The Coastal Plains Soil, Water, and Plant Research Center has been monitoring spatial yield in a test field since 1985, using a conventional corn-wheat-soybean rotation most of that time. Observations of variation in soil and crop response that correlate with yield variation suggest that crop water relations may be the key feature that causes spatial variability in yield for the Southeastern Coastal Plain. Experience with mechanistic modeling indicates that for normal weather years, the final yield is particularly sensitive to variations in soil water, presumably because the surface soil is sandy and rooting volume is limited. These conclusions, plus difficulties encountered in scheduling irrigation under a center pivot on typically variable soils, led the USDA-ARS to desi~n and build a site-specific center pivot capable of differentially irrigating lOO-m areas. A 3-tower commercial center pivot was modified by adding 39 9.2-m manifolds in 13 sections, 3 to a section. The manifolds and nozzles were sized lx, 2x, and 4x, so that octal combinations would provide up to 7x the minimum application depth for a given outer tower speed. At 50% speed, the application depths are 0 to 12.5 mm in 1.8-mm increments. A prograrnrnable controller was attached near the pivot end of the boom, so that it was proximal to but avoided the pivot control panel when the system rotated. The individual manifolds were controlled by a program residing in the progranunable controller, which obtained pivot position and other information via radio modem link with the pivot control panel. Water and nitrogen application has been accomplished using this system on a replicated field experiment. Experience gained during this phase will guide modification of a similar pivot for site-specific water, nutrient, and pesticide management on a typically variable Coastal Plain field.

Variable-rate irrigation (VRI) systems have the potential to conserve water by spatially allocating limited water resources. However, when compared to traditional irrigation systems, VRI systems require a higher level of management. In this 3-year study, we evaluated spatial irrigation management of a peanut crop grown under a VRI system using an expert system (Irrigator Pro). The irrigation management treatments evaluated were: (1) using Irrigator Pro (IP) to manage irrigation uniformly in plots with varying soils; (2) using Irrigator Pro to manage irrigation in plots based on the individual soils (IPS); (3) a treatment based on maintaining soil water potential (SWP) above −30 kPa (approximately 50 % depletion of available water) in the surface 30 cm of each soil within a plot; and (4) a non-irrigated treatment. Over the 3-year study, all irrigated treatments had significantly higher yields (4,230, 4,130, and 4,394 kg ha−1 for the IP, IPS, and SWP treatments, respectively) than the non-irrigated treatment (3,285 kg ha−1), yet the yields of the three irrigation treatments were not significantly different. Averaged over the 3-year experiment, the three treatments did not differ significantly in water usage. In the 2007 and 2009 growing seasons with below normal rainfall, the IP and IPS treatments required significantly greater total water than the SWP treatment. Overall, water use efficiency was significantly higher for the non-irrigated and SWP treatments (9.4 and 8.9 kg ha−1 per mm, respectively). The lower water use efficiency for the IP and IPS irrigation treatments (7.8 kg ha−1 per mm) was attributed to greater water applications mainly due to earlier growing season initiation of irrigation applications. However, the IP and IPS treatments maintained soil water potentials at the 30- and 60-cm depths at higher levels throughout most of the season. The two Irrigator Pro expert system treatments functioned as well as the SWP-based treatment. The Irrigator Pro expert system can be effectively used for site-specific management where management zone soils do not greatly differ. Further refinement of the expert system may be needed to improve its application in spatial irrigation applications.

A better understanding of how bermudagrass (Cynodon spp.) regrowth is influenced by production inputs will aid in advancing precision management. The objectives were to evaluate the effect of irrigation and nitrogen (N) on normalized difference vegetative index (NDVI) during regrowth and evaluate the relationship between NDVI during regrowth to forage yield. Normalized difference vegetative index (NDVI) data were collected in an experiment evaluating two harvest schedules (four or eight week), four rates of irrigation (0, 4.2, 8.4, or 12.5 mm of water each irrigation event), and three rates of N (season total of 168, 336, and 504 kg N ha(-1)). Both irrigation and N influenced NDVI of bermudagrass during the regrowth periods of both the four- and eight-week harvest schedules, but there were no irrigation by N interactions. As was expected, regrowth (as measured by NDVI) in response to irrigation was dependent on the timing and duration of rain-free periods within the regrowth periods. Generally, NDVI increased with increasing N rate at most sampling dates. Regression slopes of yield versus NDVI measured near the end of the regrowth period were higher for the eight-week harvest schedule than for the four-week schedule. Within each harvest schedule, however; slopes were similar. The high correlation coefficients between NDVI late in the regrowth period and yield suggest NDVI is a useful tool for managing bermudagrass harvests.

