on tile discharge and water quality. Agricultural Water Management, 148, 43–51. https://doi.org/10.1016/j.agwat.2014.09.017 World Meteorological Organization. (1971). Use of Weirs and Flumes in Stream Gauging, WMO Technical Note No. 117. 65 CHAPTER 4: CALIBRATING THE RZWQM2-P MODEL TO SIMULATE DRAINAGE DISCHARGE AND PHOSPHORUS LOSS IN CLAY LOAM SOIL IN MICHIGAN 4.1 Abstract Phosphorus (P) loss and transport through subsurface drainage systems is a primary focus for addressing harmful algal blooms in freshwater systems. The recent development of the phosphorus (P) routine of the Root Zone Water Quality Model (RZWQM2-P) has the potential to enhance our understanding of the fate and transport of P from subsurface-drained fields to surface water. However, there is a need to test the model under different fertilization, soil, climate, and cropping conditions. The objective of this study was to test the model's performance with daily drainage discharge, dissolved reactive phosphorus (DRP), and total phosphorus (TP) load collected from a subsurface-drained field with a clay loam soil. We calibrated RZWQM2-P using two years of measured data. Subsequently, we validated RZWQM2-P using a year and nine months of measured data. We used the Nash-Sutcliffe model efficiency (NSE) and percentage bias (PBIAS) statistics for the RZWQM2-P model evaluation. Results showed that the model performance was “good” (daily NSE = 0.66 and PBIAS = -7.16) in predicting hydrology for the calibration period. For the validation period, the hydrology prediction of the model was “very good” (daily NSE = 0.76), but it had a “satisfactory” underestimation bias (PBIAS = 23.57). The model’s performance was “unsatisfactory” in simulating DRP for both calibration (daily NSE = 0.31 and PBIAS = -61.50) and validation (daily NSE = 0.32 and PBIAS = 43.68) periods. The P model showed “satisfactory” performance in predicting TP load for both calibration (daily NSE = 0.46 and PBIAS = -32.41) and validation (daily NSE = 0.39 and PBIAS = 42.90) periods, although both periods showed “unsatisfactory” percent bias. The underperformance may have been due to the model’s inability to partition fertilizer P into different P pools under high water table or ponding 66 conditions when using daily data. In conclusion, the RZWQM2-P model performed well for drainage discharge with daily data, but further investigation is needed to improve the P component of the model. 4.2 Introduction Phosphorus (P) is extensively used in agriculture in the forms of fertilizer and manure to facilitate crop production and high yields (Sharpley et al., 2001). Precipitation and irrigation events cause significant P loss from agricultural lands to surface water, thus making it one of the major non-point sources of eutrophication in freshwater bodies (Dubrovsky et al., 2010; Kleinman et al., 2011). Eutrophication stimulates harmful algal blooms (HABs) in surface water which poses severe ecological, economic, and health concerns (Pierzynski et al., 2005). Although nitrogen (N) and P play a significant role in causing HABs in lakes and oceans, P is considered to be the limiting nutrient freshwater systems (Sharpley et al., 1994). In agricultural landscapes, P loss to surface water occurs in two pathways: surface runoff and subsurface drainage discharge (Ghane et al., 2016). P can be lost from both pathways in dissolved (i.e., dissolved reactive phosphorus, DRP) and particulate (i.e., particulate phosphorus, PP) forms, but DRP is considered to be the primary driver of HABs in freshwater systems due to its bioavailability (Macrae et al., 2021). While the current knowledge can efficiently explain P loss pathways (Pierzynski et al., 2005), it was not always the case. Due to the high P sorption capacity of agricultural soils, P was largely considered immobile (Baker et al., 1975; Radcliffe & Cabrera, 2006). Therefore, P transport via surface runoff was seen as the primary pathway, and research was focused on developing management strategies to reduce soil erosion (Ryden et al., 1974). However, in the 1990s, Sims et al. (1998) found that a significant amount of P can be lost via 67 subsurface drainage discharge, and they suggested developing management strategies that reduce subsurface transport of P. Although the installation of subsurface drainage increases the total drained water volume from a field, it significantly reduces surface runoff volume and sediment loss compared to an undrained field (Dolezal et al., 2001; Robinson & Rycroft, 1999; Skaggs et al., 1994). While the installation of subsurface drains reduces P loss in surface runoff, their role in increasing P loss from