Assessment of the riparian leaf density using different satellite imagery in a mountain stream

In the last years, remote sensing techniques may provide a useful tool in retrieving Vegetation indices (VIs) and evaluating vegetation features. Image pixel resolution can affect data analysis and results accuracy. The potential of three satellite images in retrieving the Leaf Area Index (LAI), with high and medium resolution (Pleiades 1A, Sentinel-2 and Landsat 8) was tested in two Mediterranean streams, in a nearby deciduous forest and in a winter wheat field. The Caraux-Garson, the Lambert-Beer, and the Campbell-Norman equations were used to calculate LAI from the Normalized Difference Vegetation Index (NDVI). To validate the sensor data observed LAI values were detected in-situ with the Licor LAI 2200 Plant Canopy Analyzer and compared with LAI retrieved from the satellite imagery. Generally, Pleiades 1A and Landsat 8 images performed better statistical results than Sentinel-2. The former in deciduous forest or in sites characterized by stable riparian vegetation with high canopy closure, the latter in winter wheat sites or in stream reaches where the vegetation cover was homogenous or, conversely, almost absent. Sentinel-2 images provided more accurate results in terms of the range of LAI values. Regarding the different equation used, the Lambert-Beer generally performed best in estimating LAI, especially in areas, such as deciduous forests, that are geomorphologically stable or with a denser vegetation cover.