In the southeastern region of the US., the cattle industry has a critical need for sustainable hay production. Yet this production is threatened by frequent short-term regional drought. This drought threat can be mitigated by properly managed irrigation. In this study on Tifton 85 bermudagrass, irrigation management, nitrogen fertility levels, and harvest interval were evaluated for their impact on hay quality and yield. The experimental treatments were arrayed in a split-plot design with harvest interval as the main treatment; irrigation by nitrogen (N) levels were the subplots. Treatments had four replicates and were repeated for two years. The optimal irrigation rate was set to maintain soil water potentials below -30 kPa. When needed, the full irrigation treatment received a 12.5 min irrigation application. The reduced irrigation treatments received water at rates of 0%, 33%, and 66% of the full irrigation rate. In addition, each irrigation treatment had nitrogen rates of 168, 336, and 504 kg N ha(-1). The irrigation and nitrogen treatments were harvested at four-week or eight-week intervals. Total harvests per year ranged from three to six. Over both years and for all harvests, there was no irrigation-nitrogen interaction for hay yield. Over all harvests, nitrogen significantly increased bermudagrass hay yield, nutrient concentrations, and forage quality. Forage quality was higher for the four-week harvest interval. Throughout the study forage quality was maintained within desired industry standards. When irrigation was required, it significantly increased hay yield. During these periods, the four-week and eight-week 100% irrigation treatments yielded 612 and 1600 kg ha(-1) greater, respectively, than the non-irrigated treatments. The four-week harvest interval was more sensitive to irrigation. Additionally, we observed a linear relationship between non-irrigated bermudagrass hay yields and average soil water potential. As soil water was depleted, non-irrigated hay yields decreased 31 kg ha(-1) per kPa. Timely supplemental irrigation to maintain soil water potentials above -30 kPa can increase bermudagrass yields. Thus, irrigation management should be critically assessed for its potential role in sustaining hay production in the southeastern Coastal Plain.

Availability of spatially-indexed data and crop yield maps has caused increased interest in site-specific management of crop inputs, especially water and fertilizer As commercial equipment to implement site-specific applications of water and nutrients becomes available, crop response to variable inputs and decision support systems will be required to ensure profitable crop production while conserving natural resources and protecting the environment. The objective of this research was to determine corn yield response to a range of nitrogen fertilizer and irrigation amounts on a relatively uniform southeastern Coastal Plain soil under conservation tillage. Corn was grown in a field experiment using a center pivot irrigation system that had been modified to make site-specific applications of water and fertilizer during the period 1999-2001 on a site near Florence, South Carolina. Treatments included three antecedent crop rotations (prior four years), three irrigation regimes (0, 75%, and 150% of a base rate, IBR), and four nitrogen fertilizer amounts (50%, 75%, 100%, and 125% of a base rate, NBR), and with Put. replications. As expected, corn grain yields increased with irrigation and N fertilizer Mean corn grain yields for the three-year study ranged from 6.3 to 8.9 Mg/ha for the 0% IBR avail-twin, 9.4 to 10.5 Mg/ha for the 75% IBR treatment, and 10.0 to 10.6 Mg/ha for the 150% IBR treatment. The mean corn grain yields in response to N applications ranged from 6.4 to 8.0 Mg/ha for the 50% IBR treatment, 8.6 to 9.4 Mg/ha for the 75% NBR treatment, 9.1 to 10.9 Mg/ha for the 100% NBR treatment, and 8.8 to 11.7 for the 125% NBR treatment. However, the nature of the response varied among the three years, mainly because of differences in rainfall and rainfall distribution during the growing season. Also, during the first,year there was less response to N fertilizer (7.9 to 9.1 Mg/ha) possibly because of residual soil N from antecedent soybean crop. A regression analysis indicated that the slopes of the corn yield response to increased N fertilizer application were low for both irrigated and rainfed treatments in 1999. In both 2000 and 2001, the slopes were greater for the corn yield response to increased N fertilizer In 2000, the irrigated treatments had a greater slope of the yield response for additional N fertilizer than did the minted treatments. Using an orthogonal contrast analysis, the overall yield response for the combined irrigation treatments to N fertilizer was quadratic in 1999 and 2000, and linear in 2001. These quadratic yield response's indicated that, for these conditions, a potential upper limit on production for the applied N-fertilizer and water (rainfall and irrigation) was approached. For the minted treatment, yield response to N fertilizer was linear in all three years. These results provide useful information that should be helpful in developing management strategies and decision support systems for profitable management of both water and N fertilizer on spatially-variable soils in the southeastern Coastal Plain while conserving natural resources and protecting the environment.

