The exact text we saved

This is the version exactly as it was saved, character for character. The SHA-256 hash below is a 64-character code worked out from that text. Run the same text through SHA-256 and you get the same code; change a single letter and the code comes out completely different. Each version also stores the previous version’s hash, which chains them together so no version can be swapped out quietly.

TRUVACE RECORD VERSION
record: TRV-2026-2618
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-11-12T08:04:36.330181Z
status: published
lens: g_space
sector: climate
headline: Skillful regional NWP forecasts of tropical cyclones driven by fully AI-generated boundary conditions
dek: Abstract Forecasting tropical cyclones (TCs) remains one of the most enduring challenges in numerical weather prediction (NWP), particularly in terms of storm intensity and track at medium to extended lead times. Recent advances in artificial intelligence (AI)-based global models have demonstrated strong skill in capturing large-scale atmospheric dynamics; however, their coarse spatial resolution and limited variable suites constrain their ability to produce storm-resolving forecasts critical for high-impact eve…
gain_title: In a study of 14 Northwest Pacific typhoons from 2024, an AI-driven regional system held four- to five-day track errors to about 320–340 km, versus 500–600 km for conventional regional models.
problem_title: (none)
trace_subject: (none)
gain_reading: In a study of 14 Northwest Pacific typhoons from 2024, an AI-driven regional system held four- to five-day track errors to about 320–340 km, versus 500–600 km for conventional regional models.
gain_evidence: AIFS_WRF maintains lower MAEs of approximately 320–340 km
problem_reading: (none)
problem_evidence: (none)
quick_read: Chinese researchers built a typhoon forecasting system in which the European weather centre's AI model, AIFS, supplies every starting and boundary condition for a 2-km-resolution regional weather model (WRF). Tested on 14 Northwest Pacific typhoons from 2024, it kept AIFS's strong track skill and cut intensity errors relative to AIFS alone.

At 96 to 120 hours ahead, the conventional regional systems (a WRF run driven by the ECMWF physics-based forecast, and NOAA's HAFS) had track errors above 500–600 km. The AI-driven system held errors to about 320–340 km, a 40%–50% reduction. Accurate long-range tracks matter for early warning and disaster response.

The system still struggles with rapid intensification and weakening, and intensity errors of 10–15 m/s persisted across several cases. The test covered one season and 14 storms in one basin, so how it performs in other years and regions is not shown.
limitation: Intensity errors of 10–15 m/s persisted across multiple cases, rapid intensification and weakening remain a major limitation, and the evaluation covered only 14 storms from the 2024 Northwest Pacific season.
tag: Evidence-backed gain
key_points: The system runs AIFS once from a single global analysis, then uses its output for all initial and lateral boundary conditions of a 2-km WRF model, cutting external data ingestion by nearly a factor of 20 over a 5-day forecast. | Fourteen 2024 Northwest Pacific typhoons were tested, including Gaemi, Yagi and Kong-Rey, against four configurations: AIFS, HAFS, AIFS_WRF and EC_WRF. | Between 96 and 120 hours, EC_WRF and HAFS track errors exceeded 500–600 km while the AI-driven system stayed near 320–340 km. | Intensity errors were substantially lower than AIFS alone across all lead times, especially at short ranges, and error growth was slower than in EC_WRF at longer lead times. | In 2024, AIFS cut 120-hour track error to about 300 km from about 400 km for the physics-based IFS, but had higher intensity errors than the IFS at all lead times. | For Typhoon Shanshan, the AI-driven system captured intensification, peak and weakening, while HAFS and EC_WRF advanced the storm too quickly and substantially overpredicted peak intensity. | Intensity errors of 10–15 m/s persisted across multiple cases, partly because AIFS provides only 13 pressure levels with a model top near 50 hPa.
rundown: The team targeted a known trade-off: AI global models such as AIFS predict storm tracks well but at roughly 31 km resolution they cannot resolve eyewalls and inner-core structure, so intensity suffers. Running AIFS locally and nesting a 2-km WRF model inside it, with a dynamically balanced synthetic vortex built from best-track data and spectral nudging of large-scale fields, was meant to keep AIFS's track skill while recovering storm-scale detail.

Spectral analysis of Typhoon Yagi forecasts showed the nested model kept substantially more wind energy at 50–200 km scales than AIFS, which lost fine-scale variance even relative to its starting conditions. Verification used mean absolute errors against China Meteorological Administration best-track data at 12, 24, 48, 72, 96 and 120 hours, with all WRF runs sharing identical physics to isolate the effect of the boundary forcing.

In case studies, the AI-driven system avoided the premature northward deflection EC_WRF showed for Typhoon Gaemi and the late-stage divergence in HAFS. For Kong-Rey it reproduced the observed recurvature and avoided the excessive intensity bias of the regional baselines. The authors say the setup could suit forecasting centres with limited access to conventional global model output.

The authors list remaining work: higher vertical resolution in AI models, ensemble-based probabilistic forecasts and better storm-core observations for training and validation. Rapid intensification and weakening remain a major limitation of the system.
sources:
- peer_reviewed | Environmental Research Letters | https://doi.org/10.1088/1748-9326/ae7e0a | 2026-06-16
prev: 0000000000000000000000000000000000000000000000000000000000000000

SHA-256 hash

bd1ca4bd875ea150fa1f0166e82b9222dc194dec600ff2aab3d761440abdc5d1

Previous version’s hash

0000000000000000000000000000000000000000000000000000000000000000