Tool Reviews

HiRO-ACEv2 Simulates Extreme Weather at 3 Kilometer Resolution at a Fraction of the Cost

HiRO-ACEv2 produces global 3-kilometer climate data for extreme rain, wind, and pressure at 1/300th of X-SHiELD cost, with models and demos available.

HiRO-ACEv2 Simulates Extreme Weather at 3 Kilometer Resolution at a Fraction of the Cost

Key Takeaways

  • HiRO-ACEv2 can generate global 3 km climate data for extreme rain, wind, and pressure events at about 1/300th the cost of X-SHiELD.
  • The new HiROv2 downscaling model improves extreme-event accuracy and can reproduce rare events with high-resolution statistics within 3% of X-SHiELD targets.
  • ACE and HiROv2 together can produce multi-year global high-resolution simulations whose hurricane energy and extreme-event patterns closely match X-SHiELD.
  • The system is especially useful for researchers needing many high-resolution extreme-weather datasets, but it currently only supports climates matching X-SHiELD and lacks outputs like temperature, snow, and ice.

Meta description: HiRO-ACEv2 produces global 3 km climate data for rare rain, wind, and pressure events at about 1/300th the cost of X-SHiELD, with models and demos available.

HiRO-ACEv2 Simulates Extreme Weather at 3 Kilometer Resolution at a Fraction of the Cost

Tags: HiRO-ACEv2, X-SHiELD, 3 km climate data, AI weather models, extreme weather simulation

Featured image: An abstract, friendly illustration of a coarse 100 km climate map sharpening into 3 km storm detail, with no people, no logos, and no text.

For researchers who need many versions of high-resolution rain, wind, and pressure data, HiRO-ACEv2 is the newest update to the High Resolution Output Ai2 Climate Emulator, known as HiRO-ACE. The full system can generate global 3 km climate data at about 1/300th the cost of X-SHiELD, a high-resolution global atmospheric model that explicitly simulates storms and other small-scale weather. Its new downscaling model, HiROv2, is able to reproduce even extremely rare rain, wind, and pressure events with high accuracy.

Why the update matters for extreme weather

HiRO-ACE was introduced earlier in 2026 and works in two stages. ACE is an AI model that produces a coarser 100 km view, while HiRO turns those outputs into much finer 3 km detail. The largest change in HiRO-ACEv2 is HiROv2. It can now predict several high-resolution fields at the same time rather than only rainfall, which broadens the kinds of extreme weather it can represent. It also has improved skill in predicting extreme events.

HiRO-ACEv2 Simulates Extreme Weather at 3 Kilometer Resolution at a Fraction of the Cost

The update was built around a simple goal: make extreme events come out well. New training choices improved HiROv2’s ability to reproduce the strongest rain, wind, and pressure values. When it receives coarsened X-SHiELD data, its high-resolution statistics stay within 3% of the target simulation, even for events that occur only once in a million data points.

It also estimates the total energy of tropical cyclones accurately, including storms that were not in the training set. The model can create hurricanes with realistic eyes and rain bands, although wind and pressure near the eye remain somewhat more lopsided than in X-SHiELD. It can also generate realistic strong winds moving down mountain slopes.

How ACE and HiROv2 work together

The system also includes an updated version of ACE. In a 10-year simulation, its hurricanes, average climate patterns, and overall distributions of rain, wind, and pressure match X-SHiELD well. That result suggests ACE’s coarser simulations can serve as a reliable base for long-term high-resolution climate data.

When HiROv2 and ACE are used together, a full year of global 3 km data was generated. The statistics of that year closely match X-SHiELD, even though HiROv2 was working from an ACE simulation created on its own rather than from real X-SHiELD data. In a 10-year run, the hurricane energy produced by HiRO-ACEv2 is comparable to X-SHiELD over the same period. Snapshots from that run also show a range of other extreme events that resemble those in X-SHiELD, including an example of a strong hurricane in the North Atlantic generated during the 10-year HiRO-ACEv2 simulation.

Cost savings, speed plans, and available tools

For researchers, the practical value is flexibility plus lower cost. HiRO-ACEv2 can generate many versions of high-resolution rain, wind, and pressure data for extreme events over 10 years or more. Producing global high-resolution data with the system costs roughly 300 times less than running X-SHiELD. If only one region is needed, the savings can be more than a hundred times greater, depending on the size of the area. A streamlined version of HiROv2 is also in development and could make it more than ten times faster still.

HiRO-ACEv2 Simulates Extreme Weather at 3 Kilometer Resolution at a Fraction of the Cost

To help people see what HiRO produces without running inference, a web demo was built. It shows pre-generated precipitation, wind, and sea-level pressure output from HiROv1 and HiROv2 side by side with X-SHiELD, the simulation they were trained on. The demo adds a quick visual check to the quantitative comparisons in the tech report and lets users zoom into a region of interest to judge performance there.

In the ACE data visualizer, a simulated 10 m wind speed map over Baja California shows ACE2 at 100 km resolution on the main map and HiRO-ACEv2 at 3 km in the inset. A simulated storm over the northeastern Pacific also appears: the main map shows 10 m wind speed at 100 km resolution, while the 3 km HiRO-ACEv2 inset reveals finer structure inside the storm. For deeper analysis, the models are available to download, and tutorial notebooks walk through how to run them.

What still limits the system, and what comes next

There are still boundaries. Because HiRO learned from X-SHiELD, it only works with the version of ACE tuned to match X-SHiELD. That means it cannot yet produce high-resolution data for other climates, such as a warmer future or the historical record in datasets like ERA5. The next step is to train on new X-SHiELD datasets, including one simulating a world 4°C warmer, so the system can handle different climates. Planned updates also include outputs such as near-surface temperature and snow and ice, which would allow for more uses, such as snowpack prediction.

Who should use it, and who should wait

Researchers who need many high-resolution versions of rain, wind, and pressure data for rare extreme events over 10-year periods may find HiRO-ACEv2 useful, especially when the cost of running X-SHiELD is a practical limit. Those who need climates outside the X-SHiELD match, such as ERA5-based historical records or a 4°C warmer world, should wait for planned training updates. So should anyone who still needs near-surface temperature, snow, and ice outputs.

Comments (0)

  1. No comments yet. Be the first to share what worked for you.

Leave a comment

Comments are reviewed before they appear. Your email address is not published.