Site-specific irrigation is defined as delivering different prescribed depths of water to specific areas in irrigatedfields. Since the 1990s, site-specific irrigation research has been expanded to include the delivery of water and nutrients tospecific field areas based on soil type, soil moisture status, crop needs, and other user-defined objectives. A site-specific centerpivot irrigation system was designed and installed in a field with highly variable soils of the U.S. eastern coastal plain. Thesystem consisted of 13 segments along the 140-m length of the three-tower center pivot with three delivery manifolds in eachsegment. The system was designed to apply approximately 12.5 mm of water in any selected segment when operated at 50%travel velocity. Quantifying water application depth and uniformity from the site-specific irrigation system is essential todocumenting the systems performance and interpreting experimental results. We developed a measurement system to evaluatethe water delivery rates of the irrigation system. We compared the measured water delivery from each segment of thesite-specific irrigation system to the design parameters. We found that the irrigation system was delivering water to the controlareas at rates approximately as it was designed. A total of 77 segment and manifold combinations were tested. Of these77 combinations, we found that 7 had flow rates greater than 10% different from the design flows. The manifolds with the lowerflow rates typically were more likely to differ significantly from their design values. This was most likely related to potentialclogging of the low flow nozzles that have smaller orifices. When the manifolds were used in combination, they compensatedfor each other and produced application depths near the design depths.

In the US SE Coastal Plain, adoption of site-specific farming has lagged behind that in the upper Midwest. While reasons for this may be both social and economic, it appears that the importance of the problem represented by yield variation on production fields needs to be quantified before adoption would be considered by many farm operators. Our objective was to document the severity, extent and persistence of yield variation for corn, wheat and soybean during normal production in this region. Farmer combines were fitted with commercial yield monitors to produce yield maps. Corn, wheat and soybean yields were mapped for three years on more than 4900 ha (12,000 acres). For each cooperator, crop and year, summary statistics and cumulative yield distribution functions were also developed. Yield maps showed that substantial areas had yields either well below or above the mean for the cooperator-crop-year. For instance, 25% of Cooperator A’s area had corn yields more than 30% below the mean yield of 5.06 Mg/ha and another 25% had yields more than 31% above that mean, which indicate the severity of yield variation. Variation from county to county had no consistent difference indicating that the extent of the variation is widespread. Variation was also persistent from year to year, with more than 50% of the area in 15 of 17 fields having stable yields relative to the field mean. These data show the potential importance of variable-rate management in the region and also hint at the potential environmental implications.

