A multi-sensor approach to monitor the ongoing restoration of edaphic conditions for salt marsh species facing sea level rise: An adaptive management case study in Camargue, France.

The Camargue or Rhône delta is a coastal wetland in southern France of which parts formerly devoted to salt production are undergoing a renaturation process. This study assessed a multisensor approach to investigate the link between sediment size distribution, habitat development mapped with Worldview 2, flooding durations estimated with time series of SENTINEL 2 images and elevation modelled with a LIDAR point cloud in former saltworks. A Random Forest classification algorithm was used to map the vegetation distributions of Sarcocornia fruticosa and Arthrocnemum macrostachyum, main representatives of the NATURA 2000 "Mediterranean and thermo-Atlantic halophilous scrubs (Sarcocornetea fruticosi)" habitat on the site. The best habitat map was obtained when considering the species separately. The random forest Out-of-bag errors were 1.43 % for S. fruticosa and 2.18 % for A. macrostachyum. Both species were generally distributed on different elevation and flooding duration zones considering mean values. Flooding duration was estimated with the Water In Wetland index (WIW) based on 15 Sentinel-2 scenes. Two models related to sediment grain size distribution were developed: one predicting the flooding duration and one predicting the halophilous scrub distribution. Maps of flooding duration, sediment grain size distribution and elevation highlighted two main zones in the study area: a western section with coarser sediments, shorter flooding durations and higher elevations under sea influence; an eastern section with finer sediments, longer hydroperiods and lower elevations under a historic river influence. This multidisciplinary approach offers perspectives for using space-based data over large scales to monitor changes in edaphic conditions of coastal areas facing natural and anthropogenic forcings. The results call for further investigations to predict the dynamic distribution of other coastal habitats following climate change impacts, such as sea level rise.