Little River Experimental Watershed, a keystone in understanding of coastal plain watersheds
Abstract
The US Department of Agriculture-Agricultural Research Service Southeast Watershed Research Laboratory (SEWRL) initiated a hydrologic research program on the Little River Experimental Watershed (LREW) in 1967. Long-term (52 years) streamflow data are available for nine sites, including rainfall-runoff relationships and hydrograph characteristics regularly used in research on interactive effects of climate, vegetation, soils, and land-use in low-gradient streams of the US EPA Level III Southeastern Plains ecoregion. A summary of prior research on the LREW illustrates the impact of the watershed on building a regional understanding of hydrology and water quality. Climatic and streamflow data were used to make comparisons of scale across the nine nested LREW watersheds (LRB, LRF, LRI, LRJ, LRK, LRO, LRN, LRM, and LRO3) and two regional watersheds (Alapaha and Little River at Adel). Annual rainfall for the largest LREW, LRB, was 1200 mm while average annual streamflow was 320 mm. Annual rainfall, streamflow, and the ratio between annual streamflow and rainfall (Sratio) were similar (α = 0.05) across LREWs LRB, LRF, LRI, LRJ, LRK, and LRO. While annual rainfall within the 275 ha LRO3 was found to be similar to LRO and LRM (α = 0.05), annual streamflow and Sratio were significantly different (α = 0.05). Comparisons of annual rainfall, streamflow, and Sratio between LRB and the regional watersheds indicated no differences (α = 0.05). Based upon this analysis, most regional watersheds shared similar hydrologic characteristics. LRO3 was an exception, where increases in row crops and decreases in forest coverage resulted in increased streamflow. LREW data have been instrumental in building considerable scientific understanding of flow and transport processes for these stream systems. Continued operation of the LREW hydrologic network will support hydrologic research as well as environmental quality and riparian research programs that address emerging and high priority natural resource and environmental issues.
1 INTRODUCTION
1.1 Background
Long-term hydrologic and water quality data from watersheds are a key component to building an understanding that will dictate what and where changes in watershed management need to be made to achieve improvements in ecosystem services. The US Department of Agriculture-Agricultural Research Service (ARS) has a long history of collecting data to assess and understand the condition of agricultural production regions across the US. Motivated by the Dust Bowl of the 1930s and the knowledge that long-term hydrologic data were needed to better understand relationships between rainfall, land-management, and watershed streamflow, the Senate Select Committee on National Water Resources conducted nationwide hearings to review US water resources in 1959. These hearings resulted in Senate Document 59 (US Senate, 1959) which identified hydrology and agricultural watersheds as high priority research areas. As a result of this Document, appropriations were made to establish new watershed research centers at five locations across the US. These centers have proven to be critical to building a scientific understanding of the movement and storage of water within agricultural landscapes as a function of management and climate.
One of these five locations was the Southeast Watershed Research Laboratory (SEWRL) in Georgia. Prior to selection of the watershed to be studied, extensive geologic and hydrologic assessments were conducted (Yates, 1976). After considerable study across the state of Georgia, the Little River Watershed in south-central Georgia was selected in 1965 as the watershed of study and the SEWRL was established in Tifton, GA (Bosch, Sheridan, & Marshall, 2007) (Figure 1). The original Little River Experimental Watershed (LREW) contains the primary watershed Little River B (LRB) and seven sub-watersheds (LRM, LRJ, LRK, LRI, LRF, LRN, and LRO) in a nested and paired design (Figure 1) that facilitates studying relationships of scale and land management. The LREW is located in the headwaters of the Suwannee River Basin, a major US interstate basin that originates in Georgia and empties into the Gulf of Mexico in the Big Bend region of Florida (Figure 1). The watershed is located within the US EPA Level III Southeastern Plains ecoregion. The region has a favourable climate, fertile soils, and ample rainfall to make it a productive agricultural region. Its rich surface and subsurface water resources are essential for agriculture. This region is largely characterized by low-gradient streams with sandy soils surrounding the streams. Land-use across the region is a mosaic of cropland, pasture and forage, upland forest, riparian forest, and wetlands. Although the natural vegetation across the region was predominantly longleaf pine, considerable agricultural production now occurs throughout the Southeastern Plains (Frost, 2006). The most common regional crops include cotton, peanut, and corn (Bosch et al., 2020). Considerable production forestry and pastureland remain. Much of the Southeastern Plains contains dense riparian forests. In much of the region, these riparian forests form dense stream corridors and interact with surficial aquifers (Bosch et al., 1996). Baseflow fed by these surficial aquifers is a dominant component of streamflow (Miller, 1990; Sheridan, 1997; Shirmohammadi et al., 1984). Because of the shallow nature of the surficial aquifer, evapotranspiration (ET) demand of the stream corridor riparian buffers can impact groundwater and streamflow (Bosch et al., 1996; Bosch et al., 2006).
Streamflow data have been collected from the LREW since late 1967 (Bosch & Sheridan, 2007). Water quality data have been collected intermittently since the 1970s (Bosch et al., 2020; Feyereisen et al., 2007). The LREW stream monitoring program has provided fundamental data for scientific research evaluating landscape and watershed-scale hydrologic and water quality processes, rainfall-runoff relationships, hydrograph characteristics, water yield, as well as the integrative effects of climate, vegetation, soils, and land-use. Ultimately, the data are being used to develop and test prediction methodologies and for evaluating the impacts of agricultural and conservation management alternatives on water quantity and quality in the low-gradient streams that are characteristic of the region. Continued operation of this hydrologic network also supports the extensive environmental quality, ecosystem services, and riparian research programs of the SEWRL and its cooperators. Provided here is a brief overview of some of the key results from the watershed research program.
