Urban typhoon disaster rapid prediction method based on intelligent downscaling

Abstract

The invention discloses an urban typhoon disaster rapid prediction method based on intelligent downscaling. According to the method, an intelligent downscaling model is constructed on the basis of a wind field forecasting result output by a coarse-resolution typhoon numerical forecasting mode in combination with underlying surface information, the downscaling model comprises two levels, and a high-resolution local maximum wind field simulation result considering the influence of regional environment characteristics is generated; and on the basis of constructing a building vulnerability assessment model to assess the influence of typhoon on the building, comprehensively analyzing the damage risk of the target area, and completing risk and disaster prediction. The building disaster assessment model is designed in combination with the high-resolution meteorological data, the influence of the extreme wind speed of the typhoon on the building can be efficiently assessed, and better support is provided for urban disaster prevention decision making.

Description

Technical Field

[0001] The present invention belongs to the fields of meteorological disaster prediction and urban disaster prevention, and particularly relates to a rapid prediction method for urban typhoon disasters based on intelligent downscaling. Background Art

[0002] Typhoon is one of the main natural disasters that cause damage to urban buildings. Conducting risk disaster prediction of urban buildings under typhoon weather conditions can provide early warnings of the potential risks of building damage caused by typhoons, help reduce the losses caused by disasters, and ensure the safety of people's lives and property.

[0003] Currently, the methods for urban building risk assessment mainly rely on the coupling of meteorological models and building disaster assessment models. However, the spatial resolution of traditional meteorological numerical prediction models is difficult to accurately describe the wind field characteristics in the urban microenvironment, especially in areas with complex terrain and high-density buildings. On the one hand, high-resolution numerical models require a large amount of computing resources, increasing the prediction cost and complexity; on the other hand, due to the long computing time, it is difficult to meet the demand for rapid prediction, and the response speed is slow when an actual typhoon comes, making it difficult to provide timely support for urban disaster prevention and mitigation. Moreover, the accuracy of classical building disaster assessment models designed based on coarse-resolution meteorological data also needs to be improved in the risk disaster prediction of typhoons. Summary of the Invention

[0004] The purpose of the present invention is to provide a rapid prediction method for urban typhoon disasters based on intelligent downscaling to solve the problems of insufficient spatial resolution, low computational efficiency, or insufficient physical characteristics considered in the prior art (such as traditional assessment models usually only consider wind and do not consider the additional impact of precipitation), resulting in insufficient disaster response effectiveness.

[0005] The present invention includes the following steps:

[0006] Obtain coarse-resolution typhoon meteorological forecast data;

[0007] Use the coarse-resolution typhoon meteorological forecast data to generate high-resolution wind field data and precipitation data through an intelligent downscaling model;

[0008] Use the high-resolution wind field data to generate high-resolution surface wind pressure information of urban buildings;

[0009] Design a building risk assessment model to rapidly predict the damage risk of urban buildings under typhoon weather based on the surface wind pressure data of urban buildings and high-resolution precipitation data;

[0010] Among them, the intelligent downscaling model includes at least two levels of downscaling models, and the at least two levels of downscaling models are used to generate local maximum wind field simulation results under extreme weather conditions considering the influence of regional environmental characteristics.

[0011] Preferably, the coarse-resolution typhoon weather forecast data includes wind speed, wind direction, temperature, air pressure, humidity and precipitation data.

[0012] Preferably, the intelligent downscaling model includes:

[0013] A first-level downscaling model for converting coarse-resolution typhoon weather forecast data into first-level fine-resolution meteorological data;

[0014] A second-level downscaling model for converting the first-level fine-resolution meteorological data into second-level fine-resolution meteorological data.

[0015] Preferably, a first-level fine-resolution meteorological numerical prediction model is adopted, and the model resolution needs to reach 6 km. The first-level fine-resolution meteorological data is obtained through numerical simulation.

[0016] Preferably, when training the first-level downscaling model, its input is: historical typhoon event data and coarse-resolution meteorological data for the recent complete two years; the output is: first-level fine-resolution meteorological data for the corresponding time period.