Constructed wetlands designed and properly operated for treatment of swine wastewater may enhance oxidation-reduction processes and nutrient treatment performance. The objective of this investigation was to characterize soil wetland processes related to nitrogen (N) treatment (nitrification-denitrification) and phosphorus (P) removal using soil oxidation-reduction potential (ORP) data. We evaluated three surface-flow wetland systems constructed for treatment of swine wastewater in Duplin Co., North Carolina, in 1992. Each system consisted of two 3.6- x 33.5-m cells connected in series. The three systems were planted to bulrushes, cattails, and agronomic crops (soybean in saturated soil culture and flooded rice), respectively. Soil aerobic/anaerobic conditions were determined by monitoring soil ORP at 18 sites using platinum (Pt) electrodes. Three monitoring sites were established in each wetland cell. Each site consisted of five Pt electrodes at three soil depths (0.02, 0.05, and 0.10 m) and a reference electrode. A data logger was used for hourly acquisition of soil ORP and temperature records. Hourly ORP data were averaged on a 24-h basis and corrected to standard hydrogen electrode readings (Eh). Frequency analysis of daily soil Eh showed that bulrush and soybean cells were moderately reduced (+100 < Eh < +300 mV) and anaerobic (Eh < +300 mV) about 70% of the time. However cattail and rice cells were anaerobic 100% of the time and had reduced (-100 < Eh < +100 mV) to highly reduced (Eh < -100 mV) soil conditions. These results indicate that different wetland plant species promote distinct anaerobic and reducing soil conditions. Outflow concentration of ammonia-N (NH3-N) and soluble P increased with increasing ORP values for bulrush and soybean-rice wetland cells due to lower temperatures during fall and winter but not for cattails. Denitrification enzyme activities and ORP indicated that soils in bulrush wetlands promoted better conditions for nitrification-denitrification than cattails or rice soils. However equivalent NH3-N removal rates (4.8-5.6 kg ha(-1) d(-1))for cattails and bulrush suggested that treatment occurred mostly in the water column for cattails rather than the wetland soil. Prevalent anaerobic soil conditions and soluble P outflow concentrations determined rather poor P retention capacity for all three wetlands.

Spatial variation in grain yield often justifies site-specific management. We tested maize yield response to irrigation and N using site-specific center pivots. One pivot (CP2) included twelve soil map units; the other (CP1) consisted of almost wholly the predominant soil. Under CP2, marginal N response ranged from -55 to 33 kg/kg in 1999, from -69 to 53 kg/kg in 2000, and from -31 to 46 kg/kg in 2001. Where negative, N was not the yield-limiting factor. Spatial variation on the more-uniform soil was less but not negligible. Both cases showed distinct spatial patterns. Explanations and incorporation into precision agriculture recommendations remains a challenge.

Lack of basic knowledge about spatially varying crop response to irrigation hinders optimal irrigation management and economic analysis for site-specific agriculture. The objectives of this research were to measure the mean response of corn to irrigation amounts on 12 soil map units and compare variation in the response within and among soil map units. This experiment was implemented from 1999 through 2001 with a center-pivot irrigation machine that had been modified to enable site-specific irrigation on small plots within a representative, highly variable Coastal Plain field. Four irrigation treatments (0%, 50%, 100%, and 150% of a base rate designed to hold soil water constant) and two N treatments (135 and 225 kg/ha, the recommended rainfed and irrigated rates) were imposed in 2 X 4 factorial randomized complete blocks on eight soil map units, plus randomized incomplete blocks on four additional map units. The water treatments had consistently significant main effects in the analysis of variance (ANOVA) in both linear and quadratic forms at the 1% level, and the variation within soil map units was significant at the 5% level in the latter two years and at the 1% level in 1999. Variation in yield among soil map units at any point on the response curves approximated 25% of the maximum yield in all three years. Variation in mean irrigation amounts to produce maximum yield in the eight most common map units was 61%, 61%, and 120% of the base rate amount in the three years. These data, the first such known for any soil, crop, or location in the world, have significant implications for the design, management, and economic profitability of irrigation on spatially varying soils. While the obvious design and management change would be from whole-field to site-specific approaches, even under whole-field situations, designers should consider more strongly the management zone size, range of application rates, and need for documentation.

Measurement of water stress and scheduling of irrigation are both enabled by non-contact infrared thermometers (IRTs). Technological advances have miniaturized IRTs and reduced power requirements so that inexpensive self-powered units are now commercially available. The objective of this work was to test a linear array of IRT sensors mounted on a center-pivot irrigation machine, and to use this IRT array to examine spatial variation in water stress of corn under four irrigation treatments imposed on a highly variable field with a center pivot equipped for site-specific irrigation and agrochemical application. An array of 26 IRTs was mounted on the pivot, which was run dry for a full circle on 7 days during the 1999 corn growing season. Procedures were developed to adjust for time lag during the 3.5-hr measurement period. Significant differences were obtained among the varying water treatments, as expected, but also among plots within the same soil map unit and among soil map unit means. Distinct spatial patterns, not necessarily related to the 1:1200-scale soil map, were observed. These results emphasize the necessity to consider soil water relations during the development of management recommendations for site-specific agriculture.

