Evaluation of the Soil Vulnerability Index for artificially drained cropland across eight Conservation Effects Assessment Project watersheds

Results and Discussion

Field-Scale Evaluation of the SVI. At the Arkansas Delta site, SVI vulnerability to runoff was either low or moderate depending on the hydrologic soil group because all the slopes were <0.5% (table 5, figure 2). SVI vulnerability to leaching was either moderately high or high once artificial drainage was considered (table 6). However, given the low permeability of these soils, the low slopes, and the long slope lengths for subsurface flow (Aryal et al. 2018), first principles of physics imply that subsurface lateral water movement is rather slow. Thus, the surface ditches capture mostly surface runoff, and increasing the SVI class for leaching potential by two classes, owing to the presence of artificial drainage, is misleading. Surface runoff is dominant, not leaching. A similar assessment was made in the Beasley Lake watershed. Soils in the hydrologic soil group D or C caused the SVI leaching component to be either low or moderate when artificial drainage was not considered, but moderately high or high once it was taken into account (tables 5 and 6, figure 2). A low to moderate leaching risk seemed more accurate than moderately high or high even when artificially drained because of the use of surface drainage combined with low percolation.

At the Arkansas Delta site, the SVI ratings for the Manila and Marked Tree fields were inconsistent with producers' choice of crops. The SVI rated both fields high for leaching. Yet the crops selected by the producer in these two fields reflect a difference in leaching potential: cotton (Gossypium hirsutum L.) in the Manila field contrasted with rice (Oryza sativa L.) and soybean in the Marked Tree field. Water yield in the watersheds for each field also supported producers' observations (i.e., surface discharge per unit area for similar rainfall was higher at Marked Tree than at Manila [Aryal and Reba 2017]). Higher natural soil drainage influenced the choice of the producer for a cotton crop in the Manila fields. Growing rice in a field with higher natural soil drainage would be very unlikely, owing to the challenge of retaining sufficient water on the surface of the field. Thus, the choice of a rice crop for the Marked Tree field is consistent with its lower natural drainage than the Manila field. Several factors may explain the discrepancy between the high SVI vulnerability to leaching for both fields and the differentiated perceived capacity of the soils to drain. The SVI leaching vulnerability is the same for both fields because the SVI uses the dominant hydrologic group in undrained conditions, which is D in both fields. However, the Marked Tree field has a dominant hydrologic group D, while the dominant hydrologic group of the Manilla field is C/D, meaning D in undrained conditions (table 7). The remaining components of this complex soil are in the hydrologic soil groups A or C. Even though the combined D and C/D hydrologic groups dominate in both fields, they comprise less of the field at Manilla (43%) than at Marked Tree (60%). In addition, a C/D soil may not behave the same as a D soil. Runoff and rainfall data from winter events (no crop growing) at the Manila field showed that the average value of back-calculated curve numbers was 82, which is more consistent with a bare soil in the hydrologic soil groups B than D (table 8). In presence of complex soils, the field dominant characteristics (hydrologic group and erodibility) may not be representative, and combining the SVI with other indicators of runoff and permeability may be necessary to ensure consistency with the processes occurring in the field. NRCS has already addressed this issue by requiring calculation of the SVI for all the components of a complex soil (USDA NRCS 2018). Finally, land leveling, which is common in this region, has the potential to expose subsurface layers and complicate the use of an index based on soil survey information. Field-scale application of a tool that relies on soil information derived at a larger scale may require additional sampling of soils in the field and determination of their characteristics. However, this goes against the SVI stated goal of being a quick assessment tool.

Watershed-Scale Evaluation of the SVI. Tables 5 and 6 show the distributions of the runoff and leaching components of the SVI for the cropland of these watersheds, respectively. Because artificial drainage results in only moderately high or high vulnerability, only these two vulnerability classes of the leaching component are included. In the following sections, we discuss the issues raised during the evaluation of the SVI at this scale: inconsistent soil maps across county lines, the treatment of organic soils, the differentiation of high soil vulnerability to leaching caused by highly permeable soils versus artificial drainage, the differences in leaching risk between surface and subsurface drainage, the consequences of a highly manipulated hydrology as in Upper Snake Rock, and the role of rainfall erosivity.

