Leveraging Remote Sensing, Field Surveys, and Numerical Modelling to Understand the Prek System of the Cambodian Mekong Delta - From Hydrological Processes to Ecosystem Services ; Exploiter la télédétection, les enquêtes de terrain et la modélisation numérique pour comprendre le système Prek du delta du Mékong cambodgien - des processus hydrologiques aux services écosystémiques

This thesis provides insights into the hydrological processes in the Cambodian Mekong delta, and their relationship to ecosystem services from which the local population benefits. In particular, the role of Prek channels is examined. First constructed during the period of the French protectorate in the early 20th century, Preks diverge perpendicularly from the Mekong and its deltaic distributary, the Bassac. They provide a connection between the river mainstream and its floodplains by breaching the elevated river banks. The aim of this thesis is to assess how large-scale changes in the Mekong basin, such as climate and land use change, as well as hydropower development, are reflected at a local scale in its delta. It also aims to elucidate how local infrastructure such as Preks catalyzes these processes and could help mitigate negative effects of future scenarios.To address all of these questions, a multi-method approach was chosen. First, a thorough literature review of the challenges facing the Mekong basin and its delta was carried out. Then, an analysis of long-term trends in water levels, precipitation patterns, and inundation extents and dynamics at the scale of the Cambodian Mekong delta was provided. Water level time series at four hydrological stations maintained by the Mekong River Commission (MRC) were subjected to a trend and break point analysis. Furthermore, precipitation data from three in-situ stations, as well as remotely sensed CHIRPS data were compared. Inundation extents were assessed on the basis of a new model developed in the course of this thesis: WAFL - Water Level Flood Link. WAFL establishes a correlative link between in-situ water level time series and inundation maps derived from Sentinel-1 and -2 images (2017 - 2020), accounting for flood propagation delays and attenuation. It can then be used to reconstruct past inundation extents to overcome spatial data sparsity, with good accuracy compared to historical Landsat-derived maps.Furthermore, fieldwork was carried out in a case study ...