Abstract
The Mississippi River Valley alluvial aquifer provides over 90% of irrigation water for agriculture in the Lower Mississippi River Basin. Overexploitation of the alluvial aquifer in the Cache River Critical Groundwater Area (CRCGA) in Arkansas started impacting crop production in the 2000s with previous historical data identifying sustained groundwater declines. Groundwater depletion resulted in many producers investing in on-farm reservoir-tailwater recovery systems to provide a more secure irrigation source. More focused and consistent water level measurements were necessary to identify any groundwater level changes. Groundwater trends focused on the CRCGA cone of depression were compared from two-time groupings: 1985 to 2012 and 2012 to 2019. On-farm reservoirs in the CRCGA were inventoried in 2015, and 90% of those inventoried were constructed by 2012. More frequent groundwater level data collection in the CRCGA started in 2012 as well. Groundwater levels during 1985 to 2012 had consistent decline rates of –0.17 to –0.44 m y–1. Groundwater levels during 2012 to 2019 varied with decline rates of ≤–0.10 to ≤–0.50 m y–1, but also areas of no significant change and recovery at rates of ≤0 to ≤0.25 m y–1. Trend differences between the two periods showed improved water level trends for most of the study area, with rates ranging ≤–0.30 to ≤0.50 m y–1. The measured improvements were attributed to less pumping due to greater precipitation and the use of surface water irrigation from 60 on-farm reservoir-tailwater recovery systems. Detection of these improvements is crucial to understanding the response and status of the alluvial aquifer within the agricultural area of the CRCGA. It is recommended that adding continuous groundwater monitoring stations in west Craighead and Poinsett counties would be beneficial to inform the status of the cone of depression, considering the importance of protecting the main agricultural irrigation supply for the state. If warmer temperatures occur in the future, higher temporal and spatial groundwater level data in the CRCGA will allow policymakers and water resource managers to make timely groundwater management decisions to ensure its supply for future generations.
The Mississippi River Valley alluvial aquifer, henceforth the alluvial aquifer, is the primary source for irrigation water for row crops grown in southeastern Missouri, eastern Arkansas and Louisiana, southwestern Illinois and Kentucky, western Tennessee, and Mississippi. The alluvial aquifer is an important shallow aquifer within the area, as it is the nation’s second most used aquifer for agricultural irrigation, due to its accessibility and high yield (Marston et al. 2015). It supplies 35 ML d–1 of irrigation to over 3.2 million ha that collectively produce 72% of the nation’s rice (Oryza sativa L.) as well as other important commodities (Maupin and Barber 2005; USDA NASS 2018). In a recent US Geological Survey (USGS) report, it was estimated that the alluvial aquifer generates revenues of about US$6.5 billion directly from agriculture in the Lower Mississippi River Basin (LMRB) region (Alhassan et al. 2019). This accounts for about 14% of the value of total US agriculture. Arkansas and Mississippi contain the most agriculturally productive counties in the region (Alhassan et al. 2019).
Arkansas ranks third nationally behind Nebraska and California in terms of irrigated cropland, with 88% of irrigation predominately from groundwater and the other 12% from surface water (USDA NASS 2018, 2019). Extensive overpumping for agricultural irrigation has led to large-scale declines in groundwater levels in this vital US agricultural production region (Konikow 2013). Arkansas farmers produce nearly 50% of US rice, and the state ranks in the top five cotton (Gossypium hirsutum L.) and soybean (Glycine max [L.] Merr.) producing states in the southern United States (USDA NASS 2018). Overall, agricultural productivity in the region has expanded due to a secure irrigation source from the alluvial aquifer. However, increased well development resulting in groundwater level declines threatens the future availability of the alluvial aquifer in certain regions (McGuire et al. 2019).
The alluvial aquifer is an excellent source of water due to its favorable hydrologic characteristics, and 54% of the aquifer is in eastern Arkansas (Pugh et al. 1997). It overlays Tertiary-age units and consists of Quaternary-age sand, gravel, silt, and clay deposits (McGuire et al. 2019). In some areas, the aquifer is overlain by Quaternary-age semiconfining or confining units of silt and clay, which impede groundwater recharge where present (McGuire et al. 2019). The alluvial aquifer in Arkansas is divided by Crowley’s Ridge, a structural high and erosional remnant of the Tertiary-age rock, which disconnects the eastern and western portions of the aquifer along the ridge (Schrader 2015) (figure 1). Groundwater recharge throughout Arkansas generally comes from precipitation, which slowly infiltrates into the groundwater system (Kresse et al. 2014). Precipitation in the region is relatively high at 1,200 mm y–1 for the 1981 to 2010 climate normal (Arguez et al. 2010). However, recharge rates from precipitation are predicted to be minimal at 20 to 60 mm y–1, which is a relatively small volume of recharge water compared to groundwater pumping rates (Kresse et al. 2014). Abundant recharge will occur in areas where the confining unit is very thin or absent; where the confining unit is present, recharge by lateral flow is more important (Kresse et al. 2014).
