Large-scale monitoring of vegetation properties is crucial to understand water, carbon and energy cycles. It is important to understand the vegetation response to climate and extreme events such as droughts, which are expected to become more frequent and more intense under climate change. In order to assess the evolution of the vegetation at global scale, several approaches can be used, among them optical indices or synthetic aperture radar (SAR) to estimate above ground biomass (AGB). However, optical indices are only sensitive to the green vegetation and only to the upper layer of the canopy in densely vegetated regions. In addition, they are not directly related to the vegetation water content. In contrast, the thermal emission arising from the Earth surface at microwave frequencies depends on soil characteristics such as soil temperature, soil roughness and soil moisture content which different relative contributions at different frequencies. It also depends on the vegetation water content (VWC) and structure, therefore, passive microwave radiometry is a promising tool for monitoring the vegetation hydric state and AGB on a global scale, in particular at L-band. The optical depth at this frequency, mainly due to the presence of water molecules within the vegetation and therefore known as L-band Vegetation Optical Depth (L-VOD), has been shown to be highly sensitivity to AGB (Rodriguez-Fernandez et al. 2018, Mialon et al. 2020). The frequent revisit of satellites such as European Space Agency (ESA) Soil Moisture and Ocean Salinity (SMOS), 1-3 days, makes possible to monitor changes in the VWC and the evolution of large scale carbon stocks, adding a temporal dimension to higher spatial resolution AGB maps produced with SAR sensors. The objective of this PhD work is to twofold. First, L-band brightness temperatures and L-VOD will be analyzed in comparison to other vegetation indices (Normalized Difference Vegetation Index, Leaf Area Index, Solar Induced Fluorescence, …) and VOD at higher microwave frequencies. The study will be done at different temporal and spatial scales to understand the contribution of short-term variations of VWC and the long-term AGB evolution. Secondly, methods to retrieve the AGB from L-band observations will be investigated, in particular using data driven machine learning approaches trained on in situ plots estimations of AGB.

The candidates should have have a Master degree and notions of geophysics and remote sensing and possibly applied mathematics (machine learning). They should be familiar with data analysis in Python/Matlab in a Unix environment. Good oral and writing skills in English.