1.2 Hydrology and evapotranspiration
The long-term research conducted on the LREW has helped develop a keen understanding of the hydrologic characteristics of the watershed and its relationship to rainfall patterns and land-use. The primary loss component of the water balance is ET. The average annual ET rate for LRB has been quantified as 47% to 88% of annual rainfall (Bosch et al., 2004). Evapotranspiration was determined as the difference between annual rainfall and annual water loss. Annual water losses included annual streamflow and losses to regional groundwater. Losses to regional groundwater were assumed to be 1% of annual rainfall (Sheridan, 1997). In the LREW, streamflow is typically greater in the winter and early spring than during the summer and fall months (Sheridan, 1997; Shirmohammadi et al., 1986). Analyses of long-term hydrological trends have indicated that the winter and spring months typically have higher rainfall and lower ET (Bosch et al., 2017) which create greater soil moisture and aquifer recharge, increasing surface runoff responses and groundwater contributions to streamflow. In the LREW, direct surface runoff can vary between 5% and 40% of annual rainfall while shallow subsurface interflow can vary between 2% and 35% of annual rainfall (Asmussen & Ritchie, 1969; Bosch et al., 2012; Hubbard & Sheridan, 1983).
Baseflow is particularly important in this region of rich surface and subsurface water resources that are essential for agriculture. In the LREW, baseflow was found to produce 53% of annual streamflow while stormflow was found to produce 47% (Bosch et al., 2017). In other Coastal Plain watersheds of similar size and having similar characteristics, baseflow ranging from 58% to 73% has been reported (Novak et al., 2003). Due to the highly variable transmissivity of the surficial aquifer, baseflow contributions to streamflow can be quite temporally variable. In the LREW, contributions of baseflow to streamflow have been found to vary seasonally as a function of varying ET levels and aquifer storage (Bosch et al., 2017). Baseflow was the greatest during the months from December through May (55%–57%) and the least during the months from June through November (43%–46%). Annual baseflow was found to decrease with increasing annual rainfall, indicating that during high rainfall years saturation excess driven stormflow increases in the LREW. Hydrograph analysis indicated an average stormflow duration of 7 days in the largest Little River Watershed (LRB, Figure 1), typically extended by interflow in this watershed (Bosch et al., 2017).
While row-crops across the LREW play an important economic and environmental role, the LREW landscape is dominated by a dense dendritic network of stream channels bordered by riparian forest wetlands (Asmussen et al., 1979). Within LRB, 22% of the area is covered with upland forests while 28% is covered by riparian forests (Bosch et al., 2004, Bosch et al., 2006 LandSat data). These forested areas play a key role in the LREW hydrology and water quality. Evapotranspiration for the forested portion of the watershed is considered much larger than ET from the row-crops and pasture. Bosch et al. (2014) measured transpiration rates for a forested slash pine (P. elliottii Engelm.) riparian buffer and quantified the rates to be 1114 mm for the period from April to December, or 103% of potential ET.
1.3 Land-use changes
Major land-use changes are expected to have an impact on streamflow and water quality. Several publications have indicated that forested areas, because of increased ET, consume considerably more water than do similar non-forested areas. Work as early as 1909 indicated that cutting timber within watersheds could lead to significant increases in streamflow (Bates & Henry, 1928; Evans & Patric, 1983; Hibbert, 1966). Grace III (2005) reviewed studies conducted in 13 different states in the southern region of the US to evaluate the general impact of forest operations on water quantity and quality. In all cases where water quantity data were collected, forest harvest led to increased water yields (Grace III, 2005). Trimble et al. (1987) reported a 4% to 21% decline in annual stream discharge due to a 10% to 28% conversion of cropland to forested acreage in ten different watersheds in Alabama, Georgia, and South Carolina. Each of these studies indicate increases in forested acreage will result in increased watershed ET and decreased runoff or conversely increases in cropped acreage will result in increased watershed runoff.
Some minor changes in land-cover have occurred in the LREW which may have impacted streamflow in the headwaters of the watershed (Lowrance & Leonard, 1988). Lowrance and Leonard (1988) reported an increase in streamflow in a 2212 ha LRJ from clear cutting 240 ha of riparian forest near the watershed outlet. During the 2 years following the cutting, streamflow was 66% of the annual rainfall compared to 32% in other years (Sheridan, 1997). Further research conducted by Sheridan (1997) indicated that the percent riparian forest coverage in several watersheds in South-central Georgia was significantly correlated to watershed streamflow. Bosch et al. (2006) examined land-use patterns across the entire LREW during the period from 1975 to 2003 and found that while some changes in land-use have occurred, the changes were largely insignificant at LRB. In addition, few changes have occurred in the total forested acreage in LRB during this period (Bosch et al., 2006).
1.4 Water quality
Long-term hydrologic records are particularly critical for assessing watershed-scale effects of trends in climate, land cover, and land management on water resources, and in gaining an understanding of governing watershed processes. Bosch et al. (2020) studied long-term trends in water quality in the LREW and observed that concentrations of N and P have generally remained low and stable, with some increases observed from 1980 to 1999. Loading rates of N and P, although somewhat influenced by changes in concentration, are largely dictated by changes in streamflow volume (Bosch et al., 2020). In contrast, Cl concentration appears to be steadily increasing, possibly due to increased fertilization in the LREW (Bosch et al., 2020).
Low nutrient loading in the LREW is attributed to the dense riparian buffers in the watershed which have been shown to be effective for reducing stream sediment and nutrient loading. Research conducted in the 1970s and 1980s indicated that riparian zones are effective sediment and nutrient filters (Hubbard & Lowrance, 1994). One study quantified nitrate nitrogen (NO3-N) and orthophosphate phosphorus (PO4-P) losses from cropped areas and riparian wetland zones, and estimated that 96% and 37% of the NO3-N and PO4-P was retained, utilized, and/or transformed in the heavily vegetated riparian forest (Yates & Sheridan, 1983). Another study estimated that NO3-N and PO4-P transported by streamflow represents 5% and 32% of rainfall inputs and attributed such low loads to the heavily vegetated floodplains (Asmussen et al., 1979). Reduced NO3-N levels have been reported in water leaving the riparian zone compared to the agricultural upland and the filtering of nutrients by riparian systems has been attributed to both denitrification and vegetative uptake (Lowrance, 1992; Lowrance et al., 1983; Lowrance et al., 1984). Later research focused on looking at the effects of buffer management on stream water quality. Managed three zone buffers consisting of a grass strip, a managed forest, and an unmanaged forest, were observed to reduce most nutrient loads entering the stream from an upslope field (Lowrance & Sheridan, 2005). Restored forested riparian wetlands have been shown to be effective in reducing concentrations and loads of N and P derived from agricultural land (Vellidis et al., 2003). As illustrated by Bosch et al. (2014), the surficial aquifer interacts with the root system of the riparian buffer throughout the growing season. Because of this interaction, there is opportunity for nutrient uptake, denitrification, and dilution between the upland row-crops and the stream system. More recently, Pisani et al. (2020) reported that in the LREW the replacement of forested riparian buffers with agricultural land can lead to a shift in dissolved organic matter (DOM) composition toward recently produced, low molecular weight material with low aromaticity. This altered DOM pool may ultimately affect carbon cycling and downstream water quality. Thus, these riparian systems play a key role in maintaining and enhancing the overall water quality of the Southeastern Plains stream systems.
