Agrometeorological disaster monitoring and early warning method and system based on remote sensing technology

Abstract

The invention discloses an agricultural meteorological disaster monitoring and early warning method and system based on a remote sensing technology, and the method comprises the steps: generating a multi-dimensional data set through obtaining and processing multi-source remote sensing data, and generating a comprehensive data set through weighted average fusion. Then, extracting land surface temperature data, comparing the land surface temperature data with a historical value, calculating a temperature deviation value, and generating temperature anomaly distribution data; and calculating a disaster intensity index by combining the soil humidity and vegetation index data change trend, and generating disaster intensity distribution data. The data meteorology is imported into a driving simulation system, disaster evolution is simulated, and the disaster influence range and duration are predicted. And if the prediction data exceeds an early warning threshold, generating early warning data including disaster categories, influence areas and prediction time, and generating disaster risk distribution data by using a spatial interpolation method for dynamic monitoring. According to the invention, the accuracy and timeliness of disaster monitoring and early warning are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of meteorological warning, and particularly relates to an agricultural meteorological disaster monitoring and warning method and system based on remote sensing technology. BACKGROUND

[0002] Agricultural meteorological disaster monitoring and warning is an important field to ensure food security and sustainable development of agriculture, and its research has key significance for reducing the damage of natural disasters to agricultural production. Agricultural meteorological disasters such as drought, flood, high temperature and heat damage occur frequently, which pose a serious threat to crop growth and farmers' livelihoods. Therefore, efficient and accurate monitoring and warning means are urgently needed to cope with these challenges.

[0003] At present, although some monitoring methods have been applied in the field of agricultural meteorological disasters, there are generally problems of insufficient data integration and weak dynamic response capability. Many existing solutions often rely on a single data source or static analysis, which is difficult to fully capture the multi-dimensional change characteristics before the disaster occurs, especially in the face of complex meteorological environment and regional differences, lacking real-time tracking and comprehensive assessment capability of the development process of the disaster, which limits the timeliness and accuracy of the warning.

[0004] Under this background, the core challenge in this field is how to effectively integrate multi-source data and realize dynamic monitoring and warning. First of all, the problem of weight distribution of multi-source data becomes a big difficulty. Different sources of data have significant differences in spatial resolution, time frequency and response characteristics to disasters. If the contribution of each type of data cannot be reasonably determined, it will directly affect the reliability of the monitoring results. Secondly, this problem further extends to how to build a dynamic scenario simulation system based on multi-source data to truly reflect the development trend and influence range of the disaster. These two factors are closely related. The optimization of data fusion directly determines the accuracy of scenario simulation, and the deficiency of scenario simulation will in turn limit the practicality of warning.

[0005] Therefore, how to scientifically allocate the weight of multi-source data and combine dynamic scenario simulation technology to build a precise agricultural meteorological disaster monitoring and warning system has become a key problem that needs to be solved in this research. SUMMARY

[0006] To solve the above technical problems, the present application provides an agricultural meteorological disaster monitoring and warning method and system based on remote sensing technology.

[0007] Among them, an agricultural meteorological disaster monitoring and warning method based on remote sensing technology comprises:

[0008] Obtaining initial multi-source remote sensing data, processing visible light, near-infrared, and thermal infrared band information of the initial multi-source remote sensing data to generate a multi-dimensional data set containing time identifiers and spatial location information; wherein the multi-source remote sensing data includes surface temperature data, soil moisture data, and vegetation index data;

[0009] Calculating a fusion weight coefficient for the multi-dimensional data set by a weighted average method to generate a fused comprehensive data set;

[0010] Extracting surface temperature data from the comprehensive data set and comparing it with a pre-established historical contemporaneous surface temperature average to calculate a temperature deviation value and generate temperature anomaly distribution data;

[0011] For the temperature anomaly distribution data, combining the change trend of soil moisture data and vegetation index data in the comprehensive data set, calculating a disaster intensity index by a weighted comprehensive index method to generate disaster intensity distribution data;

[0012] Importing the disaster intensity distribution data into a pre-established meteorological driving simulation system, combining a time series analysis method to simulate a disaster evolution process and generate disaster impact range and duration prediction data;

[0013] For the disaster impact range data and duration prediction data, comparing them with a pre-set disaster warning threshold value, and if the prediction data exceeds the threshold value, generating warning data containing disaster category, impact area, and prediction time information;

[0014] For the impact area information in the warning data, combining data in the comprehensive data set, generating disaster risk distribution data by a spatial interpolation method, and dynamically monitoring the disaster risk distribution data.

