When AI Meets Climate Change: A Game Between Technology and Nature

Recently, the weather has not been very “peaceful”. The sun is big this second, and the next second there will probably be heavy rain.

Extreme weather is frequent around the world, and artificial intelligence may help people cope with the effects of climate change.

Google DeepMind executive Colin Murdoch said,Artificial intelligence has the potential to accelerate world-changing innovations such as “limitless” clean energy and better weather models。

The Huawei Cloud Pangu Meteorological Model, which previously appeared in the “Nature” (Nature) journal, officially launched the official European Mid-Term Weather Forecast website on July 31, allowing the world to see the power of large domestic models to solve problems in the field of meteorology.

This article will answer your questions about how AI affects the climate and which companies and products are currently worth paying attention to in the market.

01.

How does artificial intelligence work on the climate?

Today, weather forecasting is expected to introduce artificial intelligence prediction methods. These systems can produce faster and more accurate results than traditional models, and even have the potential to change the weather forecasting industry.

Artificial intelligence models do not need to use a large number of traditional mathematical equations to calculate actual physics. Instead, they absorb large amounts of historical weather data and learn to identify patterns. Then, when provided with real-time data on weather conditions, AI can use recognition patterns to make predictions.

The researchers said,Even if current artificial intelligence prediction systems don't replace traditional models, they are still valuable in conveying information about the weather。

Over the past few years, hundreds of world-renowned companies have publicly pledged to achieve zero carbon emissions by reducing emissions and purchasing carbon offsets, and adjusted their operating plans accordingly.

As a result,Using artificial intelligence to quantify carbon emissions, fully understand the carbon footprint, optimize low-carbon decisions, and build an AI-driven carbon offsetting market has become a new business hotspot. The latter mainly applies computer vision to aerial images and sensor data, automatically estimates carbon stored in afforestation, and continuously monitors data from its carbon offsetting projects.

Whether in agriculture or industry, artificial intelligence can help monitor and reduce greenhouse gas emissions and improve the performance and stability of energy systems.

Agriculture is a major contributor to climate change, accounting for 10% to 15% of the world's greenhouse gas emissions. Modern resource-intensive agriculture often causes large amounts of resources to be wasted. The application of AI technology can improve agricultural efficiency, reduce carbon footprint, and increase food production.

In industry, for example, electricity cannot be effectively stored on a large scale, so power grids must continuously balance supply and demand in real time, and AI can achieve more efficient automation and more accurate systematization.

Using techniques such as deep learning and convolutional neural networks, we can quickly process and analyze massive amounts of meteorological observation data, satellite images, radar signals, etc., extract useful features and information from them, and provide higher quality data for the input and output of climate models;It is also possible to use techniques such as machine learning and reinforcement learning to optimize and improve climate models constructed by traditional physical equations or statistical methods, thereby reducing model errors and biases.

Relying on artificial intelligence technology to simulate the climate system can help us better understand the complexity and uncertainty of the climate system, improve the accuracy and efficiency of climate prediction, and provide scientific evidence and decision support for climate change adaptation and mitigation.

02.

Companies and products to watch

Chamas Palihapitia, a famous venture capitalist in Silicon Valley, once said,The world's new trillionaires will be born amid climate change. Combating climate change is not only a global human priority; it also contains huge business opportunities. It is also an inevitable choice for the world's common destiny.

In May of this year,Google launches Flood Hub service powered by artificial intelligenceIt combines two prediction models to estimate the amount of water flowing into the river and the estimated depth of flooding in the affected area.Users can get Google's forecast information as early as 7 days before the flood hits.

Flood Hub now covers more than 80 countries around the world, with a total of 460 million people. The data can be used by individuals or organizations for evacuation and evacuation.

Currently,About a quarter of the world's people live in flood zonesThis number is expected to rise as sea levels rise and larger hurricanes caused by climate change, and the need to accurately map flood events should not be underestimated.

