Exploitation of Airborne Hyperspectral Data to Retrieve Vegetation Chlorophyll Fluorescence
Cogliati, Sergio; Rossini, Micol; Panigada, Cinzia; Colombo, Roberto
University of Milan-Bicocca, ITALY
Monitoring plant photosynthesis using sun-induced chlorophyll fluorescence (Fs) is one of the major interests for remote sensing (RS) in recent years. The analysis of Fs together with spectral indexes related to plant greenness and status such as the widely used Normalized Difference Vegetation Index (NDVI) and Photochemical Reflectance Index (PRI) can provide detailed information about plant status and actual photosynthetic rate.
The retrieval of Fs using passive optical remote sensing is a challenge because: i) Fs is a small signal compared to the total radiance that reaches the sensor; ii) the detection of Fs requires high resolution spectra; iii) the spectral regions suitable for Fs estimation correspond to the O2-A and O2-B absorption bands, requiring a very accurate atmospheric correction. Spaceborne observation of Fs is the concept that underlies the scientific Fluorescence Explorer (FLEX) satellite mission in preparation by the European Space Agency (ESA). Prompted by FLEX, studies related to the definition of instruments and Fs retrieval algorithms rapidly increased in last years.
The object of this contribution is the retrieval of Fs exploiting airborne imageries collected by hyperspectral sensors (e.g. CASI-1500) at different times during the day to detect the diurnal changes of sun-induced chlorophyll fluorescence. Different retrieval techniques based on the Fraunhofer Line Depth (FLD) and Spectral Fitting (SF) approaches were tested at the O2-A band in order to assess the best performing algorithm at airborne level. Look Up Tables based on MODTRAN 5 RT were used to account for the atmospheric impact. Airborne-based Fs estimations at 760 nm were then compared to field measurements collected with a high resolution Ocean Optics HR4000 spectrometer (full width at half maximum of 0.1nm in the range 700-800 nm).
Results show agreement between Fs retrieved from airborne hyperspectral imagery and ground measurements with maximum values at midday and minimum values in the morning. The accuracy of the retrieval from airborne data is within the 10% of accuracy.