subsurface pathways cannot be overlooked (Bengston et al., 1995; King et al., 2015). Long-term field experiments employing management practices can investigate P loss from subsurface drainage discharge, but they are expensive and time-consuming. On the other hand, field-scale hydrologic models are inexpensive, and they provide results in a shorter time than field experiments. Models can use field experimental data as inputs and extrapolate the results across spatial and temporal scales. Over the years, significant progress has been made in developing P modules to incorporate into both field-scale and watershed-scale hydrologic models. A robust and sophisticated P model should be able to predict: (i) DRP and PP loss from both surface and subsurface pathways, (ii) P loss through macropores, (iii) plant uptake P, (iv) P transformation in soil, and (v) the effects of management practices on P loss. Although many P models exist, only a few of those can fulfill the criterion mentioned above. Askar et al., (2021) developed and tested a P module for DRAINMOD known as DRAINMOD-P that fulfills the requirements mentioned above. The authors tested their model in a site in Ohio for clay soil, which showed promising results in predicting P loss from subsurface- drained fields. In addition to developing the P module, the authors modified the macropore flow 68 component of DRAINMOD using the Hagen-Poiseuille law to accurately estimate the water and P transport through preferential flow pathways. A comprehensive literature review on P models (Qi & Qi, 2016; Radcliffe et al., 2015) suggested that ICECREAM is the most promising model to accurately predict P loss from surface runoff and subsurface drainage discharge (Tattari et al., 2001). However, ICECREAM does not have a water table based tile drainage component as estimated by Hooghoudt’s equation (Hooghoudt, 1940; Smedema et al., 2004). Instead, it estimates the sum of the water flux from the macropore and micropore at the tile drain depth to imitate tile drainage (Qi & Qi, 2016; Sadhukhan et al., 2019a). Additionally, the storage routing concept that ICECREAM uses cannot simulate upward flow, which is a problem for soils with a shallow water table (Pferdmenges et al., 2020). ICECREAM also simulates the soil matrix flow using the simple storage routing concept, but this can be improved by implementing the Richards equation (Pachepsky et al., 2003). To improve the limitations of the ICECREAM model mentioned above, Sadhukhan et al. (2019a) have developed a P module for the Root Zone Water Quality Model (RZWQM2). RZWQM2 is a field-scale process-based model that uses more robust and sophisticated approaches than ICECREAM to simulate subsurface drainage discharge from an agricultural field (Ma et al., 2011; Ma et al., 2012). RZWQM2 has been widely tested and validated across North America (Ahmed et al., 2007; Jiang et al., 2018; Ma et al., 2007; Malone et al., 2014; Thorp et al., 2008). The RZWQM2-P model performs as a single tool to simulate hydrology and P loss through surface runoff and subsurface drainage discharge pathways. Although RZWQM2-P has been tested and validated by the developers (Sadhukhan et al., 2019a), further testing is needed to evaluate the model’s performance under different soil types, 69 climate, management practices, and crop varieties. The objective of this study is to test and validate the performance of RZWQM2-P model in predicting DRP and PP loss in a clay loam soil. The outcome of this study will help RZWQM2-P model developers identify processes and subroutines within the model that need to be improved for predicting P loss from a subsurface- drained field. 4.3 Materials and Methods 4.3.1 Overview of RZWQM2-P The RZWQM2-P (version 4.2) is a field-scale, process-based, and one-dimensional (i.e., vertical in the soil profile) model that can simulate major physical, chemical, and biological processes occurring in an agricultural field. The RZWQM2-P combines two distinct models: (i) the RZWQM2 model and (ii) the P model (Sadhukhan et al., 2019a). The RZWQM2 model was first developed and then further improved by the USDA-ARS to simulate soil hydrological processes, crop production, and water quality effects under different agricultural management practices (Ahuja et al., 2000). The P routine has been recently developed and integrated with the RZWQM2 to simulate P loss through hydrologic pathways (Sadhukhan et al., 2019a). The RZWQM2 uses several mathematical equations to simulate the physical processes in the model (Ma et al., 2011; Ma et al., 2012). The modified forms of Brooks-Corey equations are used to describe the soil water