Skip to Navigation Skip to content

Regional and Mesoscale Meteorology Branch

Search the RAMMB website

Section Navigation - Skip to Section Content

Artificial Intelligence & Machine Learning Research

RAMMB is using Artificial Intelligence & Machine Learning across its broad portfolio of research projects and product development. AI techniques are proving to be a critical tool in connecting and adding value to NOAA’s data repositories from observations to models. In this research, satellite observations, radar, and model data are being combined to create new AI-ready datasets and products to aid and support NOAA’s mission.

The Branch has a long history of creating analysis-ready datasets for data fusion applications and is applying that expertise to create AI-ready datasets. Recently, the Branch worked with CIRA and the CSU Department of Atmospheric Science to create the Tropical Cyclone PRecipitation, Infrared, Microwave, and Environmental Dataset (TC PRIMED). TC PRIMED is a comprehensive 22-year (1998 to 2019) dataset of global tropical cyclone satellite observations with ancillary labeled information to aid with machine learning applications. For more information about TC PRIMED, see the TC PRIMED project page.

The RAMMB is applying a heretical approach to AI product development. Work continues to leverage long history in classic machine learning techniques through continued development and maintenance of tropical cyclone statistical–dynamical intensity aids for the National Hurricane Center, Central Pacific Hurricane Center, and Joint Typhoon Warning Center. Intermediate algorithms are being used for regression and classification problems for our partners ranging from the aviation weather community for icing and cloud classification to the modeling community through generating three-dimensional wind fields from satellite imagery for data assimilation. Deep-learning techniques are being applied to detect convection and to generate synthetic radar and passive microwave products using Advanced Baseline Imager (ABI) and Geostationary Lightning Mapper (GLM) observations from the Geostationary Operational Environmental Satellite-R (GOES-R) series of NOAA geostationary weather satellites.

In each area, explainable and trustworthy AI is at the forefront of how the Branch is developing and applying AI to problems critical to NOAA mission areas. Work on exploring explainable and trustworthy AI techniques is being done in partnership with the NOAA Center for AI and the Colorado State University researchers involved with the NSF AI Institute for Research on Trustworthy AI in Weather, Climate, and Coastal Oceanography (AI2ES). Explainable AI ensures that products are not learning spurious relationships, but relationships meaningful to geophysical problems. This step ensures that our products are trustworthy after being transitioned to operations.

We would like to thank our current partners NOAA, NSF AI2ES, US Navy (ONR & Naval Research Laboratory, Monterey):

Page Contact

Chris Slocum