1. Introduction
Crop yields vary yearly based on many factors. Some of these factors are uncontrollable by producers, such as some weather variables. Moisture is necessary for successful crop growth; however, excessive moisture can lead to yield degradation. Thus, effective water management is an important aspect of row-crop agriculture, and this includes drainage. Field drainage systems can broadly be characterized into subsurface (e.g., tiling) and surface (e.g., ditches, waterways, terraces, etc.). The purpose of establishing drainage ditches is to more quickly remove excess water from farmland, particularly during the rainy season, when soils are saturated [
1]. While subsurface drainage is very popular in the Midwestern United States., it is often not a feasible option in areas with soils high in clay content that lack permeability. Surface drainage is the only viable option in these areas. Surface drainage is also preferred over subsurface drainage when temporarily and quickly draining large quantities of surface water off land. Some types of surface drainage installations, such as grass waterways and terraces, are semi-permanent installations meant to remain in the field for many years. However, many farmers also create small temporary surface drainage ditches to drain areas in fields. These ditches are not seeded with grass, and so they do not require any additional management by a farmer. However, the ditches must be reinstalled periodically, especially after any tillage operations, and this is often accomplished during the busy spring planting season.
There is a variety of software that can take an elevation map and design subsurface drainage (e.g., Trimble WM-Subsurface, SD Drain Ditch, and Autodesk InfoDrainage). Subsurface drainage design utilizes precision design techniques and has been well-studied. Previous research on precision approaches in agricultural drainage include research on a new methodology for detecting subsurface drainage networks using electrical resistivity tomography monitoring and air injection from the outlet [
2], detection of subsurface drainage outlets with unmanned aerial vehicles (UAV) utilizing thermal infrared imagery and ground penetrating radar [
3,
4,
5,
6], and development of laser-beam automatic grade-control systems on high-speed subsurface drainage equipment [
7]. These approaches have utilized geospatial information system (GIS)-based analyses coupled with aerial photographs [
8,
9] and surface drainage feature identification using LiDAR DEM [
10]. Most of these articles [
2,
3,
4,
5,
6,
8,
9,
10] focused on detecting existing drainage systems and preserving existing systems rather than precision or data-informed methods for new surface and subsurface drainage design and construction. While different approaches for high-quality drainage network extraction in various terrains exist in other applications [
11,
12], these have yet to be fully explored in agricultural surface drainage.
The current on-farm implementation of temporary surface drainage, such as ditches, is typically performed using observations rather than data-based inferences. There is a lack of quantification of the benefits of data-informed, on-farm surface drainage. Thus, data-informed, on-farm surface drainage has the potential to improve on-farm drainage practices by optimizing drainage placement to account for spatial variation in field topography, residue state, and more. No spatial distribution of crop yield improvements has been considered in previous studies [
13,
14]. The analyses usually include the average statistics for a given field over a period of years [
15]. While subsurface drainage (tiling) is typically installed in a specific spacing pattern to cover nearly an entire field, allowing for simple efficacy, surface drainage only directly affects a field area drained by a ditch. It is crucial to explore the impact of data-informed, on-farm surface drainage ditching by considering the spatial distribution of yield over a field before making any further attempts to research automated precision ditching machines or systems. Most of the research completed in surface and subsurface drainage in agricultural systems has focused on leeching and nutrient loss studies. There are only a handful of studies that have been completed on the technical and economic sides of drainage systems such as ditch construction and management, drainage design, economic benefits, or precision approaches to drainage, particularly for surface drainage.
Early research related to ditch construction and management focused mainly on subsurface drainage. Rollin et al. [
16] tested four different thin synthetic envelope materials for use on subsurface drainage tubes in a laboratory and found that all four envelopes were successful in preventing soil from entering drain pipes [
16]. The development of practical tools, new types of installation equipment like trenchers and trenchless drainage machines, new materials for drainpipes and envelopes, and improved quality of construction were the major ditch construction and management practices that helped make rapid change possible [
17]. Other studies related to ditch construction and management included the use of geophysical and geotechnical methods that were capable of detecting buried features to find the location of agricultural subsurface drainage systems [
18], the modification of trapezoidal ditches to two-stage geometries as stable and innovative solutions for soil erosion [
19], pre-tender cost estimations [
20], the performance evaluation of a trenchless piping machine for pulling-force requirements and pipe-laying depths [
21], and the development of multi-site and multi-decadal datasets for artificially drained fields, which could potentially be used for the design of drainage systems [
22]. In approximately 1999, one of the first surface drainage ditch construction and management studies documented surface drainage configurations that were commonly used to improve land drainage [
14]. Based on outlet design considerations and aspects of pumped drains, the types, patterns, and shapes of gravity drain outlets were listed [
14].
