Abstract
This study presents a deep learning model, using the conditional generative adversarial nets (CGAN) technique, that can produce daytime visible (VIS) band information, mimicking a narrow band sensor, by combining VIS and infrared (IR) broadband measurements by different sensors. The real-observed datasets of the Geostationary Ocean Color Imager (GOCI) and meteoritical imager (MI) sensors onboard the Communication, Ocean, and Meteorological Satellite were used for training and testing our CGAN model over the Yellow Sea and Bohai Sea. The trained and tested CGAN model was then applied to generate daytime GOCI VIS and near IR (NIR) bands (0.412 to 0.865 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">μ</i> m) using daytime MI VIS (0.675 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">μ</i> m), shortwave IR (3.75 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">μ</i> m), and longwave IR bands (10.8 and 12.0 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">μ</i> m) and the differences between them as input data. GOCI and MI data were collected from January 2017 to December 2018 using 705 images of 256 × 256 pixels for the training and 44 images for the model test. The results are statistically favorable (i.e., bias = −0.013 (in a reflectance unit from 0 to 1), root-mean-square error = 0.112, mean absolute error = 0.076, agreement index = 0.945, and correlation coefficient (CC) = 0.809 for daytime reflectance in the GOCI VIS 0.49- <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">μ</i> m band) between the real GOCI VIS band observation and the CGAN-generated simulation. Our CGAN-based model showed high CC and favorable results in the GOCI VIS and NIR bands. Consequently, our study demonstrates the possibility of applying a deep learning technique to improve the temporal resolution for ocean color studies using the GOCI sensor.
Highlights
Geostationary Ocean Color Imager (GOCI) near infrared (NIR) bands such as those at 0.745 μm and 0.865 μm, showed the lowest CC and the highest root mean square error (RMSE) compared to the other GOCI VIS bands
Various ocean color sensors covering the optical spectrum onboard satellites such as MODIS, Sea-viewing Wide Field-of-view Sensor (SeaWiFS), Mediumresolution Imaging Spectrometer (MERlS), Ocean and Land Color Imager (OLCI), and GOCI, have been providing the information about the short-term and long-term variability of the oceans and constituents such as chlorophyll concentration, chromophoric dissolved organic matter (CDOM), sea ice, sea surface temperature, sediments, turbidity, currents, and quasi RGB images, Rossby waves, mesoscale eddies, storminduced effects, El Ninõ–La Ninã, net primary production (NPP), particulate organic carbon (POC), red tides, and harmful algal blooms (HABs)
Irrespective of the numerous advantages gained by using ocean color sensors, they have the physical disadvantage of depending on the presence of sunlight
Summary
O CEAN color sensors are used in a wide range of applications such as collecting data to produce information about the chlorophyll (CHL) concentration, chromophoric dissolved organic matter (CDOM), sea ice, sea surface temperature, sediments, turbidity, currents, and quasi RedGreen-Blue (RGB) images [1,2,3,4,5] as well as mesoscale open ocean processes such as Rossby waves [6], mesoscale eddies [7], storm-induced effects [8, 9], El Ninõ–La Ninã [10], net primary production (NPP) [11, 12], particulate organic carbon (POC) [13], red tides, and harmful algal blooms (HABs) [14, 15].The various ocean color sensors that are onboard satellites play an important role in measuring physical oceanographic changes and the ocean ecosystem on a global scale [12, 16,17,18,19], serving to provide knowledge concerning both the short-term and long-term variability of the oceans and their constituents.Generally, instruments used for ocean color measurement are designed to be utilized primarily within the optical spectrum in the visible (VIS) and near infrared (NIR) range of 0.38 to 0.8 μm, which is similar to the 0.4 to 0.7 μm observed by the human eye, ranging from violet to red [20]. Other ocean color sensors [22,23,24,25] include the Moderate-resolution Imaging Spectroradio-meter (MODIS) onboard the Aqua and Terra satellites, the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) onboard the Seastar satellite, the Mediumresolution Imaging Spectrometer (MERlS) onboard the Environmental Satellite Envisat satellite, and the Ocean and Land Color Imager (OLCI) onboard the Sentinel-3, which continues the heritage of MERIS with six additional bands [26] These sensors have been developed for use on low Earth orbiting satellites and are utilized to observe the global waterleaving spectral radiance of the oceans in multispectral VIS and NIR bands over time periods of approximately two to three days [27].
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
