Know more

Potential of Visible and Near infrared spectroscopy aerial acquisition with high spectral/low spatial resolution for plant phenotyping

Hello, I'm Maxime Ryckewaert, a Cifre PhD student working for Limagrain, an agricultural cooperative, within the UMR ITAP (Irstea Montpellier). My thesis topic is "Coupling visible - near infrared spectrometry and a mobile vector for vegetation characterization in phenotyping campaigns". I've been recruited to the Comic team at Irstea's UMR ITAP as part of a project to use optical instruments to detect stress in precision agriculture. The idea is to exploit the analysis of variance method called REP-ASCA on other types of data sets. In the short term, one of my missions is to transfer all the knowledge acquired during my Cifre thesis to Limagrain. Firstly, the tool and method need to be validated on numerous plots. Secondly, we plan to use the final tool in the varietal selection process.
Today, agriculture has to adapt to climate change and all its consequences, such as drought and water stress in particular, so it's important to propose new, more resistant and better-adapted varieties.
How can we identify and select them? To characterize the water stress tolerance of different corn varieties, I analyze the information contained in their reflectance spectrum, i.e. the radiation reflected by the plants. This radiation contains information on biochemical parameters such as chlorophyll and water content. The idea is to explore and develop new techniques for acquiring and analyzing this type of data.Solutions already exist, but are not adapted to crop selection. Hyperspectral imaging provides a great deal of reflectance spectrum information, but the technology requires heavy cameras and the cost of data acquisition is high. In multispectral, on the other hand, information is lost, but the sensors are light and the system is less costly.The aim of my thesis is to retain all the information on the plant's biochemical parameters, and come up with a lightweight system that can be mounted on a drone to cover several hectares.

• Starting date: 1st November 2016
• University: Université de Montpellier / Institut Agro
• PhD school:  GAIA, Filière APAB, MUSE Montpellier Université d’Excellence
• Scientific field: Process engineering, agro-resources
• Thesis management: Jean-Michel Roger (UMR Itap, Inrae)
• Thesis supervisors:  Alexia Gobrecht (UMR Itap, Inrae), Nathalie Gorretta (UMR Itap, Inrae), Fabienne Henriot (Limagrain)
•  Funding: Cifre
•  #DigitAg : Labeled PhD – Challenge 2 (Phénotypage rapide)

Keywords: Spectrometry, Phenotyping, Water stress, Chemometrics, Drone

Abstract: The operational objective of this thesis is to study the opportunities offered by the VIS/NIR spectrometry couple with an aerial acquisition system to address the needs of new tools for high-throughput phenotyping. The underlying application is to identify and describe new genotypes with better behavioural response under water-stress. From a technological point of view, it also represents a potential way to have access to the high spectral resolution of vegetal cover. However, the spectral information must be relevant and exploitable for phenotyping, and therefore, processing will be optimised by developing: • methods for extracting appropriate parameters for breeding genotype • methods that will improve in spectral/spatial resolution of signals Maintaining a high degree of spectral resolution is the key factor to produce models. In this context, it would provide answers to this following scientific question: How can a low spatial/high spectral resolution sensor be coupled with a mobile vector in order to produce high spatial /high spectral resolution information allowing to describe vegetal response under stress? Thus, the research objectives are: • To test the hypothesis that with a total spectral signature of vegetation, the crop monitoring or the extraction of phenotyping traits are more robustness and more sensitive with a small variation of genotype response under stress. • To define acquisition protocols and to develop processing methods of these spectral information to guarantee enough spectral and spatial resolution for phenotyping • To approve methodology for identification and precise characterisation of different corn cultivars under water stress.

Contact: alexia.gobrecht [AT] inrae.fr

Social networks: Resesearchgate –  Linkedin

Communications & Papers

The impact of the spatial resolution of highly resolved spectral data on pan-sharpening methods to reconstruct a hyperspectral image.
Maxime Ryckewaert, Julien Morel, Jean-Michel Roger, Alexia Gobrecht, Fabienne Henriot and Nathalie Gorretta (2017). EFITA 2017, Montpellier (France), July 2nd-6th (European conference dedicated to the future use of ICT in the agri-food sector, bioresource and biomass sector, Session 5A: Sense&Rob1: sensing, robotics and electronics for agriculture (I))

ANOVA-Simultaneous component analysis on vegetation spectra data acquired into an experimental design.
Ryckewaert Maxime, Roger Jean-Michel, Henriot Fabienne, Gorretta Nathalie, Gobrecht Alexia. HelioSpir, 27/11/2017 (communication orale.

Multivariate analysis of variance of vegetation spectra dataset included into an experimental design by using ANOVA-SCA and ANOVA-Target Projection.
Ryckewaert Maxime, Gorretta Nathalie, Henriot Fabienne, Gobrecht Alexia, Roger Jean-Michel. Conférence SFPT, 18/05/2018 (communication orale).