The SE US Coastal Plain has unique characteristics that require specialized techniques to explain yield variations and to develop management zones. This paper discusses several new methods to estimate yield variations for the development of management zones. Four techniques were developed based on the following: yield maps, black and white bare ground aerial photos, soil survey maps, and automated regular polygons. Two project fields were used for a detailed analysis of these techniques. Eight other fields were included for comparison. Results indicated that the amount of yield variation explained is related to the number of polygons used, regardless of the method used to generate the polygons. Therefore, an easily automated procedure based on regular polygons appears to be the least costly approach. Increasing the number of polygons per field reduces the size of each polygon; thus a limit will be reached at which regular polygons are not practical. Since the placement of regular polygons is arbitrary, the description of yield depends on where each polygon lands with respect to the yield variation. However, corn-based polygons showed more potential in explaining yield variation of other corn crops with fewer polygons (or fewer management zones). The prior-year corn yield maps were the preferred method of defining management zones, especially for corn followed by corn.

Crop canopy temperature is useful as an indicator of plant water stress and possibly a good measurement to use as an irrigation schedule initiator. To help determine the feasibility of using canopy temperature to control irrigation events, 26 infrared thermometers (IRTs) were mounted along the main structure of a 3- tower, 137-m center pivot irrigation machine. The center pivot provided a platform to conduct spatial canopy temperature measurements over a 6-ha field that contained 12 different soil mapping units. The IRTs were mounted in pairs, for thirteen, 9.1-m segments along the center pivot. During the 1999 corn growing season, data were acquired on eight separate days during the grain fill period of an irrigation and nitrogen fertilizer rate experiment that consisted of 396 plots arranged in randomized blocks within the 12 soil mapping units. The IRT data were collected during a single pass of the center pivot, at mid-day during mostly sunny conditions. Data were collected and stored using a data logger and a PC mounted on the center pivot. Individual canopy temperature values were stored and later corrected using calibration values for each IRT. Data were then adjusted for temporal data slew caused by time-of-day effects. Measurement techniques, data adjustment algorithms, and sample data are reported. The quadratic relationship of adjusted canopy temperature and irrigation rate, for a day when the non-irrigated plots were under stress, indicates that approximately 65 to 95% of the variation across the 12 mapping units could be explained. For data acquired two days after a 41-mm rain, the relationship explained approximately 30 to 80% of the variation across the soil mapping units. Hence, there is a possibility of determining the relative soil water status using multiple, inexpensive IRTs.

When site-specific agriculture became technologically feasible, existing crop models made computer simulation a natural choice for predicting yield under various combinations of soil, weather, and management. However, modeling for site-specific farming may require both greater accuracy and sensitivity to more parameters than current models allow. The objective of this paper was to evaluate the DSSAT V3.5 corn model, CERES-Maize, for sensitivity to parameters important to site-specific farming. The model was unexpectedly insensitive to inputs for soil type, depth to clay, nitrogen, and plant population, suggesting areas for attention. Although it was appropriately sensitive to rainfall, indicating sensitivity to soil water content is generally correct, there are known problems with the curve number procedure that calculates runoff. The runoff routine needs improvement, and a separate routine may be needed to accommodate within-field redistribution of runoff. The model also responded to maximum air temperature, but since crop temperature varies more than air temperature, perhaps crop temperature should be calculated from air temperature and water stress. Model accuracy issues aside, accommodating spatial inputs and model runs requires enhanced interfaces. These and other suggested enhancements to the model would improve its applicability for site-specific agriculture.