Faced with the pressure of carbon emissions, Google CEO Sundar Pichai also set a goal for Google to achieve zero-carbon operation by 2030.Although Google's environmental protection plans were thwarted by AI's huge heat reduction and water consumption last year, Google claims that the energy efficiency of self-customized TPU chips is far higher than that of similar Nvidia products, and that Google has a “healthy pipeline for future chips.”

Founder and CEO Hwang In-hoon emphasized the outstanding contribution Nvidia's supercomputer Earth-2 can make to climate simulation at the “Earth Virtual Engine Program (EVE)” Berlin summit held on July 3 this year.

Earth-2, as the name suggests,It is a digital twin of the Earth that can run the AI physical environment created by Modulus (Artificial Intelligence Physics Simulation Framework) at a speed of one million times in Omniverse (Nvidia's virtual world simulation engine).

This is a bit like the “Genesis Component,” a tool that Doraemon found for Nobita's summer vacation research. It can observe changes in terrestrial planets. Unlike traditional weather simulations, it can visualize the atmosphere, water resources, and land for several years in terms of physics, chemistry, and biology.

Earth-2 can achieve three miracles：The first is to simulate the climate at a sufficiently fast speed and resolution of up to micrometer to predict the impact on soil particle size;Second, AI is used to achieve high-fidelity simulation and real-time interaction with PB-level climate data, and massive data can also be pre-calculated in various ways;Third, it can interact with the Omniverse platform, visualize data, and use it by decision makers, enterprises, companies, and scientists.

Just a few days ago,IBM, open source AI platform Hugging Face and NASA released PrithVI, the world's largest open source AI foundation modelto help analyze satellite images.

PrithVi was provided by NASA with harmonized Landsat Sentinel-2 (HLS) satellite remote sensing data over the past year, optimized based on labeling data for flood and wildfire areas, and was pre-trained and further fine-tuned on IBM's WatsonX.AI basic model.

NASA predicts that by 2024, scientists will need to do researchChina faced 250,000 terabytes of data, so it signed a “Space Action Agreement” (Space Ac) with IBMt Agreement) to jointly use AI to improve this situation.

The model can be adapted to various tasks such as deforestation tracking, crop yield prediction, and greenhouse gas detection, and aims to effectively address climate change challenges and contribute to a sustainable future for the planet.

Just in July of this year, the top international academic journal “Nature” (Nature) published the independent research results of the Huawei Cloud Pangu University Model Research and Development Team — “3D Neural Networks for Accurate Mid-Term Global Weather Forecast”.thinksHUAWEI Cloud's Big Weather Model allows people to re-examine the future of weather forecasting models。

Huawei began setting up a project to build the HUAWEI Cloud Pangu Big Model in 2020, and was released in April 2021. The latest results completed the global weather forecast for the next 24 hours in just 1.4 seconds, 10,000 times faster than the current method.

Others are artificially basedAlthough intelligent prediction models, such as Nvidia's FourcastNet, can quickly achieve weather forecasting, the accuracy of predictions is far lower than traditional NWP methods.Therefore, the team creatively proposed a three-dimensional neural network adapted to geographical location to process complex meteorological data, and used a hierarchical time aggregation algorithm to reduce the number of cumulative predictions.

The wind and rain model has now been used to forecast the path of domestic typhoons. For example, in the recent typhoon “Du Surui,” an accurate prediction path was achieved, which is superior to European and American authorities.

03.

A green future is ahead

There is no doubt that we humans are facing one of the most serious challenges in the world today, and extreme climate change is already posing a huge threat to us. Fortunately, with its amazing potential, artificial intelligence has become an important partner in our fight against climate change, so we are no longer “fighting alone.”

It is important to note that artificial intelligence itself is not perfect; this technology also consumes large amounts of energy and resources, and has a certain environmental impact.

However, according to data from the International Energy Agency (IEA),Currently, 1% of the world's total electricity is sufficient to meet the energy needs of global data centers. Moreover, as technology advances and innovations, the energy efficiency and performance of artificial intelligence will continue to improve, and its negative impact on the environment will gradually decrease.

On the path to sustainable development, we need to make full use of the technological advantages of artificial intelligence to create a better and greener future with advanced awareness and sense of responsibility.