1.5 Climate change impacts
Most high-resolution climate change models predict further reductions in rainfall throughout the region, with increasing drought severity (Dai, 2010) and potential decreases in summer rainfall of 20%–30% (Mearns et al., 2003). Climate models that predict moderate increases in rainfall predict increases in ET that exceed rainfall increases and thus the net predictions are for reductions in river discharge throughout mid- and low latitudes (Mulholland et al., 1997; Nijssen et al., 2001). Recent data also suggest a trend toward more extreme rainfall events and an accompanying decrease in moderate rainfall events in the Southeastern U.S. (Dourte et al., 2015).
1.6 Need for research
- Compare hydrologic characteristics of the LREWs as a function of scale
- Compare LRB to regional surrounding watersheds
2 STUDY AREA AND METHODS
2.1 Site descriptions
2.1.1 LREW
The LREW is located in the western headwaters area of the Suwannee River Basin, centered at approximately N31° 36′ 36″ and W83° 39′ 36″ (Figure 1). The SEWRL initiated construction of a hydrologic monitoring network on the Little River in 1967. Construction of flow measurement devices and installation of the original LREW hydrologic monitoring instrumentation was completed in 1971 (Table 1). The LREW was instrumented to measure rainfall and streamflow for the primary 334 km2 LRB drainage area and seven subwatersheds that range from approximately 3 to 115 km2 (Table 1, Figure 1). The original hydrologic monitoring network has been in continuous operation since that time, with some breaks in service, upgrades, and modifications (Table 1) (Bosch & Sheridan, 2007). Rainfall and climatic data are collected in support of the hydrologic data. The supplemental climatic network currently consists of 42 rain gauges and 6 full climate measurement stations installed throughout the LREW and the Upper Suwannee River Basin (Bosch, Sheridan, Lowrance, et al., 2007). Thirty-six of these stations include soil moisture measurements in the top 300 mm of the soil profile.
| Watershed | Outlet coordinates | Data used | Data gaps | Complete years | Area (ha) | Stream order | Main Channel length (km) | Channel slope (%) | Length to width ratio (km km−1) | Drainage density (km km−2) |
|---|---|---|---|---|---|---|---|---|---|---|
| LRO3 | 31 30 27 N 83 32 40 W |
1 Dec 2007 to 31 Dec 2019 | None | 13 | 275 | 3 | 2.34 | 0.77 | 1.4 | 1.21 |
| LRM | 31 44 20 N 83 43 27 W |
1 Jan 1968 to Dec 312 019 | 1989–2002 | 38 | 262 | 2 | 2.41 | 0.35 | 4.44 | 1.73 |
| LRN | 31 31 05 N 83 35 11 W |
1 Jan 1970 to 31 Dec 2019 | 1982–2002; 2007, 2009–2010, 2018 | 23 | 1567 | 4 | 6.60 | 0.36 | 2.00 | 1.84 |
| LRO | 31 29 37 N 83 34 03 W |
1 Jan 1969 to 31 Dec 2019 | 1982–1992, 2009 | 39 | 1593 | 4 | 6.11 | 0.37 | 2.18 | 2.02 |
| LRK | 31 41 47 N 83 41 51 W |
1 Jan 1968 to Dec 312 019 | None | 52 | 1666 | 3 | 8.73 | 0.29 | 3.20 | 1.56 |
| LRJ | 31 41 33 N 83 42 08 W |
1 Jan 1968 to Dec 312 019 | None | 52 | 2212 | 3 | 10.30 | 0.25 | 2.71 | 1.60 |
| LRI | 31 40 29 N 83 41 26 W |
1 Jan 1968 to Dec 312 019 | None | 52 | 4992 | 4 | 12.71 | 0.22 | 2.24 | 1.67 |
| LRF | 31 36 17 N 83 37 53 W |
1 Jan 1969 to Dec 312 019 | None | 51 | 11 487 | 4 | 24.02 | 0.14 | 2.67 | 1.64 |
| LRB | 31 28 54 N 83 35 03 W |
1 Jan 1972 to Dec 312 019 | None | 45 | 33 433 | 5 | 39.10 | 0.10 | 4.41 | 1.60 |
| Alapaha USGS# 02316000 | 31 23 03 N 83 11 33 W |
1 Jan 2003 to Dec 312 019 | None | 17 | 171 716 | 6 | 85.00 | 0.05 | 3.00 | 1.45 |
| Little River at Adel USGS# 02318000 | 31 09 20 N 83 32 37 W |
1 Jan 2003 to Dec 312 019 | None | 17 | 149 442 | 6 | 73.00 | 0.09 | 2.10 | 1.55 |
The LREW originates just west of Ashburn, Georgia near the northwest corner of Turner County (Figure 1). The watershed flows in a generally southerly direction to its confluence with the Withlacoochee River, eventually joining the Suwannee River which empties into the Gulf of Mexico west of Gainesville, Florida. The LREW is in an area of broad floodplains, river terraces, and gently sloping uplands. Moderately wide inter-stream divides separate relatively broad valleys. The watershed is located on sandy surface soils underlain by the upper part of the shallow and relatively impermeable Hawthorne formation (Stringfield, 1966). The Hawthorne formation largely prevents percolation of surface waters into the limestones that form the Floridan Aquifer. Upland soils are primarily classified as fine-loamy (or loamy) siliceous, thermic Plinthic Paleudults (Calhoun, 1983) while bottomland soils adjacent to drainage networks are primarily loamy, siliceous, thermic Arenic Plinthic Paleaquults with some Fluvaquents and Psammaquents (Calhoun, 1983). Hydrology within the LREW is dominated by the Hawthorne which restricts downward movement of infiltrated rainfall and leads to lateral movement of the surficial aquifer and surface seeps to the stream channels (Sheridan, 1997). Because of the Hawthorne, the contribution of surface water to deep seepage is believed to be relatively small, simplifying defining the watershed budget (Bosch, Sheridan, & Marshall, 2007). Drainage of bottomland soils is poor to very poor with standing water on the surface during significant portions of the year. Locally the Floridan aquifer is confined and stream networks are generally not incised into the deeper groundwater aquifers.