[0015] Preferably, the process of obtaining initial multi-source remote sensing data, processing visible light, near-infrared, and thermal infrared band information of the initial multi-source remote sensing data to generate a multi-dimensional data set containing time identifiers and spatial location information includes:

[0016] Obtaining multi-source remote sensing data, extracting original band information from visible light, near-infrared, and thermal infrared bands, fusing surface temperature data, soil moisture data, and vegetation index data to generate an initial data set;

[0017] Correcting the initial data set by preprocessing, using a radiation correction method to eliminate sensor bias in the band information to obtain a corrected band data set;

[0018] If the corrected band data set has noise, using a Gaussian filtering algorithm to denoise the visible light, near-infrared, and thermal infrared band information to obtain a denoised band data set;

[0019] According to the denoised wave band data set, time identifiers and spatial position information are extracted, time identifiers and spatial position information are associated with wave band information by using a space-time indexing method, and a space-time associated data set is generated;

[0020] The space-time associated data set is subjected to feature extraction by a principal component analysis algorithm, and the features of the land surface temperature, soil moisture and vegetation index are fused to obtain a feature enhanced data set;

[0021] If the time series of the feature enhanced data set is complete, a long short-term memory network algorithm is used to analyze the time variation trend of the land surface temperature, soil moisture and vegetation index to obtain dynamic change characteristics;

[0022] According to the dynamic change characteristics, an interpolation method is used to complete the space-time of the feature enhanced data set to generate a multi-dimensional data set containing time identifiers and spatial position information.

[0023] Preferably, the process of calculating the fusion weight coefficient of the multi-dimensional data set by the weighted average method includes:

[0024] According to the data dimension of the multi-dimensional data set, a data set input is extracted, the data structure of each dimension is determined, and a structure data set is obtained;

[0025] By a preset fusion algorithm, the weight coefficient is calculated for each data dimension of the structure data set to obtain a fusion weight;

[0026] The weighted average method is used to combine the fusion weight to perform weighted calculation on each data dimension of the structure data set to obtain a weighted data set;

[0027] If the dimension consistency of the weighted data set meets a preset threshold value, the weighted data set is subjected to normalization processing to obtain a standardized data set;

[0028] According to the standardized data set, a data fusion algorithm is applied to generate a comprehensive data set;

[0029] The structure of the comprehensive data set is analyzed to verify the integrity of data fusion, and an output result is obtained;

[0030] Key features are extracted from the output result to generate a final comprehensive data set.

[0031] Preferably, the process of generating temperature anomaly distribution data includes:

[0032] Land surface temperature related records are obtained from the comprehensive data set, target temperature data is separated by a data screening method, and a preliminary extracted temperature data set is obtained;

[0033] According to the preliminary extracted temperature data set, the temperature mean value data corresponding to the historical same period is obtained, and the data matching method is used for one-to-one correspondence to determine the correlation between the two sets of data;

[0034] If the correlation between the two sets of data meets the preset matching condition, the difference between the preliminary extracted temperature data set and the historical same period temperature mean value data is calculated to obtain the temperature deviation data set;

[0035] For the temperature deviation data set, a statistical analysis method is used to identify deviation values that exceed a preset threshold to determine abnormal temperature data points;

[0036] The spatial distribution of the abnormal temperature data points is grid processed to obtain the regional characteristics of the abnormal distribution and determine the temperature abnormal distribution data set;

[0037] According to the temperature abnormal distribution data set, a regression analysis method is used to predict the potential change trend of the abnormal distribution to obtain a prediction result of the abnormal distribution;

[0038] If the prediction result shows that the potential change trend of the abnormal distribution exceeds a preset range, a dynamic mapping of the abnormal distribution is generated through a data visualization tool to obtain a key focus area of the abnormal distribution.