characteristic curve (Brooks & Corey, 1964). The infiltration process due to rainfall, snowmelt, or irrigation event is described by the Green-Ampt equation (Green & Ampt, 1911). Richards equation is used to simulate the redistribution of water following infiltration (Richards., 1931). Subsurface drainage discharge is simulated using the Hooghoudt’s steady-state equation (Bouwer & Van Schilfgaarde, 1963; Smedema et al., 2004) and Poiseuille’s 70 law is used to simulate the macropore flow. Daily potential evapotranspiration (PET) is estimated using the Shuttleworth-Wallace equation (Farahani & DeCoursey, 2000). Crop growth can be simulated using three crop growth simulation models which are: (i) the generic crop growth model (Hanson, 2000), (ii) the DSSAT crop growth model (Nielsen et al., 2002), and (iii) the HERMES crop growth model (Malone et al., 2017). For the simulation of P, RZWQM2-P separates the P cycle into five different soil P pools. Three of those are inorganic and two organic P pools (Jones et al., 1984). The inorganic P pools comprise the labile P pool, active inorganic P pool, and stable inorganic P pool. Among these inorganic P pools, the labile P pool is in dissolved form and it is the only pool from which P is available for plant uptake. The labile P pool is in rapid equilibrium with the active inorganic P pool, where the P is present in solid form but easily released to the soil solution. A slow adsorption and desorption process occurs between the active inorganic P and stable inorganic P. The P present in the inorganic P pool is very insoluble, therefore this pool maintains a slow equilibrium with the active inorganic P pool. The two organic P pools are the fresh organic P pool and the stable organic P pool. Mineralized P from the fresh organic P pool is added to the labile P pool and stable organic P pool. A slowly mineralized P is also added from the stable organic P pool to the labile P pool. The P model can also simulate P dynamics from manure and fertilizer application by creating four surface manure P pools and two surface fertilizer P pools.. More details on how the P model works are described by Sadhukhan et al. (2019a). 4.3.2 Site description The on-farm monitoring site is located near Blissfield, Michigan (figure 4.1). The grade of the field is 0.1%. The field area is 6.7 hectares (16.6 Acres). The soil type is Ziegenfuss clay loam, classified 71 as a poorly drained soil. The farmer uses a corn-soybean rotation in this field and applies commercial inorganic fertilizer. The date of cropping and management practices are shown in Error! Reference source not found.. The depth of subsurface drains is about 81 cm, and the s pacing is about 1005 cm. The effective radius of the four-row perforated drain pipe is estimated to be 0.7 cm (Ghane., 2022). The average soil test phosphorus (STP) concentration was 35 ppm (Bray-P) on 18 March 2018. 72 Figure 4.1. (a) geographic location of the Blissfield site. This site is part of the River Raisin watershed which directly discharges into the western basin of Lake Erie at Monroe Harbor (b) drainage layout of the study site [From Shokrana et al. (2022). Used with permission.] 73 Table 4.1. Annual management practices adopted at the Blissfield site [From Shokrana et al. (2022). Used with permission.] Year Date Management Comments practices 2018 18-May Soybean 150,000-175,000 seeds/ac planting 15-October Soybean harvest [a] 2019 09-April Inorganic Inorganic fertilizer applied at 20.2 kg/ha nitrogen, fertilizer 22.5 kg/ha phosphorus, 111.8 kg/ha potassium application 24-October Winter wheat Planting density was 70lb/ac. Wheat planted with planting a grain drill [b] 2020 07-May Winter wheat termination 12-May No-till corn 32,500 seeds/ac planting 12-May Inorganic Inorganic fertilizer applied at 6.4 kg/ha nitrogen, fertilizer 9.5 kg/ha phosphorus, 33.3 kg/ha nitrogen as Urea application ammonium nitrate, 18-June Inorganic Inorganic fertilizer applied at 140 kg/ha nitrogen as fertilizer anhydrous ammonia application 17-October Corn harvest 19-October Cereal rye planting 2021[c] 15-May Cereal rye termination 21-May Soybean 150,000 seeds/ac planting 20-October Soybean harvest 2022 03-June Soybean 190,000 seeds/ac planting [a] No crop was planted in 2019 due to wet spring. Inorganic fertilizers include nitrogen, phosphorus, and potassium fertilizers. [b] Urea ammonium nitrate was applied on 12-May 2020 and anhydrous ammonia was applied on 18-June 2020. The planting density of cereal rye is unknown. In the model, we used the same planting density as winter wheat to represent cereal rye. P fertilizer was applied in the form of P 2O5. [c] The