|Many people rely on weather forecasts to know how they should prepare for that day. It’s easier to plan for vacations or to hang out with friends when you know what kind of weather you are expecting / Photo by: RubyGoes via Flickr|
Many people rely on weather forecasts to know how they should prepare for that day. It’s easier to plan for vacations or to hang out with friends when you know what kind of weather you are expecting. Indeed, weather forecasting is a great deal in many aspects of our society. For instance, accurate weather forecasting allows farmers to make critical decisions about planting and harvesting. Airlines can maximize the use of their planes. Also, utilities can make decisions about their capacity needs during heat waves or large rainfalls.
However, there have been many times in the past when weather forecasts didn’t work. According to Phys.org, an internet news portal that provides the latest news on science, weather forecasting is a huge challenge because it’s like attempting to predict something that is inherently unpredictable. This can be explained through the “butterfly effect,” the sensitive dependence on initial conditions where a tiny change in one state of a deterministic nonlinear system can result in large differences to a later state.
The atmosphere is a chaotic system, hence, a small change in one location can have significant outcomes in some other places. A single error that develops in a forecast will rapidly grow and cause further errors on a larger scale. This is where artificial intelligence comes in. Today, many institutions and governments are using AI to improve their weather forecasting. For instance, the US National Oceanic and Atmospheric Administration (NOAA) has recently been using machine learning more to improve their forecasts.
The researchers from the NOAA discovered that AI techniques can significantly improve the prediction skill for multiple types of high-impact weather such as severe thunderstorms, tornadoes, and hurricanes. “AI methods extend easily to directly predicting impacts of high-impact weather and power generated by variable sources such as solar or wind, energy consumption in an area, or airport arrival capacity,” the authors said.
Weather Forecasting With High Accuracy
Recent reports projected that AI will be the biggest commercial opportunity for companies across the world. According to The Wall Street Journal, a US business-focused, English-language international daily newspaper based in New York City, advances in AI can potentially increase global GDP by up to 14% between 2019 and 2030. This is the equivalent of an additional $14 trillion to $15 trillion contribution to the world’s economy. Thus, it’s not surprising that researchers are trusting AI in weather forecasting. Recently, they used an AI model to develop more accurate short-term forecasts, which professor Guido Cervone at Penn State University referred to as one of meteorology’s trickiest problems. According to Cervone, they are using computational resources toward areas that are harder to predict that would help generate better short-term forecasts.
Weiming Hu, a doctoral student in geography, stated that they also utilized genetic algorithms to address complex, rapidly changing weather patterns. Instead of creating machine learning models designed to offer perfect solutions to issues of weather forecasting, it offers good solutions. “Genetic algorithms do not guarantee the best solution, but they do guarantee to find better solutions faster. In a case like predicting temperature changes, you might not care about finding the ultimate solution because it might be the difference between 29.56 degrees and 29.55 degrees. That’s probably not going to matter for the regular person,” Hu said.
One of the largest tech companies, Microsoft, is currently using AI to improve the accuracy in predicting the weather through the AI for Earth program. AI for Earth has grantees such as The Yield, SunCulture, and International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) that are working solutions for this challenge. This aims to understand how the weather will impact their farms.
The Yield utilizes sensors to measure 12 key factors, which include rain, wind, light, leaf wetness, and soil moisture. According to CNET, an American media website that publishes reviews, news, articles, blogs, podcasts, and videos on technology and consumer electronics globally, it conducts predictive modeling with the help of AI to create a seven-day weather forecast for each farm's microclimate. Also, SunCulture is using AI to make recommendations to help farmers plan. It can also be used to optimize planting, irrigation, fertilizing, and pest control.
|One of the largest tech companies, Microsoft, is currently using AI to improve the accuracy in predicting the weather through the AI for Earth program / Photo by: Raimond Spekking via Wikimedia Commons|
Predicting Severe Weather
Aside from simple weather forecasts, AI now can also recognize potential severe storms more quickly and accurately. For years, it has been difficult for meteorologists to track shapes and movements of clouds that could indicate severe storms. But, finally, researchers at Penn State, AccuWeather, Inc., and the University of Almería in Spain have developed a framework based on machine learning that can detect rotational movements in clouds.
According to Science Daily, an American website that aggregates press releases and publishes lightly edited press releases about science, the researchers analyzed more than 50,000 historical US weather satellite images, which they have labeled according to the shape and motion of "comma-shaped" clouds. The patterns of these clouds are associated with cyclone formations, leading to severe weather events such as thunderstorms, high winds, and blizzards.
After that, the researchers trained computers to automatically recognize and detect comma-shaped clouds in satellite images using computer vision and machine learning techniques. As a result, they created a model that would predict severe weather in real-time. "Because the comma-shaped cloud is a visual indicator of severe weather events, our scheme can help meteorologists forecast such events," Rachel Zheng, a doctoral student in the College of Information Sciences and Technology at Penn State and the main researcher on the project, said.
According to the researchers, their method can effectively detect comma-shaped clouds at an average of 40 seconds per prediction, with 99% accuracy. At the same time, it predicted 64% of severe weather events, which is better compared to other existing severe-weather detection methods. Professor James Wang, the dissertation advisor of Zheng, stated that this study showed the feasibility of AI-based interpretation of weather-related visual information to the research community.
Indeed, AI can bring benefits not only for businesses but also in helping us with weather forecasts. It is expected that this new way of using AI will develop further in the coming years to make weather forecasting truly reliable.
|Aside from simple weather forecasts, AI now can also recognize potential severe storms more quickly and accurately / Photo by: NOAA National Severe Storms Laboratory via Flickr|