Estimation of Leaf Area Index in vineyards by analysing projected shadows using UAV imagery

by | Dec 13, 2021 | South Africa Wine Scan

Background and aim:
A few decades ago, farmers could precisely monitor their croplands just by walking over the fields, but this task becomes more difficult as farm size increases. Precision viticulture can help better understand the vineyard and measure some key structural parameters, such as the Leaf Area Index (LAI). Remote Sensing is a typical approach to monitoring vegetation which measures the spectral information directly emitted and reflected from vegetation. This study explores a new method for estimating LAI which measures the projected shadows of plants using UAV (unmanned aerial vehicle) imagery.

A flight mission over a vineyard was scheduled in the afternoon (15:30 to 16:00 solar time), which is the optimal time for the projection of vine shadows on the ground. Real LAI was measured destructively by removing all the vegetation from the area. Then, the projected shadows in the image were detected using machine learning methods (k-means and random forest) and analysed at pixel level using a customised R code.

Results and significance of the study:
A strong linear relationship (R² = 0.76, RMSE = 0.160 m² m-2 and MAE = 0.139 m² m-2) was found between the shaded area and the LAI per vine. This is a quick and simple method, which is non-destructive and gives accurate results; moreover, flights can be scheduled during other periods of the day than solar noon, such as in the morning or afternoon, thus enabling pilots to extend their working day. Therefore, it may be a viable option for determining LAI in vineyards trained on Vertical Shoot Positioned (VSP) systems.

Vélez, S., Poblete-Echeverría, C., Rubio, J. A., Vacas, R., & Barajas, E. (2021). Estimation of Leaf Area Index in vineyards by analysing projected shadows using UAV imagery. OENO One55(4), 159–180.

This abstract is republished in its original form, with headings inserted, as permitted by the following Creative Commons licence: Creative Commons Attribution 4.0 International License

Image credit: Shutterstock

Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors