Daytime-like nighttime aerosol optical depth detection for geostationary environment monitoring spectrometer

초록

Aerosol Optical Depth (AOD) is a critical parameter for understanding the Earth's atmospheric energy balance and addressing key health-related concerns. This study presents an innovative approach for simulating virtual nighttime AOD based on daytime observations from the Geostationary Environment Monitoring Spectrometer (GEMS) with ultraviolet and visible bands onboard the GEO-KOMPSAT (GK)-2B satellite. This study set the spatial domain to 20°N–43°N and 117°E–132°E, primarily to encompass South Korea and the Yellow Sea. By employing a deep learning-based data-to-data (D2D) translation technique, we trained and tested the D2D model to learn the GEMS AOD values from infrared (IR) bands at the Advanced Meteorological Imager (AMI) onboard the co-located GK-2A satellite. We used the paired daytime GEMS AOD datasets and ten IR brightness temperature (BT) and BT difference (BTD) measurements at the GK-2A AMI sensor. The trained D2D model was subsequently applied to nighttime AMI BT and BTD data to simulate daytime-like nighttime GEMS AOD values, assuming consistency in IR datasets without reliance on solar reflection. Our proposed nighttime AOD generation model was quantitatively evaluated using daytime GEMS AOD as a reference. The results showed excellent performance, with a probability of detection of 0.888, a false alarm ratio of 0.159, a critical success index of 0.760, a Heidke skill score of 0.778, and a proportion correct of 0.895 for the case on February 25, 2022, at 05:45 UTC (14:45 Korea Standard Time). Validation of the D2D-generated nighttime GEMS AOD data, using Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) aerosol layer and Vertical Feature Mask data, as well as Aerosol Robotic Network (AERONET) observations, demonstrated excellent agreement with CALIPSO aerosol data and closely matched AOD values from AERONET. This study provides a valuable contribution to continuous AOD monitoring, offering the ability to observe AOD both day and night at 10-min intervals over East Asia, supporting atmospheric and environmental forecasting efforts. © 2024

키워드