[0017] Preferably, a second-level fine-resolution meteorological numerical prediction model is adopted, and the model resolution needs to reach 1 km. The second-level fine-resolution meteorological data is obtained through numerical simulation. The second-level fine-resolution meteorological data is used for vulnerability assessment corrected based on building data; the second-level fine-resolution meteorological numerical prediction model is a meteorological prediction model that can follow the movement of the typhoon, and the grid center position of this model is the typhoon center position in the first-level fine-resolution meteorological simulation result.

[0018] Preferably, when training the second-level downscaling model, its input is: extract the moments when the maximum wind speed of the typhoon on land reaches more than 30 m / s, and use the first-level fine-resolution meteorological data at these moments as the input; the output is: second-level fine-resolution meteorological data for the corresponding time period.

[0019] Preferably, before generating the high-resolution wind pressure information on the surface of urban buildings, it also includes:

[0020] Constructing a city CFD model;

[0021] Constructing an AI urban building wind pressure model;

[0022] Optimizing the AI urban building wind pressure model;

[0023] In the steps of constructing the urban CFD model, the logarithmic curve of wind speed and height is fitted using the wind speed data obtained from the secondary downscaling model, the wind shear is calculated, and the theoretical corrected wind speed is calculated using the equivalent roughness of the building.

[0024] In the steps of constructing the AI urban building wind pressure model, the wind field pressure on the building surface obtained by simulating the urban CFD model is used as the output to establish the AI urban building wind pressure model.

[0025] Preferably, in the designed building risk assessment model, the fragility function is used to evaluate the damage probability of the building under the action of typhoon, and the additional influence of precipitation is considered to correct the fragility function.

[0026] Preferably, based on the probability distribution of damage to each building obtained from the corrected fragility function, the dominant damage level of all buildings in the area is calculated, and the area risk disaster level result is summarized.

[0027] Advantages of the present invention: The present invention provides a rapid urban typhoon disaster prediction method based on intelligent downscaling. This method combines high-resolution meteorological data to design a building disaster assessment model, which can efficiently evaluate the impact of typhoon extreme wind speed on buildings and provide better support for urban disaster prevention decision-making, and has important significance. Description of the Drawings

[0028] Figure 1 It is a flowchart of an embodiment of the present application.

[0029] Figure 2 It is a specific flowchart of step S6 in an embodiment of the present application. Detailed Embodiment

[0030] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and other obvious variations can be thought of by those skilled in the art.

[0031] Based on the wind field forecast results output by the coarse-resolution typhoon numerical prediction model and combined with the underlying surface information, the present application constructs an intelligent downscaling model, including a two-level downscaling model, generates a high-resolution local maximum wind field simulation result considering the influence of regional environmental characteristics, and on the basis of constructing a building vulnerability assessment model to evaluate the impact of typhoon on buildings, comprehensively analyzes the damage risk of the target area to complete the risk disaster prediction.

[0032] As Figure 1 shown, a rapid urban typhoon disaster prediction method based on intelligent downscaling provided by this embodiment includes the following steps.

[0033] Step S1: Coarse-resolution meteorological simulation, used to obtain coarse-resolution typhoon meteorological forecast data.

[0034] Conduct meteorological numerical simulations covering the Chinese region with a resolution of 0.25° (about 28 km) to obtain coarse-resolution meteorological data including wind speed V 0 , wind direction θ 0 , air temperature T 0 , air pressure P a0 , humidity Hu 0 , precipitation Pr 0 data.

[0035] Step S2: First-level fine-resolution meteorological simulation, used to obtain first-level fine-resolution meteorological data.

[0036] According to the typhoon-affected areas in China, construct an independent or multiple combined first-level fine-resolution meteorological numerical forecast models with a resolution of 6 km. Through numerical simulations, obtain first-level fine-resolution meteorological data including wind speed V 1 , wind direction θ 1 , air temperature T 1 , air pressure P a1 , humidity Hu 1 , precipitation Pr 1 data.

[0037] Step S3: Construction of the first-level downscaling model, used to convert coarse-resolution typhoon meteorological forecast data into first-level fine-resolution meteorological data.

