Machine learning to infer and improve cloud properties from satellite data
Estimating cloud structures is crucial for supporting weather/climate research and aviation operations. Although satellites provide valuable global cloud measurements, accurately characterizing clouds from space has limitations. Particularly for 3D cloud information, assigning cloud base heights based on conventional satellite sensors is challenging due to the inherent bias of passive sensor measurements towards cloud top. To address this, we developed a statistical approach to estimate cloud geometric thickness and base height using NASA A-Train satellite data. This algorithm is currently operational as part of the NOAA Enterprise Cloud Algorithm Suite and is used for an improved version of Cloud Cover Layers. We are exploring various ML/AI approaches to improve algorithm performance for multi-layer cloud scenes and nighttime cloud retrievals, as well as to develop vertical cloud water profile estimation.
Leveraging CIRA’s cloud research, we are expanding our efforts to build a novel 3D cloud structure data set, rendered on a global scale. This work is part of the OVERCAST (Optical Variability Evaluation of Regional Cloud Asymmetries in Space and Time) project sponsored by the Office of Naval Research (ONR). We are investigating new blending methods and an advanced dense optical flow method to achieve a harmonized global cloud field in space and time. The use of ML/AI also plays an important role in exploring the potential of introducing a nowcasting component to advect and evolve the 3D cloud field forward in time. Research on AI-based global synthetic radar estimation and proxy nighttime cloud imagery is also ongoing. This work will be applied to multiple global satellites suitable in support of ONR and Navy operations, and will also contribute to the development activities of the International Satellite Cloud Climatology Project-Next Generation.
Noh, Y.J., Haynes, J.M., Miller, S.D., Seaman, C.J., Heidinger, A.K., Weinrich, J., Kulie, M.S., Niznik, M., and Daub, B.J. A Framework for Satellite-based 3D Cloud Data: An Overview of VIIRS Cloud Base Height Retrieval and User Engagement for Aviation Applications. Remote Sens., 14(21), 5524, Special Issue “VIIRS 2011–2021: Ten Years of Success in Earth Observations”, Nov 2022, https://doi.org/10.3390/rs14215524.
Haynes, J.M., Noh, Y.J., Miller, S.D., Haynes, K.D., Ebert-Uphoff, I. and Heidinger, A., Low Cloud Detection in Multilayer Scenes using Satellite Imagery with Machine Learning Methods, Journal of Atmospheric and Oceanic Technology, 39(3), 319-334, March 2022, https://doi.org/10.1175/JTECH-D-21-0084.1.
Lee, Y., Kummerow, C. D., Ebert-Uphoff, I. Applying machine learning methods to detect convection using GOES-16 ABI data, Atmospheric Measurement Techniques, April 2021, https://doi.org/10.5194/amt-2020-420.
Ebert-Uphoff, I., Hilburn, K. A. Evaluation, Tuning and Interpretation of Neural Networks for Meteorological Applications, Bulletin of the American Meteorological Society (BAMS), https://doi.org/10.1175/BAMS-D-20-0097.1, Aug 2020.
Hilburn, K. A., Ebert-Uphoff, I., and Miller, S. D., Development and Interpretation of a Neural Network-Based Synthetic Radar Reflectivity Estimator Using GOES-R Satellite Observations. Journal of Applied Meteorology and Climatology, https://doi.org/10.1175/JAMC-D-20-0084.1, Jan 2021.