Existing research on agricultural drainage design has also focused mainly on subsurface drainage. Rojas and Willardson [
23] estimated the allowable flooding time for surface drainage design before causing crop damage versus flooding time curves, and they developed a procedure for calculating the total drainage time, time to reach 10% aeration, and surface drainage time. Another early study conducted in England focused on subsurface drainage pipe size design, and a new design chart was published for calculating pipe diameters for pipes of available materials [
24]. Other agricultural drainage design-related research has included work completed on the effect of drain spacing on waterlogged soil and the yield of wheat [
25], drain spacing for optimal flood control [
26], the effects of perforation shapes and patterns in drainpipes and of envelope materials on drainage performance [
27], the simulation of drainage depth and spacing to optimize subsurface drainage [
28,
29,
30,
31], the use of random forest and multi-layer regression to detect subsurface drainage effects on corn plant height [
32], experimental and numerical methods to determine pipe layout parameters for water and salt drainage [
33], and the application of knitted-sock geotextile envelopes to increase drain inflow by experimentally measuring the effective radius of sock-wrapped and sand-slot pipes [
34].
There have also been limited studies on the economic benefits of agricultural drainage. Datta et al. [
35] justified the investment in subsurface drainage by the high internal rate of return. Two different subsurface drainage systems and their economic benefits were evaluated by Yan et al. [
36] to confirm the economic feasibility of a subsurface drainage system. Saikkonen et al. [
37] examined the social returns, such as economic profit and environmental impact, for crop cultivation under both surface and subsurface drainage. Using measurable social costs, it was found that the measurable social returns were higher for surface drainage than for subsurface drainage for low land qualities [
37]. This study, however, did not consider the direct economic benefits from crop improvements. An eight-year comparison of the agroeconomic benefits of open-ditch and subsurface pipe drainage in China resulted in the higher midterm economic benefits of subsurface pipe drainage [
38]. While these studies highlight the economic feasibility and profitability of subsurface drainage systems, particularly through detailed analyses of internal rates of return and social benefits, they leave the potential economic impacts of surface drainage largely unexplored.
The review of the articles discussed earlier suggests that there has been more research completed on subsurface drainage and less on surface drainage in agricultural settings. Economic benefit analyses of surface drainage are missing from the given literature, which is a research gap that can help farmers when it comes to choosing between surface drainage or subsurface drainage. The development and implementation of workflows for precision or data-informed surface drainage systems are currently lacking and are needed for optimal implementation. Exploring the yield impacts of surface and subsurface drainage systems using crop-yield data from before and after drainage construction is also lacking and is needed to accurately determine the impact of drainage on crop yield. Also, there is potential for research on the development of new equipment and technologies to automate the process of drainage ditch construction to achieve precision drainage.
The goal of this research was to investigate the effects, particularly on crop yields, of a more data-informed approach to traditional temporary surface drainage implementation. This research work was an attempt to begin addressing some of the research gaps identified in the literature by exploring the potential impact of data-informed on-farm surface drainage by using cooperator data as a case study. Specifically, the benefits of implementing data-informed on-farm surface drainage on approximately 300 acres in Central Illinois were explored by analyzing decades of data from a cooperator who had implemented data-informed drainage practices during that time period. This analysis aimed to inform current farmers and motivate future research on automated precision ditching machines or systems.
2. Materials and Methods
The dataset for this study was provided by a farmer/cooperator in central Illinois. Two 150-acre fields were used in the analysis for this study. Both fields were drained using traditional surface drainage practices until 2012, using only empirical and observational evidence by the farmer for implementation. After 2012, a data-informed approach to surface drainage was adopted, as shown in
Figure 1. To implement an initial form of data-informed on-farm surface drainage, the cooperator extracted the field topography from their tractor’s on-board GPS data. Thes data were determined to be of a sufficient resolution for the surface drainage installations. The GPS coordinates of the field areas which were known to frequently pond water after moderate rainfall were also noted. A ditch network was then designed using GIS software (ArcGIS 10.1) to create a ‘.shp’ file for the ditch paths. The ditch network was designed to route the ditch paths down natural topographic contours, with the ditch providing a flow path for the water to not disperse into the field. The areas with frequent ponding were treated as constraints in the network, with at least one ditch path required to pass through that area to drain it. Finally, the ‘.shp’ file was used to implement the drainage using a tractor with a Wolverine Ditcher (Elmers Manufacturing Ltd., Altona, MB, Canada). While this workflow contained manual steps, such as marking locations of the ponded areas and design of the ditch paths in the GIS software, these steps could be automated with additional effort. Regardless, this type of spatial data-driven surface drainage approach was the first step towards precision surface drainage, and the analysis of the resulting impact provided insights about the potential benefits of the workflow.