Alluvial aquifer depth to groundwater in (a) 1996, (b) 2000, (c) 2004, (d) 2008, (e) 2012, and (f) 2015 within the Cache River Critical Groundwater Area (CRCGA). This interpolation used US Geological Survey spring water level measurements during January to May from the same wells year to year focusing on the area of interest. This figure was created using a natural neighbor interpolation in Esri ArcGIS. (g) The extent of the alluvial aquifer in eastern Arkansas with the counties within the CRCGA and Crowley’s Ridge noted.
Increased groundwater withdrawals and limited aquifer recharge due to the presence of a confining clay layer of varying thickness have resulted in several cones of depression in eastern Arkansas. The Arkansas Department of Agriculture Natural Resources Division (ADANRD) (formally the Arkansas Natural Resources Commission [ANRC]) designated these critical groundwater study areas in eastern Arkansas (ANRC 2009). The Grand Prairie Critical Groundwater Area (GPCGA) in east central Arkansas and the Cache River Critical Groundwater Area (CRCGA) west of Crowley’s Ridge are locations with large cones of depression. These areas were predicted to reach critical declines in groundwater levels if irrigation demand continues to exceed supply (Reed 2003; Clark et al. 2011, 2013). Reed (2003) was the first to predict, based on a simulation of pumping rates continued at 1997 levels, that two critical groundwater areas in east central Arkansas and west of Crowley’s Ridge will have extreme water level declines by 2049, with some areas projected to be depleted. In order to mitigate the historical groundwater declines in the Grand Prairie Area, the US Army Corps of Engineers began construction of two surface water diversion projects that were originally initiated by the US Flood Control Act of 1950 (USACE 1998). These projects have had significant delays, and completion is not expected for decades in consideration of federal funding levels. Once in operation, these projects are expected to meet over 50% of the total irrigation demands in the GPCGA. These projects also aim to protect aquifer resources by providing continued irrigation of current agricultural crops and by reducing aquifer decline in the GPCGA (USACE 1998).
In contrast to the GPCGA, a surface water diversion project is not currently planned for the CRCGA, due to a lack of surface water resources with enough volume (ADANRD 2020). Therefore, alternatives to surface water diversion projects such as water conservation practices, surface water irrigation, and managed aquifer recharge are needed to conserve groundwater and prevent further decline of groundwater levels within the alluvial aquifer (Reba et al. 2017). Many producers within these critical groundwater areas have constructed on-farm reservoirs and tailwater recovery systems. These systems consist of a storage reservoir and a drainage ditch. The reservoir stores runoff and precipitation that is later used for irrigation, while also capturing sediments and nutrients in the process (Pérez-Gutiérrez et al. 2017; Yaeger et al. 2017; Omer et al. 2018b; Omer and Baker 2019). Many producers utilize federal assistance from the USDA Natural Resources Conservation Service (NRCS) Environmental Quality Incentives Program (EQIP) to install these systems. This water conservation practice has been implemented across eastern Arkansas with a total of 632 reservoirs in the GPCGA and 143 reservoirs in the CRCGA by 2015, which help to reduce the dependency on groundwater for irrigation (Sullivan and Delp 2012; Yaeger et al. 2017, 2018).
Many state and federal resources have been invested in collecting alluvial aquifer water level data because of its economic importance and concerns related to its depletion (Kresse et al. 2014). ADANRD, USGS, and USDA NRCS collect water level data every year across the state to compare with the historical water levels and document the growing season response to the water level of the alluvial aquifer. These data collection efforts are extremely valuable to the state and the LMRB because they aid in understanding the status and availability of the alluvial aquifer. However, these long-term annual measurements occur largely as single measurements in ~500 wells, which vary year-to-year (2014: 550 wells; 2015: 486 wells) with a measurement time between January and May (ANRC 2014, 2015). Annual water level measurements are interpreted as a general spring potentiometric surface and as year-to-year water level changes. A general trend of declining groundwater levels is evident in these critical groundwater areas, but without focused and intense monitoring, any local groundwater recovery might not be obvious.
