Improving the CERES SYN cloud and flux products by identifying GOES-17 scan anomalies using a convolutional neural network

Benjamin Scarino; David R. Doelling; Konstantin Khlopenkov; William L. Smith; Michele Nordeen

doi:10.1117/12.2594637

1 August 2021 Improving the CERES SYN cloud and flux products by identifying GOES-17 scan anomalies using a convolutional neural network

Benjamin Scarino, David R. Doelling, Konstantin Khlopenkov, William L. Smith, Michele Nordeen

Author Affiliations +

Proceedings Volume 11829, Earth Observing Systems XXVI; 1182904 (2021) https://doi.org/10.1117/12.2594637
Event: SPIE Optical Engineering + Applications, 2021, San Diego, California, United States

Abstract

The NASA Clouds and the Earth’s Radiant Energy System (CERES) project relies on top-of-atmosphere (TOA) broadband fluxes derived from geostationary (GEO) satellite imagery to account for the diurnal flux variations between the CERES observation intervals, and thereby produce a synoptic gridded (SYN1deg) product based on continuous temporal observations. Consistent broadband flux derivation depends on accurate radiative property measurements and cloud retrievals, which largely determine the radiance-to-flux conversion process. Therefore, it is important to ensure a high quality of cloud property input in order to maintain a reliable broadband flux record. In Edition 4 of the CERES SYN1deg product, a robust automated image anomaly detection algorithm based on inter-line and inter-pixel differences, spatial variance, and 2-D Fourier analysis has been successful in identifying imagery with linear artifacts, but the line-by-line inspection and cleaning process must still be performed by a human. Therefore, further automation of this quality assurance process is warranted, especially considering the excessive amount of additional cleaning necessitated by the GOES-17 Advance Baseline Imager (ABI) cooling system anomaly. As such, this article highlights advancement of the CERES GEO image artifact cleaning approach based on a convolutional neural network (CNN) for classification of bad scanlines. Once trained, the CNN approach is a computationally inexpensive means to ensure greater consistency in cloud retrievals, and therefore broadband flux derivation, based on GOES-17 measurements.

Conference Presentation

Citation Download Citation

Benjamin Scarino, David R. Doelling, Konstantin Khlopenkov, William L. Smith, and Michele Nordeen "Improving the CERES SYN cloud and flux products by identifying GOES-17 scan anomalies using a convolutional neural network", Proc. SPIE 11829, Earth Observing Systems XXVI, 1182904 (1 August 2021); https://doi.org/10.1117/12.2594637

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available