Taking a “High-Definition Picture” of the Weather: Why Do We Need Supercomputers and AI?
Hello everyone, I’m your financial journalist and economist. Today, we’re not talking about stock market fluctuations or housing prices, but about a topic that affects everyone’s safety and well-being, and one that reveals the underlying logic of a massive technology industry: weather forecasting.
You might think that weather forecasting is just about watching where the clouds are moving. But in today’s world, it’s actually a competition of extremes in computing power, algorithms, and data processing. Recently, the China Meteorological Administration and Sugon (a Chinese technology company) made a significant breakthrough: they achieved high-precision weather forecasts for the next 10 days with a resolution of 5 kilometers across the globe, all done on a domestically developed supercomputer cluster, and in record time (within 1 hour).
This sounds very technical, but put simply, it means we can finally get a “4K high-definition” picture of the Earth, rather than the previous “mosaic” images. The clearer this picture is, the earlier and more accurately we can predict whether a storm will hit your area or if a typhoon will deviate from its course.
Below, I’ll break down this news into five key points to explain what’s behind this advancement and why it’s so important.
---
1. From “Mosaic” to “High Definition”: What Does Improved Resolution Mean?
Core Logic: The finer the grid, the clearer the image—and the quicker the life-saving actions can be taken.
Previously, weather forecasts were like looking at a map with low pixel resolution. For example, if there was a 1,000-meter-tall mountain west of Zhengzhou, and our forecasting grid was 20 or 30 kilometers wide, the computer might “average” it out, showing it as a 200-meter-tall hill.
- Consequences: Air currents around the mountain would rise and condense into rain. If the computer thought it was just a small hill, it wouldn’t predict heavy rain. This is why extreme weather events like storms in Zhengzhou were often inaccurately forecasted or delayed.
- The Change: With a resolution of 5 kilometers, and eventually 3 or 1 kilometers, the computer can see the actual mountain, the shape of the clouds, and even the turbulence in the air.
- Popular Example: It’s like going from using a telescope to look at the moon (only seeing a bright spot) to using a high-power microscope to see the moon’s craters. For a typhoon, a 5-kilometer resolution can more accurately predict its intensity, not just its path.
Conclusion: Improved resolution directly determines whether we can issue timely warnings hours before a storm, reducing casualties and property damage.
---
2. Why Are CPUs No Longer Enough? The Exponential Growth of Computing Power
Core Logic: Every doubling of precision requires an 8-fold increase in computing power, which traditional computers can’t handle.
Many people think that just getting faster computers or more CPUs will solve the problem. But in weather forecasting, the increase in computing demand is exponential, not linear.
- Mathematical Example: Doubling the resolution from 10 kilometers to 5 kilometers increases the number of grid points by 4 times horizontally. Adding vertical resolution multiplies the total computing load by 8 times. Further refinements could increase it by 16 times.
- Real-World Challenge: Experts estimate that using traditional CPU architectures would require nearly 100,000 CPU cores to achieve 5-kilometer global forecasts in 2 hours. This is practically impossible, not only due to high costs but also because the speed of data transfer between CPUs is too slow.
- Solution: We need to switch from traditional CPU-based parallel computing to GPU (Graphics Processing Unit) heterogeneous computing. GPUs are excellent at handling thousands of small tasks simultaneously, which is perfect for the massive, repetitive calculations in weather models.
Conclusion: This is more than just a simple computer upgrade; it’s a revolution in computing architecture. We need “multi-core monsters” like GPUs to handle this exponential growth in computing demands.
---
3. AI Is No Longer Just an “Assistant”: The New Trend of Super-Intelligent Integration
Core Logic: Scientific computing + artificial intelligence = more accurate forecasts.
In the past, AI was used for auxiliary tasks in weather forecasting, such as data processing and post-processing. But now, AI has moved to the core of physical simulations.
- Why AI Is Needed: Weather forecasting has a natural limitation: we can’t install temperature and humidity sensors in every corner of the world. Observational data is incomplete.
