Water Use

In 2015, groundwater withdrawals in the State of Delaware totaled 170 million gallons per day (Mgal/d) (Dieter and others, 2018), an 11.1 percent increase from the 2010 estimate of 151 Mgal/d (Masterson and others, 2016). The allocation of Delaware groundwater withdrawals is summarized in figure 3.

Hydrogeologic Setting

The unconfined surficial aquifer thickens from north to south and lies over several confining units and confined aquifers (Denver and Nardi, 2016). Flow paths within Delaware’s surficial aquifer are relatively short, with a majority of estimated groundwater ages between 30 and 50 years (Sanford and Pope, 2013). Younger groundwater is present near the surface of the aquifer, whereas older groundwater from upgradient recharge areas is present at depth in the aquifer beneath the younger water (Ator and Denver, 2012). Flow paths in the confined aquifers are much longer, corresponding to much older groundwater ages (Sanford and Pope, 2013). Recharge to the surficial aquifer occurs over most of the land’s surface because of the sandy nature of the overlying soil and aquifer sediment, with mean annual recharge estimated between 14 and 17 inches per year (Sanford and Pope, 2013). Recharge to the surficial aquifer can occur throughout the year and has a relatively equal distribution between the growing (March to September) and non-growing seasons (October to February) (Stahl and others, 2020). Recharge to the aquifer is possible even with the high evapotranspiration demand of summer, as groundwater levels respond to precipitation during intense rainfall events (Denver and others, 2018). Descriptions of geologic formations that compose the surficial aquifer and lower confining units and aquifers may be found in Fleming and others (2017).

Groundwater Chemistry

Groundwater chemistry is influenced by the dissolution of minerals in aquifer sediments, inputs from the land surface, and reduction-oxidation (redox) conditions. Relatively insoluble siliciclastic sediments dominate the Delaware surficial aquifer and their dissolution results in naturally dilute groundwater with specific conductance values less than 60 microsiemens per centimeter (µS/cm) and nitrate concentrations of less than 0.4 mg/L (Denver, 1989; Hamilton and others 1993; Ator, 2008). Due to its proximity to the land’s surface, groundwater of the surficial aquifer is susceptible to transport of chemicals from the land surface through the soil zone during recharge events. As a result, land management practices are commonly the dominant driver of shallow groundwater quality in the surficial aquifer; for example, salt and salt-brines applied to reduce ice on roadways can dissolve and infiltrate the soil, causing an increase in groundwater concentrations of sodium and chloride ions (Ator and Denver, 2012; Fleming and others, 2017). In 2014, wells sampled near urban areas and high road density also had higher specific conductance related to sodium, potassium, and chloride (Fleming and others, 2017).

In agricultural areas, soil amendments such as fertilizer, manure, and lime are applied to boost soil fertility and meet crop growing needs. Macronutrients such as nitrogen, phosphorous, calcium, magnesium, and sulfur, as well as minor nutrients such as boron and manganese, are commonly applied to improve soil fertility (Abaye and others, 2006). Sources of nitrogen and phosphorus that are applied to cropland include commercial fertilizers and poultry litter. Historically, most of the nitrogen delivered to cropland came from manure, but concerns related to phosphorus buildup in soil from these inputs resulted in restrictions on manure application (U.S. Department of Agriculture Natural Resources Conservation Service, 2013). As the allowance for manure applications decreased within the state, the nitrogen required for plant nutrition was supplied through the application of commercial fertilizers. Crop fields in Delaware also receive soil acidity treatments in the form of calcium and magnesium rich limestone. The combination of the soil fertility treatments used in modern agriculture leads to groundwater chemistry being dominated by magnesium, calcium, and nitrate in areas influenced by agriculture (Denver, 1989; Hamilton and others, 1993; Böhlke 2002; Fleming and others, 2017).

