Translation Exercise·Weather Forecast and Artificial Intelligence

/ SIon | 子虚栈

Original link: https://blog.si-on.top/2023/weather-forecast/

Big advances in weather forecasting, the stars are still shining – three things that will bring the full potential of weather forecasting

Maniac (computer code name:狂人), a computer designed in Princeton after World War II, can perform 10,000 times per second ultra-fast (Blistering) calculations. This extraordinary power is used primarily for two purposes: modeling thermonuclear explosions and Earth’s weather. These are also the two most consequential applications that the maker of that machine could have imagined.

The calculations that once took狂人1.38 billion years (bn-years) to perform can be completed in an hour on the fastest computer today. Despite the increased power and ambit of modern supercomputers, when they are devoted to weapons and weather, there is still a big gap in their capabilities. For every life that lives in the undercurrent of fear (Dread) every day, the contribution they made through the hydrogen bomb design is really very little. But it’s finding practical applications everywhere in using the weather to predict what to wear around the world.

According to the research of the World Bank and other relevant institutions: Numerical Weather Prediction (NWP) [1] makes the world an annual profit of 16.2 billion yuan. Its success can be confirmed not only from any modern farm or military commander, but also from your everyday clothes (Fabric). No smartphone lacks sunshine ☀ ,wind ? ,cloud ☁ The freehand icon [2] . Umbrellas according to the weatherman’s advice ? Staying at home is no longer about hope trumping experience. [3]

NWP+AI, NB
The application of machine learning and other forms of artificial intelligence will take weather forecasting even further. Those supercomputers that are used for numerical weather forecasting to calculate the next day’s weather based on the current state; Trillions of data operations. With simple training based on past weather data, machine learning systems can more or less match their forecasts, at least in some respects (albeit not comprehensively). If the development of AI has some guidance elsewhere, then this is only the beginning.

What’s more, in some cases, artificial intelligence methods seem to be able to reveal behaviors of the weather that numerical weather prediction (NWP) cannot calculate alone. Artificial intelligence burns less money and attracts more newcomers to the weather industry. They are expected to provide precise customized needs for customers and fresh ideas to open the door to new markets.

There are three things that need to be done the most to achieve the greatest possibility.

  1. The first is to ensure that healthy competition does not damage the infrastructure.

Most government organizations manage NWP, and they put a lot of effort into obtaining observations from all over the world for the accurate representation of weather they need. The cost of their investment can be offset by selling high-value predictions to specific markets.

To do their best work, AIs need to be trained on these representations. There is little suspense that these “good jobs” will be sold for less than some modern weather forecaster products. So a stopgap must be found so that newcomers can get the data they need to train the AI ​​without losing too much of the current weather forecasters in the industry [4] . To do otherwise would threaten the carefully crafted systems [5] they use to translate observations and computations into the datasets on which artificial intelligence and the world depend, at least for now.

  1. The second thing to do is to combine artificial intelligence and digital computing to fight climate change.

At the moment, it is not possible to run climate models at the resolution (accuracy) of weather forecasts. New hardware built for AI systems may help (chipmaker Nvidia is interested [6] ). AI can also be used to find patterns in the predictions produced by these models, making them even more informative. Also serves as an interaction layer that makes it easier for non-experts to understand.

  1. Better access to information is a must now before this becomes a big problem.

In 2019, the Global Commission on Climate Adaptation reported that 24-hour advance notice of damaging weather events could reduce losses by 30 percent, and that investing $800 million in early warning systems in developing countries could avert annual losses of $3 billion to $16 billion. Therefore, the World Meteorological Organization will achieve “National Early Warning” as a priority by 2027. According to Petteri Taalas, the head of the organization, three out of every four people in the world have mobile phones, but only half of the countries have disaster warning systems, which is outrageous.

How to save lives? No breakthrough is needed to solve this problem, just some modest investments, detailed planning, focused discussions, and enough political will to overcome inevitable institutional obstacles. This is not a Promethean effort by the founder of MANIAC; it neither sets the world on fire nor simulates the mechanics of a burning world. But it should save tens of millions of lives and millions of livelihoods.

  1. Numerical weather prediction (numerical weather prediction) refers to solving the equations of fluid dynamics and thermodynamics describing the weather evolution process through large-scale computer numerical calculations under certain initial and boundary value conditions based on the actual conditions of the atmosphere, and predicting certain weather conditions in the future. The method of atmospheric motion state and weather phenomenon in time period. –Baidu Encyclopedia ↩ ︎

  2. Redolent of means reminiscent of …, here free translation as “freehand icons of various weather” ↩ ︎

  3. It means that the weather forecast can already accurately predict the weather, and no longer need to rely on experience and hope. This sentence is difficult to understand, the original sentence is: “Deciding to leave an Umbrella at home on a forecaster’s advice is no longer necessarily a triumph of hope over experience” ↩ ︎

  4. The original sentence is Pocket, translated into “industry” is better than “in the bag”. ↩ ︎

  5. Is the person who wrote this sentence crazy? Layer upon layer. “To do otherwise could threaten the meticulously set up systems they use to turn observation and computation into the data sets on which the AIs and the world rely, at least for the time being.” ↩ ︎

  6. Linus: Nvidia, fuck you ? ! ↩ ︎

This article is transferred from: https://blog.si-on.top/2023/weather-forecast/
This site is only for collection, and the copyright belongs to the original author.