What does this research mean for the field?

A neural network framework trained on mixed-pixel physical simulations can accurately retrieve far-red sun-induced chlorophyll fluorescence from GOME-2 satellite data, producing measurements consistent with TROPOMI. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

This research aims to develop a neural network framework for accurately retrieving sun-induced chlorophyll fluorescence from GOME-2 satellite data.

May 31, 2026Open Access

Global retrieval of GOME-2 sun-induced chlorophyll fluorescence via neural networks trained on mixed-pixel simulations

Key Points

This research aims to develop a neural network framework for accurately retrieving sun-induced chlorophyll fluorescence from GOME-2 satellite data.
Developed a simulation-trained neural network framework for far-red SIF extraction.
Utilized mixed-pixel simulations combining SCOPE and MODTRAN for informed model training.
Compared gridded neural network-derived SIF across two GOME-2 sensors and TROPOMI.
Achieved consistent far-red SIF retrieval across two GOME-2 sensors.
Demonstrated strong correlation between neural network outputs and TROPOMI measurements.
Validated model performance with simulations, indicating robustness in mixed-pixel scenarios.

Abstract

• A simulation-trained neural network framework retrieved far-red SIF from GOME-2. • Mixed-pixel SCOPE + MODTRAN simulations supported physically informed model training. • Gridded NN-SIF was consistent across two GOME-2 sensors and with TROPOMI.

Global retrieval of GOME-2 sun-induced chlorophyll fluorescence via neural networks trained on mixed-pixel simulations

Key Points

Abstract

Cite This Study