← Back
NVIDIA Blog

University of Manchester Uses NVIDIA Earth-2 to Forecast Air Pollution Across the UK

•5 min read•
#nvidia#compute#inference
University of Manchester Uses NVIDIA Earth-2 to Forecast Air Pollution Across the UK
Level:Advanced
For:AI Engineers
✦TL;DR

The University of Manchester leveraged NVIDIA’s Earth‑2 family to build a UK‑wide air‑pollution forecast, training the CorrDiff generative downscaling model on the Isambard‑AI supercomputer in just two days and deploying inference on a single‑GPU‑desk‑top DGX Spark. The workflow used a year of hourly simulated pollution data to produce a 2‑3 km² resolution model, and the newer StormCast model adds time‑dependent forecasting directly from observations. This demonstrates that high‑resolution, near‑real‑time air‑quality modeling can move from national supercomputers to desktop‑scale hardware, opening rapid scenario analysis for health and policy applications.

⚡ Key Takeaways

  • Training the Earth‑2 CorrDiff model on Isambard‑AI completed in two days using an eight‑GPU node.
  • The model operates at a spatial resolution of 2‑3 square kilometers across the UK.
  • Earth‑2 StormCast enables time‑dependent forecasts that ingest live air‑quality observations.
  • Inference can be run on a DGX Spark personal AI supercomputer, fitting on a desk.
  • The workflow required a year’s worth of hourly UK pollution simulation data as training input.
💡 Why It Matters

Engineers can now generate fine‑grained pollution forecasts with desktop‑scale GPUs, dramatically reducing compute costs and turnaround time for health‑impact assessments and policy scenario testing.

✅ Practical Steps

  1. Train an Earth‑2 CorrDiff model on the Isambard‑AI supercomputer using a year of hourly UK pollution simulation data.
  2. Deploy the trained model for inference on a DGX Spark workstation to obtain 2‑3 km² resolution forecasts.
  3. Extend the pipeline with Earth‑2 StormCast to incorporate real‑time air‑quality observations for time‑dependent predictions.

Want the full story? Read the original article.

Read on NVIDIA Blog ↗

More like this

AI’s GPU focus meets a CPU reality check at Dell and AMD

SiliconANGLE AI•#agents

NVIDIA RTX Spark Compact Desktops Out Now

NVIDIA Blog•#compute

Helping AI text detectors keep up with changing conditions

MIT News AI•#llm

Hugging Face and OpenRouter join forces to make open-weight models easy to run

Hugging Face Blog•#inference

EXPLORE AI NEWS

Daily hand-picked stories on LLMs, RAG, agents and production AI — curated for engineers who ship.

BROWSE NEWS

GET THE WEEKLY DIGEST

Join engineers getting the Monday signal-over-noise AI breakdown. No spam, unsubscribe anytime.

LEARN AI ENGINEERING

Curated courses, research papers, repos and tutorials built for engineers leveling up in AI.

START LEARNING