Calibration of Sentinel Land Surface Phenology with Ground Phenological Observations for Agricultural Applications
Climate extremes have led to serious damages in Austria’s agriculture during the last years. The worst case so far was the frost damage to fruit trees and vineyards in 2016, where in Styria alone, damages equivalent to 200 Mio. Euros were reported.
Damage quantification is currently based on rough estimations from farmers or generalized calculations based on statistical figures or local drone surveys. The results from these quantification methods typically show deficiencies in either availability, accessibility, quantity and/or quality.
Both public institutions and insurance companies therefore often lack an objective view on the dimensions of the susceptible areas before and damaged areas after a frost event. In order to provide the efficient monitoring and the required overview, the assessment of the severity and extent of damages due to weather events will be improved. For this purpose, remote sensing data from Europe’s earth observation satellites Sentinel-2 and Sentinel-3 will be used to generate detailed phenological information and join it with meteorological models for improved damage forecasts.
Phenological information is currently mainly based on MODIS data with a resolution between 250 and 1000 m. By combining Copernicus Sentinel satellite imagery and phenological ground-truth data, it will be possible to improve the resolution to 10 m and to optimize damage forecast and assessment. The integration of ground phenology (GP) and land surface phenology (LSP) will increase accuracy and reduce manual survey costs.
Innovative fusion techniques will provide daily phenological data at a high spatial resolution through the powerful platform allowing to draw conclusions for Austria’s small-structured agricultural landscapes and will enable the information on the phenological status in combination with meteorological data and models, which will be used to better predict areas of damage.
In the project PhenObserve, two prototypes based on phenological parameters will be developed:
- The real-time determination of the phenological stage of fruit tree crops (bud swelling, bud break-out, start of flowering, etc.), which in combination with the meteorological data and forecasting models will enable to delineate the regions with the higher potential frost damage;
- The comparison of multi-temporal phenological data of an observation year with the data of a "standard year" is enabling the detection and identification of the maize fields suffering from drought stress at an early stage, prediction and taking measures to improvement.