Retrieval of maize canopy fluorescence and reflectance by spectral fitting in the O2–A absorption band

Abstract

Canopy level chlorophyll fluorescence and reflectance of maize were retrieved simultaneously by using spectral fitting (SF) techniques applied to canopy and reference upwelling radiances measured on the ground in the O2–A atmospheric absorption band by means of a ground-measurements-based (GMB) method, using a white reference panel. This method was inspired by the Fluorescence Experiment (FLEX) mission concept, which is expected to provide the user community with a top-of-canopy radiance product, as well as sufficient data on atmospheric conditions to enable the simulation of a white reference panel radiance, after which the ground-based method can also be applied by the users of FLEX data. For the retrieval, a coupled surface–atmosphere radiative transfer model was also used to simulate the canopy radiance in specific atmospheric conditions and to quantify fluorescence and reflectance variables by using a second method based on the canopy radiance simulation (CRS), which uses the canopy radiance measurements only. The CRS method does not require any cross calibration of reference measurements, and is extremely useful when a reliable reference cannot be found. Part of the mathematical functions that modeled reflectance and fluorescence were recently used by the authors to perform simulations of observations from space. Simulations of the retrievals for both methods were performed at two different spectral band widths of 9nm and 20nm to evaluate the accuracy limits for a signal to noise ratio equal to 300:1. These simulations demonstrated an enhanced accuracy as compared to previously reported retrievals on the ground, and indicated that the CRS model can indeed be successfully applied for the retrieval of fluorescence. In the retrievals from measurements, the two intervals were compared to better evaluate the combined influence of the atmospheric conditions and forward modeling spectral accuracy on the CRS method. The 20nm interval was also used to evaluate the possibility of retrieving the bi-directional and hemispherical–directional reflectances in the viewing direction of the canopy and surroundings. Lastly, the narrower 9nm interval delivered the most accurate simulations and was chosen for comparing the retrievals obtained by means of the two different methods. From this comparison fluorescence retrieved by means of the CRS method resulted higher (about 5%) than that retrieved with the GMB method by means of the same mathematical functions, while the retrieved reflectances were very similar. The methods presented here demonstrate that fluorescence can be retrieved even when atmospheric and surface information is limited.