During the past decade, there has been increasing interest in applying water and chemicals to crops based on need or yield potential rather than applying uniformly to the entire field. While ground-driven variable-rate chemical application equipment is now being used, most irrigation systems continue to apply nominally uniform water depths. Our objective was to make variable-rate irrigation applications possible by developing a digitally controlled metering device. The device consists of a reservoir that is alternately filled and emptied at a rate determined by a digital pulse from an external source and requires pressurized sources of water and air. The flow rate can be altered by changing the cycle duration and frequency, by changing air and water pressure, or by exchanging the reservoir with one of different volume. Tests with prototypes indicate reproducible flow rates for a range of operating pressures and discharge cycle durations. Various sprinklers or nozzles may be attached to the outlet if specific distribution patterns are desired. Additionally, the metering device can be used in a wide variety of applications with a variety of fluids or gases for variable-rate flow or injection of a fluid into either another fluid or gas.

Normal‐ and late‐planted cotton (Gossypium hirsutum L.) often differ in fiber properties, especially those properties related to fiber secondary wall characteristics. This field study was conducted to (i) determine the effect of planting date on fiber properties of bolls at two flowering times, and (ii) determine the relationship between fiber properties and canopy photosynthesis during development of those bolls. Cotton (‘Stoneville 453’) was planted on 3 May and 3 June in 1995 and 3 May and 31 May in 1996. Canopy photosynthesis was measured 10 to 12 times on sunny days from initial flowering through the end of the season. Fiber properties were determined on first sympodial position bolls that bloomed during the first and fourth week of flowering (WOF). Maximum canopy photosynthesis was 21% higher in 1996 than in 1995 and lint yield was 22% greater in 1996 than in 1995. Within each year, average maximum canopy photosynthesis did not differ between planting dates, although yield was approximately 30% lower for the late planting date each year. Bolls from the first WOF generally had lower lint percent, higher short fiber content, lower elongation, and lower whiteness index than bolls from the fourth WOF. Micronaire, immature fiber fraction, and fiber cross‐sectional area were linearly related to the amount of canopy photosynthesis that occurred from 15 to 45 d after flowering. Our results are consistent with the hypothesis that assimilate supply influences cotton fiber properties associated with secondary wall characteristics.

Efficient management of irrigation and N-fertilizer requires knowledge of soil variability and crop response to these inputs. Automated measurements can be used to indicate plant stress throughout the growing season and may help indicate timing and amount needed of these inputs. In 1999, a 6-ha center pivot, modified to provide site-specific irrigation, was used to impose irrigation and N-fertilizer treatments on 396 separate corn (Zea mays) plots (9 m by 9 m) on highly variable soils. Treated plots were arranged in randomized complete blocks where space allowed and incomplete blocks where space was limited. Two N-fertilizer rates (135 and 225 kgN/ha) and four irrigation rates (0%, 50%, 100%, and 150% of a base rate) were applied within each of twelve soil mapping units representative of the southeastern Coastal Plain. The four irrigation treatments, averaged across N-fertilizer and soils, produced corn grain yields of 6.4, 8.8, 10.1, and 10.7 Mg/ha. Canopy temperatures were measured on eight separate days using non-contact infrared thermometers (IRTs). Soil moisture of the surface 6 cm was measured on two days when canopy temperatures were measured. Preliminary analysis suggests this information will be useful to researchers and precision farming innovators interested in managing spatial variation, especially within the southeastern Coastal Plain and similar sandy areas.

This study integrates field collection, crop modeling, and remote sensing to assess spatial variability in biophysical properties of a soybean crop. These tools are used to describe and predict leaf area index (LAI), intercepted photosynthetically active radiation (PAR), biomass, and yield for the 1998 growing season at the USDA-ARS Coastal Plains Soil, Water, and Plant Research Center. LAI and yield data collected in the field are compared with LAI, PAR, biomass, and yield modeled with CROPGRO-Soybean. Field and modeled data are then compared to six dates of SPOT 4 satellite imagery and associated vegetation indices (NDVI, SR, SAVI, and TSAVI). When averaged over the growing season, vegetation indices captured the spatial variability of observed LAI and simulated LAI, PAR, and biomass. Vegetation indices from individual dates were less successful, possibly due to the coarse spatial resolution of the SPOT imagery, or due to shortcomings in the index calculations. SPOT imagery did not capture the spatial variability of observed yield.