In this study, the historical alluvial aquifer data were used to understand the long-term groundwater trends within a subsection of the CRCGA in Craighead and Poinsett counties, areas centered on the cone of depression west of Crowley’s Ridge. As noted previously, many producers implemented on-farm reservoir-tailwater recovery systems for alternate irrigation, with many of these systems coming into operation during the 2000s. At that time as well, the USDA Agricultural Research Service (ARS)-Delta Water Management Research Unit (DWMRU) groundwater monitoring well network started collecting monthly water level data from active and abandoned wells across Craighead and Poinsett counties, west of Crowley’s Ridge. Producer collaborators allowed access to these wells. With these more consistent and intensive water level data since the implementation of on-farm reservoirs, water table fluctuations will be more fully documented to identify any groundwater recovery.
The objectives of this research are to describe the local alluvial aquifer trends in order to document any recent groundwater response and to clarify any further cone of depression development west of Crowley’s Ridge within a critical groundwater area. This work stresses the importance of the more frequent and consistent data collection in northeast Arkansas to improve the understanding of water level trends in the alluvial aquifer, which will contribute to sustainable use of the resource.
Materials and Methods
Study Area. The study area was a subsection of the CRCGA in eastern Arkansas, specifically Craighead and Poinsett counties west of Crowley’s Ridge. This area is about 1,500 km2. Much of the area is cultivated in rice, soybean, or corn (Zea mays L.) (USDA NASS 2018). The predominate soil type in the area is a silt loam of the Hillemann, Calloway, Calhoun, or Henry soils (National Cooperative Soil Survey 2020). Between 1996 and 2015, on average, 4 ± 1 reservoirs were constructed per year, corresponding to a cumulative reservoir surface in the CRCGA of 60 ± 18 ha y–1 in 2015 (Yaeger et al. 2018). In 2015, on-farm reservoirs in Poinsett County and Craighead County totaled 69 and 36, respectively, with 60 reservoirs located in the focused study region (Yaeger et al. 2017).
Methodology. Three different sets of groundwater data were used in this study. First, USGS well measurements from the alluvial aquifer in eastern Arkansas were used to characterize the development of the cone of depression over two decades. Then, select USGS data from within the smaller study region were used to calculate groundwater trends during the 1985 to 2012 period (Group A). The USGS wells were first narrowed down to those in the study area, and then to locations containing at least eight years of data for analysis. The select USGS wells were then 24 wells for 1985 to 2012 and 6 for 2012 to 2019. Wells with years-long gaps in data or inconsistent measurement data were not included. Lastly, select USGS and USDA ARS-DWMRU well data from within the focused study region were used to determine groundwater trends from the 2012 to 2019 period (Group B). The select USDA ARS-DWMRU wells had at least two years of monthly groundwater level data from 2012 to 2019. On-farm reservoirs in the CRCGA were inventoried in 2015, and 90% of those inventoried were constructed by 2012 (Yaeger et al. 2018). More frequent water level data collection in the CRCGA by the USDA ARS-DWMRU started in 2012 as well.
Discrete well measurements were collected from the USGS National Groundwater Monitoring System (https://waterdata.usgs.gov/nwis) from wells measured during the spring (January to May) of 1996, 2000, 2004, 2008, 2012, and 2015. The spring timeframe is pre-irrigation with water levels generally recovered from pumping in the previous irrigation season. The interpolation input data were from 238 wells within the alluvial aquifer throughout eastern Arkansas. Yaeger et al. (2017) describes the water level interpolation method using 2015 data, and this publication is the first to illustrate the interpolations as a time series. Not all 238 wells were measured after 2015, such as 192 wells and 168 wells out of the 238 wells were measured in 2016 and 2017, respectively. Therefore, the timeline was not extended past 2015. Water level measurements were interpolated to form a groundwater surface map in ArcGIS version 10.4 (ESRI, Redland, California, United States), using the natural neighbor method (Yaeger et al. 2017; ADANRD 2020). The interpolation was cropped to the CRCGA, with 111 well points within this area. This simple interpolation method is a proportional area-weighted method as the water level data were limited in some areas. The goal was to present data in a simplistic manner and minimize the introduction of artifacts into the result.
Precipitation and temperature data were analyzed to determine if there was a significant change in climate between the two monitoring periods, Group A 1985 to 2012 and Group B 2012 to 2019. Monthly total precipitation and mean temperature data were collected from the National Oceanic and Atmospheric Administration (NOAA) National Climatic Data Center historic observation database in Jonesboro, Arkansas, from 1985 to 2019 (Jonesboro 2 NE Station USC00033734). These parameters were first compared on an annual basis for normal distribution between the two periods using a t-test, represented by the test statistic t with the degrees of freedom noted in subscript (Gosset 1908). Monthly values for temperature and precipitation were compared between periods with either a t-test or a Mann-Whitney test (Gosset 1908; Mann and Whitney 1947). Monthly data were combined within each period to create mean monthly climate curves, which were then compared statistically using a Kolmogorov-Smirnov (KS) test (Kolmogorov 1933; Smirnov 1948). The nonparametric KS test was applied to determine if monthly precipitation or temperature distributions changed between periods.
