Brenda Tubana, Dennis Burns, Marilyn Dalen and Ralph Frazier

A vast amount of literature collectively estimates that more than 60 percent of applied nitrogen fertilizer is lost from the soil and plant systems. This translates into costly losses in field crop production and increased threats to water quality through groundwater nitrate contamination and nitrogen enrichment of surface water. The most logical approach to minimize losses is to manage nitrogen fertilizer by applying the right rate in the right location when it is needed most by the crop. However, recommendations derived from soil and plant tissue testing have a long turnaround time, limiting their use for on-the-go and as-needed nitrogen fertilizer applications.

Louisiana’s most productive lands along the alluvial floodplain of Mississippi and Red rivers are farmed to a wide array of field crops. More than 10 years ago, cotton occupied about 900,000 of these acres. While cotton acreage has steadily declined in the state, it does not change the fact that effective management of nitrogen fertilizer remains an essential component of its year-to-year success.

In 2008, a research project funded by Cotton Inc. began with an objective of integrating remote sensing technology for managing nitrogen fertilizer in cotton. The first three years of research were invested in collecting sensor readings and yield data, which were immediately followed by establishing a working algorithm that computes cotton nitrogen requirements based on real-time sensor readings. In 2012, the U.S. Department of Agriculture provided funding support to conduct a large-scale test to compare the performance of remote sensing and variable rate technology (VRT) with the standard practice of a one-time, uniform application of nitrogen fertilizer at the early growth stage of cotton. The test was conducted in Tensas Parish (Figure 1).

The remote sensor/VRT system uses sensors for applying liquid nitrogen fertilizer at the base of cotton plants, six rows at a time (Figure 2). The other components of system are a computer, a rate controller, a flow meter, a flow control valve and a boom control bar.

The soil types in these locations are diverse and have several multiple soil productivity management zones. The plot size was large, almost equivalent to an acre, for three replications; the total area could represent a relatively small cotton field. The net return from nitrogen fertilizer application was one of the performance measures used to compare remote sensor/VRT systems and the farmer’s standard nitrogen practice.

Six out of nine site-years showed benefits from using the remote sensor/VRT system, with an average $34.14 per acre net return from nitrogen fertilization (Figure 3). The largest gain was $69.47 per acre in 2014 at Site B, which was attributed to both increased yield of 60 pounds per acre and an average nitrogen fertilizer application of only 89 pounds per acre, which saved fertilizer cost by $27.47 per acre. In this scenario and at Site A in 2013, the average nitrogen fertilizer rates were less than the farmer’s standard practice yet resulted in higher yield. This is because the system can distribute a precise amount of nitrogen fertilizer where it was needed in the field. The remote sensor/VRT system’s ability to adjust the fertilizer rate is not limited by the concept of “producing more with less”; rather, the system can also detect high potential yield, which requires more nitrogen fertilizer — for example Site B in 2013 and Site A in 2014. In other cases, lower amounts of nitrogen were applied to offset potential yield losses in Sites A and C in 2012. Each instance led to a higher net return when compared with the farmer’s standard practice.

Applying lesser amounts of nitrogen fertilizer has implications on water quality and the environment in general. The remote sensor/VRT system applied an average 30 pounds of nitrogen per acre, which was less than the farmer’s standard practice. This reduction did not result in substantial yield losses. In fact, two sites attained even higher yield (Table 1). Looking on the other side of the story, nitrogen applied at the same rate would have been lost from the soil.

Clearly, the remote sensor/VRT system has the potential to improve the profitability of cotton production in Louisiana, but its adoption is limited by many factors. Among them are the cost of the technology and the requirement for training. One way of facilitating the use of as-needed nitrogen application in cotton is to use another method of acquiring crop information that will produce similar outcomes as the remote sensor/VRT system but is easier to use and less expensive.

A drone equipped with digital camera offers a faster way of collecting aerial images to measure normalized difference vegetation indexes (NDVI) over larger production areas and landscapes. NDVI measures the greenness of plant biomass, indicating the density of the vegetation. The speed in acquiring crop information enables farmers to identify problems more quickly and act in a timelier manner to avoid yield losses.

In 2016, two locations were selected for concurrent collection of NDVI using the remote sensor/VRT system and aerial images from a drone. All aerial images were tagged and processed using software to generate an NDVI map. The objective was to evaluate the use of aerial image-based NDVI as a reliable indicator of cotton nitrogen health status. This was done by comparing the maps and values of NDVI of more than 7,000 data points between these two platforms. Distribution of values was visually the same between aerial NDVI and the remote sensor/VRT system NDVI; clusters of high values were found at the same location on both maps (Figure 4). While the aerial-NDVI readings had a wider range of value than the remote sensor/VRT system, there was a moderately strong linear relationship between these two groups of NDVI values. The successful establishment of the procedure for generating NDVI maps using drones equipped with digital camera will make remote sensing a more powerful tool in improving nitrogen fertilizer management in cotton production.

Brenda Tubaña holds the Jack E. and Henrietta Jones Professorship in the School of Plant, Environmental and Soil Sciences. Dennis Burns is an extension agent in Tensas Parish. Marilyn Dalen is a graduate student in the School of Plant, Environmental and Soil Sciences. Ralph Frazier is an extension agent in Madison Parish.

(This article appears in the fall 2017 issue of Louisiana Agriculture.)

Figure 1. Locations of the large demonstration plots for remote sensing and variable nitrogen rate technology in cotton conducted in Tensas Parish from 2012 to 2016.

Figure 2. Remote sensor and variable nitrogen rate applicator running in cotton field in one of the demonstration plots in Tensas Parish.

Figure 3. Net return in dollars per acre from the amounts of nitrogen fertilizer recommended and applied by using remote sensor and variable rate technology, using farmer’s standard nitrogen practice ($0) as reference.

Figure 4. Comparison of maps and values of normalized difference vegetation index from remote sensor variable rate technology and aerial images.