The upland watershed divides are nearly level, very gently sloping or undulating. River channel slopes are generally less than 0.1%, whereas upland side slopes generally range up to 5% (Yates, 1976). These types of watersheds have been classified as floodplain swamps (Kitchens Jr. et al., 1975), floodplain wetlands (Kibby, 1978), seasonally flooded wetlands (Johnston et al., 1984), forest wetlands (Leitman et al., 1983), and blackwater swamp systems (Wharton, 1978). General land-use within the LREW is a mixture of row-crop agriculture, pasture and forage, upland forest, and riparian forest. Land cover over the entire LREW was classified as 50% forest, 41% mixed agricultural, 7% urban, and 2% water in 2003 (Bosch et al., 2006). Sub-watersheds range from approximately 25% to 60% row-crops (Bosch et al., 2006; Sheridan, 1997). The agricultural fields are typically less than 40 ha in size and nested among forested drainages (Bosch et al., 2004). Agricultural cropping rotations have changed over the decades from peanut, corn, soybean, winter wheat, and tobacco, to cotton, peanut, and vegetable crops (Feyereisen et al., 2008). Forested lands consist of pines in the upland areas and hardwoods in the dense riparian vegetation in the flat, broad swamp areas. Rainfall in the region is poorly distributed and often occurs as short-duration, high-intensity convective thunderstorms (Bosch et al., 1999).
The Virginia V-notch weir (Ree & Gwinn, 1959) was selected to fit the design constraints in the watershed (Bosch & Sheridan, 2007). The devices were designed so that flows would be contained within the V-notch center portion of each weir 90% to 95% of the time. Because of the broad, flat floodplains characteristic of the region, flow measurement installations on the LREW were located at road crossings. Measurement station coordinates along with physical characteristics and record periods are listed by watershed in Table 1. Each flow measurement site consists of a fixed control (or weir) for constricting and measuring streamflow, steel-sheet piling cutoff walls across the stream channel, guide walls or wingwalls to direct streamflow across the control device, a concrete apron for energy dissipation immediately downstream from the control, and stilling wells hydraulically connected to stream sections immediately above and below the control. At all culvert sites except LRM, weirs were located between the outer ends of the upstream culvert wing walls, approximately 3 m upstream from the culvert. At site LRM, weirs were placed inside the downslope end of two small box culverts. The original design flow measurement control section at all sites other than LRM consisted of a horizontal 0.41 m width weir with a V-notch center section. The width of the horizontal weir and the depth of the V-notch vary from site to site. Additional details on the weir construction are available in prior publications (Gwinn, 1974; Sheridan et al., 1995; Yates, 1970; Yates, 1976).
Free overfall over the weir crest, without interference from the tailwater or the downstream water pool is typically required for highly accurate flow measurements. To provide information on periods when tailwater levels could potentially impact flow control structure ratings, that is, when submerged flow conditions exist, both upstream and downstream water surface elevations are recorded. Stilling wells are connected hydraulically to the upstream and the downstream sides of the structures. Rating curves for each flow measurement structure were developed based on theoretical considerations, results of laboratory model testing, and an intensive field streamflow measurement program (Bosch & Sheridan, 2007). Stream stage-discharge ratings developed from the field measurements were used in conjunction with the model-based ratings to develop stage-discharge rating curves for converting recorded stream stage data to instantaneous flow rates. Correlations were developed by plotting the log of measured discharge versus the log of upstream depth above the V-notch center section, or head. Further details on the structures, rating curves, data collection, and data analysis can be found in prior research (Bosch & Sheridan, 2007; Hubbard et al., 1990; Rawls & Asmussen, 1973; Sheridan, 1997).
2.1.2 LRO3
Little River watershed O3 (LRO3) was not part of the original, historical LREW, added to the network in June of 2006 (Table 1). The 275-ha watershed is located at the headwaters of LREW subwatershed O (LRO) and drains a watershed with a high livestock population operated by the University of Georgia. About 74% of the University farm lies within watershed LRO3. Within LRO3, pastures and hay land occupy 92 ha, crop land 61 ha, riparian woodland 75 ha, ponds 10 ha, and facility buildings and associated landscapes and farm roads occupy 37 ha. A dairy research facility accommodates approximately 200 cows and occupies a small section at the north-central part of LRO3, partially outside of LRO3. A large part of the pastures is utilized by beef cattle. A beef research facility handles approximately 250 cows and is situated at the upper end of LRO3.
2.1.3 Alapaha and Little River at Adel River basins
Characteristics of the Alapaha River near Alapaha, GA (USGS site 02316000) and the Little River at Adel, GA (USGS site 02318000) are shown in Table 1. The Alapaha lies within Hydrologic Unit Code (HUC) 03110202 while the Little River at Adel lies within HUC 03110204. Daily streamflow data were downloaded for these sites from the USGS dataset (https://waterdata.usgs.gov/ga/nwis/rt, accessed August 3, 2020). Data from 2003 to 2019 were used for both sites. Daily data were aggregated into annual data and compared to the LRB data.
