CN-121981548-A - Agricultural typhoon disaster damage assessment method based on public data and machine learning

CN121981548ACN 121981548 ACN121981548 ACN 121981548ACN-121981548-A

Abstract

The invention relates to an agricultural typhoon disaster loss assessment method based on public data and machine learning, which comprises the following steps of S1, obtaining provincial typhoon disaster data, typhoon paths and near-center minimum air pressure through statistics annual bill or public data query The method comprises the steps of (1) researching agricultural statistical data of an area, S2 dividing provincial disaster condition data into municipal or county scale according to a maximum wind speed radius empirical formula and a typhoon body influence range, S3 estimating local agricultural typhoon loss based on a disaster area of crops in the disaster condition data according to the municipal or county scale agricultural statistical data, S4 selecting various disaster causing factors to normalize the data, dividing the data into a training set and a testing set, inputting the divided training set data into a machine learning model for training, carrying out agricultural loss assessment, and solving the problems that disaster damage conditions of smaller scale of regional county cannot be assessed in current typhoon disaster assessment and agricultural typhoon disaster loss assessment is lacked.

Inventors

FANG PINGZHI
HE HAO
TANG JIE
NIU XIAOJING

Assignees

上海亚太台风研究中心
清华大学

Dates

Publication Date: 20260505
Application Date: 20260126

Claims (9)

1. An agricultural typhoon disaster damage assessment method based on public data and machine learning is characterized by comprising the following steps: s1, acquiring provincial typhoon disaster data, typhoon paths and near-center minimum air pressure through statistics annual bill or public data query Agricultural statistics of the area of investigation; S2, dividing provincial disaster data into municipal or county scale according to a maximum wind speed radius empirical formula and a typhoon body influence range by means of the assumption that the typhoon body influence range is 4 times of a maximum wind speed radius and crop planting areas are uniformly distributed; s3, estimating local agricultural typhoon loss based on the disaster area of crops in disaster data according to the urban or county agricultural statistical data; S4, after normalization processing is carried out on the data by selecting various disaster causing factors, dividing the data into a training set and a testing set, inputting the divided training set data into a machine learning model for training, and carrying out agricultural loss assessment.
2. The agricultural typhoon disaster damage assessment method according to claim 1, wherein step S1 obtains provincial typhoon disaster data, typhoon routes, near-center minimum air pressure according to tropical cyclone optimal route set data, statistical annual certificates, government officials and news platform public data sources And agricultural statistics of the research area, wherein typhoon disaster data comprises disaster area under a jth site typhoon Area of disaster Area of extinction Direct economic loss The typhoon path is the longitude and latitude path of the relevant typhoon Agricultural statistics include yield of the kth crop of the ith city/county Monovalent Area of planting 。
3. The agricultural typhoon disaster damage assessment method according to claim 2, wherein the specific method of step S2 provincial disaster condition data division is as follows: s21, supposing that the influence range of the typhoon body is circular, the influence range of the jth typhoon body at the t moment Calculated from the following formula: wherein j and t respectively represent typhoon numbers and moments; The influence area indicating the maximum wind speed radius of the typhoon at that time is calculated by the following equation: In the formula, Representing the air pressure difference between the typhoon center and the ambient atmosphere; Representing the latitude of the typhoon center; the area of the region affected by typhoons at the city level or the county level is calculated by the following formula, Wherein i represents a city/county number; representing the affected areas of the ith city/county and the jth typhoon at the t moment; Representing the administrative division scope of the ith city/county; Representing the maximum range of the ith city/county affected by the jth typhoon, and taking 0 when the typhoon affected range is not intersected with the administrative division of the city/county in space; s22, assuming that all crops are approximately uniformly distributed, dividing provincial disaster data into municipal or county levels according to the following formula: In the formula, The disaster area of the jth typhoon is represented; the method and the system are characterized in that the method and the system are used for representing the disaster area of the ith city/county under the jth typhoon, M represents the total number of administrative division of the city level or the county level, and the total number is determined according to the research area, and the disaster area and the harvest-free area of the administrative division of the city/county level can be obtained in the same way.
4. The agricultural typhoon disaster damage assessment method according to claim 3, wherein the disaster area of the city/county level administrative division in step S22 is expressed as: In the formula (I), in the formula (II), Representing the disaster area of the typhoon in the j-th field; Representing the disaster area of the ith city/county in the jth typhoon; The area of the administrative division at the city/county level is expressed as: In the formula (I), in the formula (II), Representing the area of extinction of the jth typhoon; Represents the area of extinction of the ith city/county in the jth typhoon.
5. The method for estimating a loss of agricultural typhoons according to claim 4, wherein the specific method for estimating an agricultural typhoons loss in step S3 is that, assuming that all crop planting areas are uniformly distributed locally, the agricultural loss of typhoons in a certain city/county level administrative division under a typhoon in a certain farm is expressed as: In the formula, Representing the agricultural loss of the ith city/county in the jth typhoon; , , Respectively representing the yield, unit price and planting area of the kth crop in the ith city/county, and acquiring related data according to a local statistical annual survey and price detection report, wherein N represents the type of the crop; representing the area of reduced yield of the ith city/county under the jth typhoon, wherein the area of reduced yield is obtained according to the definition of disaster area, disaster area and harvest-free area.
6. The agricultural typhoon disaster damage assessment method according to claim 5, wherein the yield reduction area calculation formula in step S3 is as follows: In the formula, , , , Disaster area, yield reduction area, harvest area and disaster area of the ith city/county under the jth typhoon; , the ratio of the area of extinction and the area of disaster recovery to the area of disaster recovery is shown, respectively.
7. The agricultural typhoon disaster damage assessment method according to claim 1, wherein the plurality of disaster causing factors in step S4 include a maximum center wind speed V, a closest distance D, a center wind pressure P, a maximum wind speed radius R M , an influence time h, an influence area S, and an absolute disaster recovery area ratio α s during the influence of typhoons, and in order to avoid the scale influence of different features, normalization processing is required for the data: In the formula, , , Representing variables respectively And its maximum and minimum values, x represents the normalized variable.
8. The method according to claim 1, wherein in step S4, 80% of samples of the city/county disaster data obtained by division are used as training sets, 20% are used as test sets, and the machine learning model is one of a Random Forest (RF) evaluation model, a Support Vector Machine (SVM) evaluation model, a BP neural network evaluation model, and XGBoost evaluation model.
9. The agricultural typhoon disaster damage assessment method according to claim 8, wherein the machine learning model in the step S4 is XGBoost model, and the agricultural damage assessment is carried out by calling XGBoost algorithm in sklearn library to perform model training.

