Date of Award
Doctor of Philosophy (PhD)
Environmental Engineering and Earth Sciences
Dr. Ahmad Khalilian, Committee Chair
Dr. Mike Marshall
Dr. Joe Maja
Dr. Haibo Liu
Dr. Dara Park
Irrigated land in South Carolina has increased at a rate of about 4,047 hectares per year over the past two decades. With this increase in irrigated land comes the ability to gain higher yields by applying nutrients through irrigation water. Therefore, many growers apply nutrients through irrigation systems, known as fertigation.
On average, South Carolina growers apply approximately 100 kg/ha nitrogen (N) on cotton for a total of 32 million kg annually which totals $4.4 million annually. To stay competitive in the global market, it is increasingly important for growers to reduce crop input costs while maximizing yields. For example, a 20% reduction in N usage could save South Carolina cotton growers over $3.7 million annually.
Applying proper fertilizer rates is a major management decision for Southeastern U.S. producers. In this region, considerable soil variation (texture and water holding capacity) and other major factors within production fields affect crop production including fertilizer management strategies. Therefore, uniform application of N fertilizer over the entire field in this region can be both costly and environmentally questionable.
Several researchers have developed sensor-based algorithms and guidelines for variable-rate N management for this region. However, currently, there is no variable-rate fertigation equipment available to apply a correct amount of N where it is needed within a field through an overhead irrigation system. Therefore, the first goal of this study was to develop a variable-rate N application system that works independently of irrigation water flow and can be retrofitted onto an overhead irrigation system (conventional or variable-rate) for site-specific fertilizer application. The variable-rate fertigation system (VRFS) uses the pulse width modulation technique to apply precise rates of N based on prescription maps. The system is controlled by custom software developed at Clemson University, SC, USA.
The application system closely followed design specifications and can apply different rates of N ranging from 0 to 135 kg/ha and could easily be retrofitted on an existing overhead irrigation system (uniform-rate or variable-rate). The VRFS was completely independent of the amount of irrigation water being applied to a location in a field and could apply fertilizers based on crop needs.
The average application errors for the nozzle flow uniformity tests was 0.1%. The pulses with modulation results were promising with an overall average error of 1.8%. The system was capable of following prescription maps with an average N application rate error of less than 1.8% for all N rates. There was a strong correlation (R2 = 0.9996) between target and actual N application rates.
Additionally, there are no practical decision-making tools available for variablerate application of N through overhead sprinkler irrigation systems, the predominant row crop irrigation system in South Carolina. Therefore, field tests were conducted on cotton (Gossypium hirsutum L.) during the 2016 and 2017 growing seasons to 1) adapt the Clemson sensor-based N recommendation algorithms from single side-dress application to multiple applications through an overhead irrigation system; 2) develop correlations between plants “Normalized Difference Vegetation Index” (NDVI) measured using a commercially available optical sensor (GreenSeeker®) and those measured by an unmanned aerial vehicle (UAV), and c) to compare sensor-based VRFS with conventional nutrient management methods in terms of N use efficiency (NUE) and crop responses on three soil types.
Two seasons of testing Clemson’s N prediction calculation to apply multiple applications of N was very promising. The multiple applications of N compared to the growers’ method (even though much less N was applied) had no adverse impact on yields in either growing season. There was no difference in cotton yields between 101 and 135 kg/ha N applications in either management zone. Also, there were no differences in yield between sensor-based, multiple N applications and conventional N management techniques. In relation to comparisons of the sensor methods only applying N in three or four applications, statistically increased yields compared to single or split applications in 2016. Applying N in 4 applications, statistically increased yields compared to single, split or triple applications in 2017.
When the sensor-based methods were compared to the growers’ methods averaged over four treatments the sensor-based N applications reduced fertilizer requirement by 69% in 2016 and 57% in 2017, compared to growers’ conventional method. When comparing N rates among the four sensor-based methods (three or four) applications, increased N rates by 22.41 kg/ha in 2016 and 25.77 kg/ha in 2017 compared to single or split applications but increased the cotton lint yields by 272.36 and 138.98 kg/ha, for 2016 and 2017, respectively.
There was a positive correlation between the applied N rates and the leaf N concentration in cotton leaves. When more N was applied the more leaf N content was found in the plant’s leaves. However, this had no adverse impact on yields because the sensor-based methods applied significantly less N but three and four sensor-based applications yields were not significantly different from treatments that received 101 and 135 kg N/ha.
Plant height was significantly less on the sensor-based methods compared to the growers practice. Statistically there was no difference in boll count between treatments 101 and 135 kg N/ha and four sensor-based N applications. Cotton biomass samples were collected and were not significantly different from any of the treatments.
Utilizing an UAV to measure plant NDVI and subsequently calculate plant N requirements is promising. A drone was calibrated against ground based GreenSeeker® optical sensors. NDVI values measured with the sprayer-mounted GreenSeeker® were correlated with those measured using an UAV. There was a strong correlation between the GreenSeeker® and UAV.
Williams, Phililp, "Development of a Sensor-Based, Variable-Rate Fertigation Technique for Overhead Irrigation Systems" (2018). All Dissertations. 2176.