2.2 Rainfall data
Rainfall data for each of the LREWs were taken from watershed weighted daily values for each watershed (Bosch, Sheridan, & Marshall, 2007). For LRO3 there was only a single rain gauge located within the watershed, RG64 at N 31° 30′ 17.05″ and W 83° 31′ 58.33″. Rainfall data for Alapaha were obtained from the University of Georgia Alapaha climate station (http://weather.uga.edu/mindex.php?variable=HI&site=ALAPAHA, accessed August 3, 2020). Rainfall data for Little River at Adel were taken from the watershed weighted rainfall LRB which is in the headwaters of the watershed.
2.3 Land features
To assess land features within the LREW digitized boundaries were used (Sullivan et al., 2007). For the Alapaha and the Little River at Adel, groupings of USGS HUC10 regional watersheds that closely approximated the two USGS watersheds from which streamflow data were obtained were used (Figure 1). For the Alapaha, HUC10 watersheds 0311020201, 0311020202, 0311020203, and 0311020204 were used. For the Little River at Adel, watersheds 0311020401, 0311020402, and 0311020403 were used. Land cover was assessed using the USDA, National Agricultural Statistics Service (NASS) 2018 cropland data layer (Boryan et al., 2011). For both Alapaha and Little River at Adel, confidence in the 2018 land cover classification was 71% (mean = 69.7%). Error in the land cover classification indicates the confidence that the land cover at a given pixel was accurately classified (Liu et al., 2004) and was extracted from the Confidence Layer for the 2018 Cropland Data Layer (https://www.nass.usda.gov/Research_and_Science/Cropland/Release/, accessed September 21, 2020).
Irrigated areas across Alapaha, Little River at Adel, LRB, and LRI were quantified using the 2017 MIrAD dataset (Brown et al., 2019). The MIrAD dataset uses a combination of MODIS, USDA NASS county level statistics for irrigated land area (https://www.nass.usda.gov/AgCensus/index.php, accessed September 21, 2020) and land cover mask from the National Land Cover Dataset (U.S. Department of the Interior, U.S. Geological Survey, 2014). The dataset utilizes 250 m pixels and was judged unsuitable for watersheds below the scale of LRI. For the smaller LREW subwatersheds, an irrigation data layer based upon digitization of National Agriculture Imagery Program 2017 aerial imagery was used (US Department of Agriculture, Farm Service Agency, Aerial Photography Field Office, 2017). For comparison purposes, irrigated acreage within LRB, LRF, and LRI were quantified using both methods. Water body features across the studied watersheds were characterized using the National Hydrology Dataset (NHD) (U.S. Geological Survey and National Geospatial Program, 2018). The NHD characterizes lakes and ponds, reservoirs, and swamps as water bodies. The data used for this analysis were mapped at a 1:24 000 scale.
2.4 Data analysis
Data were grouped and statistical comparisons across the grouped data made using one-way analysis of variance (ANOVA) tests. Subsequent post-hoc analysis of differences determined through the ANOVA were made using the paired t-test. Significance of the linear correlations were tested using the Pearson's correlation coefficient. For the LREW, watersheds LRM, LRN, and LRO3 were excluded from the statistical test because of reduced records for those watersheds (Table 1). LRO3 was compared separately to LRO and LRM using the same record years as those available for LRO3. The Alapaha and Little River at Adel were compared to LRB using the record period available for Alapaha, 2003–2019.
3 RESULTS
3.1 Rainfall and streamflow metrics
3.1.1 Little River Experimental Watersheds
While year to year variability exists in the observed streamflow, long-term average annual rainfall and streamflow were fairly consistent across the LREWs (Table 2). Average annual rainfall ranged from a low of 1166 mm for LRO3 to a high of 1222 mm for LRK. Average annual streamflow for the LREWs ranged from a low of 270 mm for LRM to a high of 496 mm for LRO3. Sratio ranged from a low of 21% for LRM to a high of 42% for LRO3. The relatively low Sratio observed for the studied watersheds indicate a high degree of moisture recycling via ET that occurs in the region (Savenije, 1996). While there were differences in the magnitude of these metrics, no statistical differences (α = 0.05) were found between the rainfall, streamflow, or Sratio for watersheds LRB, LRF, LRI, LRJ, LRK, and LRO. Watersheds LRM, LRN, and LRO3 were excluded from this statistical analysis due to fewer years of record (Table 1).
| Watershed | Average annual precipitation (CV) (mm) | Average annual streamflow (CV) (mm) | Average annual ratio of streamflow/precipitation (CV) (%) | Regression (Equation 6) | ||
|---|---|---|---|---|---|---|
| Slope | Rainfall threshold (mm) | Correlation coefficient r2 | ||||
| LRO3 | 1166 (24.6) | 496 (50.3) | 42 (47.5) | 0.47 | 117 | 0.30 |
| LRM | 1208 (17.5) | 270 (53.8) | 21 (46.7) | 0.48a | 650 | 0.50 |
| LRN | 1214 (16.1) | 347 (44.3) | 28 (37.5) | 0.60a | 638 | 0.57 |
| LRO | 1186 (19.0 | 309 (53.4) | 25 (44.7) | 0.53a | 612 | 0.51 |
| LRK | 1222 (17.8) | 365 (51.0) | 28 (41.0) | 0.77a | 749 | 0.81 |
| LRJ | 1221 (17.3) | 383 (54.5) | 30 (44.1) | 0.87a | 780 | 0.77 |
| LRI | 1220 (17.3) | 399 (50.0) | 31 (38.9) | 0.85a | 749 | 0.80 |
| LRF | 1222 (16.2) | 349 (45.3) | 27 (35.3) | 0.69a | 717 | 0.75 |
| LRB | 1200 (15.9) | 320 (48.9) | 26 (40.3) | 0.64a | 711 | 0.64 |
| Alapaha | 1151 (16.2) | 244 (52.8) | 21 (51) | 0.52a | 503 | 0.67 |
| Little River at Adel | 1182 (19.5) | 260 (55.8) | 21 (47.0) | 0.38 | 678 | 0.30 |
- Note: Coefficient of variation (CV) shown in parenthesis.