Two groupings of monitoring well data were used in the groundwater trend analyses, Group A 1985 to 2012 and Group B 2012 to 2019. Group A consisted of USGS historical monitoring well data gathered from the National Groundwater Monitoring System (https://waterdata.usgs.gov/nwis). The USGS wells within the CRCGA study region that had at least eight full years of continuous measurements between 1985 and 2012 were included. The eight-year minimum was chosen based on the annual or semiannual nature of measurements in these wells, with eight years typically providing the minimum number of data points for robust trend analysis. In total, 24 USGS monitoring wells within western Craighead and Poinsett counties with between 8 and 27 years of data were included in Group A (table 1).
Well monitoring data range within each group.
Group B consisted primarily of USDA ARS-DWMRU monitoring wells where monitoring began between 2012 and 2016. Water levels in these wells were measured manually with an e-tape, an In-Situ Inc. Rugged Water Level Tape 200 electronic sounding device (In-Situ Inc., Fort Collins, Colorado) or a Solinst Model 101 Water Level meter (Solinst, Georgetown, Ontario, Canada). A USGS study reported electronic-tape models to be accurate to roughly ±0.02 m per 30 m without additional calibration, and a similar accuracy was assumed for the USGS and USDA ARS-DWMRU measurements (Jelinski et al. 2015). Groundwater level measurements were taken on a monthly schedule and recorded relative to a marked measuring point on each well. There was a mean of 36 days between measurements where monthly measurements were occasionally missed due to severe weather, farming activities, etc. Three replicate measurements were collected at each well, with the mean measurement recorded as the monthly water level. Because Group B’s measurement density was much higher than Group A, the 24 USDA ARS-DWMRU wells with at least two years of continuous monthly data were included in the trend analysis (table 1). Additionally, six Group A wells with data extending to 2019 were included in Group B to strengthen the analysis.
Mann-Kendall (MK) tests were used as an initial assessment of the long-term groundwater trend (Mann 1945; Kendall 1975). The MK analysis assesses whether there is a monotonic trend, either nonincreasing or nondecreasing, in a time series. There is no assumption of linearity in the MK analysis. In general, manual measurements for an entire year were used in the calculation. The effects of nearby pumping were occasionally observed in the regularly monitored wells, creating subseasonal fluctuations in water level. However, severe drawdown measurements collected during pumping events were assessed based on Cook’s Distance (D) to verify if the measurement impacted the trend calculation (Cook 1979). Cook’s D identifies outliers by measuring the influence of each data point on the linear regression analysis. There is no defined value of Cook’s D that marks an outlier; it depends on the data set, goals of the analysis, and other factors. No data points were removed from any of the Group A wells or the USGS wells used in Group B, due to less temporal consistency in the measurements. Outliers from 12 wells in the Group B USDA ARS-DWMRU wells were removed with the Cook’s D range from about 0.25 to 0.80. For 11 of these 12 wells, three or fewer points were removed to account for 1.4% to 6.5% of the measurements. For the other well (CC#7), six points were removed, to account for 11.3% of the points. These wells were active pumping wells or have pumping wells within about 15 m and recorded significant drops each summer. These measurements were removed because they would impact the trend calculation and not be representative of long-term trends. However, especially when more than four to five full years of data were available, subtle fluctuations from summer pumping activity typically did not impact analysis. Because of pumping and seasonal groundwater fluctuations, the trend results were not always equivalent to the change in depth over the measurement period, which is dependent on the time of year for the starting measurement. The R Program (R Core Team 2018) was used to run MK tests on all wells in each group, which were assessed using Mann Kendall’s tau and a p-value (α = 0.05). Significant results indicated the existence of a decline or recovery trend, and insignificant results suggested no water level change.
Simple linear regression (SLR) analysis using the R Program was applied after the MK analysis to understand the nature and degree of the detected trends (R Core Team 2018). Linearity was evaluated visually and with residual plots. Regressions with a p < 0.05 were considered significant with the slopes (output in meter per day and converted to meter per year) signifying a trend in the groundwater level.
A series of maps were created using the SLR analysis results to visualize the groundwater trend differences between Group A and Group B. First, the locations of Group A and B wells were spatially plotted in ArcGIS (ArcGIS Pro 2.2.0, 2018) using the available coordinate information. For Group A, these locations were gathered from the USGS National Groundwater Monitoring Network. For Group B, locations were plotted based on high-accuracy survey data collected using a Sokkia GRX2 GNSS receiver base station (Sokkia, Olathe, Kansas).