Inconsistent Soil Maps across County Lines. The two leaching maps for the Upper Big Walnut Creek watershed (figure 3) illustrate soil surveys that are not consistent from one county to another. The SSURGO soil surveys for the different counties were completed at different times, and a distinct demarcation can be observed in the hydrologic soil group classification. In particular, two soil map units had the same name (Centerburg silt loam) in Morrow County in the north and Delaware County, but were characterized as hydrologic soil group C for undrained condition in Morrow County and D in Delaware County. These differences in hydrologic soil group affected the SVI estimates of vulnerability to leaching. We expect that NRCS will address these differences during future revisions of the soil surveys, but SVI users should be aware of the potential for these discrepancies.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Vulnerabilities to runoff, leaching, and leaching under artificial drainage as classified by the Soil Vulnerability Index (SVI) at surface drained sites: (a, d, h, k) SVI runoff component, (b, e, i, l) SVI leaching component when artificial drainage is not considered, and (c, f, j, m) SVI leaching component when artificial drainage is considered. (g) Map for the Upper Choptank watershed differentiates cropland with mineral and organic soils.

Treatment of Organic Soils. Approximately 18% of the Upper Choptank cropland soils are organic soils (figure 2). These soils can have >50% organic matter in the top soil horizon, which varies in thickness from 5 to 20 cm. For the most part, those are hydric soils, and while most of these organic soils underlie wetlands, some were drained and are productive cropland. Artificially drained or not, these soils result in high SVI vulnerability to leaching. Yet, even when drained, hydric soils tend to remain wet during wet years because of properties such as high clay content and location in low lying areas of the landscape. Crops grow poorly during wet years on these soils, in spite of artificial drainage. For example, when wet spring conditions delay planting, producers switch to soybeans because of their shorter growing season. Thus, the high leaching rating conferred by the SVI to these soils, which is equivalent to the leaching rating for well-drained soils, is exaggerated.

Without artificial drainage, the hydric soils in the poorly drained upland areas increased denitrification and herbicide degradation, resulting in improved water quality of surface runoff (Hively et al. 2011). With artificial drainage, these wetland ecosystem services may be reduced, thus affecting the quality of the surface and subsurface runoff. Hence, artificial drainage of these depressions does have a negative impact on overall losses from the field, but through surface runoff, not leaching. However, presence of artificial drainage affects the leaching component of the SVI but leaves the runoff component untouched.

Artificial Drainage versus Permeable Soils. In artificially drained conditions, the risk of leaching was either moderately high or high because artificial drainage raises the risk by two classes, effectively lumping four potential risk classes into only two (table 2). In order to differentiate the leaching risks caused by high permeability of soils from that caused by artificial drainage, we ran the leaching component of the SVI twice: once assuming no artificial drainage, and once assuming that all cropland was artificially drained (table 6).

A moderately high or a high leaching risk is predicted by the SVI in one of the following conditions: the land is artificially drained, soils are organic, soils are in the hydrologic soil group A or soil group B, slope ≤12%, and erodibility <0.24. Greater soil permeability increases the risk of NO₃ leaching to groundwater; artificial drainage increases the risk of NO₃ transport to streams or lakes via drains or ditches. The two processes affect different water resources, have different flow paths, and should be addressed separately. Thus, it is important to distinguish the two sets of conditions when they both co-exist, as was the case in the Upper Choptank and in some of the Arkansas Delta sites. At these two sites, all the SVI leaching risks were either moderately high or high under artificially drained conditions. Yet, the SVI leaching risks without artificial drainage ranged from low to high (figure 2), showing the location of high permeability soils, which pose risks to groundwater resources. Calculating the SVI with and without artificial drainage is necessary for correct interpretation of the results. For the Beasley Lake (figure 2), and Upper Big Walnut Creek, Cedar Creek, Walnut Creek, and South Fork of the Iowa River (figure 3) watersheds, the SVI leaching risks without artificial drainage were low or moderate because soils were in hydrologic soil groups C or D, or because slopes were high enough that surface runoff was the primary flow path. Calculating the SVI with and without artificial drainage conditions showed the increase in vulnerability caused by artificial drainage. NRCS has already addressed this issue by splitting the leaching component into two: the leaching component, which describes vulnerability to leaching without considering artificial drainage, and the managed leaching component, which does consider artificial drainage (USDA NRCS 2018).

The situation is slightly different in the Upper Snake Rock watershed (figure 2), for which artificial drainage is a necessity because the amount of irrigation water diverted into the canals does not exactly match the irrigation requirements and some water drains back to the river along with runoff from surface irrigated fields (Bjorneberg et al. 2008). If irrigation methods should switch completely from furrow to sprinkler irrigation, surface drainage may not be necessary because sprinkler irrigation has higher irrigation efficiency (80%) than furrow irrigation (60%; James 1988). In this watershed, determination of the likelihood of artificial drainage does have to take irrigation method into account, which the SVI does not.

