Meiyu period 222

July 8 -15, 222.

Because the temperature in each region is different, the time to enter and leave the plum blossom in 222 is also different. However, plum blossoms usually enter in mid-June and leave in the first half of July, lasting about 2 days. But there are also cases of late recruitment and late departure of plum blossoms. For example, in the Meiyu area in 22, plum blossoms enter the plum blossom early and leave the plum blossom late, lasting for a long time. At that time, Zhejiang officially entered the rainy season at the end of May, ten days earlier than before.

Generally speaking, the rainy season in 222 will start in early June and end in early July, lasting about 2 days. It is expected that all localities will officially enter the plum blossom around June 1 this year, and the plum blossom time will be in mid-July. When the rainy season comes, we must pay attention to things at home, check more and don't get moldy.

Taizhou, Jiangsu _ Meiyu in 222

1. When did Jiangsu enter Meiyu in 222?

The rainy season in Jiangsu officially entered May on June 23rd, 222. According to the latest press conference held by Jiangsu Meteorological Observatory and Nanjing Meteorological Observatory, it was announced that Nanjing officially entered the rainy season on June 23rd. In addition, the area south of Huaihe River in Jiangsu Province is also expected to enter Mei on June 23, so the rainy season in Jiangsu this year is Thursday, June 23.

1. Is Jiangsu a late bloomer this year?

it belongs to late plum blossom. Because the average plum blossom day is June 19, this year's plum blossom day is June 23, which is a bit late. Because the Meiyu belt swings from north to south, there are many strong convective weather, with obvious intermittent precipitation and staged high temperature. At the same time, Huaibei will also enter a rainy period from June 23.

2. What's the weather like in rainy season in Jiangsu in May this year?

According to the latest introduction by the chief forecaster of Jiangsu Meteorological Observatory, the high temperature weather in our province will continue in the early rainy season this year, and strong convective weather will become more frequent. After June 24, it is expected that there will be short-term heavy precipitation, thunderstorms and even hail weather in the central and northern parts of Jiangsu Province, so more precautions should be taken. It is estimated that there will be two obvious rainfall processes in Jiangsu in the coming week, from the night of 22nd to 24th and from 27th to 28th respectively. From 23 to 26, there were strong convective weather such as short-term heavy precipitation, thunderstorms and small hail. There is high temperature above 35℃ in the north-central part of the 22nd, along the Yangtze River, in southern Jiangsu, along Huaihe River and Huaibei River from 24th to 25th.

3. What is the amount of plum rain in Jiangsu this year?

the average amount of plum rains is 2-26mm. In the meantime, the average rainfall in Huaibei area is 17-23mm, which is more than normal.

2. When will Jiangsu plum blossom in 222?

According to the latest introduction by the chief forecaster of Jiangsu Meteorological Observatory, it is estimated that May will blossom in mid-July 222. The duration of plum rains in Jiangsu in recent years is as follows:

1. In 221, Jiangsu Meteorological Observatory issued a forecast of plum rains, and some areas south of Huaihe River officially entered plum rains on June 13th.

2. In 22, the rainy season in Jiangsu began on June 9th and ended on July 21st, and lasted for 43 days.

3. In 219, Jiangsu entered the rainy season from June 18th to July 21st. The total length of the Meiyu period is 33 days, which is longer than the normal Meiyu period of 23 to 24 days.

4. The rainy season in Jiangsu in 216 lasted for 32 days.

Generally speaking, the rainy season in Jiangsu Province in 222 officially entered the Mei season on June 23rd, and generally came out in July. According to the latest weather forecast in Jiangsu Province, plum blossoms will appear in early July this year.

Has the Huangmei Day passed in Wuxi? 222

The rainy season in 222 appears from late May to late June. Because the annual rainy season occurs during the two solar terms of mango and light summer, this year's mango is June 6 and the light summer is July 7.

Therefore, it is predicted that the rainy season in the middle and lower reaches of the Yangtze River in China will begin in early June, but according to the time of plum blossom in previous years, it is not uniform and will be separated by a few days. For example, Shanghai entered Mei on June 1th in 221; Suzhou, Jiangsu entered Mei on June 1th, and the area south of Huaihe River entered Mei on June 13th.

attention.

plum entry criteria in p>222: the average temperature exceeds 22℃ for five consecutive days, and only four rainy days can be considered as plum entry. According to the recent weather forecast in Shanghai, it has not officially entered Mei, and the lowest temperature is still between 16 and 18 degrees.

epidemic trend chart from p>22 to 222

big data epidemic observation: has the national epidemic peak passed?

Research on Tengjing Macro Finance Trend

222-12-2317: 23 From Beijing

Tengjing Macro Express

December 23, 222

Big data epidemic observation: Has the national epidemic peak passed?

—— Based on Tengjing AI high-frequency simulation and prediction

Tengjing high-frequency and macro research team

Highlights of this issue:

In view of whether the prediction is accurate or not and whether the national epidemic has peaked, we have added daily data of subway passenger traffic in 28 cities for auxiliary judgment. The lack of non-netizen samples may lead to biased prediction results.

