UAV (or Drone) technology is an exciting new technology that has taken off (pardon the pun!) over the last couple of years. This is especially true in the surveying and mapping industries where the benefits are a no-brainer. For example in mining, stockpile and pit volume calculations can take several teams of surveyors several days to complete, but with UAVs only one man and a couple of hours is all that is needed to do the same work.
And so I believe that UAV technology can also benefit the agriculture industry. Take a look at the Agriculture page on 4D Delta's website by clicking here to see some of the applications in broadacre farming. One benefit I see is mapping broadacre crops at peak-biomass using near-infrared (NIR) imagery to get an estimate of how much yield to expect at harvest. Being able to confidently estimate yield several months before harvest would put the farmer in a strong position when it comes to negotiating prices for their produce.
In September last year we got the opportunity to image a broadacre farm at peak-biomass using the senseFly eBee kitted with an NIR imaging sensor. We captured approximately 125ha in two flights at a resolution of 4cm per pixel (that's really good by the way!), see Image 1. From that imagery we converted it into a Normalised Difference Vegetation Index (NDVI) map, Image 2, which gives an indication of plant biomass.
Image 1: near-infrared map of a broadacre farm
Image 2: NDVI map created from the NIR imagery
Later in the year, around early December, the crop was harvested and we got access to the yield data from the harvester, see Image 3.
Image 3: Yield map from harvester
In the farm management software SMS we gridded the NDVI & Yield datasets into 10m lots and then performed a correlation analysis between them (Note that the comparable area is the purple boundary in both Images 2 & 3). The result was a correlation value of 0.467. In correlation analysis a result of 0 would be no correlation, 0.3 - low correlation, 0.5 - medium correlation, 0.7 - high correlation, and 1 - perfect correlation. So our results showed that NDVI mapping at peak-biomass has the potential of being a good indicator of final yield.
Looking at the two datasets there are some areas where high/low yield correlates strongly to the biomass given in the NDVI image and some areas that do not. For the latter, the area that is most noticeable is at the eastern boundary of the comparison area where biomass at peak-biomass was high but yield was very low.
There are a myriad of factors that can affect final yield after determining biomass at peak-biomass including rainfall, soil moisture retention, disease, pestilence, and so on. I think in this case that these factors affected the overall correlation from peak-biomass to final yield. But how much these factors do affect final yield is a reason in itself to get imagery flown. Wouldn't you want to know why a potentially high yielding area of your paddock turned into a dust bowl in a matter of months?
This study is only the beginning but I think it shows the potential of UAV technology to give farmers a good estimator of final yield and/or to determine why high-potential crops fail.
I'd love to hear what you think about our study so send an email to email@example.com with your comments.