Network Design

The shallow aquifer well network is composed of wells previously used to monitor pesticides (Blaier and Baxter, 2000) and water quality within the surficial aquifer (Koterba and others, 1995; Shedlock and others, 1999; Debrewer and others, 20074612). The original 48 wells were selected owing to their shallow screened depth (11–34 feet), proximity to agricultural areas, and their likely high oxygen content (Fleming and others, 2017). Much of the previously reported and documented spatial variability in groundwater nitrate concentrations on the Delmarva Peninsula included results from networks with wells in both oxic and anoxic groundwater (Debrewer and others, 2007). The sampling of the shallow aquifer network in 2014 and 2019 was designed to represent shallow, oxic groundwater conditions in Delaware’s surficial aquifer. In 2019, samples were collected between October and December for a direct comparison to the sampling time frame in 2014. Wells with higher oxygen content were sought for this network as low dissolved oxygen leads to the removal of nitrogen from water through microbial activity. The original 48 wells sampled in 2014 (Fleming and others, 2017) were revisited in 2019 to assess for resampling following USGS protocols (USGS, variously dated). An initial reconnaissance of available wells from the 2014 sampling event was completed and three wells were determined to be unusable for sampling; one well was found dry (Pf41-02), one was destroyed (Fa45-07), and one well owner could not be reached for permission (Oc21-03). A well identified as Ib32-08 was added to the group of wells to sample (table 1). A total of 46 groundwater wells were sampled for this study.

Groundwater Sample Collection and Analysis

Groundwater samples were collected using methods outlined in the USGS National Field Manual for the Collection of Water-Quality Data (USGS, variously dated) and the sampling protocol outlined in Fleming and others (2017). All groundwater samples were collected using Teflon tubing and a 0.45-micrometer capsule filter inside a clean sampling chamber. Filtered water samples for major inorganics analysis were preserved using nitric acid to a pH below 2. All samples were analyzed for field parameters, nutrients, major ions, and stable isotopes of water including hydrogen and oxygen, and the δ¹⁵N and δ¹⁸O isotopes of the nitrate.

All samples were maintained at a temperature below 4 degrees Celsius in a sealed cooler during shipment to the laboratory. Samples from all wells were analyzed for major ions and nutrients at the USGS National Water Quality Laboratory (NWQL) in Denver, Colorado using methods described in Fishman (1993), and samples for stable isotopes were analyzed at the USGS Stable Isotope Laboratory using methods described in Coplen and others (2012) and Révész and others (2012). Sampling results from this study are available for comparison to sampling results from past and future studies, data collected during this study are available through the Water Quality Portal (National Water Quality Council, undated).

Quality Control

Equipment blanks, field blanks, and sequential replicate samples were collected following protocols described in Koterba and others (1995) to estimate potential contamination bias and measurement variability from water-quality data-collection processes. An equipment blank was collected prior to the commencement of sampling. Four field blanks and five replicates were collected during field activities at selected wells.

The field blanks were collected to ensure that sample collections and processing did not result in contamination. No nutrients or major ions were detected in the field blanks. Replicate samples measure the combined precision of sampling and laboratory analysis procedures. Replicate samples demonstrated consistent chemistry with their paired environmental sample. A relative percent difference (RPD) of 20 percent between environmental and replicate sample chemistry was used as an indication of variability from sampling procedures for this study. In the 2019 sampling event, the only instances of greater than 20-percent RPD occurred near the reporting limit of analytic instrumentation.

During this study an issue was discovered in quality assurance at the NWQL for a single dissolved orthophosphate (OP) result for one site. An error in the calibration procedure of the analytical instrument led to a result that the NWQL suggested was biased low, meaning the actual OP concentration was likely higher. The affected site, Nf52-02, had an OP concentration of 0.051 mg/L. Although the 2019 OP concentration from well Nf52-02 was withheld from the statistics table, it’s concentration of 0.051 mg/L is nearly identical to the well’s 2014 concentration of 0.05 mg/L.