Surface versus Subsurface Artificial Drainage. The SVI considers artificial drainage as a flag without any consideration of the type of drainage. In general, the SVI vulnerability to leaching was appropriate for subsurface drainage systems. In the Upper Big Walnut Creek watershed, all the cropland was either in the low or moderate leaching risk class when artificial drainage was not considered, because of the C and D soil hydrologic groups (figure 3). Artificial drainage raised the SVI risk of nutrient losses by leaching to moderately high or high (figure 3, table 6). The high SVI vulnerability to leaching is consistent with the fact that most of the subsurface drainage in this watershed occurs for 9 to 10 months in the year, may drop in July to August as crops mature and water table draws down, and starts to flow again in October (Williams et al. 2015). In the South Fork of the Iowa River and Walnut Creek watersheds, the SVI captured the vulnerability of these soils to leaching very well. The SVI leaching index reflects the importance of this issue as it describes the dominant low vulnerability to leaching in the watershed without artificial drainage, which increases to moderately high or high after accounting for the drainage system (figure 3). On the other hand, the vulnerability to runoff predicted by the SVI was initially considered too high because there is no allowance for the fact that subsurface drainage reduces surface runoff volumes and peak runoff rates because of greater available water holding capacity (Skaggs et al. 1994). However, surface management practices are useful to control erosion at the drainage outlet so that the outflow does not erode the bank of the receiving stream. Leaving the SVI runoff vulnerability untouched highlights the need for such practices.

In the Cedar Creek watershed, SVI classification of leaching risks resulted in more than 50% of the land in the low vulnerability category and about 45% of the land in the moderate risk class when artificial drainage was not considered (figure 3). However, 80% of the cropland is subsurface-drained (Heathman et al. 2009). As a result, the entire watershed was classified in the high and moderately high vulnerability class for leaching (figure 3, table 6). This led to a high risk for both runoff and leaching for >50% of the crop fields in the watershed. This reflects accurately the conditions experienced by producers: high vulnerability to erosion and gully formation, and high vulnerability to nutrient loss through leaching (Smith et al. 2008).

In case of artificial surface drainage, the SVI vulnerability to runoff and leaching were not appropriate. Actual leaching risks were lower than indicated by the SVI, and runoff risks were higher than predicted, as shown in the following examples. In the Upper Choptank watershed, surface ditches drain numerous surface depressions, which would be wetlands if not artificially drained (Fenstermacher et al. 2014; McCarty et al. 2008). These soils cannot be used to grow crops unless they are artificially drained. In comparison to subsurface drains, surface ditches are more likely to transport sediment from the fields and be subject to streambank degradation without proper banks maintenance. Draining surface depressions actually increases the total volume of runoff to the streams (Lee et al. 2016). The SVI underestimated the vulnerability to runoff in the watershed because it did not take into account the runoff drained from these depressions. Depression areas, characterized by slope <2%, were all in a low vulnerability class for runoff. Yet they do contribute to runoff, sediment, and nutrient stream loads because of the surface ditches (Lee et al. 2016). On the other hand, the SVI predicted high leaching vulnerability owing to artificial drainage. Yet these depressions are drained via surface rather than leaching processes. The moderately high or high SVI vulnerability for leaching from these drained depressions reflected the risks to stream water quality, but attributed it to leaching processes rather than runoff processes. That could be acceptable if the purpose of the SVI was to determine an overall risk of farming in terms of water quality, i.e., if the runoff and leaching components of the SVI were grouped together to provide an overall risk assessment. However, if the runoff and leaching components are interpreted separately to guide the type of conservation needs on that land, the low runoff risk and high leaching risk estimated by the SVI could mislead conservation planners.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Vulnerabilities to runoff, leaching, and leaching under artificial drainage as classified by the Soil Vulnerability Index (SVI) in watersheds characterized by the presence of subsurface drains: (a, d, g, j) SVI runoff component, (b, e, h, k) SVI leaching component when artificial drainage is not considered, and (c, f, i, l) SVI leaching component when artificial drainage is considered.

