Why Are Service Providers Slow to Adopt Crop Sensors for Nitrogen Management?

Why Are Service Providers Slow to Adopt Crop Sensors for Nitrogen Management?

Recently, we have seen important discussions about key aspects of precision agriculture. One of these articles, about the value in variable rate nutrition, explains how there is no easy recipe or universal method to do it successfully, even more so when talking about variable rate application of nitrogen. The dynamics of nitrogen in soil and crop response to nitrogen fertilizers is quite complex, varying from place to place and year to year, making it a challenge to apply the optimum rate.

Advertisement

In another article, the author raises general questions about some missing pieces in precision agriculture. Among the questions, there is a discussion about how complex the prescriptions for variable rate application should be. Simple prescriptions, based only on soil testing, for example, are cheaper to implement and easy to understand, but may not account for important factors related to crop response to fertilizers. Recommendations that are more complex will need more knowledge of the area and some local calibration of algorithms.

For nitrogen, time is also very important. Ideally, topdressing applications should consider in-season data to better account for temporal variability, and for that, we need a workflow of data collection, data processing, decision making, and application usually within one week. Scientists have been developing remote sensing tools to make this feasible. Active crop canopy sensors, usually known as NDVI sensors, are the state-of-the-art remote sensing technologies. These sensors have been available for some time, mostly focused on real-time variable rate application of nitrogen. However, there is still low adoption of this technology by farmers. This raises an important question: if conventional methods are not good enough to make variable rate prescriptions for nitrogen application, why haven’t service providers and farmers adopted crop canopy sensors as a better alternative?

A recent paper published in Field Crops Research journal by Colaço and Bramley provides some insights to answer this question. The paper is a review on the use of crop sensors to improve nitrogen management in grain crops. Overall, the studies using crop sensors to guide variable rate nitrogen reported fertilizer savings of 5%-45% with little effect on grain yield. Economical evaluations reported impacts on profit usually ranging between losses of US$ 30 ha−1 and profits of US$ 70 ha−1, with an average profit of US$ 30 ha−1. The lack of consistent evidence of economic benefits is one factor that limits adoption by farmers. About 25% of studies reported economic losses from sensor-based nitrogen applications, but there are some concerns with the methodology used that may explain the results. The following paragraphs include a summary of the authors’ comments plus some personal opinions.

The strategies for using the sensors are divided in two groups: real-time prescription based on sensor values and a local reference; and real-time redistribution based on sensor values and a pre-defined average rate. In the first one, the concept of reference strips is used to determine the deficit of nitrogen in plants that received a normal rate of nitrogen when compared to plants that received nitrogen fertilizer in abundance. In this group, the algorithm includes two variables, the adjustment of the average rate as a function of crop response in that specific place and conditions, and the variable rate application according to field spatial variability.

The second concept is simpler to implement by using a virtual reference concept, which is obtained as the sensor performs readings in the area. Despite the advantage of not needing the N-Rich reference strips, this strategy assumes that the planned average rate is optimum, and only redistributes the fertilizer in the field. In both scenarios of using crop sensors, the shape of the algorithm used to convert sensor values into prescription rates depends on crop type, development stage, and assumptions of underlining limiting factors. All of this will influence the results obtained with variable rates.

What is observed in many studies is that the greatest gains in the use of sensors have come from the better estimate of the average optimum rate, rather than from the variable rate itself. In this scenario, the use of other tools like predictive models of crop response based on historical data and weather forecasts can enhance the chances of success.

Real-time-VRA-cotton

Real-time variable rate application of topdress nitrogen fertilizer in cotton and the vegetation index variability map obtained by crop canopy sensors.

Regarding the limitations in methodologies used, most experiments are carried out in small plots, which makes it difficult to evaluate the effects of spatial variability. The allocation of reference strips within the field is also a problem because field variability must be considered. Better results can be obtained using on-farm-trials and management zones, so that reference strips can be replicated in strategic locations, using rates close to the optimum recommended by modeling tools rather than very high rates as sometimes used in N-Rich strips.

One thing that sometimes seems to be forgotten is that the ROI of using variable rate technology is strictly related to the magnitude of spatial variability. There is still no widely used precision agriculture opportunity index to compare how variable are the experimental fields used. Most researchers fail to describe and consider this. In relatively homogeneous fields the improvement of nitrogen management using crop sensors can be expected to be much less important than applying the optimum average rate. In fields with large variations, nitrogen may not be the more important yield limiting factor, therefore the use of a single algorithm is unlikely to provide good results.

In conclusion, there is still a lot to be learned about nitrogen management in each situation and this is an area of research that can be improved by new technologies such as big data and machine learning. It is evident that more data including historical information and multiple sensors, coupled with complex models, are needed to describe crop response to nitrogen application. Moreover, this needs to be simple to use, otherwise the technology will not be adopted.