What happened
NOAA's Hurricane Research Division developed the Statistical Hurricane Intensity Prediction Scheme (SHIPS), which first ran in real time in 1990. Its developers write that it uses a linear regression technique with predictors from climatology, persistence, the atmosphere and the ocean. In a technology assessment dated 14 December 2023, the U.S. Government Accountability Office (GAO) states that SHIPS outperformed most operational hurricane intensity models through the mid-2010s. GAO reports that in 2021 it was the best of 48 models evaluated by the National Hurricane Center (NHC) at longer lead times for Pacific hurricanes. GAO writes that SHIPS uses a simple form of machine learning to predict rapid intensification. GAO did not verify the forecasts. It relays the results of NHC, which uses SHIPS. NHC's report of 25 April 2022 states that SHIPS with inland decay was the best individual model at the longer forecast periods in the eastern North Pacific. Its comparison table covers six individual models and four consensus aids. The ranking covers one season in one basin. No source measures how SHIPS changed NHC's official forecasts or warnings.
Sources
-
Report, by the organization that used it.
Cangialosi, J. P. National Hurricane Center Forecast Verification Report: 2021 Hurricane Season. NOAA National Hurricane Center, 25 April 2022, p. 13.
Among the individual models, HMNI was the best performer for the short lead times while DSHP was best at the longer forecast periods.
-
Descriptive report, by authors with no tie to the system.
U.S. Government Accountability Office. Artificial Intelligence in Natural Hazard Modeling: Severe Storms, Hurricanes, Floods, and Wildfires. Technology assessment GAO-24-106213, 14 December 2023, p. 25.
SHIPS has historically outperformed most operational hurricane intensity models through the mid-2010s and still provide skillful intensity guidance. In 2021, it was the best of 48 models evaluated by the National Hurricane Center at providing intensity guidance for hurricanes in the Pacific at longer lead times.
-
Descriptive report, by the makers. Context or contrary evidence.
Regional and Mesoscale Meteorology Branch (NOAA/NESDIS) and Cooperative Institute for Research in the Atmosphere, Colorado State University. Statistical Tropical Cyclone Intensity Forecast Technique Development. Web page, undated, checked 1 November 2026.
The SHIPS and STIPS models use a linear regression technique, with the impacts of land applied as a correction in a post-processing step.
Details
AI system
- Workflow
-
SHIPS combined predictors from climatology, persistence, the atmosphere and the ocean to forecast changes in maximum sustained winds. The National Hurricane Center used SHIPS with other models to help forecasters predict hurricane intensity. NHC compared the intensity models in a homogeneous verification of the 2021 eastern North Pacific season.
Follow-up
- Conditions for replication
-
Atmospheric forecast models must supply the predictors. Storms must not intensify rapidly, since SHIPS forecasts have not performed well in those cases.
Entry details
- Sector
-
Weather forecasting / tropical cyclones.
- Purpose
-
NOAA's Hurricane Research Division built SHIPS to give the National Hurricane Center objective guidance for predicting the intensity of tropical cyclones.
31 more questions not yet checked
model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). SHIPS gave the best long-range Pacific intensity guidance in 2021 (AIBD-1128). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-1128.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)