In Iowa (Walnut Creek and South Fork of the Iowa River watersheds), the prairie potholes also contribute a substantial proportion of runoff. Saleh et al. (2007) estimated that pothole area occupied 10% of the watershed, but 57% of the total surface runoff directly flowed into potholes. Without drainage, these potholes would retain water from poorly drained areas for long periods (Saleh et al. 2007). With artificial drainage through surface intakes and underground pipes, these areas contribute not only runoff but also NO₃, dissolved P, and herbicides to the receiving streams (Tomer et al. 2010) because there is no time for herbicide degradation or denitrification. However, these depressions are not a source of sediment, as those settle out around the surface intake. Instead, streams, which are beyond the scope of the SVI, are the major sources of sediment—and an important source of total P (Tomer et al. 2010).

While the SVI runoff component correctly reflected the low risk for soil erosion from cropland in Walnut Creek and the South Fork of the Iowa River watersheds (low vulnerability to runoff), it underestimated the risk of P loss because it did not consider the pathway from potholes to the streams via surface inlets. The high ranking of the SVI leaching component could counter-balance this low classification as long as the two components are interpreted together. However, it would be misleading, as the P loss from drained potholes is primarily a surface runoff process with a subsurface delivery mechanism from the pothole to the stream rather than a leaching process through the soil and a subsurface drain.

Upper Snake Rock: A Special Case. In the Upper Snake Rock watershed, local water management complicated soil leaching potential vulnerability classification. When artificial drainage was not considered, the dominant leaching vulnerability in the region was moderate (figure 2), owing to most soils being in the hydrologic soil group C. When artificial drainage was considered, high vulnerability was dominant. While there is no subsurface drainage in the area, there is a lot of manipulation to collect surface and percolated water. Yet, the transport of sediment and sediment-bound nutrients in Upper Snake Rock is essentially a surface process, and the increase of the SVI leaching component by two classes because of this drainage was misleading, implying pollutant transport by leaching instead of surface processes. Generally, the leaching vulnerability without artificial drainage seemed to be the most representative of the region. On the other hand, the large tunnels that drain the perched water table caused by years of furrow irrigation do contribute to the transport of nutrients and agricultural chemical inputs to the stream (Lentz et al. 2018). Overall, the SVI was not very useful in this watershed because of the numerous modifications and complexity of the water conveyance system.

In addition, irrigation can have significant impacts on surface water quality and leaching. As seen in the Upper Snake Rock watershed, surface irrigation has caused significant leaching to the point of creating a manmade shallow aquifer. Irrigation may not be a concern in the Upper Mississippi and Ohio-Tennessee River basins where the SVI was originally developed but is present in many other cropland areas of the United States. Thus, along with artificial drainage, the SVI should consider whether irrigation is likely to increase leaching, which is a function of irrigation method, amount, and timing.

Role of Rainfall Erosivity. In the Upper Big Walnut Creek watershed, the SVI runoff potential overestimated the vulnerability to runoff. In intensively drained regions of Ohio, runoff remains very small or does not occur during most rainfall events (Pease et al. 2018). However, when significant runoff occurs, about two to three times each year, it is associated with high levels of sediment and nutrient loss (Pease et al. 2018). Vulnerability to runoff and erosion depends on rainfall intensity and amount (King et al. 2009, 2015), and in Ohio, those are relatively low as indicated by the rainfall runoff factor (R-factor) values ranging from 95 in the Northwest to 155 in the Southwest (USDA ARS 2013). Conversely, in the Beasley Lake watershed, the SVI appeared to underestimate inherent soil vulnerability to runoff. The low slopes (<2%) resulted in a low or moderate vulnerability to runoff according to the SVI (figure 2, table 5). Despite very low slopes (<2%) in this watershed and for most fields in the lower Mississippi Delta region (R-factor of 350), runoff-induced sediment losses are significant during the high winter and spring rainfall events (Cullum et al. 2010; Wren et al. 2019). The primary factors influencing soil erosion in this landscape include high intensity rainfall, low soil permeability and thus high runoff potential, and relatively low soil organic C. The SVI considers permeability through the hydrologic soil group, and organic C is indirectly represented, via the effect it has on soil erodibility (K-factor; Wischmeier and Smith 1978). Indeed, the K-factor is >0.32 for most of the cropland in this watershed. Low C content increases soil erodibility and the likelihood and amount of sediment and associated nutrient loss during high rainfall intensity events. However, the SVI does not consider rainfall intensity and the low slopes (<2%) resulted in a low or moderate SVI vulnerability to runoff. Introducing some measure of rainfall intensity into the SVI is needed to improve predictions and avoid a false sense of low vulnerability to runoff processes. Given the focus on erosion and soil degradation, an appropriate method could be to use the R-factor.

