Sentinel-2 data för karakteristik av säsongsvariationen hos markvegetation
The project aims at developing and testing methodology for estimating vegetation seasonality at high spatial resolution from Sentinel-2 data, and to assess the scientific value of these products within the fields of phenology and photosynthetic carbon uptake. Sentinel-2 will generate data of very high spatial (10-60m), temporal (3-5 days) and spectral (24 bands) resolutions. This combination will allow for a range of new applications and products to be developed that will be invaluable for vegetation monitoring. Seasonal dynamics of vegetation has so far only been possible to observe at coarse resolutions (100’s of meters to kilometers), making it impossible to draw accurate conclusions about the seasonal development of individual landscape units. This has led to considerable difficulties when calibrating data against field measurements, and for developing accurate models of seasonal carbon uptake. The poor temporal coverage of higher-resolution data (SPOT, Landsat etc.) has hampered the use of these data for assessing vegetation dynamics.
We aim to develop a software platform for extracting seasonal dynamics from Sentinel-2 data, based on the existing TIMESAT platform. This software platform has been in use for >10 years and has proven to be useful for time-series analysis of coarse-resolution satellite data. The platform will need to be modified in several respects in order to allow for accurate and rapid analysis of data, and will be optimized for Sentinel-2 data with respect to accuracy and speed, in collaboration with Lund university high-performance computing centre, LUNARC. In calibrating and validating the new methods we will utilize data from an existing network of optical sensors with year-round multispectral measurements, located in the Nordic countries and Africa.
We further aim to develop and test our methods with two application areas in mind: phenology and carbon uptake of the vegetation. We have long experience in both fields and can capitalize on good field data and broad scientific collaboration. With respect to phenology it is the improved possibility to calibrate satellite measurements against ground observations that is particularly important. Spectral measurements, as well as manual field observations of phenology, are generally carried out in areas that are too small for coarse-resolution satellites, but that match the pixel size of Sentinel-2 very well. We expect that this improved matching will improve the precision of phenological parameter extraction, to enable better understanding of climate-vegetation interactions. Carbon uptake, estimated from flux tower measurements, is the other field where we think important break-troughs can be made with Sentinel-2. The footprint area of the flux-towers have up till now not been properly characterized with respect to seasonal vegetation development. However, with Sentinel-2 data the footprint areas can be accurately modeled and a better understanding of the carbon flux data be achieved.