[0038] Statistically analyze historical typhoons affecting China, select an appropriate amount of typhoon event time, and the recent complete two-year period. Use the coarse-resolution meteorological data (from Step S1) during this period as input, combined with the differences in surface information between first-level fine-resolution and coarse-resolution, including terrain elevation differences and roughness ratios. Use the first-level fine-resolution meteorological data (from Step S2) during this period as output to construct and train the first-level downscaling model. Achieve that the input of coarse-resolution meteorological data can obtain first-level fine-resolution meteorological data.

[0039] Step S4: Second-level fine-resolution meteorological simulation, used to obtain second-level fine-resolution meteorological data.

[0040] To effectively reduce the computational amount and ensure sufficient data in the typhoon area, establish a meteorological forecast model that follows the movement of the typhoon. The model grid range is usually set to 180 km × 180 km, and the resolution needs to reach 1 km. The center position of the model grid is the typhoon center position in the first-level fine-resolution meteorological simulation results (from Step S2 or S3). Through numerical simulations, obtain second-level fine-resolution meteorological data including wind speed V 2 , wind direction θ 2 and precipitation Pr2 。

[0041] Since the wind speed is affected by multiple meteorological elements, in the secondary fine-resolution meteorological simulation, complete primary fine-resolution meteorological data including wind speed V 1 , wind direction θ 1 , air temperature T 1 , air pressure P a1 , humidity Hu 1 , precipitation Pr 1 data are still required as input. However, at this time, only wind data and precipitation data need to be output for vulnerability assessment corrected based on building data.

[0042] Step S5: Construction of a secondary downscaling model for converting primary fine-resolution meteorological data into secondary fine-resolution meteorological data.

[0043] According to the selected influential historical typhoons, extract the moments when the maximum wind speed of typhoons on the Chinese mainland reaches more than 30 m / s. Use the primary fine-resolution meteorological data (from Step S2 or S3) at these moments as input, combine with terrain information, roughness information, and building information, and use the secondary fine-resolution meteorological data (from Step S4) during this period as output to construct and train a secondary downscaling model. Achieve obtaining secondary fine-resolution meteorological data from the input primary fine-resolution meteorological data.

[0044] Furthermore, the construction logic of the primary downscaling model in this embodiment is to perform mapping in a fixed area, which is mainly used to improve the resolution to capture convective effects and is a conventional practice. Since the secondary fine-resolution meteorological simulation requires accurately capturing the wind and precipitation characteristics in the typhoon area and requires a very high resolution, directly using the fixed-area downscaling method requires a large amount of computation, and the duration of typhoons in each grid area is not long, and the non-typhoon data in a large number of areas is redundant. Therefore, it is necessary to track the typhoon as the center and use a moving grid.

[0045] Step S6: Risk disaster prediction. Based on meteorological data, conduct building vulnerability assessment in combination with building information to complete the prediction of urban risk disasters. The specific steps include: S6-1 Construction of an urban CFD model, S6-2 Construction of an AI urban building wind pressure model, S6-3 Optimization of the AI urban building wind pressure model, S6-4 Building vulnerability assessment, and S6-5 Regional risk disaster assessment, as shown in Figure 2 。

[0046] Step S6-1: Construction of an urban CFD model.

[0047] Build a CFD model of the city with a grid range of approximately 5 km × 5 km and a resolution of approximately 5 m. The model uses the corrected wind data at the second-level fine resolution (from step S5), including the wind speed V 2f and the wind direction θ 2 as the initial conditions and boundary conditions, and outputs the wind field pressure P on the building surface.

[0048] In a preferred example, the specific correction method of the wind data is as follows:

[0049] 1. Use the wind speed data V 2 obtained from the second-level downscaling model (the wind speeds at different heights are different, let k be the layer number, and the total number of layers is N), fit a logarithmic curve of wind speed versus height, and calculate the wind shear α:

[0050]

[0051]

[0052]

[0053] where V is the velocity variable symbol and H is the height variable symbol.

[0054] 2. Use the equivalent roughness z B of the building to calculate the theoretical corrected wind speed V 2f1 :

[0055]

[0056] where z 0 is the environmental roughness.

[0057] 3. Combine the empirical parameter , such as taking 0.1, to calculate the corrected wind speed V 2f :

[0058]

[0059] Step S6-2: Construction of the AI urban building wind pressure model.

