How to make flood forecasts blow the whistle in advance but never leave ten
In 1998, China had only 3,300 flood reporting stations, and now it has grown to more than 120,000. Now, with the help of the hydrological information exchange system, it only takes 10-15 minutes to gather information from more than 120,000 flood reporting stations; it adopts meteorological and hydrological forecast coupling, With human-computer interaction technology and other methods, the accuracy of our country has been continuously improved, and the accuracy of forecasting major rivers in the south can reach 90%.
Our reporter Tang Ting
Since June, heavy rains and floods have occurred frequently in southern my country, and floods have been severe in some areas. For a flood on the way, predicting the time and peak value of its transit as early as possible is undoubtedly of great significance for the early deployment and formulation of a defense plan.
"On July 2nd and 17th, the Yangtze River experienced floods No. 1 and No. 2 this year. We made accurate flood peak forecasts two days in advance. On July 17th, Taihu Lake had over-standard floods. We did it 5 days in advance. Out of the pre-judgment that the Taihu Lake may exceed the standard flood, an accurate water level forecast was made 2 days in advance." Liu Zhiyu, deputy director of the Hydrology and Water Resources Monitoring and Forecasting Center of the Ministry of Water Resources, introduced in an interview with a reporter from the Science and Technology Daily. This year, as of July 22 , The national hydrological department issued a total of 1,034 rivers and 1,579 cross-section operation forecasts for more than 207,000 stations, providing strong support for active flood prevention.
So, how is flood forecasting achieved? How to simulate and derive the process from rainfall to flood formation through related models? A reporter from Science and Technology Daily conducted an interview on this.
Simulation of flood formation
The rapid increase in river water volume caused by heavy rains accompanied by a sharp rise in water levels is called torrential floods. Most of the major floods in the rivers of our country are torrential rain floods, which mostly occur in summer and autumn, and may also occur in spring in some areas in the south.
“Scientists establish a flood forecasting system to simulate the process from rainfall to flood formation, and then predict and forecast the development trend of floods.” said Zhang Dawei, a professor-level senior engineer at the Disaster Mitigation Center of China Institute of Water Resources and Hydropower Research: “How much rain does it rain? How rainfall forms runoff on the ground and how the runoff moves on slopes and gullies after it is formed are all factors that need to be considered in flood forecasting."
"When rainfall falls on the ground, it first moves vertically, after deducting losses such as evaporation and filling depressions, if the soil infiltration capacity is exceeded, super-seepage runoff will be formed on the ground. The water that seeps into the soil moves in the soil and continues to infiltrate the ground. After the soil water is full, the excess water will form full runoff on the ground. All surface runoff flows into the river network along a certain slope, soil water and groundwater. This is the general process of rainfall runoff in the basin." Liu Zhiyu said.
The surface runoffs formed in various parts of the river basin gather at the observation section of the river course according to its distance. When the nearby surface runoff reaches this section, the river flow begins to increase, and the water level rises accordingly. This is when the flood rises. As the surface runoff in the distance of the river basin flows into the river one after another, the flow and water level continue to increase. When most of the high-intensity surface runoff converges on the observation section of the river, the flood flow increases to the maximum.
It is understood that the process of deducting the loss from rainfall and turning into net rain is called the runoff process; the net rain flows into the river network along the slope and then flows through the river network to the drainage section of the basin. This process is called the basin flow process.
Combined with the rainfall data in the basin, hydrological forecasters can forecast the basin runoff and confluence. Then combined with runoff calculation and confluence calculation to calculate rainfall runoff in the basin, there are generally two common methods. One is to use the rainfall-runoff correlation graph to calculate the net rainfall, and then use the unit line and other confluence method to calculate the runoff process at the outlet section of the basin. Another commonly used method is to use some complete hydrological conceptual models for calculations, such as the Xin'anjiang model and the northern Shaanxi model. The data that needs to be input are mainly rainfall data and previous soil water content data.
After the rainfall runoff merges from the tributary into the main stream, it needs to be deduced and predicted through the flood forecast of the river section. "The commonly used methods at this stage include the corresponding water level (flow) method and the Muskingen method. The input data is the water level (flow) process data of the upstream section." Zhang Dawei introduced. In addition, the hydrodynamic model can also be used to calculate the flood evolution process, but the hydrodynamic model requires very detailed river topographic data, and this method cannot work in areas with incomplete data.
Real-time correction to improve forecast accuracy
The flood forecasting model basically determines the relevant parameters first, and then combines the observed historical hydrological data to forecast the flood. Such forecasting schemes will inevitably produce certain errors in real-time forecasting.
When analyzing the causes of errors, Zhang Dawei believes that the hydrological system of the river basin is a very complex giant system, and the models or methods used for forecasting usually adopt a series of assumptions and generalize the real world conditions; rainfall station Monitoring data or rainfall data obtained by other means cannot fully reflect the true temporal and spatial distribution of rainfall; many small water conservancy projects in the basin may not be able to reflect in the model in time after major changes in the scheduling or underlying surface.
"In order to further improve forecast accuracy, forecasters usually use rainwater information flood and hydrological emergency monitoring information to make real-time corrections to flood forecasts to improve forecast accuracy as much as possible." Liu Zhiyu, for example, currently has 170 locations in the piedmont plain of North China. Multiple emergency monitoring points can make rolling correction forecasts based on upstream monitoring information to improve the accuracy of downstream river forecasts. The application results on Zhanghe and Hutuohe rivers show that this "reporting by measurement and supplementary reporting" technical means can increase the accuracy of flood forecasting by about 10% on average.
In fact, improving the accuracy of flood forecasting and extending the forecast period have always been the pursuit of hydrological forecasters. In recent years, the Ministry of Water Resources has taken strong measures to promote the development of hydrological forecasting technology. Liu Zhiyu introduced that in 1998, there were only 3,300 flood reporting stations in China, and now it has grown to more than 120,000. In 1998, it took two hours to collect the information from 3,300 flood reporting stations with the use of telegraph transmission, but now it is relying on water regime information exchange. The system collects information from more than 120,000 flood reporting stations in just 10-15 minutes. Using methods such as meteorological and hydrological forecasting coupling and human-computer interaction technology, the accuracy of flood forecasting in China has been continuously improved, and the accuracy of forecasting for major southern rivers can reach 90%.
Liu Zhiyu said that at present, technologies such as multi-source information fusion, data assimilation, ensemble forecasting, artificial intelligence, big data, and distributed simulation have made great progress, and will be further applied to flood forecasting business in the future to improve flood impact forecasting and risk warning capabilities , Provide strong support for flood and drought disaster prevention.