Evaluation of UAV imagery for mapping Silybum marianum weed patches

Authors: TAMOURIDOU, A - Aristotle University Of Thessaloniki; ALEXANDRIDIS, T - Aristotle University Of Thessaloniki; PANTAZI, X - Aristotle University Of Thessaloniki; LAGOPODI, ANASTASIA - Aristotle University Of Thessaloniki; KASHEFI, JAVID - European Biological Control Laboratory (EBCL); MOSHOU, D - Aristotle University Of Thessaloniki

Submitted to: International Journal of Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/15/2016
Publication Date: 11/6/2016
Citation: Tamouridou, A., Alexandridis, T., Pantazi, X., Lagopodi, A., Kashefi, J., Moshou, D. 2016. Evaluation of UAV imagery for mapping Silybum marianum weed patches. Journal of Chemical Ecology. http://dx.doi.org/10.1080/01431161.2016.1252475.

Interpretive Summary: Mapping of weeds is useful for planning control strategies and monitoring their effectiveness; however it is often difficult and expensive to perform. Milk thistle (Silybum marianum) is a spiny weed with large rosettes that often invades fallow wheat fields in Greece. We used an unmanned aerial vehicle (UAV) to obtain digital images using a multispectral camera (green–red–near-infrared) to map the distribution and abundance of milk thistle. This technique was up to 87% accurate when using a 1-meter (39 inch) resolution for pixels. However, failure to discriminate milk thistle from some other visually similar plants (animated oat, poison hemlock) caused an overestimation of the area infested by milk thistle. More sophisticated analysis, such as using artificial neural networks or machine learning algorithms may improve accuracy. The use of UAVs and multispectral cameras can provide relatively quick, effective mapping of milk thistle, which can help land managers decide if and where to apply treatments to control this invasive weed.

Technical Abstract: The invasive weed, milk thistle (Silybum marianum), has the tendency to grow in patches. In order to perform site-specific weed management, determining the spatial distribution of weeds is important for its eradication. Remote sensing has been used to perform species discrimination, and it offers promising techniques for operational weed mapping. In the present study, the feasibility of high-resolution imaging for S. marianum detection and mapping is reported. A multispectral camera (green, red, near-infrared) mounted on a fixed wing unmanned aerial vehicle (UAV) was used for the acquisition of high-resolution images with pixel size of 0.1 m. The maximum likelihood (ML) classifier was used to classify the S. marianum among other weed species present in a field, with animated oat (Avena sterilis) being predominant. The three spectral bands and the texture were used as input to the classifier. The scale of the mapping was varied by degrading the image resolution to evaluate classification performance, with 1 m resolution providing the highest classification accuracy. The classification rates obtained using ML reached an overall accuracy of 87% with a K-hat statistic of 74%. The results prove the feasibility of operational S. marianum mapping using UAV and multispectral imaging.