29 May 2023 in Remote Sensing of Environment
10 September 2024 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
The successful operation of airborne and space-based spectrometers in recent years holds the promise to map solar-induced fluorescence (SIF) accurately across the globe. Machine learning (ML) can play an important role in this effort, but its application to SIF retrieval methods is in part hindered by the need for time-consuming radiative transfer modelling to account for atmospheric effects. In this work, we address this difficulty and develop a fast and accurate physics-based ML emulator of at-sensor radiances around the O $_{2}$ -A absorption band for the space-based DESIS and the airborne HyPlant spectrometers. Different ML models are trained on an extensive set of simulated spectra encompassing a wide range of atmosphere, geometry, surface and sensor configurations. A fourth-degree polynomial model is found to perform best, presenting errors at or below 10% of typical SIF at-sensor radiances and a prediction time per sample spectrum of 10-20 $\mu$ s. Using data acquired with the HyPlant instrument, the proposed model is also shown to be able to match very closely the measured spectra. We illustrate how to improve further the accuracy of the emulator and how to generalize it to other sensors using the particular case of ESA's FLEX space mission. Our findings suggest that physics-based emulators can be efficiently used for the development of ML-based SIF retrieval methods by generating large training data sets in short time and by enabling a fast simulation step for self-supervised retrieval schemes.
22 July 2024 in Optics Express
LALeonardo M. A. Alvarado
+15
This study presents what we believe is the first extensive assessment of the water reflectance products from the German hyperspectral Environmental Mapping and Analysis Program (EnMAP). We evaluate EnMAP’s standard normalized water leaving reflectance [ρ
W
]
N
over 17 water sites in the first two years of the mission. The EnMAP [ρ
W
]
N
standard product is generated by a dedicated water atmospheric correction (AC) called the Modular Inversion Program (MIP). The quality of the [ρ
W
]
N
retrievals was assessed using in situ hyperspectral measurements and Aerosol Robotic Network - Ocean Colour (AERONET-OC) multispectral measurements. The results showed very good agreement between in situ hyperspectral match-ups and EnMAP [ρ
W
]
N
, with an underestimation of EnMAP of −17.37% (bias, β) and an error (ϵ) of 23.75% at 418 – 797 nm. Two other AC processors were also investigated: the polynomial based algorithm applied to MERIS (Polymer) and the atmospheric correction for OLI lite (Acolite). The intercomparison exercise between the three AC methods applied to EnMAP data using the hyperspectral match-up dataset showed better statistical metrics for MIP (ϵ=23%,β=−17.37%) compared to Polymer (ϵ=42.20%,β=−2.43%) and Acolite (ϵ=97%,β=97%). The superior performance of MIP was further confirmed by the validation results obtained with the multispectral match-up dataset; MIP retrievals show good agreement with in situ measurements at the majority of study sites. Conversely, Polymer and Acolite retrievals tended to overestimate, especially in clearer waters as the Lampedusa study site.
30 November 2010 in The Astrophysical Journal
We report on the observation of the region around supernova remnant G65.1+0.6 with the stand-alone MAGIC-I telescope. This region hosts the two bright GeV gamma-ray sources 1FGL J1954.3+2836 and 1FGL J1958.6+2845. They are identified as GeV pulsars and both have a possible counterpart detected at about 35 TeV by the Milagro observatory. MAGIC collected 25.5 hours of good quality data, and found no significant emission in the range around 1 TeV. We therefore report differential flux upper limits, assuming the emission to be point-like (<0.1 deg) or within a radius of 0.3 deg. In the point-like scenario, the flux limits around 1 TeV are at the level of 3 % and 2 % of the Crab Nebula flux, for the two sources respectively. This implies that the Milagro emission is either extended over a much larger area than our point spread function, or it must be peaked at energies beyond 1 TeV, resulting in a photon index harder than 2.2 in the TeV band.
25 August 2025 in Remote Sensing of Environment