fall 2021 season was very wet, so no cover crop was planted during that time. 74 4.3.3 Data collection Soil water characteristic information is one of the most important inputs for the simulation of hydrology in RZWQM2. Overall, the soil profile is delineated into five layers. We used the HYPROP2 instrument (Meter Group, Inc.) to automatically determine the output parameters of the soil water characteristics curve (SWCC) for the topsoil layer (Shokrana & Ghane, 2020). However, it was challenging to collect undisturbed soil samples from deeper layers using the soil sampling tools provided with HYPROP. Therefore, we acquired the soil physical and chemical properties data from the GSSURGO database for the deeper soil layers (table 4.2) (Soil Survey Staff, Natural Resources Conservation Service, n.d.). The bulk densities (ρb) of the first three layers were measured in the lab by the scoop method (Peck, 2015) and the last two layers were collected from the GSSURGO database. The specific density of soil was assumed to be a constant value of 2.65 g/cm3 (Jury & Horton, 2004). In addition, soil texture data were analyzed by the Bray-P method in the soil, plant, and nutrients (SPNL) lab at Michigan State University. All the SWCC parameters such as residual water content, saturated water content, field capacity, permanent wilting point, and initial saturated hydraulic conductivity were acquired from the GSSURGO database. 75 Table 4.2. Input soil hydrologic properties data for Ziegenfuss clay loam soil [From Shokrana et al. (2022). Used with permission.] Depth of Bulk Porosity Sand Silt Clay Residual Saturated Field Permanent soil density[ (cm3/cm3) (%) (%) (%) water water capacity, wilting layer[a] b] , ρb content, content, ϴfc point, ϴpwp (cm) (g/cm3) ϴr ϴs (cm3/cm3) (cm3/cm3) (cm3/cm3) (cm3/cm3) 0-23 1.37 0.483 27.2 35.8 37.0 0.113 0.390 0.282 0.195 23-76 1.46 0.449 27.2 37.3 35.5 0.110 0.360 0.267 0.195 76-102 1.50 0.434 25.7 40.8 33.5 0.108 0.350 0.282 0.211 102-170 1.72 0.351 35.0 33.0 32.0 0.104 0.320 0.268 0.208 170-205 1.90 0.283 34.0 37.0 29.0 0.098 0.240 0.195 0.157 [a] The depth to the impermeable layer is 170 cm. [b] The ρb of the first three layers were measured in our lab and the last two layers were collected from the GSSURGO database. The depth to the impermeable layer is 170 cm. We started monitoring at the on-farm site on October 1st, 2018 and continued until June 30th, 2022 (i.e., the study duration is 3 years and 9 months). The hourly weather data used to run the RZWQM2-P are precipitation, air temperature, solar radiation, wind speed, and relative humidity. All these data were collected from our on-site ATMOS-41 weather station (Meter Group, Inc.). Missing weather data due to equipment malfunctioning were replaced from the Enviroweather database (https://enviroweather.msu.edu/). In case both ATMOS-41 and Enviroweather databases failed to provide data, those were collected from the National Oceanic and Atmospheric Administration (NOAA) database (https://www.ncei.noaa.gov/). We used the NOAA station at Lenawee County Airport (Adrian, MI, USA) to collect weather data. A control structure was installed to regulate the water table inside the field using stop logs or weir boards (Gilliam et al., 1979). HYDROS-21 sensors were placed both upstream and downstream of the control structure to collect water level data. Daily water samples were collected year-round in 1000-mL plastic containers using a Teledyne ISCO autosampler based on a daily composite sampling strategy with six aliquots per day (Dialameh & Ghane, 2022). The water samples were retrieved and brought back to the lab on a weekly basis. Afterward, the 76 samples were analyzed for DRP within 24 hours of retrieval. The samples were first filtered with a 0.45-µm filter before analyzing for DRP. The TP analysis was performed on unfiltered water samples within 7 days of retrieval, using the alkaline persulfate digestion procedure (Patton & Kryskalla, 2003). We used the Gallery Discrete Analyzer (Thermo Fisher Scientific) which uses the colorimetric technique for analyzing samples. Estimation of subsurface drainage discharge is one of the essential parts of this study because the nutrient load is a function of the drainage discharge. We used our own calibrated V- notch weir discharge equation as the primary method for estimating drainage discharge (Shokrana & Ghane., 2021). This equation is used until the head of water reaches the maximum height of the V-notch (i.e., 22.1 cm). Once the head of water goes above the maximum height of the V-notch, an area-velocity sensor (TIENET 350 area velocity sensor) was used to estimate the drainage discharge. 