---
title: "HiRO-ACEv2: Simulating extreme weather at 3 km resolution for a fraction of the cost"
description: "HiRO-ACEv2 generates global 3 km climate data for extreme weather at roughly 1/300th the cost of the X-SHiELD model, with even larger savings for single-region runs. The system pairs a coarse 100 km AI model (ACE) with the new HiROv2 downscaler that simultaneously predicts rain, wind, and sea-level pressure instead of just rainfall, and has improved skill at reproducing extreme events. When fed coarsened X-SHiELD data HiROv2 matches the target’s high-resolution statistics within 3% for events as rare as one in a million data points, and accurately estimates tropical cyclone energy for storms never seen during training. Over a 10-year simulation the full system’s hurricane energy and extreme event snapshots compare well with X-SHiELD. However, HiRO cannot yet handle other climates because it depends on X-SHiELD-specific ACE tuning; future work includes training on warmer-world datasets and adding temperature and snow outputs."
---

# HiRO-ACEv2: Simulating extreme weather at 3 km resolution for a fraction of the cost

[Hugging Face](https://yomu.fyi/company/hugging-face) · Kyle Wiggers · Oct 14, 2026

**Type:** Announcement

## Summary

HiRO-ACEv2 generates global 3 km climate data for extreme weather at roughly 1/300th the cost of the X-SHiELD model, with even larger savings for single-region runs. The system pairs a coarse 100 km AI model (ACE) with the new HiROv2 downscaler that simultaneously predicts rain, wind, and sea-level pressure instead of just rainfall, and has improved skill at reproducing extreme events. When fed coarsened X-SHiELD data HiROv2 matches the target’s high-resolution statistics within 3% for events as rare as one in a million data points, and accurately estimates tropical cyclone energy for storms never seen during training. Over a 10-year simulation the full system’s hurricane energy and extreme event snapshots compare well with X-SHiELD. However, HiRO cannot yet handle other climates because it depends on X-SHiELD-specific ACE tuning; future work includes training on warmer-world datasets and adding temperature and snow outputs.

## Context

X-SHiELD is a computationally expensive high-resolution global atmospheric model that explicitly simulates storms and other small-scale weather. HiRO-ACEv2 aims to produce similar climate data at a fraction of the cost using AI.

## Approach / What changed

The HiRO-ACEv2 system uses a two-part architecture: a coarse 100 km resolution AI model (ACE) to generate base climate fields, and a downscaling model (HiROv2) that sharpens those outputs to 3 km resolution while predicting multiple variables simultaneously. HiROv2 training was improved to better represent extreme events by learning from high-resolution X-SHiELD data.

## Takeaways

- HiROv2 can predict multiple high-resolution fields (precipitation, wind, sea-level pressure) simultaneously, improving upon the single-field output of HiROv1.
- When using coarsened X-SHiELD as input, HiROv2 matches the target high-resolution statistics to within 3% for events as rare as one in a million data points.
- The full HiRO-ACEv2 system generates global 3 km climate data at roughly 1/300th the cost of X-SHiELD; savings are even greater for single-region simulations.

## FAQ

### What are the main improvements in HiROv2 compared to HiROv1?

HiROv2 can simultaneously predict multiple high-resolution fields (rain, wind, pressure) instead of just rainfall, and it has improved skill in predicting extreme events.

### How much cost savings does HiRO-ACEv2 offer over X-SHiELD?

Producing global high-resolution data with HiRO-ACEv2 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 area.

### Can HiRO-ACEv2 be used for future climate scenarios?

Not yet, because HiRO learned from X-SHiELD and only works with the version of ACE tuned to match X-SHiELD. The team plans to train on new X-SHiELD datasets, including one simulating a world 4 degrees Celsius warmer, to support different climates.

**Tags:** [Extreme Weather](https://yomu.fyi/topic/extreme-weather), [Machine Learning](https://yomu.fyi/topic/machine-learning)

- Source: [Hugging Face](https://huggingface.co/blog/allenai/hiro-acev2)
- Source URL: https://huggingface.co/blog/allenai/hiro-acev2
- Ingested by Yomu: 2026-10-14T18:00:15.070Z

[Read original post](https://huggingface.co/blog/allenai/hiro-acev2)