The files provided for analysis were in ‘.shp’ format and contained geolocations, elevations, timeframes, moisture levels, crop yields on dry and wet bases by weight and by volume, and other machine- and flow-related variables in the attribute table. For yield analysis, the ‘yield volume on dry basis (bushels/acre)’ variable was selected. Yield data were acquired during harvest for the given year using a combine yield monitor. The data provided also included the elevation and shape files for the drainage ditch for each field and the corn and soybean yields for each field for the years 2008 to 2021, allowing for some examination into the before- and after-effects of the data-informed on-farm surface drainage.
Figure 2a shows a sample field with marked areas of water accumulation and
Figure 2c shows the implementation of the topographical information in creating the ditch layout. Although there were some forms of empirical drainage before the data-informed approach of drainage, the before and after drainage in this study refers to the years before and after 2012 when the data-informed on-farm drainage was implemented. For corn, the before-drainage years were 2008, 2009, and 2011 for Field 1, and they were and 2008 and 2010 for Field 2. After-drainage included the years 2013, 2015, 2016, 2018, and 2020 for Field 1 and 2016, 2018, and 2020 for Field 2. For soybeans, the before-drainage for Field 1 included the year 2010. After-drainage for Field 1 included the years 2012, 2014, 2017, 2019, and 2021. Before-drainage for Field 2 included the years 2008 and 2011. After-drainage for Field 2 included the years 2013, 2015, and 2017.
To plot and analyze the crop yields, ArcMap version 10.8.1 was used. For both crops, initially, a yield map was generated for each field for each year. Those maps were then normalized by dividing all yield data by the average yield of the entire field for the given year. The data were categorized based on the year of drainage construction, i.e., the average normalization was completed before- and after-drainage for crop yields for both fields. For a comparative yield analysis by drainage-affected areas, the likely impacted areas of drainage for a subset of ditches were isolated from each field based on the results from the previous step and the flow direction was plotted as vectors, and their average values before and after drainage were calculated and summarized in a table.
Figure 3 explains the procedure carried out for this analysis, where elevation data were used to fill the hydrology feature, which was then used to calculate the flow accumulation direction.
Figure 4 shows the isolated affected drainage areas for both fields. Three affected areas were isolated from the fields to study the effects of drainage based on the flow direction. These isolated affected areas (IAAs) were manually constructed using the polygon feature of ArcMap. The yields of both crops for both fields before and after drainage were clipped to these IAAs to analyze the crop yields and compare them to the crop yield for the entire field. After isolating these areas, descriptive statistics (minimum, maximum, mean, and standard deviation) were calculated for the clipped yields on these areas before and after drainage, and the percentage changes in these statistics were determined.
3. Results and Discussion
The geospatial distribution of yield was plotted using ArcMap for both before and after drainage to show the impact of data-informed on-farm surface drainage.
Figure 5 and
Figure 6 show the satellite images and the elevation plots for field 1 and field 2, respectively. Before the drainage ditch was constructed, the yields in areas with lower elevations were relatively lower compared to areas with higher elevations, as can be seen in
Figure 7. After the drainage ditch was constructed in the year 2012, the spatial distribution of crop yield showed that the lower elevation areas had better yields compared to the scenario before the drainage. Comparing the yield maps side by side, as shown in
Figure 7,
Figure 8 and
Figure 9 below, the construction of a data-informed on-farm surface drainage ditch improved the relative corn yield compared to the traditional/empirical surface drainage, especially in the areas circled by dotted lines. However, due to the availability of only one year of data prior to the data-informed drainage construction, drawing any conclusions about the average normalized soybean yield for Field 1 is inappropriate (
Figure 8). Similar trends of improved yields in areas of drainage could be observed in the case of soybean yields for both fields, suggesting that the drainage system positively impacted overall crop productivity.