- a Slope is significant (α = 0.05).
The observed annual rainfall, streamflow, and ET patterns over the 45-year period of record for LRB is shown in Figure 2. Evapotranspiration was determined as the difference between annual rainfall and annual water loss; with annual water losses including annual streamflow and regional groundwater losses equal to 1% of annual rainfall (Sheridan, 1997). As would be expected, streamflow is responsive to increases in rainfall. While a slight decrease in watershed weighted rainfall at LRB has been observed from 1972 to 2019, the trend is not significant (p = 0.32, α = 0.05). Similarly, a slight decrease has been observed in LRB streamflow. The trend is also not significant (p = 0.17, α = 0.05) as was not the trend in the Sratio (p = 0.12, α = 0.05). Similar patterns were observed for the other LREW due to the statistical similarity observed in rainfall and streamflow across the subwatersheds.
As with the rainfall and streamflow data, considerable variability was observed with the year to year annual Sratio for LRB, illustrated by the high CV of 40.3 (Table 2). The variability of the streamflow and Sratio were greater than that observed for the rainfall (Table 2), agreeing with the results of Sheridan (1997) where a shorter record was examined for the same watersheds. Sratio is responsive to increasing levels of annual rainfall (Figure 3). Increased precipitation in this watershed leads to saturation throughout the surficial aquifers and the floodplain, eventually leading to saturation excess flow throughout the floodplain. However, the relationship between annual Sratio and annual rainfall peaks and flattens at annual rainfall above 1250 mm, at a Sratio of approximately 35% (Figure 3). This pattern is repeated for each of the historical LREWs (data not shown). This can likely be attributed to the seasonal patterns of rainfall in this region (Figure 4). While considerable rainfall occurs from December through March, the region also experiences high rainfall during the months from June through August. While the high rainfall from December through March typically yields high streamflow, that from June through August typically does not (Figure 4). High temperatures and vegetation growth during the summer period leads to high seasonal ET, resulting in low streamflow. During years where high rainfall rates are observed much of this occurs from June through August from convective thunderstorms (Bosch et al., 1999). While some of these storms can produce significant streamflow, it is typically not sustained. The Sratio during years with high annual rainfall would be expected to be less responsive to increases in rainfall.
3.1.2 Little River O3
Little River O3 is of interest because of the higher than normal livestock population in the watershed. Here we compare rainfall and streamflow in LRO3 to similar LREWs. Two watersheds were selected for comparison, LRO due to physical proximity to LRO3 and LRM due to similarities in scale. Periods of record from LRO and LRM were extracted to match those of LRO3, 2007–2019. ANOVA tests indicated there were no significant differences between the rainfall for these three watersheds over this period of record (p = 0.12, α = 0.05). Significant differences were found for the streamflow and the Sratio for the three watersheds (α = 0.05). Post-hoc tests indicated a significant difference between the streamflow and Sratio from LRO3 and both of the other two watersheds in the comparison (α = 0.05). No significant differences in streamflow or Sratio were found between LRM and LRO. The average annual streamflow for LRO3 for this period of record was 496 mm while it was 245 mm at LRM and 249 mm at LRO. The LRO3 Sratio averaged 42% over the period of record while it averaged 21% for LRM and 25% of LRO. These data indicate that the rainfall-runoff response from LRO3 is different than LRO and LRM.
3.1.3 Regional watersheds
Data from the largest LREW, LRB, was compared to the two larger regional watersheds, the Alapaha River at Alapaha, GA and the Little River at Adel, GA. Both of these are in physical proximity to the LREW, have similar characteristics, but are approximately five times the size of LRB (Figure 1, Table 1). Here we compare the rainfall and streamflow characteristics of the three for the period from 2003 to 2019, matching the period of available data from the Alapaha watershed.
No statistical difference was found between the rainfall for the three compared watersheds for the period examined (p = 0.89, α = 0.05). Average annual rainfall for the Alapaha watershed was 1151 mm. Because the same rainfall data were used for LRB and the Little River at Adel, both were 1182 mm. No statistical difference was found between the streamflow for the three compared watersheds for the period examined (p = 0.72, α = 0.05). Average annual streamflow for the period was 244 mm for Alapaha, 260 mm for the Little River at Adel, and 284 for LRB (2003–2019 period). Similarly, no statistical difference was found for the Sratio of the three watersheds (p = 0.81, α = 0.05). The long-term average for the Sratio was 21% Alapaha, 21% for the Little River at Adel, and 23% for LRB. Based upon these results, the three watersheds exhibit similar rainfall-runoff characteristics.
The Alapaha and Little River at Adel watersheds display similar relationships between the annual rainfall and the Sratio to that of LRB (Figure 3). All Sratios typically fail to increase above a threshold of annual rainfall approximately equal to 1250 mm. However, the peak Sratios observed for the Alapaha and Little River at Adel watersheds are approximately 25%–30% versus the 35% observed for LRB.
