Monitoring the Application of Urea to Rice

Fertilizers represent one of the highest input costs for agricultural operations and one of the most difficult to manage. Nitrogen can be especially challenging, if applied in appropriate amounts, this application can increase yield. But if applied in excess, in addition to wasting money, it can lead to leaching and contamination of nearby streams and lakes.

To assist in the administration of urea (nitrogen) from rice fields in Australia, researchers at Deakin University are testing the use of multispectral images. These researchers use a MicaSense RedEdge sensor on their aerial platform (drone), giving them the ability to collect detailed information in a field as often as they need to monitor the health of the crop.

For one of their tests involving a rice field, the researchers completed six different flights over a three-month period, capturing calibrated data using their aerial system with a RedEdge sensor. The calibration allowed researchers to compare these different data captures, automatically compensating for the effects that different sunlight conditions can have on the data.

The data collected from one of the flights is shown below. This flight was completed two weeks after the application of urea by machinery, before the application of permanent water in the rice field. Both graphic data displayed were captured from the same flight, the first image is an NDVI map of the data while the second image is an NDRE map.

It does not appear that there are any problems when the field is analyzed using the NDVI map, a vegetation index commonly used to assess the vigor of the plant. In addition to some known problem areas in the corner of the field, the NDVI map does not show any areas of concern after the application of urea.

After a subsequent review using the NDRE map, it becomes evident that there was a problem, the spreader did not distribute the fertilizer evenly.

Why doesn’t the NDVI map show this variability? As the application of nitrogen has not yet affected plant growth and biomass (leaf area) and NDVI values ​​correlated with the amount of biomass or leaf cover of the plant. As a more immediate effect of poor nitrogen application, the chlorophyll content in the leaves fell in the affected areas. NDRE is a much better indicator of chlorophyll levels than NDVI is, which is why the NDRE map shows the problem much more clearly.

Researchers at Deakin University monitored this rice field for the rest of the harvest. By surveying the calibrated data generated by RedEdge and analyzed in Atlas MicaSense, they tracked changes in yield over time, monitoring NDVI and NDRE values ​​for lines that had low or high levels of nitrogen due to the application of non-uniform urea in the beginning of the harvest.

The results show that vegetation indices such as NDRE (enabled by the use of narrowband multispectral sensors such as RedEdge) can detect the application of non-uniform urea much earlier than NDVI. In this case, almost two months earlier.

At the end of the harvest, the NDVI images show the scale of variability in crop health. In this growth phase, NDVI is a good indicator of general biomass and, consequently, the yield of the rice crop. The difference in the application of urea that NDRE detected at the beginning of the season led to a significant drop in crop performance.

Advanced multispectral imaging provides valuable information that serves as triggers to take corrective actions based on this early detection of problems. For example, nitrogen is often applied to rice during panicle initiation, when the panicle (or head) of rice begins to form in the sheath. This application meets the nitrogen requirements of the rice plant during flowering, making the amount and time of application critical. Multispectral imaging can be used to assist in decisions about nutrient application rates and time, maximizing crop yield.