Segmentation-based multi-pixel cloud optical thickness retrieval using a convolutional neural network

Nataraja, Vikas; Schmidt, Sebastian; Chen, Hong; Yamaguchi, Takanobu; Kazil, Jan; Feingold, Graham; Wolf, Kevin; Iwabuchi, Hironobu

doi:https://doi.org/10.5194/amt-15-5181-2022

Articles | Volume 15, issue 17

https://doi.org/10.5194/amt-15-5181-2022

© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

Cloud, Aerosol and Monsoon Processes Philippines Experiment...

https://doi.org/10.5194/amt-15-5181-2022

© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 15, issue 17

Research article

|

14 Sep 2022

Research article |

| 14 Sep 2022

Segmentation-based multi-pixel cloud optical thickness retrieval using a convolutional neural network

Vikas Nataraja, Sebastian Schmidt, Hong Chen, Takanobu Yamaguchi, Jan Kazil, Graham Feingold, Kevin Wolf, and Hironobu Iwabuchi

Data sets

Radiance and Cloud Optical Thickness from Large Eddy Simulations over the Sulu Sea V. Nataraja, S.Schmidt, H. Chen, and T. Yamaguchi https://doi.org/10.5281/zenodo.7008103

Model code and software

Cloud-Retrieval-CNN: v2.0.0 V. Nataraja https://doi.org/10.5281/zenodo.7055057

Trained Model Weights for Retrieval of Cloud Optical Thickness using Radiance V. Nataraja https://doi.org/10.5281/zenodo.7013101

Short summary

A convolutional neural network (CNN) is introduced to retrieve cloud optical thickness (COT) from passive cloud imagery. The CNN, trained on large eddy simulations from the Sulu Sea, learns from spatial information at multiple scales to reduce cloud inhomogeneity effects. By considering the spatial context of a pixel, the CNN outperforms the traditional independent pixel approximation (IPA) across several cloud morphology metrics.