Monitoring alpine grassland dynamics with optical sensors on multiple spatial scales

Abstract

After forested areas grasslands are the second largest land cover type in the Alps ranging from intensively managed meadows to semi-natural pastures for livestock. Although grasslands are well studied from an ecological point of view, the changing climatic condition combined with the high diversity and management of grasslands require a continuous monitoring of their status. Recently, advances in the availability and detection frequency from optical sensors has been accomplished through new sensors (Copernicus Sentinels), alternative platforms (e.g. UAV) or further adapted technology (e.g. repeated digital imagery) which are helpful to link optical responses to real biophysical processes. We produced the NDVI index from four different sensors: Sentinel-2 MSI, Phenocams, Spectral Reflectance Sensors (SRS) and a spectroradiometer. As test sites we used four different meadows in South Tyrol during the period of 2017. We then analysed how measurements on different S2-pixels are represented by the respective scale and what the strengths and weaknesses of each sensor are. First results show clear differences between the NDVIs acquired from the diverse scales. Although intense filtering processes the linear correlation among the Sensors is lower than expected (R2 0.46 to 0.92) and especially the NDVI saturation level is reached at different periods as well as different maxima. This has an impact on the combined use of the same index from different sensors on analysing vegetation dynamics (e.g. stress, drought), anthropogenic interaction (e.g. irrigation, harvesting) and for the prediction of biophysical parameters (e.g. LAI).