A raster surface of the groundwater trend was created for Group A using the natural neighbor tool in the ArcGIS Interpolation toolset. This process was completed using the trend rate results for each well from SLR analysis, with any insignificant trends represented as 0.0 m y–1. Because the Group A wells were spread across a larger area than the Group B wells, the raster interpolation also covered a larger region. This allowed a change-in-trend analysis to be conducted using the two raster outputs. Using the raster math tool in the ArcGIS 3D Analyst toolset, the difference in groundwater trends between the two monitoring periods was calculated by subtracting the Group A raster from the Group B raster. The resulting raster represented a change in the groundwater trend with positive values indicating a decreasing rate of decline, and negative values indicating an increasing rate of decline.
Results and Discussion
Cache River Critical Groundwater Area Decline and Recharge. Annual USGS spring measurements from wells within the CRCGA were interpolated to create the depth to groundwater maps in figure 1. The effects of pumping on the cone of depression are evident over this 19-year period. The cone of depression west of Crowley’s Ridge elongated, widened, and deepened over this time, as withdrawals continued at unsustainable rates compared to recharge. Historically, two cones of depression, which combined into a single depression by 2002, started forming west of Crowley’s Ridge in the 1980s—one depression centered in St. Francis County and another spread across Craighead, Poinsett, and Cross counties (Reed 2004). ANRC (2009) noted as early as 1984 that areas of south-central Craighead County west of Crowley’s Ridge had a critical saturated thickness of <6 m, described as ~50% of the predevelopment saturated thickness. More recently, the single depression remains centered over Poinsett County west of Crowley’s Ridge, and extends into Craighead and Cross counties. Yaeger et al. (2018) described the alluvial aquifer saturated thickness percentage change in relation to on-farm reservoir construction within the CRCGA and GPCGA. After 1996, 91% of the 78 reservoirs in the CRCGA were in areas with a remaining saturated thickness of 50% or less (Yaeger et al. 2018). Many new reservoirs (63% in CRCGA) were constructed on previously productive cropland. The next most common land use, representing 15% of new reservoirs constructed in CRCGA, was a combination of a field edge and a ditch, stream, or other low-lying area. It was suggested that reservoir construction in the CRCGA was likely motivated by recent aquifer declines and the increasing difficulty of obtaining groundwater (Yaeger et al. 2018).
This region receives sufficient annual precipitation for agriculture, but the timing and amount does not always coincide with irrigation needs. Alluvial aquifer declines west of Crowley’s Ridge were measured as early as the 1950s, at a rate of –0.3 m y–1 (Albin et al. 1967). Between 1950 and 2017, there was a 12-fold increase in irrigated land in Arkansas (Reba and Massey 2020). Groundwater use from the alluvial aquifer continued to rise to keep up with increases in irrigated area as well as growing irrigation dependence for historically dryland crops such as corn, soybean, and cotton (Kresse et al. 2014). More specifically in 2010, Poinsett County reported the largest water use from the alluvial aquifer, while also being one of the highest rice producing counties in the state (USDA NASS 2014; Kresse et al. 2014). Continuous flooding remains the standard irrigation practice for rice, accounting for the bulk of the agricultural water use in the area. Some producers use alternate rice irrigation methods such as intermittent flooding with multiple inlet rice irrigation (MIRI), which has been shown to reduce irrigation applications and field runoff (Martini et al. 2013; Massey et al. 2014; Reba et al. 2017).
The presence of the confining clay cap and lack of significant surface water sources diminish annual recharge rates. Godwin (2020) estimated the confining unit thickness using an interpolation of well logs in Craighead and Poinsett counties west of Crowley’s Ridge in the CRCGA. This interpolation was to update the previous confining unit mapping work completed by Gonthier and Mahon (1993) and Saucier (1994). Overall, the thickness of the confining unit in Craighead and Poinsett counties west of Crowley’s Ridge varied between 1.4 to 15 m, with no clear spatial trend, while much of the area was estimated with a thickness between 4.5 to 6.0 m (Godwin 2020). The thickest portions were hypothesized to be associated with paleochannels of the Cache River, with thicker portions also along the western edge of Crowley’s Ridge (Godwin 2020). With the confining unit at least 3 m thick in Craighead and Poinsett counties west of Crowley’s Ridge, lateral recharge via surface water should be a more important contributor to the groundwater recharge. However, the Cache River flowing on the western portion of the CRCGA cone of depression does not appear to contribute to groundwater level recovery (figure 1). Other lower order streamflow systems occur in Craighead and Poinsett counties. The headwaters of two rivers begin in western Poinsett County; however, only the downstream reaches are gauged by the USGS. Two small creeks flow off the southwest edge of Crowley’s Ridge through western Craighead County, and neither are ranked by the USGS with a flow category. Irrigation withdrawals and recharge from these surface water sources are unequal, resulting in continued decline of the alluvial aquifer.
