Link the day and night: A deep learning framework to retrieve global nighttime AOD from VIIRS DNB

Meng Zhou; Jun Wang; Xi Chen

doi:10.1109/LGRS.2026.3713835

Back

Link the day and night: A deep learning framework to retrieve global nighttime AOD from VIIRS DNB

Journal article

Peer reviewed

Link the day and night: A deep learning framework to retrieve global nighttime AOD from VIIRS DNB

Meng Zhou, Jun Wang and Xi Chen

IEEE geoscience and remote sensing letters

07/14/2026

DOI: 10.1109/LGRS.2026.3713835

View Online

Abstract

The unique Day/Night Band (DNB) on the Visible Infrared Imaging Radiometer Suite (VIIRS) has transformed nighttime environmental monitoring. Yet its daytime measurements, collecting concurrently with VIIRS narrowband visible and near-infrared channels (e.g., blue, red, and near infrared), remain underutilized despite their rich atmospheric information content. Here we leverage these daytime observations to train a machine-learning framework that uses the DNB as a spectral "bridge" to transfer knowledge from day to night, enabling fast, global retrievals of nighttime aerosol optical depth (AOD) over oceans and rural land. We validate year-2020 nighttime AOD against CALIOP over oceans and AERONET Lunar observations over land. The neural-network retrieval achieves Pearson correlations of 0.75 (ocean) and 0.72 (rural land), demonstrating the robustness and feasibility of this architecture. By exploiting the co-acquired daytime DNB and narrowband AOD retrieval for training and then applying the learned DNB-AOD relationship to nighttime DNB, our approach fills substantial observational gaps in nocturnal aerosol coverage, providing around-the-clock constraints on aerosol transport at global scales. This study highlights the underused value of daytime DNB paired with multispectral VIIRS bands and establishes an efficient pathway for large-scale nighttime AOD retrievals.

Clouds

Geometry

Machine Learning

Aerosols

AOD

DNB

Lighting

Modeling

Moon

Oceans

Reflectivity

retrieval

Surfaces

Training

VIIRS

Details

Title: Subtitle: Link the day and night: A deep learning framework to retrieve global nighttime AOD from VIIRS DNB
Creators: Meng Zhou - National Aeronautics and Space Administration
Jun Wang - University of Iowa
Xi Chen - University of Iowa
Resource Type: Journal article
Publication Details: IEEE geoscience and remote sensing letters
DOI: 10.1109/LGRS.2026.3713835
ISSN: 1545-598X
Publisher: IEEE
Grant note: 80NNSC21L1976 / NASA’s Terra, Aqua, and SNPP program 80NSSC21K1628 / Future Investigators in NASA Earth and Space Science and Technology (FINESST) program 80NSSC21K1494 / NASA’s Modeling and Analysis Program (MAP)
Language: English
Electronic publication date: 07/14/2026
Academic Unit: Electrical and Computer Engineering; Civil and Environmental Engineering; Iowa Technology Institute; Physics and Astronomy; Chemical and Biochemical Engineering
Record Identifier: 9985183122002771

Metrics

1 Record Views