Ink weather, more possibilities for mining weather data

Hungry, Baidu take-out and other take-out platforms are based on the solution provided by the ink weather to carry out manpower deployment and adjust delivery time. For companies and individuals, the value of short-term forecasts is reflected in the fact that “no rain for the next hour” is far more valuable than “it may be rain today”.

"On the day of fog on the sea, the horizon was misunderstood."

The heavy fog in Hainan during the Spring Festival this year has still made many tourists still vividly remembered. This once-in-a-year 67- year-old fog once caused tens of thousands of vehicles to remain in the sea, and the price of air tickets from Haikou to major domestic cities even soared to 10,000 yuan.

Although the impact of the extreme weather caused by the disruption of traffic has passed, the resulting thinking on the status quo of domestic meteorological services continues.

All along, since the weather forecast involves national security, it must be uniformly issued by the official authoritative meteorological department. After years of development, China's meteorological business has made considerable progress. According to statistics, there are currently nearly 60,000 automatic weather stations in China , covering all cities in the country and 96% of rural areas. In 2017 , the China Meteorological Administration was recognized by the World Meteorological Organization as the World Meteorological Center.

However, there are certain "shortcomings" in the weather forecast of public welfare nature. Taking Hainan's heavy fog as an example, Zhu Dingzhen, a member of the CPPCC National Committee and chief engineer of the Meteorological and Film Center of the Public Meteorological Service Center of the China Meteorological Administration during this year’s National Two Conferences, once stated that this event reflects from one aspect the existence of public weather-related weather forecasts mainly in The “face” forecast cannot meet the individual needs of the people and it is difficult to help determine the possible impact of weather conditions on travel, and calls for the meteorological industry to absorb the strength of various market players.

In fact, current meteorological service platforms such as ink-based weather are playing the role of “short board” for public weather forecasting, meteorological information interpreters based on meteorological data, and consumer meteorological product providers.

Meteorological data of Ink Weather, in addition to cooperation with professional meteorological agencies ( such as GFS of the United States Meteorological Agency , EC of the European Meteorological Administration, and China Meteorological Administration ) , obtains data of the order of 500 G per day , in addition to the time-lapse community that went online in 2012 . It has also become one of the main sources of meteorological data. Ink weather has 500 million users. Each day, there are hundreds of thousands of weather photos uploaded in the time scene community. This way, the massive real-time and real-world meteorological data and picture resources have reached 100 million.

The long-term accumulated meteorological industry knowledge and abundant data also provide suitable application scenarios for artificial intelligence. Ink weather uses cutting-edge technologies such as artificial intelligence to apply to traditional meteorological fields. Based on massive data and meteorological knowledge, machine learning can explore the historical trends of these meteorological data and analyze meteorological data to achieve minute-level, kilometer-level short-term forecasts. As the amount of data increases, multi-dimensional statistical analysis will make the forecast more accurate. The data analysis of ink weather is mainly two parts: one is the short-term forecast of 0-2 hours and the short-term forecast of 2-8 hours, and the other is the mid- to long-term forecast of 8 hours to 15 days.

Among them, the short-term forecast function complements the “short board” of forecast on the “surface” of the weather forecast in terms of personalization and accuracy. Users can arrange trips in advance according to the short-term forecasting function of ink weather, and even get a more accurate weather condition in the next two hours. Imagine if you arranged a trip on the southeastern coast, please take a good vacation and buy an airline ticket to book a good hotel. However, because of a sudden typhoon, all the plans are disrupted. Is this short-term service experience more effective? What about value?

Of course, the mining of weather data is not just about this. Meteorology is closely related to food, clothing, and shelter. The provision of information based on people's daily lives is also of great value. Taking the ink weather as an example, it also provides users with more refined services in terms of air quality, clothing, and travel. For example, the AQI index based on fog and incendiary queries helps users understand the PM2.5 value so as to protect them; The on-line allergy index helps allergies avoid allergies; the flowering forecasting function provides viewing time for many travel enthusiasts.

It is small to all aspects of people's lives, and it is related to corporate business. Precise short-term weather forecasting can also help many companies to avoid losses caused by weather factors. For example, the famous 7-11 convenience store will conduct detailed product management by analyzing weather information. They found that sales of donuts are related to the weather. Weather factors may even affect chocolate taste, creamy taste, or ordinary sweetness. Circle sales are high. Through this meticulous single-product management, 7-11 convenience stores keep stocks to a minimum while not being out of stock, increasing profits to a maximum.

In China, with the rise of online dating vehicles and take-away platforms, more and more companies are also realizing the influence of weather factors on business operations. For example, like hungry, Baidu’s takeaway platforms like Baidu have cooperated with meteorological service platforms such as ink-based weather, and according to the solutions provided by them, human resources are deployed according to weather conditions, and delivery times for platforms are adjusted to avoid weather-related problems. Loss.

Source: "China Meteorological News" April 16, 2018 third edition.