Title: Sensing nitrogen requirements for irrigated and rainfed cotton Authors
Submitted to: Applied Engineering in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: June 9, 2014
Publication Date: N/A
Interpretive Summary: Current nitrogen (N) management for crop production often results in inefficient use of N fertilizer that can lead to economic losses and environmental contamination. A field experiment was conducted in 2010 - 2012 at the University of Missouri Fisher Delta Research Center to investigate using crop canopy sensors to better synchronize N application and crop N demand for cotton. Plots with total applied N ranging from none to the local recommended rate and application timings from preplant to 10 days after first flower were sensed on irrigated and rainfed cotton. Rainfall varied among the years, which is typical for the Mid-South region. Similarly, which in-season and end-of-season factors were significant varied considerably among years and none was consistent for all three years. However, as more data are collected the relationship between in-season (sensors) and end-of-season (yield and quality) observations should improve. This research is important for improving N recommendations for cotton. Producers will benefit from fewer excess N applications resulting in wasted fertilizer, energy, and labor, and everyone will benefit from less environmental contamination contributing to problems like the hypoxic zone in the northern Gulf of Mexico.
Technical Abstract: Crop canopy sensor-based nitrogen recommendations have been established for grain crops including corn and wheat, but less information exists for using these sensors in cotton. Further, little is known about the impact of crop water stress on sensor-based recommendations. To address these problems, a field experiment was conducted in 2010 - 2012 at the University of Missouri Fisher Delta Research Center to investigate the performance of crop canopy sensors and the timing of sensor data collection for both irrigated and rainfed cotton receiving different nitrogen (N) rates and timings of N application. Eleven N treatments were included, with total N ranging from 0 – 134 kg ha-1 and application timings from preplant to 10 days after first flower. A data collection platform including sensors for visible and near-infrared canopy reflectance, crop height, and canopy temperature was driven through the plots once preflower and once postflower each season. While growing degree days indicated that air temperatures were similar all three years, rainfall varied, which is typical for the Mid-South region. Which in-season (i.e., sensor readings) and end-of-season (i.e., yield and quality) factors were statistically significant varied considerably among years. For irrigation: normalized difference vegetative index (NDVI) - calculated from the reflectance measurements, height, temperature, yield, turnout, micronaire, length, uniformity and strength all exhibited significant differences, but none were consistent over all three years of the study. Similarly, NDVI, height, yield, turnout, and micronaire exhibited significant N-treatment effects but none were consistent over all three years. The lack of N response for some of the sensor data was affected by the fact that not all of the N treatments had been applied when the sensor data were collected, especially for the preplant readings. When the data from all three years were combined in a correlation analysis, only correlations between yield and both postflower NDVI and height and between uniformity and both postflower NDVI and temperature had |r| > 0.5. However, as more data are collected and additional variables are investigated the relationship between in-season and end-of-season observations should improve. While these data demonstrate that in-season sensor measurements can detect differences in water and nitrogen status for cotton that result in end-of-season differences in yield and fiber properties, further research, including evaluation of additional sensors, is needed before real-time N-application algorithms that account for the effects of drought stress can be established.