We sat down with Sundararaman “Gopal” Gopalakrishnan, Ph.D., the senior meteorologist and leader of the modeling team that developed NOAA’s newest hurricane forecast model – the Hurricane Analysis and Forecast System – that is expected to go into full operation at NOAA’s National Weather Service in late June. Gopalakrishnan leads the Hurricane Modeling and Prediction Program at NOAA’s Atlantic Oceanographic & Meteorological Laboratory, (AOML) Hurricane Research Division. He is also the development manager for NOAA’s Hurricane Forecast Improvement Program (HFIP)
1. What are the major ways this forecast model differs from the hurricane forecast models NOAA has used previously?
The Hurricane Analysis and Forecast System brings together the best of our existing hurricane forecast models and adds key research advancements to create more accurate, higher-resolution forecast information both over land and ocean that can save lives and protect property. The foundational component for this model is its moving nest, which allows the model to zoom in on hurricanes across the Atlantic and Pacific basins. In the future, the model will also allow forecasters to track multiple hurricanes at once, which has been shown to improve the accuracy of our forecasts of track and intensity, or wind speeds.
The moving nest is like a high-definition TV that allows us to zoom into areas of a hurricane such as the eyewall and bands of intense rain. With a resolution down to 1.2 miles or 2 kilometers in a model with a general resolution of 7 miles or 12 km, we can better predict wind speeds and precipitation amounts. The moving nest was developed at AOML in coordination with NOAA’s Geophysical Fluid Dynamics Laboratory (GFDL) FV3 modeling group.
The the Hurricane Analysis and Forecast System also has improved physics gleaned from high-resolution radar and flight level observations collected by AOML scientists during flights by NOAA Hurricane Hunter Aircraft P-3 in recent NOAA hurricane field campaigns. The data have enabled us to improve our understanding of the structure of hurricanes and their representation in the new model.
In the 1990s, we focused on improving the track of hurricanes. Since the early 2000s, we have also focused on improving predictions of intensity, or hurricane winds. This need was heightened by Hurricanes Katrina and Rita in 2005. The public demand continues for more accurate information about winds, rainfall, tornadic threats over land and storm-surge, demonstrated by hurricanes such as Michael (2018), Laura (2020) and Ian (2022).
2. What observations are used in the model? Are some of the newer observations from uncrewed systems being used in the model?
The new model uses satellite observations, radar data, NOAA Hurricane Hunter Aircraft and U.S. Air Force Reserve flight-level data of winds, moisture, temperature and Global Positioning System radio occultation observations, a remote sensing technique to measure the atmosphere. The AOML team has also used invaluable turbulence measurements from Hurricane Hunter flights to improve the model’s physics. There are still gaps in the model, especially in our understanding of the area where the ocean and atmosphere meet, a key area called the boundary layer, where hurricanes gain and lose strength through an exchange of heat and energy between the ocean and atmosphere. New observations from uncrewed ocean and aircraft systems will definitely help in the future to improve our understanding of this area and of our model.
3. Tell us about how the new model was created? Who worked on this?
The model is a perfect example of teamwork. It was jointly created by AOML’s Hurricane Modeling and Prediction Program and NOAA’s National Weather Service Environmental Modeling Center (EMC) in collaboration with GFDL and NOAA’s Cooperative Institute for Marine & Atmospheric Studies at the University of Miami. A major component of the physics used to represent the boundary layer relies on observations from NOAA Hurricane Hunter P-3 aircraft. The testing and evaluation was jointly done by EMC and AOML. HAFS was created to aid forecasters with better tools and model products for providing guidance.Consequently, the forecasters at National Hurricane Center were involved from the developmental stages of the model all the way to its evaluation and implementation. HAFS, which is part of the Unified Forecast System, is also a great example of community-based collaboration on model development and the streamlining of the operational transition process.
4. What future improvements to the model are you planning that will help emergency managers and the public?
We are at the starting point of the next generation of hurricane forecast modeling.The initial operational capability is expected to replace the Hurricane Weather Research and Forecast System (HWRF) and the Hurricane in a Multi-scale Ocean-coupled Non-hydrostatic Model (HMON). Running the experimental version of HAFS from 2019 to 2022 in near real time, we have already seen a 10-15% improvement in track predictions compared to the best hurricane model today, HWRF. This season, these two older models will also run in parallel with HAFS as we complete the full transition.
NOAA plans to implement the basin-scale HAFS in 2025 and 2026, which is expected to improve prediction of interactions between several tropical cyclones as well as the prediction of how storms behave once they make landfall. This will aid forecasters at NHC with improved products of winds, rainfall and tornado threats inland. NOAA’s Hurricane Forecast Improvement Program is also supporting the development of the HAFS ensemble system with a focus on incorporating risk communication research to create more effective watch and warning products in future.
To continue this work we need to retain our motivated and talented workforce as well as to bring in new skilled researchers to ensure we have a diverse workforce. We also need more high performance computing power to help us accelerate the progress we are making.
For more information, please contact Monica Allen, director of public affairs for NOAA Research, at email@example.com or 202-379-6693.