Data Analysis

A nonparametric matched pair test, the Wilcoxon Signed Rank (WSR) test (Helsel and others, 2020), was applied to compare nutrient or major ion concentrations between 2014 and 2019 for wells that had been resampled. The Wilcoxon Rank Sum (WRS) test for unpaired samples was used to determine statistical difference between the Agricultural, Mixed, and Urban Groups. Resampled wells in 2019 were also grouped based on previously assigned geochemical classification described in Fleming and others (2017) for agricultural, mixed, and urban water types. Stable isotopes of water sampled in 2019 were examined and compared against modeled isotope data (Bowen, 2018) to estimate the source of sampled water. Nitrogen isotope ratios from sampled nitrate were evaluated to identify sources of nitrate in groundwater, including inputs from synthetic fertilizers, manure, septic discharge, or natural processes (Böhlke, 2002).

Spatial Analysis

Spatial analysis was used to relate observed water quality to potential influences at the land surface. Agronomic survey and census data related to the acreage of various crops and years of production were provided by USDA-NASS (undated). Spatial datasets of estimated enrichment ratios of stable isotopes within precipitation were used to estimate approximate season of groundwater recharge (Bowen, 2018)

Field Parameters

Samples had a median specific conductance of 220 µS/cm and ranged from 80 to 667 µS/cm in 2019 (table 2). The Urban Group had higher specific conductance concentrations than the Mixed Group (WRS p<0.05) and the Agricultural Group (WRS p=0.09). The median specific conductance of all samples in 2019 were slightly lower than 2014 but the difference was not statistically significant. The differences within groups between the 2014 and 2019 sampling events were not statistically significant.

In 2019, samples had a median pH of 5.2 and ranged from 4.4 to 6.7 (table 2). The Urban Group had a higher median pH of 5.6 than the Mixed Group with a median of 4.85 (WRS p<0.05) and the Agricultural Group with a median 5.05 (WRS p=0.07). There was no statistically significant difference between median concentrations in 2014 and 2019 or, within groups between 2014 and 2019.

Dissolved oxygen values for the 2019 sampling event ranged from 1.7 to 9.6 mg/L and had a median concentration of 5.3 mg/L (table 2). Although there were slight differences between groups, the differences were not statistically significant (WRS p>0.05). Although the 2019 sampling event had lower median than the 6.3 mg/L found in 2014, this difference was also not significantly different (WRS p=0.2).

Nitrate

In 2019, as in 2014, nitrate was the dominant form of nitrogen found within the shallow aquifer network of the Delaware Coastal plain. Samples from the entire network had a median nitrate concentration of 5.09 mg/L in 2019, with a range from 0.45 to 30.7 mg/L. When the wells were split into their geochemical groups and compared, water sampled from the Agricultural Group had a median nitrate concentration of 10.7 mg/L, higher than that of the Mixed Group (4.95 mg/L; WRS p<0.05) and the Urban Group (1.75 mg/L; WRS p<0.0001); the difference between the Mixed and Urban Groups was also statistically significant (WRS p<0.05) (table 2; fig. 4). The difference between groups in 2019 was similar to the 2014 sampling, where nitrate concentrations between groups were significantly different from one another (Fleming and others, 2017; WRS p<0.05). When the respective Agricultural, Mixed, and Urban Groups of 2014 and 2019 were compared, no significant changes were found (fig. 4). The median nitrate concentration of all samples in 2019 was slightly higher than that observed in 2014, but not statistically significant (WSR p=0.56; table 2). Eleven samples collected in 2019 had nitrate concentrations above 10 mg/L, the U.S. Environmental Protection Agency (EPA) primary drinking water quality threshold (EPA, 2024a). All but 2 of the 11 samples from the Agricultural Group were higher than the threshold.

