Principal Investigator: Jianwu Wang (University of Maryland, Baltimore County)
This project develops an extensible platform that combines collocated satellite observations, 3D radiative transfer simulations, deep learning methods, and cloud computing to generate training datasets and deep learning models for global cloud property retrieval. By working with Visible Infrared Imaging Radiometer Suite (VIIRS) data from the polar-orbiting Suomi National Polar-orbiting Satellite (Suomi NPP) and Advanced Baseline Imager (ABI) data from the Geostationary Operational Environmental Satellite-16 (GOES-16), this project explores a generic framework useful for cloud property retrieval of different satellites. The work will help develop and benchmark different algorithms for satellite cloud remote sensing.
Project Objectives
- Generate high-quality cloud bulk physics property retrievals, including cloud mask and cloud phase, via deep learning models that allow for joint representation learning from multi-sensor heterogeneous data
- Generate realistic cloud microphysics/optical property retrievals, including Cloud Optical Thickness (COT) and Cloud Effective Radius (CER), via 3D radiative transfer simulations and deep learning models
- Develop cloud computing based Big Data processing and analytics services for scalable cloud retrieval algorithms in the cloud
Update
Clouds cover about two-thirds of Earth’s surface and play a critical role in our climate system, and have a fundamental influence on the energy, water, and biological cycles. Currently, satellite-based remote sensing is the only way to observe clouds on a global scale. For these reasons, cloud observations have always been a major task of NASA’s Earth Science endeavor.
In the latest NASA Decadal Survey, cloud observations have been given top priority for NASA’s missions. Numerous satellite sensors have been developed to observe and retrieve cloud properties. They can be largely divided into two groups: active sensors such as those used by the Cloud Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) and CloudSat missions, and passive sensors such as the Moderate Resolution Imaging Spectroradiometer (MODIS), the Visible Infrared Imaging Radiometer Suite (VIIRS), and Advanced Baseline Imager (ABI). The advantages of active sensors include their capability of resolving the vertical location of cloud layers and better performance during nighttime and over polar regions. On the other hand, passive sensors have a much better spatial sampling rate.
Machine learning (ML)-based algorithms have brought revolutionary changes to almost every aspect of our lives. ML is also increasingly used in NASA’s satellite remote sensing algorithms. For most machine learning, especially deep learning (DL)-based algorithms, high-quality training datasets are critical.
In this project, we propose to generate highly accurate labels for passive satellite remote sensing of cloud properties by combining the following advances/opportunities: 1) complementary retrievals collected from collocated active and passive sensors, 2) advances in 3D RT simulation and DL methods that mitigate the biases and uncertainties in 1D RT assumptions and enable the multi-pixel “3D cloud property retrieval”, 3) advances in deep learning to enable knowledge transferring among multiple domains/sensors and learning from simulation results to solve the so called “inverse problem”, 4) advances in Big Data and cloud computing techniques to enable efficient processing and analysis of the huge, and growing, archive of Earth observations.
Major Accomplishments
- Scalable satellite collocation data and toolkit. Multiple collocation data, including CALIPSO-VIIRS and CALIPSO-ABI, have been generated. We have received its approval for the New Technology Report (NTR) and the Software Release Request (SRS) with case number GSC-18900-1. The corresponding toolkit at Github is now open-source
- 3D radiative transfer (RT) simulation data. We have generated multiple 3D simulation data for synthetic cloud fields, such as fractal clouds and Large-Eddy-Simulation (LES) clouds
- Deep learning models for cloud property retrieval. Two major deep learning models have been developed for cloud property retrieval. Experiments show our models outperform existing physics and deep learning approaches. We have received approvals for the New Technology Report (NTR) with case numbers GSC-19020-1 and GSC-19108-1. Detailed information can be found in our papers below and the source codes will be open-sourced later at Github: AI-4-atmosphere-remote-sensing
For More Information
Machine Learning for Cloud Remote Sensing
Publications and Presentations
Seraj A., Mostafa, M., Wang, J., Holt, B., Purushotham, S., & Wang, J. (2023). CNN Based Ocean Eddy Detection Using Cloud Services. Accepted by the International Geoscience and Remote Sensing Symposium (IGARSS 2023), 2023.
Wang, X., Guo, P., Li, X., Wang, J., Gangopadhyay, A., Busart, C.E., & Freeman, J. (2023). Reproducible and Portable Big Data Analytics in the Cloud. Accepted by the IEEE Transactions on Cloud Computing. doi:10.1109/TCC.2023.3245081
Huang, X., Wang, C., Purushotham, S., & Wang, J. (2022). VDAM: VAE based Domain Adaptation for Cloud Property Retrieval from Multi-satellite Data. Proceedings of the ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems 2022 (ACM SIGSPATIAL 2022), Article No.: 107, pages 1–10. doi:10.1145/3557915.3561044
Sun, Z., Sandoval, L., Crystal-Ornelas, R., Mousavi, S.M.., Wang, J., Lin, C., Cristea, N., Tong, D., Carande, W.H., Ma, X., Rao, Y., Bednar, J.A., Tan, A., Wang, J., Purushotham, S., Gill, T.E., Chastang, J., Howard, D., Holt, B., Gangodagamage, C., Zhao, P., Rivas, P., Chester, Z., Orduz, J., & John, A. (2022). A Review of Earth Artificial Intelligence, Computers & Geosciences, 159, 105034. doi:10.1016/j.cageo.2022.105034
Huang, X., Ali, S., Wang, C., Ning, Z., Purushotham, S., Wang, J., & Zhang, Z. (2020). Deep Domain Adaptation based Cloud Type Detection using Active and Passive Satellite Data. Proceedings of the 2020 IEEE International Conference on Big Data (BigData 2020), pages 1330-1337. doi:10.1109/BigData50022.2020.9377756