Big data is not perfect, and the application of big data for macroeconomic forecasting is not perfect. We analyzed why Google flu trend failed. The reasons may include: the media's extensive coverage of Google's flu trend has led to changes in people's search behavior, which in turn will affect the forecast results of GFT.

At present, the epidemic situation in China may not have reached its peak, but the peak process may be advanced. With the help of subway passenger traffic data, we can judge that Beijing, Shijiazhuang, Wuhan, Chongqing and other cities have passed the peak of the epidemic, while Chengdu, Tianjin, Changsha, Nanjing, Xi 'an and other cities have not yet reached the peak.

1. Is the forecast accurate? Mutual verification between expectation and reality

In the last issue of the report "Big Data Epidemic Observation: Central Cities Take the Lead in Heading for the Peak", we analyzed and gave the prediction that the epidemic situation in some cities in Beijing and Hebei has reached an "inflection point", and Chengdu, Kunming and other cities will gradually peak. According to Baidu search index data, Beijing Baidu's "fever" search index continued to decline, and the "cough" search index peaked after "fever", which basically confirmed the prediction of our model. However, we also noticed that the nationwide "fever" index peaked on December 17, 222. Does this mean that the national epidemic situation peaked? If so, this data is different from the judgment of some epidemic prevention experts before and after the Spring Festival. Some experts also believe that the national epidemic may not have reached its peak, but the process has been shortened.

However, according to ByteDance's "massive calculation", Tik Tok's "fever" search index peaked on December 17th, but the headline "fever" search index is still fluctuating upward. In the prediction of Zhihu's "Data Emperor", which is widely spread in the circle of friends, most provinces and cities have reached the peak of infection around December 2, 222. So, what many researchers want to confirm is whether the new infections in a single day nationwide have reached the peak on December 23, 222. Some people think that the prediction is very accurate, which is more consistent with their perception of the epidemic on the Internet these days. Some people think it's not accurate. They think that all the relatives and friends around them are yang, and the progress bar is less than half. There is a big difference between the personal feeling and the predicted results.

At the same time, we noticed that around December 16th, 222, the "fever" search index of almost all cities and provinces in China ushered in a pulse growth of "first rising and then suppressing", and the subsequent daily data was never higher than the value on the 16th. Does this mean that the most difficult phase of the epidemic has passed? Through data mining and modeling analysis of Baidu and headline epidemic disease search engine data, it can provide important reference for the future trend judgment of epidemic situation. However, we understand that in order to quantitatively evaluate the progress of the epidemic, more data need to be introduced.

Because there is no authoritative data as a reference, the predictions of various epidemic situations are only based on intuitive, inferential or deductive models with parameters, and the predictions are inaccurate, and there is no objective authority to compare the results. Therefore, it is difficult to objectively measure the accuracy of the predictions. All the audience and readers who participated in this prediction can only verify the prediction results through microscopic data and the spread degree of the surrounding epidemic situations. The order of infection among different groups in a city and the rhythm of peak infection in different cities will have different understandings of the accuracy of the predictions.

the model has limitations, the applicability of logical assumptions, and the lack of authoritative data for verification. Isn't it necessary to predict? Thomas kuhn and karl popper launched the most influential confrontation on the concept of "philosophy of science" in the 2th century. They all questioned the basic premise of science from the philosophical point of view in their own way. Kuhn's "The Structure of Scientific Revolution" points out that even if the results predicted by the existing paradigm have counterexamples in reality, the existing scientists will not think that there is something wrong with its paradigm; Only when a new scientific paradigm that can replace the existing paradigm appears and a certain number of counterexamples are reached can the existing scientific paradigm be falsified and the scientific revolution occur. From a critical point of view, denying the forecasting process is also a process of discovering new forecasting methods.

The most famous view of philosopher karl popper, admired by george soros of Quantum Fund, is that science is conducted through "falsifiability"-people can't prove that the hypothesis is correct, or even get the evidence of truth through induction, but if the hypothesis is wrong, they can refute it. According to Popper, only theoretical systems that can be falsified by experience should be given real scientific status. Therefore, Popper advocates bold assumptions, and constantly tries to make mistakes and revisions by falsification, instead of putting forward hypotheses, and then looking for evidence to support his theory everywhere. Falsification is also a way of thinking that Soros has been advocating and practicing.

Second, the subway passenger volume is an important auxiliary observation index for the peak of the epidemic

Therefore, we start from the epidemic and return to the economy to verify the peak of the epidemic from multiple dimensions. The passenger volume of subway is undoubtedly a good observation index. The passenger volume of a city with subway is affected by several factors: 1. Travel control; 2. Travel willingness; 3. The convenience of subway.

From the data point of view, Beijing and Shanghai, as the two cities with the highest number of subways in China, also have the highest average daily passenger traffic. The subway data reflect the level of the epidemic, and the daily data of subway passenger traffic is delayed by 1-3 days, which is relatively timely. From the point of view of data collection, the subway data comes from the automatic collection of Internet of Things equipment, with little influence from manual intervention, and the data is fully objective, which can be regarded as the second major epidemic.