[0060] Since this embodiment mainly focuses on cities in fixed siting areas, significant short-term changes in the cities are usually not considered. Therefore, one or more sets of urban CFD models can be established according to the urban areas to be analyzed, and the meteorological data with secondary fine resolution for training the secondary downscaling model can be classified according to different typhoon levels to establish a constructed typhoon environment dataset. The wind data in this dataset is used as the input of the urban CFD model after data correction, and the wind pressure on the building surface simulated based on the urban CFD model is used as the output to establish an AI urban building wind pressure model, so as to quickly obtain the wind pressure on the urban building surface according to the results of the secondary downscaling model in the follow-up.

[0061] Step S6-3: Optimization of the AI urban building wind pressure model.

[0062] When the number and types of simulated urban areas are rich enough, the geometric parameters of the buildings in the area (such as height, width / length-width ratio, building orientation, relative building position, etc.) can be further used as features to directly input the meteorological data with secondary fine resolution and the building layout conditions, and directly obtain the wind pressure distribution on the urban building surface under the building layout conditions.

[0063] Step S6-4: Assessment of building vulnerability.

[0064] After obtaining the high-resolution wind pressure map, the fragility function is applied to evaluate the damage probability of the building under the action of typhoon. The specific process is as follows: According to the attributes of the building such as height, structural system, age, functional use, etc., the buildings are divided into limited categories (such as multi-story brick-concrete residential buildings, frame office buildings, high-rise steel structures, etc.), and each category of building is matched with a set of fragility functions.

[0065] The fragility function in this embodiment describes the probability that a building reaches or exceeds a certain damage level under a given disaster load (such as wind pressure), and the expression of its fragility function is as follows:

[0066]

[0067] Where P b represents the probability that the building damage level D reaches or exceeds level d k under a given wind pressure (unit: Pa); μ represents the median wind pressure when reaching damage level d k ; β represents the standard deviation of the lognormal distribution (reflecting uncertainty); Φ represents the cumulative distribution function of the standard normal distribution; the building damage level D has a total of 4 levels, including no damage, slight, moderate, severe, and complete destruction, and d k is a specific level (such as "moderate").

[0068] Heavy precipitation accompanied by strong winds will increase the likelihood of building damage. Therefore, in this embodiment, the classical fragility function is improved again, and a precipitation correction factor λ is designed. The value of this factor is related to the precipitation amount (data from step S5) and the building damage level D. For example, the default value of λ is 1. When the precipitation amount is greater than 25 mm, λ for slight, moderate, severe, and complete damage increases by 0.2; when the precipitation amount is greater than 50 mm, λ for moderate, severe, and complete damage increases by 0.3 again; when the precipitation amount is greater than 100 mm, λ for severe and complete damage increases by 0.5 again; when the precipitation amount is greater than 250 mm, λ for complete damage increases by 1. The specific values need to be adjusted according to the actual situation and regional conditions. The corrected fragility function P bf has the following expression:

[0069]

[0070] For each building, according to the corrected fragility function matched to its category, use the formula to calculate the damage probability of each level: Take the level corresponding to the maximum probability as the predicted damage level.

[0071] Step S6-5: Regional risk disaster assessment.

[0072] Based on the probability distribution of the damage of each building obtained from the corrected fragility function, then statistically calculate the proportion of the building group at each damage level on the regional scale. According to the damage level probability of each building, use the maximum probability method (select the level with the highest probability as the damaged level of the building) to calculate its dominant level, and summarize the dominant damage levels of all buildings in the region. This result is the regional risk disaster level result.

[0073] Since the prediction method in this embodiment does not require a large number of on-site investigations or physical modeling, it has three major advantages: high efficiency, low cost, and scalability. It can rely on existing data and models, without expensive on-site measurements or wind tunnel experiments, quickly complete the damage prediction of a large range of building groups, and can use remote sensing and building parameters to achieve the classification and risk prediction promotion of large-scale building groups. It is especially suitable for scenarios such as typhoon disaster emergency warning, risk census, insurance claim assessment, and urban planning support at the city level or regional level, providing scientific and efficient technical support for disaster prevention and mitigation.