[0039] Preferably, the disaster intensity index is calculated by a weighted comprehensive index method, and the process of generating the disaster intensity distribution data includes:

[0040] Obtain the respective change trend data from the temperature abnormal distribution data, soil moisture data and vegetation index data;

[0041] The temperature deviation value, soil moisture change value and vegetation index change value are calculated by a standardization processing method to obtain a normalized change trend data set;

[0042] According to the normalized change trend data set, the weight coefficients of each parameter are determined;

[0043] The weight coefficients of the temperature deviation value, soil moisture change value and vegetation index change value are calculated by an entropy method to obtain a weight coefficient set;

[0044] According to the weight coefficient set and the normalized change trend data set, the weighted comprehensive index is calculated to obtain a weighted comprehensive index value;

[0045] According to the weighted comprehensive index value, initial disaster intensity distribution data is generated;

[0046] If the weighted comprehensive index value is greater than a preset threshold, it is determined as a high-intensity disaster area, and first disaster intensity distribution data is generated;

[0047] The spatial interpolation method is used to process the missing area for the first disaster intensity distribution data.

[0048] The spatial smoothing processing is performed on the data using the Kriging interpolation algorithm to obtain second disaster intensity distribution data.

[0049] The spatial mapping is performed on the second disaster intensity distribution data by fusing geographic information data.

[0050] The raster processing method is used to associate the disaster intensity index with the geographic coordinates to generate the final disaster intensity distribution data.

[0051] The data compression method is used to store the final disaster intensity distribution data.

[0052] The Zstandard compression algorithm is used to compress the data to obtain the compressed disaster intensity distribution data.

[0053] Preferably, the process of generating disaster impact range and duration prediction data includes:

[0054] The format conversion and cleaning are performed on the disaster intensity distribution data to obtain a standardized input data set.

[0055] The spatio-temporal features are extracted based on the standardized input data set, and the long short-term memory network model is used for time series analysis to obtain the disaster intensity change trend.

[0056] The meteorological factors are combined based on the disaster intensity change trend, and input into the pre-established meteorological driven simulation system to generate the disaster evolution path.

[0057] If the intensity of the disaster evolution path exceeds the preset threshold, the impact range is calculated by the spatial interpolation algorithm to obtain the disaster impact range data.

[0058] The duration is predicted by the linear regression model through the time step analysis of the disaster evolution path to obtain the disaster duration data.

[0059] The geographic information system is fused based on the disaster impact range data and duration data for visual processing to generate the disaster prediction distribution map.

[0060] The high-risk area features are extracted from the disaster prediction distribution map to generate the disaster warning information.

[0061] Preferably, the process of comparing the disaster impact range data and duration prediction data with the preset disaster warning threshold includes:

[0062] The disaster impact range data and duration prediction data are further formatted and cleaned to obtain a standardized prediction data set;

[0063] Comparing the normalized prediction data set with the preset warning thresholds item by item, if any indicator in the prediction data set exceeds the threshold, it is marked as high-risk data, and a high-risk data set is determined;

[0064] Extracting corresponding disaster categories and affected area information based on the high-risk data set, and classifying disaster types to obtain classified disaster category information;

[0065] Based on the classified disaster category information, combined with the predicted time data, a time series analysis tool is used to integrate the time dimension to determine the release time point of the warning information;

[0066] According to the release time point and disaster category information, the affected area data is associated with the disaster category to generate structured warning information content;

[0067] Based on the structured warning information content, geographic information processing tools are used to spatially map the affected area data to obtain a visual distribution of warning areas;

[0068] The boundary information of high-risk areas is extracted from the visual warning area distribution, recorded through the data storage module, and the final warning information data set is determined.

[0069] Preferably, the process of generating disaster risk distribution data by a spatial interpolation method and dynamically monitoring the disaster risk distribution data includes:

[0070] The standardized regional dataset is fused with the comprehensive dataset and a weighted average method is used to generate a fused dataset.

[0071] For the fused data set, a Kriging interpolation method is used to generate disaster risk distribution data to obtain an initial risk distribution map;

[0072] If there is data missing in the initial risk distribution map, the missing area is filled by interpolating adjacent points to obtain a complete risk distribution map;

[0073] Through time series analysis, the changing trend of the complete risk distribution map is extracted to obtain the dynamic change characteristics of the risk;

[0074] Based on the dynamic change characteristics of the risk, the sliding window method is used to calculate the risk fluctuation value and determine the stability of the risk change trend;

[0075] If the risk fluctuation value exceeds the preset threshold, the dynamic monitoring system is triggered to update the risk distribution data and obtain the real-time risk distribution status.

