19
2022
-
08
Smart Grid Forecasting: A Revolutionary Shift from Stations to Grid Points
Author:
According to the China Meteorological Administration's plan, by the end of December this year, the unified data-source "one-network" gridded forecasting service for China's weather prediction will officially go live. This network boasts a spatial resolution of 5 km × 5 km and can provide hourly updates of weather forecasts for the next 10 days, enabling the public to access location-based, highly detailed meteorological services anytime and anywhere. This marks a significant shift in China's weather forecasting, moving from traditional station-based predictions to grid-based, advanced forecasting methods. Let’s take a comprehensive look at intelligent grid forecasting and explore this exciting new service together.
According to the China Meteorological Administration's plan, by the end of December this year, the unified data-source "one-network" gridded forecasting service for China's weather forecast will officially go live. This network boasts a spatial resolution as fine as 5 km × 5 km and can provide hourly updates of weather forecasts for the next 10 days—specifically, every 3 hours. As a result, the public will have access to location-based, highly detailed weather services anytime, anywhere. This marks a significant shift in China's weather forecasting, moving from traditional station-based predictions to grid-based, advanced forecasting methods. Let’s take a comprehensive look at intelligent grid forecasting and explore this exciting new service together.
Grid Chapter
Although "smart grid forecasting" is a term that has only recently entered the public eye, its technological development actually dates back much further.
Weather forecasting is steadily evolving from qualitative and descriptive forecasts toward more digital, gridded predictions. For instance, previously, meteorological agencies issued weather forecasts for towns and cities that included only weather conditions—such as precipitation, temperature ranges, and wind speed and direction—for just over 2,400 urban areas, with forecasts updated only three times a day. As a result, both the temporal and spatial accuracy of these forecasts were relatively low, failing to meet the growing demands of various industries and the public.
In 2012, the National Meteorological Center introduced a new forecasting product: refined, high-resolution forecasts specifically tailored for major cities. This initiative provided detailed 24-hour weather predictions for provincial capital cities and separately listed municipalities across the country, updating every six hours. Precipitation was even reported down to the millimeter level. Yet, despite these improvements, the forecasts still lacked sufficient granularity to fully satisfy the needs of different sectors and the general public.
Demand drives business growth. Around 2010, the concept of "gridded forecasting" was introduced into China's refined weather prediction services. But how can we understand it? Think of it like the Earth's latitude and longitude grid—China, along with each individual city, can be divided into countless 5km x 5km, or even 1km x 1km grids. The public essentially lives within these grid cells, and weather conditions naturally vary from one grid to the next.
Grid-based forecasting is conducted for each of these individual grids. Compared to the original point-based forecasts, it offers much greater spatial resolution and targeted accuracy. Take Beijing's weather forecast as an example: Previously, the city’s overall weather conditions—such as temperature and precipitation—were represented solely by data from a single location: the Nanjiao Observatory. However, with the implementation of grid-based forecasting, Beijing’s weather is no longer reflected by just one fixed point. Instead, meteorological services and weather predictions for Beijing can now be precisely tailored to every distinct grid across the entire city. Currently, some provinces are even capable of delivering weather forecasts accurate down to the minute. And in terms of spatial coverage, over ten provinces can already provide forecasts with resolutions as fine as 3 kilometers, 2.5 kilometers, or even finer.
The sophistication of gridded forecasting isn’t just reflected in its spatial precision—it also shines through its ability to be updated and released more frequently, even on a daily basis. Previously, a single weather phenomenon was typically covered in a day’s forecast; now, with gridded forecasting, it’s possible to provide nationwide, 3-hour-by-3-hour predictions for up to 10 days ahead. This means that at any given moment, the public can instantly check the specific weather conditions for their exact location, gaining clear insights into key meteorological factors such as temperature, precipitation, and wind patterns.
In addition to forecasting conditions over land grids, the meteorological department has also divided China’s responsibility sea areas into multiple 10-kilometer-by-10-kilometer grids, providing detailed forecasts for maritime visibility, strong offshore winds, and other key elements.
Grid-based forecasting not only refines the models further but also delivers more detailed and comprehensive forecast content.
In the original forecast products, the public was most familiar with three key elements: temperature, wind, and weather phenomena. However, today’s land and ocean forecast products have been refined to include four major categories comprising 18 distinct meteorological elements.
▲ Category One These are the essential elements, including temperature, precipitation, precipitation type, wind, cloud cover, and relative humidity. Previously, 12-hour forecasts only depicted a single weather phenomenon, but now we have more detailed, hourly forecasts—specifically, forecasts every three hours—that can capture weather conditions at different time intervals.
▲ Category Two It refers to environmental meteorological elements, including fog, haze, sandstorms, visibility, and more.
▲ Category Three It refers to hazardous weather elements, including short-term heavy rainfall, thunderstorms, thunderstorm winds, hail, and other severe convective weather forecasts.