Relationships Between Specific Conductance, Nitrate, and Chloride

In 2014, specific conductance was positively correlated with concentrations of chloride in the Urban Group and both chloride and nitrate in the Agricultural Group (Fleming and others, 2017). Data from 2019 showed the same relationship between chloride and specific conductance in the Urban Group of wells, where a linear regression indicated that specific conductance described a high degree of variability in chloride concentrations (r²=0.87). Wells within the Urban Group maintained a significantly higher median concentration of chloride than wells in the other groups (WRS P < 0.05). Correlations between specific conductance and nitrate among groups varied from weak (r²=0.041) in the Urban Group to moderate (r²=0.68) in the Agricultural Group. When data from 2014 and 2019 were combined and the linear regression was repeated, the relationships within groups remained consistent (fig. 5). A multivariate regression using both specific conductance and magnesium as explanatory variables increased the r² value to 0.82 within the Agricultural Group (table 3). The power of specific conductance and magnesium to predict nitrate concentrations decreased in the Mixed and Urban Groups (table 3), which have lower magnesium concentrations than the Agricultural Group (WRS p<0.01; table 2).

Orthophosphate

Groundwater concentrations of OP were generally low; with 18 of the 46 samples above the detection limit of 0.004 mg/L (table 2). Most OP detections occurred in the Agricultural and Urban Groups where almost half of the samples collected had a detection of OP. Orthophosphate concentrations within groundwater are generally low as OP is easily bound to ionic exchange sites on soil and sediment (Sharpley and others, 2013). Soils that have received repeated phosphorus applications may experience a saturation of available binding sites and OP may be released to subsurface transport and groundwater (Kleinman and others, 2007). Although OP concentrations measured in the 2019 sampling were generally low, three sites had OP concentrations above the recommended regional criteria for total phosphorus of 0.031 mg/L (EPA, 2000). The three sites with high OP concentrations in 2019 also had OP concentrations above 0.031 mg/L in the 2014 sampling effort. Orthophosphate concentrations at or above eco-region specific criteria are likely the result of sediments that are near or above saturation of adsorbed phosphorus (Domagalski and Johnson, 2011). In comparison, the Choptank River at Greenboro, Maryland, surface-water gaging station (station 01491000), which has a majority of its watershed in Delaware, had a median OP concentration of 0.041 mg/L and a mean of 0.048 mg/L for all environmental samples collected from 2014 to 2019, respectively (USGS, undated).

Chloride, Sodium and Chloride-to-Bromide Ratios

Chloride observed in 2019 had a median concentration of 17.75 mg/L, which was less than the 2014 median of 27.15 mg/L but not significantly different (WRS p=0.2; table 2). Wells within the Agricultural Group had a similar chloride concentration in both sampling years; the Mixed and Urban Groups had lower concentrations in 2019 compared with that in 2014 (table 2). Unlike in 2014, there were no samples which exceeded the chloride EPA Secondary Drinking Water Standard of 250 mg/L in 2019 (EPA, 2024b). In both 2014 and 2019, samples with high sodium also had higher chloride concentrations (fig. 6A). Wells from the Mixed and Urban Groups had consistently higher concentrations of sodium and chloride regardless of spatial position (WRS P <0.05; fig. 6A); this relationship suggests that proximity to urban land use and roadways is a stronger indicator of sodium and chloride concentrations than latitude in Delaware.

Previous studies have used the mass ratio of chloride to bromide as an indicator of chloride source (Mullaney and others, 2009). The ratio of chloride to bromide generally found in the Agricultural well Group (fig. 6B) are similar to agricultural areas surveyed by Mullaney and others (2009), which received both animal manure and potash fertilizer (which contains high concentrations of potassium). In 2019, values from the Urban Group (fig. 6B) were indicative of land that received deicing treatments (Mullaney and others, 2009). In areas with dilute groundwater, ratios of chloride to bromide are relatively low (below 1000) and increase as anthropogenic sources of chloride are added (Mullaney and others, 2009). The chloride-to-bromide ratio increases because anthropogenic sources of chloride such as deicing agents and water softeners have high concentrations of chloride and little bromide.