[0076] The application also provides an agricultural meteorological disaster monitoring and early warning system based on remote sensing technology, comprising:

[0077] A data acquisition and processing module is configured to acquire initial multi-source remote sensing data, process visible light, near-infrared and thermal infrared band information of the initial multi-source remote sensing data, and generate a multi-dimensional data set containing time identifiers and spatial position information; wherein the multi-source remote sensing data comprises ground temperature data, soil moisture data and vegetation index data.

[0078] A data fusion module is configured to calculate a fusion weight coefficient for the multi-dimensional data set by a weighted average method, and generate a fused comprehensive data set.

[0079] A temperature anomaly analysis module is configured to extract ground temperature data from the comprehensive data set, compare the ground temperature data with a pre-established historical contemporaneous ground temperature average value, calculate a temperature deviation value, and generate temperature anomaly distribution data.

[0080] A disaster intensity calculation module is configured to calculate a disaster intensity index by a weighted comprehensive index method for the temperature anomaly distribution data in combination with change trends of soil moisture data and vegetation index data in the comprehensive data set, and generate disaster intensity distribution data.

[0081] A disaster simulation module is configured to import the disaster intensity distribution data into a pre-established meteorological driving simulation system, combine a time series analysis method, simulate a disaster evolution process, and generate disaster impact range and duration prediction data.

[0082] An early warning generation module is configured to compare the disaster impact range data and duration prediction data with a pre-set disaster early warning threshold value, and if the prediction data exceeds the threshold value, generate early warning data containing disaster category, impact area and prediction time information.

[0083] A risk monitoring module is configured to generate disaster risk distribution data by a spatial interpolation method for impact area information in the early warning data in combination with data in the comprehensive data set, and dynamically monitor the disaster risk distribution data.

[0084] Compared with the prior art, the application has the following advantages and technical effects:

[0085] The application discloses a disaster monitoring and early warning method based on multi-source remote sensing data, calculates temperature anomaly and disaster intensity index by fusing multi-dimensional data such as ground surface temperature, soil moisture and vegetation index, predicts the disaster evolution process in combination with a meteorological driving simulation system, and generates disaster influence range and duration prediction data. When the prediction data exceeds the early warning threshold, the application generates early warning data containing the disaster category, influence area and prediction time, and generates high-resolution disaster risk distribution data by using a spatial interpolation method for dynamic monitoring. The application realizes comprehensive utilization of multi-source remote sensing data, improves the accuracy and timeliness of disaster monitoring and early warning, and provides a scientific basis for disaster prevention and control decision-making. BRIEF DESCRIPTION OF DRAWINGS

[0086] The accompanying drawings, which form a part of the present application, are intended to provide further understanding of the present application and serve as an aid in explaining the principles of the present application and its implementation, and are not intended to impose any undue limitations upon the present application. In the drawings:

[0087] Figure 1 A method flowchart of an embodiment of the present application;

[0088] Figure 2 A system structure schematic diagram of an embodiment of the present application. DETAILED DESCRIPTION

[0089] It should be noted that the embodiments and features in the present application can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0090] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a group of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that shown herein.

[0091] Embodiment one

[0092] As Figure 1 shown, the present embodiment provides an agricultural meteorological disaster monitoring and early warning method based on remote sensing technology, comprising:

[0093] Obtaining initial multi-source remote sensing data, processing visible light, near-infrared and thermal infrared band information of the initial multi-source remote sensing data to generate a multi-dimensional data set containing time identification and spatial position information; wherein the multi-source remote sensing data includes ground surface temperature data, soil moisture data and vegetation index data;

[0094] The fusion weight coefficient of the multi-dimensional data set is calculated by using a weighted average method to generate a comprehensive data set after fusion;

[0095] Extracting the ground temperature data from the comprehensive data set, comparing with the pre-established historical synchronous ground temperature average value, calculating the temperature deviation value, and generating the temperature anomaly distribution data;

[0096] For the temperature anomaly distribution data, combining the change trend of the soil moisture data and the vegetation index data in the comprehensive data set, calculating the disaster intensity index through the weighted comprehensive index method, and generating the disaster intensity distribution data; wherein the weighted comprehensive index method is defined as the disaster intensity index being equal to the weight coefficient of the temperature deviation value multiplied by the temperature deviation value plus the weight coefficient of the soil moisture change value multiplied by the soil moisture change value plus the weight coefficient of the vegetation index change value multiplied by the vegetation index change value;

[0097] Importing the disaster intensity distribution data into the pre-established meteorological driving simulation system, combining the time series analysis method, simulating the disaster evolution process, and generating the disaster impact range and duration prediction data;

[0098] For the disaster impact range data and the duration prediction data, comparing with the pre-set disaster warning threshold value, if the prediction data exceeds the threshold value, generating the warning data containing the disaster category, the impact area and the prediction time information;

[0099] For the impact area information in the warning data, combining the data in the comprehensive data set, generating the disaster risk distribution data through the spatial interpolation method, and dynamically monitoring the disaster risk distribution data.