- AI’s Role: AI learns patterns from historical data, filling in the gaps where observations are missing. It’s like an experienced meteorologist who can predict the weather in an unobserved area based on surrounding conditions.
- Super-Intelligent Integration: Modern approaches like the MCV model combine traditional scientific computing (accurate physical equation solving) with AI (fast predictions and gap filling).
- CPU/GPU Division of Labor: CPUs handle logical tasks, while GPUs handle the massive calculations.
- Precision Differences: Traditional computing requires high precision (64-bit), but AI models often use lower precision (16-bit/8-bit) for speed. Modern hardware (like the Sugon 8000 cluster) supports both, maximizing efficiency.
Conclusion: Future weather forecasting will combine “physical laws” with “data intelligence.” AI makes forecasts faster and more accurate, handling complex non-linear relationships that traditional math formulas struggle with.
---
4. The Big Test for Domestic Computing Power: From “Usable” to “User-Friendly”
Core Logic: Independence and control are not just political goals; they reflect technical prowess.
A highlight of this news is that this achievement was made using domestically developed computing power.
- Background: Weather forecasting is crucial for national security. If we rely on imported computing power, our systems could be paralyzed in a supply crisis.
- Achievement: The MCV model developed by the China Meteorological Administration was run on the Sugon 8000 cluster (a domestically produced supercomputer with 100,000 GPUs) and completed 5-kilometer forecasts for the next 10 days in 1 hour.
- Significance:
1. Proof of Capability: It shows that domestic chips and clusters can meet international standards.
2. Performance Comparison: The U.S. currently uses 12.5-kilometer resolution, European centers use 9-kilometer, while China has reached 5 kilometers and is moving towards 3 kilometers, placing it in the international forefront in terms of precision.
3. Ecosystem Development: This isn’t just for weather forecasting; the Sugon 8000 cluster is also used for protein research, molecular dynamics, and new drug development, demonstrating its versatility in high-performance computing.
Conclusion: This breakthrough indicates that China is on par with global leaders in high-end scientific computing and may even lead in certain areas, such as high-resolution weather forecasting.
---
5. The Future Trend: Specialized Chips and Ten-Thousand-GPU Clusters
Core Logic: The surge in computing demand is creating new hardware opportunities and investment prospects.
The news mentions two key trends that are important for the tech industry and investors:
1. Rise of Specialized Chips (ASICs):
- General-purpose GPUs are flexible but not the most efficient for specific tasks like weather forecasting.
- Customized chips for weather and geophysical exploration could significantly improve performance.
- Benefits: Reduced costs, increased stability, and faster speeds.
- Business Opportunities: Chip design, custom computing hardware, and high-performance storage will see growth.
2. Ten-Thousand-GPU Clusters as the Standard: As resolution improves to 1-3 kilometers, computing power demand will continue to soar.
- Ten-thousand-GPU clusters will become standard in large research and weather institutions.
- Cross-Industry Impact: This demand extends to oil and mineral exploration, materials science, AI model training, and robotics, all of which require similar high-performance computing.
Conclusion: The advancement in weather forecasting is a reflection of the upgrade in computing infrastructure. It means not only more accurate forecasts but also a comprehensive upgrade in high-performance computing (HPC), artificial intelligence (AI), and semiconductor chips. This represents a trillion-dollar tech market.
---
Summary
This news might seem to be about better weather forecasts, but it’s actually about the strengthening of China’s technological foundation.
- For the Public: It means earlier and more accurate warnings for storms and typhoons, providing better protection for our lives and property.
- For Industries: It indicates that super-intelligent integration (supercomputing + AI) is the future direction of scientific research, with traditional computing being transformed by AI.
- For the Economy: It signifies a huge demand for the domestic computing industry chain (chips, servers, cluster software), from weather forecasting to medicine and materials science. High-performance computing is the new “water and electricity” of the era.
So, the next time you see a weather forecast, remember that behind it are thousands of GPUs working tirelessly, AI filling in data gaps, and Chinese scientists striving for the ultimate high-definition view of the weather.