Plenary 1 2018

Monitoring natural ecosystems using multi / hyperspectral imaging and physical modeling

Abstract: Monitoring status and changes in function and composition of ecosystems is an important challenge: operational applications are awaited by ecologists and deciders in order to identify solutions against the accelerating erosion of biodiversity. Remote sensing is a critical source of information to build such monitoring system, and imaging spectroscopy proved its capacity for the estimation of biodiversity indices over various types of ecosystems, even highly heterogeneous and complex tropical forests. Multispectral satellite imagery such as Sentinel-2 data compensates less spectral information by high frequency of revisit and a capacity to perform regional mapping. Therefore an important work is needed in order to identify the potential of individual data sources as well as their synergies.

Physical modeling is particularly relevant tool to help understand the physical processes leading to information acquired with imaging spectroscopy: it can be used for the identification of key biophysical properties influencing the radiometric signal, and in order to improve existing methods, including generalization ability of data driven image processing algorithms using machine learning or statistical approaches. It can also play a role in the identification of the synergies among sensors, as well as the preparation of future satellite missions.

This presentation will introduce recent advances in physical modeling of vegetation, from leaf scale to canopy scale. It will discuss about possibilities to integrate lidar information and field spectroscopy in order to generate simulations of different types of images acquired over heterogeneous vegetated ecosystems, and perspectives for the exploitation of current data sources and the preparation of future missions.