Description

Agricultural typhoon disaster damage assessment method based on public data and machine learning Technical Field The invention belongs to the technical field of natural disaster risk assessment, relates to an agricultural typhoon disaster damage assessment method based on public data and machine learning, and particularly relates to an agricultural typhoon disaster damage assessment method based on public data considering disaster damage of lower administrative regions, which is used for assisting coastal typhoon disaster prevention and reduction, agricultural insurance and other businesses in China. Background The evaluation analysis on typhoon disasters mainly focuses on both risk evaluation and disaster evaluation. The risk evaluation mainly carries out qualitative or quantitative evaluation on the possibility and risk of typhoon disasters according to regional typhoon information, such as intensity, distribution rules, paths and the like, and by combining population distribution, economic level, social development, topography and the like of a research region. The evaluation is generally performed based on a natural disaster systematic theory method from aspects of disaster causing factor dangers, disaster-tolerant environment vulnerability, disaster-tolerant body exposure and the like. The disaster-causing factors refer to factors causing typhoon disasters, such as wind speed, air pressure, precipitation and other meteorological factors, the disaster-causing environment refers to the disaster-causing factors and the external environment where disaster-bearing bodies are located, such as topography, and the like, and the disaster-bearing bodies are objects with natural disasters, such as population, cultivated land, houses and the like. And grading or evaluating the typhoon disaster risk of the disaster area through functions or indexes related to the three. The disaster condition evaluation mainly evaluates losses caused by typhoons such as economic losses, casualties and the like, and mainly comprises pre-disaster evaluation, namely, a disaster damage model is built according to technologies and methods such as historical data, simulation, machine learning, GIS and the like, so that casualties, economic losses and the like possibly suffered by typhoons in a certain area under the influence of events are analyzed or predicted in advance. And secondly, disaster in-disaster and post-disaster assessment, namely, during the development process of the disaster or after the end of the disaster, measuring the disaster intensity, the disaster forming place and the disaster condition characteristics at the first time, monitoring and measuring and calculating the loss caused by the disaster, and formulating scientific and reasonable disaster prevention and relief policies aiming at different disaster intensities and disaster-stricken areas to guide related departments and rescue institutions to respond quickly. Most of the existing researches focus on risk evaluation for a certain area, and disaster risk of the area is analyzed by classifying risk levels. A few researches on disaster assessment are mainly performed directly according to historical disaster data disclosed in a concerned area, the characteristics of typhoon disasters, strength, range, frequency, space-time distribution and the like of loss conditions are analyzed through methods of statistics, classification and the like, and a potential disaster damage model is constructed through analysis of data characteristics of disaster damage rate, frequency and the like. However, these public data-based evaluations have some disadvantages that (1) disaster damage evaluations in large areas such as provincial level and the like are usually performed, disaster damage situations in smaller scales in regional and county level cannot be evaluated, and (2) most of disaster damage grades or direct economic loss evaluations are performed, and typhoon disaster damage evaluations aiming at agriculture are performed rarely. Disclosure of Invention In view of this, the invention provides an agricultural typhoon disaster damage assessment method based on public data and machine learning, which can divide disaster data into smaller scales (in this method, municipal/county administrative division) according to the influence range of typhoon bodies, further estimate agricultural typhoon damage according to local agricultural statistics data, and train an agricultural disaster damage assessment model based on the machine learning method in order to solve the problems that the disaster damage condition of smaller scales of county level cannot be assessed in the current typhoon disaster assessment and the more detailed real-time emergency decision cannot be supported. In order to achieve the above purpose, the present invention provides the following technical solutions: An agricultural typhoon disaster damage assessment method based on public data and machine learning compr