Synthesis and Proposed Improvements. In this study, soil vulnerability to runoff from artificially drained land was either underestimated or appropriate. Conversely, vulnerability to leaching was either overestimated or appropriate. Reasons for each assessment were different and are summarized below, along with possibilities on how to address the discrepancies between SVI classification and our knowledge of these watersheds.

In cases of subsurface drains placed within or below the root zone to lower the water table, the consensus was that the SVI estimated the runoff and leaching risks accurately, including with the two-class vulnerability increase to accommodate artificial drainage impacts. In cases of surface artificial drainage of surface depressions or very flat land, with surface ditches or with surface inlets that bring the water to subsurface drains, the consensus was that the SVI underestimated the risk of runoff and overestimated the leaching risk. Reasons were that the runoff component of the SVI did not account for the surface runoff that otherwise accumulated in surface depressions. Surface artificial drainage results in increased surface runoff delivered to streams compared to what would have happened without artificial drainage. Surface drainage also affects leaching minimally because drainage ditches do not lower the water table significantly.

By treating all artificial drainage in the same way, the SVI implies that draining the soil profile with surface ditches surrounding or crossing the field is as effective at lowering the water table within the soil profile—and as risky in terms of NO₃ transport—as subsurface drainage systems. In addition, it may confuse conservation planners using the SVI as to what processes are responsible for the risks. Surface drainage transports runoff and associated dissolved and particulate contaminants to receiving streams (Skaggs et al. 1994) while subsurface drainage systems increase the risk of NO₃-N and dissolved P transport via subsurface water movement (Tomer et al. 2010; King et al. 2015; Williams et al. 2015). A possible SVI modification would be to differentiate surface from subsurface artificial drainage and increase the risk class of the runoff or leaching component, depending on what type of drainage is used.

Raising the leaching risk class by two levels also resulted in less differentiation of vulnerabilities to leaching in a given area. Distinguishing high vulnerabilities to leaching caused by inherent soil properties or the presence of artificial drainage is important in order to plan for appropriate practices. In case of field assessment, the farm operator or the conservationist is probably aware of the processes and conditions in that field. However, at watershed or regional scale, both conditions leading to high leaching vulnerabilities may be present but not distinguished by the SVI. For example, there may be cropland underlain by very permeable soils (hydrologic soil group A or B) next to very poorly drained soils (hydrologic soil group D) that require artificial drainage. The SVI classifies both sets of conditions as highly vulnerable to leaching, but each set requires different methods to address that vulnerability. NRCS has implemented what we did here: applying the leaching component of the SVI twice—once in undrained conditions and once in artificially drained conditions (USDA NRCS 2018). The results differentiate the causes of vulnerability to leaching: highly permeable soils and artificial drainage.

Finally, this analysis highlights factors that affect vulnerability but are not represented in the SVI. At the watershed or regional scale, presence or absence of artificial drainage in each field is not necessarily available, requiring assumptions to be made. In this analysis, we assumed that all cropland was artificially drained. In the most recent version of the SVI, NRCS determines the need for artificial drainage based on NRCS drainage class (USDA NRCS 2018). Additional factors (e.g., shallow topsoil) may need to be considered. For example, poorly drained soils in northern Missouri are typically not artificially drained because the shallow top soil depth makes irrigation a requirement associated with artificial drainage, and marginal crop yields make investment in these practices more challenging. At field scale, limitations exist in the presence of a complex soil. Components of a complex soil can belong to different hydrologic soil groups and may have very different K-factors, which produce different estimates of the vulnerability of the field to runoff and leaching. Thus, SVI assessment for each component of that complex soil are necessary for full evaluation of the potential risks. Limitations exist at both scales when the hydrologic system is heavily manipulated.

The assessment of the SVI in Beasley Lake watershed highlighted the role of rainfall and rainfall intensity. The SVI was developed with data from the Upper Mississippi and Ohio-Tennessee River basins, thus in a region where rainfall amounts and intensities vary but are within a more limited range: around 20 to 35 mm h⁻¹ for a one year, one hour storm in the SVI development region compared to 35 to 56 mm h⁻¹ in Mississippi (Hershfield 1961). Given similar soil and topographic properties, a field is significantly more vulnerable to soil and nutrient loss by runoff in Mississippi than in Ohio because of the greater rainfall intensities that occur there. Vulnerability to runoff was considered too high in Ohio and too low in Mississippi, in part because of rainfall intensities. Incorporating some form of rainfall intensity into the SVI algorithm, possibly via the R-factor, would eliminate this risk. In regions where irrigation is a frequent practice, the SVI should consider the impact of irrigation on leaching and surface water quality.