Comparison of visual vegetation indices from aerial images to measure turfgrass health using small unmanned aircraft
H.J. Sommer: The Pennsylvania State University
<div>Normalized Difference Vegetation Index (NDVI) is difficult to measure over large areas of golf courses using handheld devices, and cost-effective multispectral cameras are still not widely available. To facilitate using standard red-green-blue (RGB) cameras carried by small unmanned aircraft (UA), ten traditional visual vegetation indices (VVI) and three new VVI were computed from UA images of turfgrass. The VVI values were then correlated to handheld NDVI ground-truth measurements to assess their efficacy in predicting turfgrass health. Initial correlations were developed for A4 creeping bentgrass in which plots were stressed over a 14 day study by mowing at different heights. Correlations were validated for a second study of A4 creeping bentgrass in which the plots were subjected to water stress over 30 days. The new VVI include L*a*b* color chromaticity defined by the <a href="https://en.wikipedia.org/wiki/International_Commission_on_Illumination">International Commission on Illumination</a> (CIE) and a novel optimization method to tailor VVI for specific species. Visible Atmospherically Resistant Index (VARI) was the most effective among traditional indices. The new General Visible Vegetation Index (GVVI) using optimal RGB weighting coefficients performed better than VARI. Direct linear correlation to L*a*b* chromaticity components had almost identical efficacy as GVVI. These VVI allow expanded use of UA to scout golf courses and identify plant pathology for early identification/intervention.</div>
|