▲ Category Four These are marine meteorological elements, specifically reflected in forecasts as offshore strong winds, offshore visibility conditions, and maritime weather phenomena, among others.
This forecast grid, with its increasingly high spatiotemporal resolution, has already become the meteorological department's most critical forecasting product for operational use. In the future, all forecast service products from the meteorological agency will be derived directly from this grid—allowing them to "pull out" exactly what’s needed for any given situation.
With the advancement of gridded forecasting, the content of these products has become increasingly comprehensive and diverse, while the volume of data continues to grow exponentially. Although the public may not directly use these products, the National Meteorological Center provides them as foundational tools to meteorological agencies and service units at all levels. Leveraging the highly detailed gridded forecasts, local meteorological departments then integrate location-based services, web platforms, mobile applications, and other innovative delivery methods to offer the public tailored, localized weather services and products—right at their doorstep. Currently, several provinces are even using refined gridded forecasts to help citizens plan their travel arrangements, going so far as to provide hourly updated weather forecasts specifically designed for daily commutes.
Smart Chapter
"AlphaGo" defeated Ke Jie, the long-time world No. 1 Go player, in three consecutive matches, leaving humanity both secretly worried and eagerly anticipating the future of artificial intelligence. This holds true for many industries—including weather forecasting, where one of the core features of intelligent grid-based weather prediction is its advanced AI capabilities.
Among all cutting-edge meteorological technologies that showcase intelligence, numerical forecasting and ensemble forecasting stand out as the most critical. High-resolution intelligent grids rely heavily on high-resolution regional numerical prediction models for support. Currently, China operates four operational high-resolution models, including GRAPES, the global numerical forecast model developed by the National Meteorological Center of China, as well as high-resolution models independently developed by the regional meteorological centers in Beijing, Shanghai, and Guangdong.
How can we place them on a single platform for easy access and use? The China Meteorological Administration has established a Numerical Forecasting Cloud in Shanghai (currently being upgraded into an Intelligent Grid Forecasting Cloud), hosting these four core weather forecasting models in the cloud. This allows meteorological agencies across the country to quickly share high-resolution numerical forecast products via the Numerical Forecasting Cloud client. During the "access and utilization" process, provincial-level meteorological departments carry out data processing, diagnostic analysis, interpretive applications, and interactive corrections of the models, ultimately generating their own provincial-level grid forecast products. Since these locally produced grid forecasts will become a key component of the future nationwide intelligent grid forecasting system, this "cloud" platform holds immense significance.
In the future, the meteorological department will also develop intelligent forecasting technologies that combine physical mechanisms with big data mining and numerical prediction applications. On one hand, mathematical statistics derived from numerical prediction models will be used to create sophisticated forecasting models and methods. On the other hand, data mining and machine learning techniques—leveraging big data technology—will enable researchers to explore advanced deep-learning forecasting models or even forecast robots. Some may wonder: if we rely entirely on machine learning and artificial intelligence, how exactly will smart forecasting evolve in the future?
Indeed, "intelligence" poses a challenge to forecasting. Take medicine as an example: in recent years, the U.S. has developed disease-diagnosis methods entirely based on machine learning and deep learning. This February, IBM’s robotic doctor Watson came to China for a "consultation," prescribing treatments within 10 seconds—and was immediately endorsed by five senior, experienced human chief physicians on-site, significantly boosting diagnostic and treatment efficiency. Similarly, in the world of Go, the "AlphaGo" team doesn’t include any professional Go players; instead, its core members are scientists and engineers with backgrounds in artificial intelligence. Today, AlphaGo has moved far beyond the stage of being "fed" vast amounts of game records—it now relies solely on deep learning to continuously improve and evolve.
However, meteorology differs from both medicine and Go in one key aspect: Medicine studies the human body as a closed system, while Go, though internally complex, still follows relatively simple rules. Meteorology, on the other hand, involves an even more intricate system, influenced by countless factors. For instance, even if conditions like water vapor and humidity are perfectly in place for rainfall, without "condensation nuclei" in the air, rain simply won't fall. Similarly, in the case of sandstorms, winds that are too strong will only cause them to "race" through the stratosphere—preventing the formation of actual dust storms altogether.
Therefore, "intelligence" does not mean forecasters are completely "useless" in this process. Forecasters excel thanks to their extensive experience and keen understanding of critical weather patterns. In the early stages of intelligent forecasting, forecasters' years of expertise can be used to "train" machines and models. However, as models become increasingly sophisticated and AI capabilities continue to advance—leaving less room for human intervention—some forecasters will shift their focus toward research and development of new technologies, while others will primarily transition into meteorological service roles, such as providing guidance on key, turning-point weather events.