Figure: Shanghai subway passenger volume

▲ Data source: Wind, Tengjing AI economic forecast

The above figure shows the passenger volume data of Shanghai subway from December 219 to the present, with the Wuhan epidemic in early 22, the Shanghai epidemic in April 222 and the national epidemic in December 222 being more obvious. As the subway passenger volume follows the principle of high on Mondays to Fridays and low on Saturdays and Sundays, the daily data information is somewhat redundant. Later, we can filter short-term intraday data fluctuations by comparing the average data of Zhou Du.

Figure: Shanghai subway passenger volume

▲ Data source: Wind, Tengjing AI economic forecast

Comparing the passenger volume of Beijing subway, it can also be seen that in April 222, Shanghai subway stopped for about seven weeks, although Beijing did not stop, the average passenger volume of Zhou Du subway decreased from 8 million in the past three years to less than 1 million. It is worth noting that the passenger volume of Beijing subway after September 222 is obviously lower than that of Shanghai. On the one hand, it is an epidemic, on the other hand, Beijing subway needs to check the nucleic acid for 72 hours on the whole network, which was further shortened to 48 hours on November 24, and this policy was lifted on December 5.

Figure: Beijing subway passenger traffic

▲ Data source: Wind, Tengjing AI economic forecast

Figure: The 7-day moving average of subway passenger traffic in ten major cities is highly consistent

▲ Data source: Wind, Tengjing AI economic forecast

Based on this data, we think that the epidemic peak in Beijing has passed, but the overall epidemic peak in China is not as shown by Baidu search index and headline index. We have established a four-stage data model to help verify whether cities have reached the peak. As shown in the following figure, the passenger traffic of Beijing, Wuhan, Chongqing, Shenyang, Shijiazhuang, Lanzhou and Kunming subways has stabilized and rebounded, and is currently in the fourth stage; Chengdu, Tianjin, Changchun, Zhengzhou, Guangzhou, Xiamen, Shenzhen, Xi 'an, Shanghai, Nanjing and other cities are still in the third stage of reaching the peak. Because the moving average may bring data lag, later, we did a test with real data.

figure: epidemic spread process

▲ data source: tengjing AI economic forecast

figure: subway passenger traffic in some cities in China

note: the top ten cities refer to: Beijing, Shanghai, Guangzhou, Chengdu, Nanjing, Wuhan, Xi' an, Suzhou, Zhengzhou and Chongqing, the same below.

▲ data source: Wind, tengjing AI economic forecast

in the progress of the epidemic in days, if the subway travel data picks up on that day, we should mainly look at two data, the first is the year-on-year and the second is the ring-on-ring ratio.

according to the daily data, the subway trips in Beijing are in the upward phase, both month-on-month and year-on-year, which is consistent with the judgment of peaking. Other cities that may peak are Wuhan, Chongqing and Chengdu. The passenger volume of Shanghai, Guangzhou, Nanjing, Suzhou, Xi 'an and other subways continues to decline, which indicates that the epidemic is still in the process of reaching its peak.

Figure: Subway passenger traffic in some cities in China

▲ Data source: Wind, Tengjing AI economic forecast

Due to the serious year-on-year decline in subway passenger traffic data, we judge that the epidemic situation in Shanghai, Guangzhou, Nanjing, Xi 'an, Suzhou, Zhengzhou and other cities is still in the process of reaching its peak, while Beijing, Wuhan and Chongqing have turned positive year-on-year, and it is expected that the epidemic peak has passed.

Figure: Subway passenger traffic volume in 28 cities and Zhou Du year-on-year

▲ Data source: Wind, Tengjing AI economic forecast

Third, how do expectations interact with reality?

There are many experiences after the liberalization of epidemic control. No matter the rhythm of epidemic peak, the impact on consumption and labor participation rate, there are many countries that can refer to it. This undoubtedly gives us some expectations, and the liberalization of 1.4 billion people is different from that of medium-sized population countries. Domestic infectious disease experts also said in various media that the epidemic will peak in the first quarter of next year before and after the Spring Festival, releasing such a signal that the future will peak. But from the perception of Beijing and most cities, it seems that the epidemic peaked earlier than our cognition, so what will go wrong?

Policy indicator failure: Goodhart's Law

When most Internet participants know that Baidu search index can indirectly represent the epidemic situation, it may be inaccurate. To some extent, it is the embodiment of Goodhart's Law in the epidemic situation. Goodhart's law comes from the statement of Charles goodhart, a British economist, which means that when a policy becomes a goal, it will no longer be a good policy. One of the explanations is that once a social indicator or economic indicator becomes an established goal to guide macro-policy making, it will lose its original information value.

There is no doubt that the Baidu Epidemic Index is still effective with a high probability when most people don't know its importance. The connotation logic is that the search volume of big data indirectly reflects the spontaneous online search behavior of most residents, and "fever" search is the same thing as being positive and symptomatic to some extent. However, in the case that both the official media and the self-media are reporting, this indicator will lead to more searches, and these searches have nothing to do with the epidemic itself, but Internet flows.