TY  - JOUR
AU  - Morata, Miguel
AU  - Siegmann, Bastian
AU  - Morcillo-Pallarés, Pablo
AU  - Rivera-Caicedo, Juan Pablo
AU  - Verrelst, Jochem
TI  - Emulation of Sun-Induced Fluorescence from Radiance Data Recorded by the HyPlant Airborne Imaging Spectrometer
JO  - Remote sensing
VL  - 13
IS  - 21
SN  - 2072-4292
CY  - Basel
PB  - MDPI
M1  - FZJ-2021-04065
SP  - 4368 -
PY  - 2021
AB  - The retrieval of sun-induced fluorescence (SIF) from hyperspectral radiance data grewto maturity with research activities around the FLuorescence EXplorer satellite mission FLEX, yetfull-spectrum estimation methods such as the spectral fitting method (SFM) are computationallyexpensive. To bypass this computational load, this work aims to approximate the SFM-based SIFretrieval by means of statistical learning, i.e., emulation. While emulators emerged as fast surrogatemodels of simulators, the accuracy-speedup trade-offs are still to be analyzed when the emulationconcept is applied to experimental data. We evaluated the possibility of approximating the SFM-likeSIF output directly based on radiance data while minimizing the loss in precision as opposed to SFMbasedSIF. To do so, we implemented a double principal component analysis (PCA) dimensionalityreduction, i.e., in both input and output, to achieve emulation of multispectral SIF output basedon hyperspectral radiance data. We then evaluated systematically: (1) multiple machine learningregression algorithms, (2) number of principal components, (3) number of training samples, and(4) quality of training samples. The best performing SIF emulator was then applied to a HyPlantflight line containing at sensor radiance information, and the results were compared to the SFM SIFmap of the same flight line. The emulated SIF map was quasi-instantaneously generated, and a goodagreement against the reference SFM map was obtained with a R2 of 0.88 and NRMSE of 3.77%.The SIF emulator was subsequently applied to 7 HyPlant flight lines to evaluate its robustness andportability, leading to a R2 between 0.68 and 0.95, and a NRMSE between 6.42% and 4.13%. EmulatedSIF maps proved to be consistent while processing time was in the order of 3 min. In comparison, theoriginal SFM needed approximately 78 min to complete the SIF processing. Our results suggest thatemulation can be used to efficiently reduce computational loads of SIF retrieval methods.
LB  - PUB:(DE-HGF)16
UR  - <Go to ISI:>//WOS:000718576000001
DO  - DOI:10.3390/rs13214368
UR  - https://juser.fz-juelich.de/record/902156
ER  -