Using Vegetation Indices to Determine Peanut Maturity
Peanuts are a major crop in the United States with 10 states, mainly in the Southeastern US, producing the majority of the peanut crop. Peanuts are an extremely important crop in Georgia, which ranks 1st in the nation in total production. In 2008 over 685,000 acres of peanuts were planted yielding over 2 billion pounds. One of the more challenging aspects of peanut production is determining when to harvest. Unlike most other field crops, peanuts are produced underground rather than above ground where a visual assessment can be made. Digging too early or too late can be devastating. There have been many efforts over the years to develop a system for farmers to effectively determine peanut maturity and when to harvest.
In the 1980’s a system for peanut maturity testing was developed at the University of Georgia. This system requires the use of a Peanut Profile Board which combines qualitative and quantitative assessment of maturity. A sample of peanuts are removed from the field, then using the “hull scrape method” part of the outer layer of the peanut is scraped off, revealing the color of the mesoderm underneath. This can either be done with a pocket knife or a pressure washer. Then peanuts are placed on the profile board according to their color in order to determine days until digging, a recommendation for when peanuts will be mature enough to harvest.
While this method has been implemented on many farms throughout the US, it a very objective means of determining maturity. The goal of our research is to determine the feasibility of using vegetation indices (VI’s) to determine peanut maturity. VI’s are mathematical ratios of the amount of light reflected by plant canopies at specific wavebands (green, red, near infrared (NIR), etc.) Reflectance is a measure of the percentage of sunlight reflected by plants at those wavelengths and can be measured with tractor-mounted sensors like the Crop Circle® (Figure 1).
We conducted the study on 14 peanut varieties. The peanuts were planted in a replicated random block design intended for a planting date/peanut maturity experiment being conducted (Figure 2). Two different planting dates were used: April 20, 2009 and June 08, 2009.
We monitored the reflectance response of all 14 varieties weekly from August 18 until digging. This resulted in 5 reflectance data sets from the first planting date and 10 from the second planting date. Peanut samples were collected and maturity measured using the hull-scrape method 5 times for the first planting date and 6 times for the second planting date (Figure 3). The peanuts from the first planting date (April 20) were inverted on September 18 while the peanuts from the second planting date (June 08) were inverted on October 23.
The reflectance data were used to calculate six different VIs with potential for predicting peanut maturity. The response of the VIs over time was graphed with the goal of identifying a pattern which indicates maturity. For example, in Figure 4 you see the response of the NLI (non-linear vegetation index) over time for 4 of the 14 varieties evaluated. Each cluster of bars represents the NLI response for that variety for the 5 weeks prior to when the peanuts were inverted. The last bar in each cluster represents the NLI response immediately prior to inverting. As you can see in the graph, the rate at which NLI decreased as the peanuts matured leveled off. This leveling off appears to be a good indicator of peanut maturity. The numbers above the bars indicate the days to digging estimated by the hull-scrape method. For the set of data associated with the April 20 planting date, there is a clear relationship between VI response and peanut maturity.
Because of deteriorating weather conditions in late October 2009, the peanuts from the June 08 planting date were inverted a week to 10 days before even the earlier maturing varieties reached full maturity. So even though we see the same general trends for the June 08 planting date as shown in Figure 4, additional research is needed to verify our observations.
An unexpected finding of our work has been that the results of the hull-scrape method are quite inconsistent as the peanuts approach maturity. If you look carefully at the days-to-digging numbers in Figure 4, you will see that for 3 of the 4 varieties, the predicted days to digging stayed about the same across three sampling periods (14 days difference). There are many potential reasons for this including human error in properly sorting the peanuts and in-field variability (the point of doing this study). However, all our samples were collected from the same small plot, so in-field variability is an unlikely cause.
In conclusion, NLI and two other VIs appear to have great potential for serving as indicators of peanut maturity for many varieties. To verify that the 2009 observations are repeatable, the experiment must be repeated during 2010.
This work has been supported by grants from the Georgia Agricultural Commodity Commission for Peanuts and Hatch and State funds allocated to the Georgia Agricultural Experiment Stations.
Download the 2009-2010 Final Report to the Georgia Peanut Commission
Project Leader: George Vellidis
Contact Info: email@example.com
Affiliation: University of Georgia
P.O. Box 748
Tifton, GA 31793