For the soybean yield in Field 2, the left half of the field had a lower yield compared to the right half before drainage. Also, it can be observed from the elevation map (
Figure 6b) that the left portion of the field was at a higher elevation (but relatively flat) than the right portion of the field. The drainage in this field was therefore constructed in a left-to-right (west–east) direction, as shown in
Figure 9b. This improved the yield on the left part of the field was expected. Some small patches of low yields, however, were observed in this field after drainage, especially in the middle and right parts of the field (
Figure 9b). This suggests that the flow direction of the water should be considered in addition to the topo map when constructing a drainage ditch for better efficiency.
A comparison of the yields between each isolated affected area and the average for the field provided insights into the relative impact of the data-informed on-farm drainage installations. Comparing the yield changes for each isolated affected area to the entire field helped isolate some effects such as seed and practice improvements over time. The area covered for each isolated affected area studied was calculated using the Geometry Tool in ArcMap. For Field 1, the three isolated affected areas covered 4.85, 6.10, and 2.66 acres, respectively. For Field 2, the left half of the field was used for corn and the right half of the field was used for soybeans (as shown in
Figure 6a) in alternation with wheat, and hence, there were three different isolated affected areas studied for corn and three other different isolated affected areas studied for soybeans, as shown in
Figure 4. The isolated affected areas for corn covered 4.84, 2.37, and 7.10 acres, respectively, and the isolated affected areas for the soybeans covered 1.70, 11.83, and 1.53 acres, respectively, for field 2. There was some inherent bias in this comparison because the average total field yields also included the yield impacts of the isolated affected areas; however, it was still a meaningful comparison.
Figure 10 shows the percentage changes in mean yields for the periods before and after drainage for the entire field and each isolated affected area for both corn and soybean. For corn in Field 1, there was an 18.3% increase in average yield across the entire field for the period after the data-informed on-farm surface drainage. Out of the three isolated areas (IAA1, IAA2, and IAA3), the percent changes in corn yield were 26.1% and 26.5% for IAA1 and IAA2, respectively, which were higher than the percent change in corn yield across the entire field. However, the percent change in corn yield for IAA3 was only 15.9%, which was less than the increase for the entire field. Similar positive impacts were observed in the case of soybean yields in Field 1, albeit with lower magnitudes. For soybean in Field 1, there was a 5.5% increase in average yield across the entire field for the period after the data-informed on-farm surface drainage. For the three isolated areas (IAA1, IAA2, and IAA3), the percent changes in soybean yield were 14.6% and 14.5% for IAA1 and IAA2, which were higher than the percent change in soybean yield across the entire field. However, the percent change in soybean yield for IAA3 was only 0.8%, which was lower than the yield improvement for the entire field.
For corn in Field 2, there was a 13.9% increase in average yield across the entire field for the period after the data-informed on-farm surface drainage. For the three isolated areas (IAA1, IAA2, and IAA3), the percent changes in corn yield were 21.4%, 40.2%, and 26.7%, respectively. All IAA improvements were higher than the increase for the entire field, likely indicating the impact of the data-informed on-farm surface drainage. For soybean in Field 2, there was a 23.1% increase in average yield across the entire field for the period after the data-informed on-farm surface drainage. For the three isolated areas (IAA1, IAA2, and IAA3), the percentage changes in soybean yield were 38.5%, 23.9%, and 42.2%, respectively. These percent changes in soybean yield were higher than the percent change across the entire field, indicating a net positive impact from the data-informed on-farm surface drainage.
Historically, corn and soybean yields in the United States and other countries have increased over time, primarily due to improvements in genetics. Between 2009 and 2016, United States corn yields increased by an average of 9.84% per acre [
39], while soybean yields increased by 13.33% per acre [
40]. These years represent the median period before and after the data-informed drainage ditch construction, which was implemented in this study. Thus, the corn yields for the case study fields improved more than the national average during the same period. Relative yield improvements varied between fields for soybean. In Field 2, we observed a 23.1% increase in soybean yield—higher than the national average of 13.33%. However, Field 1 showed a soybean yield increase lower than the national average. It should be noted that for Field 1, the pre-data-informed drainage data were limited to a single year (2010), which may have influenced this outcome.
This study presents a significant advancement in the field of surface drainage practices in United States row crop agriculture, particularly in areas where subsurface drainage is not viable. Unlike previous research that often relied on anecdotal evidence or generalized recommendations, this case study employed a robust, data-informed approach by analyzing over one decade of yield data. Future research could greatly benefit from integrating economic assessments to provide farmers with a more comprehensive understanding of the cost–benefit dynamics involved in adopting data-informed surface drainage practices.