Agricultural drainage ditches and on-farm reservoirs are widespread throughout the CRCGA; however, minimal infiltration is expected from these surface waters. Much of the local landscape has been transformed for agricultural production, drainage improvements, and levee and ditch construction. This, along with the presence of the confining unit and a depth to groundwater of up to 40 m, does not increase expectations that the drainage ditches are connected to the alluvial aquifer to provide recharge. It is unknown how much leakage from on-farm reservoirs results in groundwater recharge. However, the purpose and construction of these reservoirs aim for minimal infiltration loss to maximize storage and support a shift to using more surface water irrigation.
Groundwater Trends and Time Comparisons. The mean annual temperature was significantly higher in Group A (15.6°C ± 2.34°C) than in Group B (15.0°C ± 1.0°C) (t34 = 38.176, p < 0.001) (figure 2). In contrast, the mean annual total precipitation was significantly different between the two analysis periods, with Group B having higher mean annual precipitation (1,244 ± 262 mm) than Group A (1,015 ± 228 mm) (t34 = 24.931, p < 0.001). Monthly distributions of mean precipitation (D = 0.995, p < 0.001) and mean temperatures (D = 0.948, p < 0.001) were also significantly different based on KS tests and t-tests for every month (tables S1 and S2 in supplemental materials). The differences in both temperature and precipitation are consistent throughout most of the year (figure 2). Group B precipitation amounts were greater for the spring/recharge months compared to Group A, as well as during the growing season months. These differences in precipitation between the two periods could impact potential recharge via infiltration and lateral recharge, replenishment of water in reservoirs for later irrigation, and support a lesser need for supplemental irrigation. In contrast, if one period had significantly higher average temperatures, as was the case during Group A, the loss of recharge water to evaporation could be greater, likely resulting in the necessity for groundwater irrigation during the growing season, which could potentially contribute to groundwater level declines.
Monthly temperature (°C) and precipitation amounts (mm) for data periods—Group A, 1985 to 2012 (clear box) and Group B, 2012 to 2019 (gray box). Data were collected from the National Oceanic and Atmospheric Administration National Climatic Data Center historic observation database in Jonesboro, Arkansas (Jonesboro 2 NE Station USC00033734) (https://www.ncdc.noaa.gov).
The groundwater trends in the study area varied greatly by period of analysis and location. In all cases, the SLR results agreed with the results of the MK analysis (table S3). R2 values for wells with significant trends ranged from 0.16 to 0.99, with a median of 0.78 (table S3). The MK analysis for Group A indicated significant trends in all cases (p < 0.02), which was confirmed by SLR analysis (p < 0.02) (table S3). The MK analysis detected significant trends for all but eight well sites, and the insignificant well data trends were all in Group B (0.06 < p < 0.90) (table S3). The SLR analysis for these eight wells agreed in all cases (0.05 < p < 0.81), and the groundwater trends for these sites were no change (0 m y–1) (table S3). Similarly, MK and SLR analysis corroborated in all cases for Group B.
Groundwater level trends were consistently linear within both observation periods, although some wells exhibited seasonality (table S3). All Group A wells exhibited trends of declining groundwater levels ranging from –0.17 to –0.44 m y–1 (table 2). Eleven of the 30 wells in Group B exhibited declining trends in groundwater level, ranging from –0.05 to –0.75 m y–1 (table 2). Eleven other wells in Group B had a reversal of declines with increases in the groundwater elevation of 0.04 to 0.16 m y–1 calculated by SLR at these locations (table 2). Eight wells in Group B had no significant change in the groundwater level trend (table 2).
Groundwater level trends (m y–1) for specific wells within Group A and B with bolded values indicating declines, italicized values indicating recoveries, and NS indicating no significant trend. The US Geological Survey wells are listed with their respective National Water Information System site numbers; other wells are the USDA Agricultural Research Service Delta Water Management Research Unit wells.