Silica

Silica in groundwater comes from the dissolution of silicate minerals that dominate soil and aquifer sediments and has previously been related to estimated groundwater age (Denver and others, 2018). In 2014, 18 samples from the shallow network were analyzed for estimated groundwater age using sulfur hexafluoride (SF₆; Fleming and others, 2017). Figure 7A shows the relationship between silica concentrations and estimated groundwater age and shows a wide range of silica concentrations indicated similar groundwater age and a weak relationship driven by an outlying data point. The high variation in silica to groundwater age is likely because of the variation in aquifer properties within the shallow well network compared to that of an individual field (Denver and others, 2018). In addition, a WRS test found significant differences between the Urban and Mixed Groups, and the Urban and Agricultural Groups in both 2014 (fig. 7B) and 2019 (fig. 7C). The differences observed between these groups suggest that, although silica may be a reliable surrogate for age on a local scale, there is insufficient evidence to estimate groundwater age from silica on a state-wide basis. The differences between geochemical groups suggest that land use may also influence silica concentrations. Despite the weak relationship between silica concentrations and relative groundwater age calculated from SF₆ concentrations, the relative ages (fig. 7A) of samples collected from this network suggest that all water sampled was modern in nature.

Isotopes

Stable Isotopes of Water

The isotopes δ²H and δ¹⁸O are naturally occurring variations of hydrogen and oxygen and are found in low concentrations within water molecules. These molecules are heavier than their counterparts δ¹H and δ¹⁶O because they contain additional neutrons within the nucleus. The relative enrichment of water with δ²H and δ¹⁸O isotopes is determined by physical processes such as evaporation and precipitation. The ratios of isotope occurrence in water are known as “enrichment ratios” and they can indicate the source and possibly timing of groundwater recharge from precipitation. Previous studies on stable isotopes indicate that recharge may occur year-round in Delaware (Stahl and others, 2020), which is supported by recharge estimates by Sanford and Pope (2013).

Stable isotope samples analyzed from the shallow aquifer network were compared against monthly isotope data modeled from the Isomaps Program (fig. 8; Bowen, 2018). Water sampled during 2019 fell within the range of previously sampled groundwater on the Delmarva Peninsula. One well, Ph34-15, stood out from the other samples owing to its low enrichment of both δ²H and δ¹⁸O compared to the rest of the network (fig. 8). This well is close to the Delaware inland bays (fig. 1) and it is possible that the water is influenced by them; oceanic sources of water are generally less enriched than water from continental land masses (Stahl and others, 2020). When compared with the Water Isotope Program data, the Delaware shallow isotopes were in the middle of the range of values estimated by the Isomaps data, suggesting that the water sampled was from mixed time periods and or sources. This would be expected as water samples from a well with a 3-foot screened interval, such as those in this network, which represent a composite of water from several recharge events that occurred over a longer time frame than a single recharge event.

Nitrate Isotopes

Nitrate isotopes δ¹⁵N and δ¹⁸O are naturally occurring isotopes of nitrogen and oxygen. Their abundance is determined by the sources of nitrogen and biological transformations such as denitrification (Böhlke and Denver, 1995). Sources of nitrogen added to the landscape and in the atmosphere provide the basis for δ¹⁵N concentrations in water. Synthetic nitrogen fertilizers (such as urea or ammonium sulphate) are created from atmospheric N² gas and generally range from −3 per mil (‰) δ¹⁵N to +7‰ δ¹⁵N (Michalski and others, 2015). A survey of 270 different synthetic fertilizers sold in the United States in 2015 indicated that 80 percent of fertilizers had −3‰ δ¹⁵N to +3‰ δ¹⁵N (Michalski and others, 2015). Once in the soil profile, nitrification processes alter the commercial fertilizer such that the commonly found range of isotopes are −5‰ δ¹⁵N to +5‰ δ¹⁵N (Michalski and others, 2015). Other sources of nitrogen that come through biological pathways, such as manure or septic systems, typically have higher δ¹⁵N (Kendall and others, 2007). Previous literature indicates that manure and septic systems have a range of 10‰ δ¹⁵N to 20‰ δ¹⁵N (Böhlke and Denver, 1995; Kendall and others, 2007). Denitrification can produce nitrate values between 10‰ and 30‰ δ¹⁵N and similarly raise δ¹⁸O isotopes fractions compared to the groundwater pool of nitrate (Kendall and others, 2007).