Slopes of the linear regression of Equation 6 indicate how streamflow will respond to rainfall once the rainfall threshold is obtained. The slopes for the studied watersheds varied from a low of 0.38 for the Little River at Adel to a high of 0.87 for LRJ (Table 2). All slopes were found to be significant at α = 0.05 except those for LRO3 and Alapaha. Both LRO3 and Alapaha had shorter periods of record. In addition, streamflow from LRO3 was not closely related to rainfall totals. Examination of the slopes for the LREWs indicate a difference between the watersheds in the northern portion of the watershed (LRK, LRJ, and LRF) which range from 0.77 to 0.87 to those in the southern portion of the watershed (LRN, LRO, and LRB) which range from 0.53 to 0.64. Rainfall thresholds varied from a low of 177 for LRO3 to a high of 780 for LRJ (Table 2). While the Little River at Adel watershed begins to produce annual runoff at a rainfall threshold similar to the other studied watersheds, the slope of 0.38 indicates an ability to store water throughout the year without significant increases to streamflow. This is also illustrated by the lower maximum Sratio observed for this watershed (Figure 3). This can likely be attributed to the 375 ha Reed Bingham State Park reservoir near the outlet of the watershed (Figure 1). However, similar behaviour by the Alapaha watershed indicates the larger floodplains and riparian areas in these watersheds also influence the streamflow at the higher annual rates. LRO3 begins to produce runoff at a very low rainfall threshold compared to the other LREWs, but produces streamflow at just 47% of annual rainfall compared to the other watersheds which generally produce streamflow at higher percentages (48%–87%). LRJ has a rainfall threshold similar to the other LREWs but produces rainfall at a high percentage of annual rainfall once this rainfall threshold is reached, indicating less ability to store water in this watershed. Similar characteristics appear to be shared by LRK (Table 2). The regression analysis indicates a different hydrologic behaviour for watersheds LRO3 with a low rainfall threshold and the Little River at Adel with a low slope.
3.2 Land features
Land-cover was characterized for each watershed utilizing the NASS 2018 data (Table 3). Some differences in land-cover exist within these watersheds. The Alapaha watershed is more heavily forested (51% of total area), as are the northern LREWs (LRM 53%, LRK 57%, LRJ 53%, and LRI 53%). Row crop coverage across the studied watersheds is from 37% to 55%. The least row crops are found in the northern LREWs, LRK (37%), LRJ (37%), and LRI (37%), as well as the Alapaha (37%), corresponding to the increases in forest cover. LRO3 contains a greater fraction of pastures and herbaceous wetlands (10%) and row crops (55%), likely a function of the livestock within the watershed. Combined, these features indicate an increase in cropped area in the southern LREWs accompanied by a decrease in forest coverage. The larger watersheds studied, Alapaha, Little River at Adel, and LRB, have similar land-cover characteristics (Table 3).
| Watershed | Water bodiesa (%) | Row cropsb (%) | Pastures/herbaceous wetlandsb (%) | Forest and shrublandb (%) | Urban/residentialb (%) | Irrigated landc (%) | Irrigated landd (%) |
|---|---|---|---|---|---|---|---|
| LRO3 | 4 | 55 | 10 | 28 | 6 | ND | 10 |
| LRM | 0 | 43 | 1 | 53 | 2 | ND | 3 |
| LRN | 3 | 53 | 6 | 35 | 5 | ND | 15 |
| LRO | 4 | 48 | 8 | 37 | 6 | ND | 9 |
| LRK | 1 | 37 | 2 | 57 | 3 | ND | 11 |
| LRJ | 3 | 37 | 4 | 53 | 4 | ND | 8 |
| LRI | 2 | 37 | 4 | 54 | 4 | 10 | 9 |
| LRF | 4 | 38 | 7 | 47 | 7 | 11 | 12 |
| LRB | 8 | 43 | 7 | 43 | 5 | 15 | 13 |
| Alapaha | 7 | 37 | 5 | 51 | 5 | 11 | ND |
| Little River at Adel | 6 | 41 | 6 | 45 | 7 | 14 | ND |
- Note: All values are expressed as a function of total watershed area.
- Abbreviation: ND, no data.
- a NHD water features.
- b NASS 2018 land-cover.
- c MIrAD data.
- d SEWRL digitized data based upon 2017 aerial imagery.
The fraction of each watershed covered by water bodies is fairly consistent across the LREW and Little River at Adel (0%–8%) (Table 3). The NHD characterization of water bodies used here includes swamplands ((U.S. Geological Survey and National Geospatial Program, 2018). The larger watersheds have elevated water body coverage (6%–8%). As the scale of the watersheds increase the floodplains increase, accompanied by an increase of swampland leading to the observed increase in water bodies for the larger watersheds. Swamplands within LRB, Alapaha, and Little River at Adel make up 62%, 87%, and 62%, respectively, of the total water bodies in those watersheds (data not shown). Swampland in the other studied watersheds typically is less than 30% of the total water body coverage. These swamplands are largely composed of the broad floodplains of the watershed, likely accompanied by a large increase in riparian buffers at the larger scales. Sheridan (1997) also reported an increase in open water for the larger watersheds in the LREW.
Irrigated land across the LREW varies from a low of 3% in LRM to a high of 15% in LRN utilizing the SEWRL digitized data (Table 3). Irrigated acreage shows a slight increase moving from the northern portion of the LREW (8%–11%) compared to the southern portion (9%–15%) (Table 3). Similar irrigated acreage was found for the Alapaha (11%) and the Little River at Adel (13%) to the LRB (15%) utilizing the MIrAD data. For comparison purposes, irrigated acreage within the LRI, LRF, and LRB were characterized using both the MIrAD and SEWRL digitized data. The two methods yielded results within +/− 2%, indicating the two methods yield comparable results.
In aggregate, key characteristics in the study watersheds that may impact hydrology include: (1) a greater fraction of row crops and pastures within the LRO3, (2) an increase in row crops accompanied by a decrease in forest land as you move from north to south in the LREW, (3) a greater percentage of area in swampland within the larger watersheds, LRB, Alapaha, and Little River at Adel, and (4) similar land features in the LRB, the Alapaha, and the Little River at Adel.
4 DISCUSSIONS
Average annual precipitation for the studied watersheds ranged from 1151 to 1222 mm, matching general patterns for the region (PRISM Climate Group, 2004). Average annual runoff for the studied watersheds ranged from 244 to 496 mm, within the expected range of 305 mm for this region (Gebert et al., 1987). Using an average annual precipitation for the region of 1200 mm and an average annual runoff of 305, the expected Sratio for the region is 25%, within the ranges observed here (Table 2).
