Assessing macrophyte seasonal dynamics using dense time series of medium resolution satellite data

Abstract

Thanks to the improved spatial and temporal resolution of new generation Earth Observation missions, such as Landsat 8 and Sentinel-2, the potential of remote sensing techniques in mapping land surface phenology of terrestrial biomes can now be tested in inland water systems.\n\nWe assessed the capabilities of dense time series of medium resolution satellite data to deliver quantitative information about macrophyte phenology metrics, focusing on three temperate European shallow lakes with connected wetlands, located in Italy, France and Romania.\n\nLeaf area index (LAI) maps for floating and emergent macrophyte growth forms were derived from semi-empirical regression modelling based on the best performing spectral index, with an error level around 0.11 m2 m-2. Phenology metrics were computed from LAI time series using TIMESAT code and used to analyse macrophyte seasonal dynamics in terms of spatial patterns and species-dependent variability. Peculiar patterns of autochthonous and allochthonous species seasonality across the three study areas were related to the environmental characteristics of each area in terms of ecological and hydrological conditions.\n\nIn addition, the influence of satellite dataset characteristics - i.e. cloud cover thresholding, temporal resolution and missing acquisitions - on phenology timing metrics retrieval was assessed. Results have shown that with full resolution (5-day revisit) time series, cloud cover can bias phenology timing metrics by less than 2 days, and that reducing temporal resolution to 15 days (similar to Landsat revisit) still allows for mapping the start and peak of macrophyte growth with an error level around 2-3 days.