Currently, the relevant team from China Meteorological Administration has already collaborated with Tsinghua University, the Chinese Academy of Sciences, and other institutions to advance research and development in artificial intelligence technologies. Meanwhile, the national-level meteorological agency has jointly established a prototype system team for intelligent forecasting services. Additionally, a group of seasoned chief forecasters from the National Meteorological Center, along with young, research-oriented forecasters and IT engineers, has come together to form a cutting-edge big data and AI-driven forecasting team. These initiatives will not only significantly boost the development of intelligent grid-based forecasting but may also bring unexpected breakthroughs to weather forecasting services.
Service Section
From June 21 to 24, Beijing experienced average rainfall of 92 millimeters across the city. Yet, for those who received warnings around the same time, the heavy rain didn’t arrive simultaneously in their immediate vicinity. On the evening of the 21st, rain first began in Hebei, Tianjin, and southern Beijing, sparking widespread rumors almost immediately. Then, on June 22nd, people endured another long day of anticipation—until late that night, when the torrential downpour finally arrived as promised.
During this process, it’s easy to see that while weather forecasts are accurate, they don’t yet fully align with public needs—some areas are already experiencing rain, while others haven’t seen a drop yet. Bridging this gap between forecasts and public demand is precisely the challenge that intelligent grid forecasting aims to address. The initial goal of intelligent grid forecasting is to divide the entire country into 5-kilometer-by-5-kilometer grids, updating forecasts every hour based on this detailed grid system. And if significant weather events occur, the update frequency will even increase further to ensure timely and precise information for everyone.
In other words, if it’s still the same rain event, no matter where in Beijing you are, the weather forecast you receive will be specific to the 5-kilometer grid area you’re currently in. For instance, if you’re located in a “grid” on the southern side, the forecast you get will differ from what someone in a “grid” farther north—or even in a nearby grid to the southwest or southeast of you. All you need to do is prepare yourself accordingly before the rain arrives in your exact location. From another perspective, this approach also helps prevent unnecessary panic and keeps rumors from spreading uncontrollably.
Some might ask: With such fine-grained resolution, can accuracy still be guaranteed? From the perspective of meteorological science, at the same technological level, increasing both temporal and spatial resolution typically leads to a decline in forecast accuracy. However, during the pilot phase and trial operation of the intelligent grid system, the meteorological authorities have prioritized ensuring that the accuracy of the smart grid forecasts—not only matches but actually surpasses—that of previous station-based forecasts.
In fact, despite the risk of slightly reduced forecast accuracy, the meteorological department is determined to shift from traditional forecasting methods and develop high-spatio-temporal-resolution gridded forecasts—driven by the very needs of weather services. Only in this way can we finally break the puzzling phenomenon of "accurately predicting rain elsewhere, but missing it entirely where I am."
Of course, the smart grid forecast doesn’t just provide you with rainfall predictions—it actually delivers precise weather services tailored to your computer’s IP address and mobile device location. Beyond rain, it also includes information on wind, temperature, humidity, cloud cover, visibility, and more.
Compared to previous forecasts, the gridded forecast will be entirely fine-scale, quantitative, and digital—providing significant support for forecasters in issuing impact forecasts for severe weather events. Leveraging this advanced, intelligent grid, forecasters can dramatically improve the accuracy of their predictions regarding downstream impacts from geological hazards, torrential rain-induced flooding, heatwaves, and prolonged droughts.
Additionally, thanks to the more detailed grid forecasts, the weather-related services you enjoy—whether in agriculture, transportation, or tourism—will become even more precise. For instance, in maritime meteorological services, the refined ocean grid forecasts provide ships at sea with highly accurate, route-specific weather information, offering robust data support for safe and secure ocean navigation.
With the advancement of meteorological technology, the figures of 5 kilometers and 3 hours are steadily shrinking—meaning that grid-based forecasts will become increasingly precise. Starting from last year, the meteorological authorities have also been developing global grid forecasts. As a result, even if you travel abroad or conduct business in countries along the Belt and Road Initiative, you’ll still be able to enjoy the thoughtful weather services provided by China’s meteorological agencies.
Weather,Precision optimization
Annual Gala | Auspicious Dragon Brings in Spring, Huateng Soars on the Wind
As the Auspicious Dragon ushers in spring, HuaTeng soars boldly on the winds of change
2024-02-03
The camp is officially open! HuaTeng Microwave's "Young Seed Training Camp" kicks off today!
To continuously build a dynamic organization, cultivate outstanding talent, and develop a strong pipeline of future leaders
2023-12-20
Huateng Microwave Debuts at the 2023 China Meteorological Modernization Technology Expo
On March 29, the three-day "2023 China Meteorological Modernization Technology Expo,"
2023-03-31
For any cooperation or support please contact:
Bella Yang
E-mail: yangqing@htmw.cn / xinqing_2011@foxmail.com
Tel/Whatsapp: +86 188 1011 7965
Follow the official account