The nitrogen isotopes measured in nitrate from water samples in 2019 indicate the main source of nitrogen measured in the shallow aquifer network appears to be a mixture of synthetic fertilizer and manure in all groups (fig. 9, table 4). There was no significant difference between groups when comparing nitrate isotopes. Partial denitrification was evident in some of the water samples owing to their greater enrichment of both nitrate isotopes (fig. 9). Greater enrichment of nitrate isotopes can occur as microbes preferentially metabolize isotopes with less mass (Kendall and others, 2007). Water from well Gc14-04 had the highest δ¹⁵N and δ¹⁸O values; this well is located near wetlands and had a relatively low dissolved oxygen value of 1.8 mg/L in 2019 compared to the rest of the network. Despite the relatively low oxygen content and the enriched isotopic signature, the 2019 sample from Gc14-04 contained 6.1 mg/L nitrate.

Table 4.    

Summary Statistics for stable isotopes of water and the nitrogen and oxygen isotopes of nitrate from the 2019 sampling of the Delaware Shallow Aquifer Network, samples are split into geochemical groups.

[Geochemical groups were identified by Fleming and others (2017). N, number of samples; min, minimum; max, maximum; δ, delta]

Table 4.    Summary Statistics for stable isotopes of water and the nitrogen and oxygen isotopes of nitrate from the 2019 sampling of the Delaware Shallow Aquifer Network, samples are split into geochemical groups.

Constituent	All samples				Geochemical groups
					Agricultural				Mixed				Urban
	N	Min	Median	Max	N	Min	Median	Max	N	Min	Median	Max	N	Min	Median	Max
Nitrate isotopes
δ15N	46	−1.97	5.29	29.76	15	3.49	5.6	29.76	13	−1.97	2.48	9.19	17	0.08	5.21	9.35
δ18O	46	−3.08	1.345	13.43	15	−0.89	1.43	13.43	13	−0.96	1.37	6.03	17	−3.08	1.15	5.98
Stable isotopes of water
δ2H	46	−43.9	−38.475	−16.24	15	−42.36	−39.39	−29.58	13	−43.9	-36.36	−35.02	17	−41.58	−38.28	−16.24
δ18O	46	−7.23	−6.56	−3.89	15	−7.23	−6.6	−5.49	13	−7.12	-6.44	−5.78	17	−6.98	−6.56	−3.89

Factors Influencing Groundwater Chemistry

The Delaware Shallow Aquifer Network was designed to sample shallow groundwater across the State of Delaware to better understand spatial and temporal variation in groundwater chemistry through repeated sampling over time. Sampling the recently formed groundwater within the shallow well network would demonstrate changes in water chemistry driven by land management (fig. 7). However, results from the second sampling of the network in 2019 suggest that there was little change in nutrients and most other major ion concentrations.

Age-dating techniques determined that the groundwater sampled during this study generally formed within 10 years of the sampling time (fig. 7A). Assuming the wells not analyzed for age-dating characteristics had similar ages owing to their depth and placement in the aquifer, the sampling of this network included groundwater that recharged to this aquifer between 2004 to 2014 for the first round of sampling and 2009 to 2019 for the second (fig. 10). The range of corn yields over each 10-year period indicates that there were factors influencing nutrient utilization of crops, which varied inter-annually. In 2014, for example, the water samples reflected years where Delaware corn yields varied from 99 bushels per acre in 2007 to 200 bushels per acre in 2014. This variation in yield is likely related to variable growing conditions and not drastic changes in farming practices in the state. This inter-annual variation in crop production may have implications for water quality on shorter time scales, as previous research has shown considerable variation in groundwater chemistry within the growing seasons (Denver and others, 2018).