All 24 well locations from the Group A had steady declines in groundwater elevation of ≤–0.17 to ≤–0.44 m y–1 (figure 3). During the 1985 to 2012 period, Group A sites CC#23, CC#24, P#5, and P#6 all experienced linear declines exceeding –0.4 m y–1 (table 2). For example, the depth to water table at CC#24 increased from 25.6 m in March of 1988 to 35.9 m in March of 2012, a decline rate of –0.74 m y–1. During the same period, sites closer to Crowley’s Ridge and along the western edge of the study area reported less severe declines. The depth to the water table at site CC#35, on the western edge of Crowley’s Ridge, increased from 22.8 m in April of 1994 to 25.5 m in December of 2003 at an average rate of –0.17 m y–1 (table 2). Lower decline rates were associated with areas outside of the central cone of depression—near Crowley’s Ridge and western Craighead County. The study area within Poinsett County had consistent declines in groundwater elevation. The two areas of higher decline rates are evident (figure 3). The first is located at the south end of the study area in Poinsett County. This is generally the center of the CRCGA cone of depression to 2012 and is characterized by three high-decline wells. The second area is centered on CC#23 and located in southern Craighead County. This area is associated with the elongation and expansion of the cone of depression as also seen in figure 1.
Groundwater level trend (m y–1) for Group A during 1985 to 2012 from the US Geological Survey monitoring well measurements. Wells with more than eight years of records were used in this interpolation. This interpolation was created using the natural neighbor tool in the Esri ArcGIS Interpolation toolset.
Group B groundwater trends during 2012 to 2019 exhibited decline, no change, and recovery of water levels (figure 4). The groundwater level for some wells declined at rates of ≤–0.10 to ≤–0.50 m y–1, similar to Group A rates. Of the 30 wells with data during this period, one Poinsett County well exhibited a greater decline (PC#14: –0.75 m y–1) than the rates observed in Group A. Most wells had either no significant change in groundwater level (8 wells) or showed recovery (11 wells), with rates ranging from ≤0 to ≤0.25 m y–1 (table 2 and figure 4). The recovery area was largely contained in Craighead County with specific well recovery rates from 0.06 m y–1 (CC#10 and CC#14) to 0.32 m y–1 at CC#21 (table 2 and figure 4).
Groundwater level trend (m y–1) for Group B 2012 to 2019 from the US Geological Survey and USDA Agricultural Research Service Delta Water Management Research Unit monitoring well measurements. Wells with two or more years of continuous data were used. This interpolation was created using the natural neighbor tool in the Esri ArcGIS Interpolation toolset.
The comparison of the groundwater trend maps from each period revealed a significant shift in trend patterns in the study area (figure 5). Water level changes in the groundwater trends post-2012 ranged ≤–0.30 to ≤0.50 m y–1 (figure 5). A vast majority of the study area had a decreasing rate of decline, resulting in an improvement in water levels (figure 5 and table 2). The northern section of the study area in Craighead County had the greatest water level improvements with many wells showing recovery. The area near the center of the CRCGA cone of depression in Poinsett County and at the eastern edge of the study area were the only places to show an increase in decline rates.
Change in the groundwater level trend (m y–1) between Group A 1985 to 2012 and Group B 2012 to 2019. The polygons spread across the area are on-farm reservoirs. Groundwater trends for both periods were calculated from US Geological Survey and USDA Agricultural Research Service Delta Water Management Research Unit well measurements.
An explanation for the water level trend differences between Group A and Group B is a shift to using on-farm reservoir-tailwater recovery systems for irrigation. Omer et al. (2018a) described the water budget for tailwater recovery systems in the Mississippi Delta and estimated that these systems offset 15% of the annual alluvial aquifer’s deficit through surface water irrigation and groundwater seepage. Figure 5 depicts the locations of these systems within the study area. These systems were installed due to the critical groundwater levels in the area, with the purpose of reducing the dependence on groundwater for irrigation. These systems are increasing surface water storage, which helps resolve temporal disconnections between groundwater supply and demand (Scanlon et al. 2012). The depletion areas at this time seem to be localized to Poinsett County, at the center of the cone of depression, with the county historically reporting the greatest groundwater withdrawals from the alluvial aquifer (ADANRD 2020). Focused conservation efforts at local, state, and federal levels will need to be continued in this area to ensure the sustainability of the alluvial aquifer.
If further depletion occurs within the CRCGA, agricultural production will suffer the greatest impact, which will lead to effects on the state’s economy. Without a secure irrigation source, agriculture may be converted to more dryland crops. Rice production would likely decrease, or production costs would increase due to pumping from deeper wells. Producers could turn to pumping from the deeper Sparta aquifer; however, pumping costs would be higher, and pumping rates would likely be unsustainable.