No statistical differences were found in the rainfall, streamflow, or Sratio in the historical LREWs. Statistical differences were found for LRO3 streamflow and Sratio when compared to LRM and LRO. These observed differences appear to be related to the greater fraction of row crops and pastures, and reduced forest coverage, in LRO3. As discussed, increased forest coverage typically increases overall watershed ET and reduces runoff. As seen with the comparison to LRO, these differences do not carry over into the larger watershed into which LRO3 drains which has land characteristics similar to the remainder of the LREW (Table 3). Sheridan (1997) found that riparian area coverage was more effective than upland land use for predicting annual runoff rates, with runoff decreasing as riparian area coverage increases. Because the swamplands in the floodplain area and riparian coverage increases for the larger watersheds annual streamflow for these watersheds would be anticipated to decrease. This research indicates some decrease in annual streamflow for the larger studied watersheds than for the smaller watersheds (Table 2), although comparisons across just the LREWs indicated no statistical differences.
While differences in the row crop and forest coverage occur across the northern and southern portions of the LREW, these differences are not large enough to impact streamflow or Sratios. A trend across the region is the increased use of center pivot irrigation on row crop areas. The fraction of irrigated land also increases in the southern portion of the LREW (Table 3). Surface water ponds throughout the watershed used to supply irrigation water are typically installed on first order streams. These ponds are frequently supplemented with groundwater from the deeper Floridan Aquifer. High volumes of irrigation would thus be expected to influence runoff by introducing inputs in addition to rainfall. While the fraction of irrigated land in LRN is greatest, this does not appear to have an impact on streamflow or Sratio in that watershed. Because most of the irrigation occurs during the summer period when ET is very high and landscape conditions are typically dry, the increased irrigation does not appear to impact streamflow at the levels of irrigated land observed here.
Similarly, while differences in row crop, irrigation, forest, and water bodies of the Alapaha exist, these differences were not large enough to significantly impact streamflow or Sratios. There is some indication of reduced flows due to the higher fraction of forest coverage, but these differences were not statistically different and could be caused by slightly reduced rainfall in the Alapaha (Table 2). Land features and streamflow of the Little River at Adel into which the LREW drains, appear quite similar to the LRB.
Prior research has compared the water yield of the Little River Experimental Watersheds to that of other ARS watersheds across the United States (Renard, 1977; Sheridan, 1997). Comparisons of the water yield as a function of drainage area using the LREWs along with the Alapaha and Little River at Adel are presented in Figure 5. Similar results were found to those presented by Sheridan (1997). The water yield-drainage area relationship observed here is consistent with that of a drainage network that intercepts surface runoff and lateral subsurface flow from uplands and has low deep seepage losses (Sheridan, 1997). A slight decrease in water yield is observed as the watershed area increases (Figure 5) (trend not found significant, α = 0.05). Based upon the general trend obtained from the fitted line to the log–log plot, the water yield decreases from the smallest watershed studied here (LRM) to the largest studied (Alapaha) the water yield decreases from 390 to 274 mm. This decrease indicates increasing consumptive water use with increasing open water/wetland area in the region, attributed to increased ET that occurs with increasing relative open water and wetlands. It must be noted that considerable variability was observed between the water yields from LRM and LRO3. These watersheds also varied considerably in their land characteristics. The implication is that land management can have a significant impact upon the hydrologic flow characteristics of the smaller watersheds in the region. This may carry over to larger watersheds if the portion of the land effected is large enough.
5 CONCLUSIONS
Comparisons of rainfall, streamflow, and Sratio indicate the long-term characteristics of the historical LREWs are statistically similar. This helps build an understanding between regional climatic patterns, landscape features, and streamflow. The statistical similarity may not carry over into examinations of daily or annual patterns where differences in stormflow and cropping practices are expected to exhibit temporal variability. Summer storms across the region can be highly variable (Bosch et al., 1999), leading to isolated stormflow from smaller regions. As illustrated by the statistical differences in streamflow and Sratio for LRO3, significant changes in crop and forest coverage can alter long-term characteristics of streamflow. Here, increases in row crops and decreases in forest coverage within LRO3 resulted in increased streamflow. These differences do not carry over into the larger watersheds into which LRO3 drains with more typical land features. While larger irrigated acreage was observed within LRN of the LREW, the levels of irrigation observed within the watershed were not found to impact streamflow.
One of the goals of this analysis was to characterize the similarity or dissimilarity between the LREW and other regional watersheds. Comparisons between the LRB and the Alapaha and Little River at Adel indicated statistical similarity across these three watersheds. While differences in row crop, forest, and water body acreage were observed for the Alapaha watershed, these differences did not impact the overall streamflow. This would indicate a level of resilience within these Coastal Plain watersheds to changes in land cover. As watersheds increase in scale, annual streamflow decreases (Figure 5), likely a function of increasing floodplain and riparian coverage. While the decrease was not statistically significant when the three watersheds were compared, or when relationships between watershed area and streamflow examined, grouping of all of the studied watersheds points to a relationship between scale and streamflow.
Continued operation of the LREW hydrologic network supports hydrologic research as well as environmental quality and riparian research programs. While considerable understanding has been established from prior research, time has illustrated the continued value of the long-term data. Additional work is required to better quantify irrigation inputs in these watersheds and their role in water budgets. These data have been critical to validating watershed scale natural resource models. As our understanding of these systems evolves and computing resources improve, these models become more accurate and sophisticated. Data from these watersheds will continue to prove their value in the future.
ACKNOWLEDGEMENTS
This research is a contribution of the USDA Agricultural Research Service (ARS) Gulf Atlantic Coastal Plain Long-Term Agroecosystem Research site. The authors are grateful for partial funding for this project through the USDA-NRCS Conservation Effects Assessment Project. The authors are also grateful for the assistance of the many scientists and field and laboratory technicians who have supported the research.
Open Research
DATA AVAILABILITY STATEMENT
Databases described in each of these reports are available from the SEWRL web site: https://www.ars.usda.gov/southeast-area/tifton-ga/southeast-watershed-research/ Daily data are also available for direct download from the SEWRL maintained web site, http://radio.tiftonars.org/archived_data.htm and the STEWARDS (Sustaining the Earth's Watersheds, Agricultural Research Data System) database, https://www.nrrig.mwa.ars.usda.gov/stewards/stewards.html (Steiner et al., 2008; Sadler et al., 2008).