Shallow groundwater chemistry may vary over short timeframes as there is a short travel distance between land management activities and the groundwater reservoir. Due to the variable nature of management activities and timing of precipitation, recharge events can concentrate or dilute water chemistry. Variations in groundwater nitrate concentrations on a seasonal basis were shown by Denver and others (2018), who demonstrated near Bucks Branch in Delaware that nitrate-nitrogen concentrations were the highest in both soil and very shallow groundwater during the main growing season. The increased nitrate concentrations in soil and groundwater at Bucks Branch were related to nutrient application timing and precipitation events (Denver and others, 2018). The higher concentrations near the water-table surface decreased later in the growing season as more recharge reached the aquifer. One of the monitoring wells included in this study, Ng45-02, had daily conductance and nitrate values collected from 2014 to 2020 and 2016 to 2020, respectively (fig. 11, fig 12.). The variability in conductance demonstrated at Ng45-02 demonstrates relatively rapid changes in water quality. In figure 11, greater depth to groundwater indicates a relatively dry period, while wetter conditions lead to higher groundwater levels. Daily conductance data from Ng45-02 showed that a single recharge event may either increase ionic concentrations or dilute them (fig. 11). Increases in groundwater level frequently corresponded to increases in specific conductance (fig. 11). Owing to the strong relationship (r²=0.92) between specific conductance and nitrate at this site (fig. 12), we may infer that nitrate concentrations were also variable. Discrete samples of nitrate at Ng45-02 showed an increase in maximum observed concentrations from 13.1 mg/L in 2014 to 18.3 mg/L in 2019. This increase was likely caused by a change in cropping practice from consistent hay production to a corn and soybean rotation in 2014. This example from Ng45-02 illustrates how nitrate concentrations can change in groundwater in a relatively short timeframe when agricultural practices change significantly. Although samples collected in 2014 and 2019 targeted the same timeframe (October–December), there were differing antecedent moisture conditions leading into the sampling event.

The Palmer Drought Severity Index (PDSI) reflects the relative availability of moisture at a given point in time (Palmer, 1965). Positive PDSI values indicate wetter-than-average conditions and negative PDSI values indicate drier-than-average conditions. During the sampling events for 2014 and 2019, Delaware started the calendar year with relatively wet conditions (figs. 11, 13) that dried out over the summer growing season. This pattern of greater dryness towards the end of the growing season is consistent with daily groundwater elevations reported from Ng45-02 (fig. 11). In 2019, the calendar year started relatively wet compared to 2014 (fig. 13) but by late spring in 2014 and 2019, PDSI values were decreasing in the State of Delaware (fig. 13). During and after periods of typical nitrogen applications, the state experienced similar hydrologic conditions but there was greater antecedent moisture in 2014 during the period of sampling (fig. 13).

Although the groundwater chemistry is variable, the wells within the shallow aquifer network maintained relatively consistent concentrations of nitrate, magnesium, and chloride, which were defining characteristics of the geochemical groups in 2014 and 2019. Wells that had high nitrate and phosphorus in 2014 also had high nitrogen and phosphorus in 2019; the same was true for chloride concentrations. The sampling method of the shallow aquifer network was designed for a relatively short timeframe to control for seasonal variability in groundwater variability and changes in antecedent moisture conditions (fig. 13). This relatively short sampling window provided a synoptic of water quality across the state. Despite variation in groundwater chemistry between sampling events appearing limited, repeated sampling of groundwater networks allows for the detection of a trend despite variable weather signals (Lindsey and others, 2023). Continued sampling of the shallow aquifer network should indicate changes in long-term shallow groundwater chemistry.