Recommendations. Additional groundwater recovery could result from prioritizing federal funding and assistance for the use of water conservation practices in critical groundwater areas. One such example is the Arkansas Groundwater Initiative (AGWI). This project partners the USDA NRCS with the USDA ARS-DWMRU, USGS, and ADANRD to target water conservation approaches in the GPCGA and CRCGA to address critical groundwater declines (USDA NRCS 2019). Financial and technical assistance will be provided to producers to accelerate voluntary installation of conservation practices with the overall goal of sustaining agriculture through stabilizing groundwater levels. Funding priority will be determined through a tiered system with producers that utilize water conservation practices in areas experiencing the most severe declines such as the GPCGA and CRCGA being of the highest priority for federal assistance.
Based on the results of this study, it is recommended to add continuous groundwater stations in conjunction with targeted implementation of water conservation practices. Groundwater level monitoring stations should span north to south through the middle of west Craighead and Poinsett counties with a smaller number spanning east to west across central-west Poinsett County. This would be beneficial to track the further development of the cone of depression, considering the importance of protecting the main agricultural irrigation supply for the state. If warmer temperatures occur in the future, these water level data will be in high demand for timely groundwater management decisions. Water managers could use such data to act effectively to help protect groundwater resources for future generations.
Drier summer conditions and variable precipitation could exacerbate growing season irrigation use resulting in greater water level declines and minimize recovery as seen in Group A. If declines revert to Group A rates, the alluvial aquifer is unlikely to be able to sustain the groundwater quantity that it has in the past. The implications of a return to the Group A decline rate would be increased pumping costs due to lowered water levels, decreased well yields due to reduced saturated thickness, potential increased pumping interference between neighboring farms if current pumping rates are continued at a reduced aquifer storage capacity, and increased risk of depletion in the deeper Sparta aquifer (ANRC 2009).
Summary and Conclusions
The Mississippi River Valley alluvial aquifer provides over 90% of irrigation water for agriculture in the Lower Mississippi River Basin. Overexploitation of the alluvial aquifer in the CRCGA in Arkansas started impacting crop production in the 2000s with previous historical data identifying sustained groundwater declines. Many producers in this region have been proactive both independently and with assistance from USDA NRCS in implementing on-farm reservoir-tailwater recovery systems in response to declining groundwater levels. More focused and consistent water level measurements were necessary to identify any groundwater level changes. Groundwater trends focused on the CRCGA cone of depression were compared from two-time groupings: 1985 to 2012 and 2012 to 2019. Groundwater levels during 1985 to 2012 had consistent decline rates of –0.17 to –0.44 m y–1. Groundwater levels during 2012 to 2019 varied with decline rates of ≤–0.10 to ≤–0.50 m y–1, but also areas of no change and recovery at rates of ≤0 to ≤0.25 m y–1. Trend differences between the two periods showed improved water levels for most of the study area, with rates ranging ≤–0.30 to ≤0.50 m y–1. The measured improvements were attributed to less pumping due to greater precipitation and the use of surface water irrigation from 60 on-farm reservoir-tailwater recovery systems. Detection of these improvements is crucial to understanding the response and status of the alluvial aquifer within the agricultural area of the CRCGA. It is recommended that adding continuous groundwater monitoring stations throughout west Craighead and Poinsett counties would be beneficial to track the further development of the cone of depression, considering the importance of protecting the main agricultural irrigation supply for the state. If warmer temperatures occur in the future, higher temporal and spatial groundwater level data in the CRCGA will allow water resource managers to make timely groundwater management decisions to help ensure its supply for future generations.
Supplemental Material
The supplementary material for this article is available in the online journal at https://doi.org/10.2489/jswc.2022.00170.
Acknowledgements
This research was supported by Cotton Incorporated Core Funding project number 13-786, a nonassistance cooperative agreement with Arkansas State University grant number 58-6024-6-006, and the USDA Agricultural Research Service (ARS). The authors are grateful for the collaboration with area producers. The authors acknowledge the technical assistance of Lindy Alexander, Hadley Coleman, W. Jonathon Delp, Rush Evans, Patrick Leppold, Josh McNatt, and Richard Smith of the USDA ARS-Delta Water Management Research Unit (DWMRU). This manuscript was improved by the comments from three anonymous reviewers. There are no conflicts of interest for any author. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the USDA, which is an equal opportunity provider and employer.
The data repositories for the US Geological Survey groundwater level data are available at https://waterdata.usgs.gov/nwis. The climate data are available at https://www.ncdc.noaa.gov. The USDA ARS-DWMRU groundwater level data are available through USDA Ag Data Commons, Leslie, Deborah L; Reba, Michele. (2020). Data from: Groundwater trend comparison during 1985 to 2019 in a critical groundwater area of northeastern Arkansas. Ag Data Commons. https://doi.org/10.15482/USDA.ADC/1519425.
- Received October 6, 2020.
- Revision received April 29, 2021.
- Accepted May 10, 2021.
- © 2022 by the Soil and Water Conservation Society