4.3.4 Initialization of P pools Once the SWCC parameters and drainage discharge data were collected, the next step was to initialize the P pools before starting the model simulations. The soil samples were collected from the Blissfield site and were sent to the Soil and Plant Nutrient Laboratory (SPNL) at Michigan State University for nutrient analysis. The soil samples were collected up to 90 cm from the surface. The results from the laboratory provided an initial estimate of the amount of Bray-P in the top 90 cm of the soil. Afterward, the initial amount of labile P pool for the first two layers (0-23 cm and 23-76 cm) were estimated from the total amount of Bray-P present in the top 30 cm of the soil. For the third (76-102 cm) and fourth layer (102-170 cm), we assumed that the concentration of Bray-P was 1 ppm and 0.25 ppm, respectively. We also assumed that the 77 initial amount of labile P pool in a specific layer is the same as the Bray-P amount in that layer. Since we did not have any soil test data for the deepest soil layer (170 cm - 205 cm), the initial amount of labile P was assumed as zero for that layer. This value is important in estimating the active P amount in a soil layer. The stable inorganic P pool is assumed to be 4 times the active P amount in a soil layer (Jones et al., 1984). RZWQM2-P assumes that 90% of the crop residue remains on the surface or first layer, and the remaining 10% goes to the second layer. The fresh organic P pool in a soil layer was assumed to be 0.03% of the crop residue amount in a soil layer (Noack et al., 2012). A detailed description of the method used to calculate different P pools can be found in (Sadhukhan & Qi, 2018). The input data provided for the initialization of P pools are shown in table 4.3. Table 4.3. Input data for P pools to initiate the P model [From Shokrana et al. (2022). Used with permission.] Depth Soil Soil Amount Initial P sorption Active P Crop Initial Initial Fresh of soil organic organic of labile P coefficient[b] amount residue stable stable organic layer matter carbon Carbon (kg/ha) (kg/ha) amount inorganic organic P (cm) (%) (%) (kg/ha) [a] (kg/ha)[c] P pool P (kg/ha) (kg/ha) (kg/ha) 0-23 3.00 1.74 54827 65.1 0.36 114.0 4500 455.8 548 1.35 23-76 0.75 0.44 48268 19.8 0.36 34.5 500 138.0 483 0.15 76-102 0.50 0.29 44370 3.9 0.36 7.0 0 28.3 444 0 102-170 0.50 0.29 84796 2.9 0.36 5.0 0 21.2 848 0 170-205 0.43 0.25 97141 0 0.36 0 0 0 971 0 [a] The initial labile P in a specific layer was assumed to be the same as Bray-P for that layer. The total amount of Bray- P for the top 30-cm soil layer was 84.94 kg/ha. The initial labile P amounts for the third and fourth layers (i.e., 76- 102 cm) were calculated from the Bray-P concentrations for those layers based on field measurements. The initial labile P for the last layer (i.e., 176-205 cm) was assumed to be zero. [b] This P sorption coefficient is a unitless parameter that was estimated from the (Williams et al., 2008). [c] We assumed that 90% of the crop residue remains on the surface or the first layer and the remaining 10% remain on the second layer. Since corn was planted in the year before the warm-up period of simulation, we assumed the corn residue amount as 5000 kg/ha. 4.3.5 Model calibration and validation Our two-year calibration period started on October 1, 2018 and ended on September 30, 2020. The validation period was one year and nine months, ranging from October 1, 2020 until 78 June 30, 2022. While choosing the calibration period, we ensured that it covered both wet and dry years of the study period (Skaggs et al., 2012). The annual precipitation for the water year (i.e., water years started from October 1 of the current year and ended on September 30 of the following year) 2019, 2020, and 2021 were 112.65, 73.83, and 85.71 cm, respectively. The calibration was performed in the following order: soil water dynamics, surface runoff, drainage discharge, evapotranspiration, plant water uptake parameters, and then recalibrated in the same order (Ma et al., 2012). Once we were satisfied with the hydrology calibration, we started calibrating DRP and TP loss through surface runoff and drainage discharge. 4.3.5.1 Model performance statistics We used the Nash-Sutcliffe efficiency (NSE) and percent bias (PBIAS) values to evaluate our model performance. Each model component was evaluated and categorized as very good, good, satisfactory, and unsatisfactory (Moriasi et al., 2007, 2015). For hydrology calibration and validation, the model performance was considered “very good” when NSE>0.75 and PBIAS < ±10%, “good” when 0.65 0.65 and PBIAS < ±15%, “good” when 0.50