[0074] The basic principles, main features and advantages of the present invention have been shown and described above. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and what is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, various changes and improvements will occur to the present invention, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A rapid prediction method for urban typhoon disasters based on intelligent downscaling, characterized in that: The method comprises the following steps: Obtain coarse-resolution typhoon weather forecast data; Using the coarse-resolution typhoon weather forecast data, high-resolution wind field data and precipitation data are generated through an intelligent downscaling model; Using the high-resolution wind field data, high-resolution urban building surface wind pressure information is generated; Design a building risk assessment model to quickly predict the damage risk of urban buildings in typhoon weather based on urban building surface wind pressure data and high-resolution precipitation data; Among them, the intelligent downscaling model includes at least two levels of downscaling models, and the at least two levels of downscaling models are used to generate local maximum wind field simulation results under extreme weather conditions taking into account the influence of regional environmental characteristics.

2. The method for rapid prediction of urban typhoon disasters based on intelligent downscaling according to claim 1 is characterized in that: The coarse-resolution typhoon weather forecast data includes wind speed, wind direction, temperature, air pressure, humidity and precipitation data.

3. A method for rapid prediction of urban typhoon disasters based on intelligent downscaling according to claim 1 or 2, characterized in that: The intelligent downscaling model includes: The first-level downscaling model is used to convert the coarse-resolution typhoon weather forecast data into the first-level fine-resolution meteorological data; The second-level downscaling model is used to convert the meteorological data with the first-level fine resolution into the meteorological data with the second-level fine resolution.

4. The method for rapid prediction of urban typhoon disasters based on intelligent downscaling according to claim 3 is characterized in that: A first-level fine-resolution meteorological numerical forecast model is used, and the model resolution needs to reach 6km. The meteorological data of the first-level fine-resolution is obtained by conducting numerical simulation.

5. The method for rapid prediction of urban typhoon disasters based on intelligent downscaling according to claim 4 is characterized in that: When the first-level downscaling model is trained, its input is: historical typhoon event data and recent coarse-resolution meteorological data for a complete two-year period; and its output is: first-level fine-resolution meteorological data for the corresponding time period.

6. The method for rapid prediction of urban typhoon disasters based on intelligent downscaling according to claim 3 is characterized in that: A second-level fine-resolution meteorological numerical forecast model is used, and the model resolution needs to reach 1 km. The second-level fine-resolution meteorological data is obtained by conducting numerical simulation. The second-level fine-resolution meteorological data is used for vulnerability assessment based on the revised building data; The second-level fine-resolution meteorological numerical forecast model is a meteorological forecast model that can move with a typhoon, and the grid center position of the model is the typhoon center position in the first-level fine-resolution meteorological simulation result.

7. The method for rapid prediction of urban typhoon disasters based on intelligent downscaling according to claim 6 is characterized in that: When training the secondary downscaling model, its input is: extracting the moments when the maximum wind speed of the typhoon on land reaches 30m / s or more, and taking the meteorological data of the first-level fine resolution at these moments as input; the output is: the meteorological data of the second-level fine resolution of the corresponding time period.

8. The method for rapid prediction of urban typhoon disasters based on intelligent downscaling according to claim 1 is characterized in that: Before generating high-resolution urban building surface wind pressure information, the method further includes: Construction of urban CFD model; AI urban building wind pressure model construction; AI urban building wind pressure model optimization; In the step of constructing the urban CFD model, the wind speed data obtained by the secondary downscaling model is used to fit a logarithmic curve of wind speed and height, wind shear is calculated, and the equivalent roughness of the building is used to calculate the theoretically corrected wind speed; In the step of constructing the AI ​​urban building wind pressure model, the wind field pressure on the building surface obtained by simulating the urban CFD model is used as output to establish the AI ​​urban building wind pressure model.

9. The method for rapid prediction of urban typhoon disasters based on intelligent downscaling according to claim 8, characterized in that: In the risk assessment model for designing buildings, a fragility function is used to assess the probability of damage to buildings under the action of typhoons, and the additional impact of precipitation is considered to correct the fragility function.

10. The method for rapid prediction of urban typhoon disasters based on intelligent downscaling according to claim 9, characterized in that: Based on the probability distribution of damage to each building obtained by the modified fragility function, the dominant damage level of all buildings in the region is calculated, and the regional risk disaster level results are summarized.

Images