[0100] Further, the process of obtaining the initial multi-source remote sensing data, processing the visible light, near-infrared and thermal infrared band information of the initial multi-source remote sensing data, and generating a multi-dimensional data set containing time identifier and spatial location information includes:

[0101] Obtaining multi-source remote sensing data, extracting original band information from visible light, near-infrared and thermal infrared bands, fusing ground temperature data, soil moisture data and vegetation index data, and generating an initial data set;

[0102] Correcting the initial data set through preprocessing, using a radiation correction method to eliminate sensor bias in the band information, and obtaining a corrected band data set;

[0103] If the corrected band data set has noise, a Gaussian filtering algorithm is used to denoise the visible light, near-infrared and thermal infrared band information, and a denoised band data set is obtained;

[0104] According to the denoised band data set, extracting the time identifier and the spatial location information, using a space-time indexing method to associate the time identifier and the spatial location information with the band information, and generating a space-time associated data set;

[0105] The characteristic enhanced data set is obtained by feature extraction on the spatio-temporal correlation data set through a principal component analysis algorithm, fusing the characteristics of the land surface temperature, soil moisture, and vegetation index.

[0106] If the time series of the characteristic enhanced data set is complete, a long short-term memory network algorithm is used to analyze the time variation trend of the land surface temperature, soil moisture, and vegetation index to obtain dynamic change characteristics.

[0107] According to the dynamic change characteristics, an interpolation method is used to complete the spatio-temporal of the characteristic enhanced data set to generate a multi-dimensional data set containing time identifiers and spatial location information.

[0108] For example, when acquiring multi-source remote sensing data, the original data of visible light, near-infrared, and thermal infrared bands can be obtained through a satellite image platform. Assuming that the data comes from an agricultural monitoring scene in a certain area, the coverage area is 100 square kilometers, and the time range is from January to December 2023. The visible light band is used to capture the color information of the ground, the near-infrared band is used to analyze the health status of the vegetation, and the thermal infrared band reflects the distribution of the ground temperature. These band data combined with the land surface temperature, soil moisture, and vegetation index form an initial data set. In the preprocessing stage, the radiation correction method can be used to eliminate the sensor bias.

[0109] For example, by comparing historical calibration data and current band data, the brightness value deviation of the visible light band is adjusted to ensure the accuracy of the data. After correction, if noise is found in the data, a Gaussian filter algorithm can be used to denoise the band information. Assuming that the noise mainly comes from sensor jitter, the details of the band data are made clearer through smoothing processing, thereby improving the reliability of subsequent analysis.

[0110] Specifically, when extracting time identifiers and spatial location information, the specific date and latitude and longitude coordinates can be labeled for each image, for example, the image of June 1, 2023, is associated with the location of north latitude 30 degrees and east longitude 120 degrees. Using a spatio-temporal indexing method, these information is bound with the band data to form a spatio-temporal correlation data set, which is convenient for subsequent data query by time or spatial dimension. In the feature extraction link, the principal component analysis algorithm can be used for dimension reduction and feature fusion.

[0111] For example, the multiple variables of the land surface temperature, soil moisture, and vegetation index are compressed into a few principal components, retaining more than 90% of the information amount, to generate a characteristic enhanced data set. This not only reduces data redundancy, but also highlights key features and improves analysis efficiency. For time series analysis, if the characteristic enhanced data set is complete, a long short-term memory network algorithm can be used to analyze the dynamic change trend.

[0112] For example, based on 12 months of data, the rising trend of surface temperature in summer and the growth pattern of vegetation index in rainy season are predicted to obtain dynamic change characteristics. This helps to understand the periodicity of environmental changes. In the spatio-temporal completion stage, interpolation methods can be used to fill in missing data.

[0113] For example, if a region lacks soil moisture data at a certain time point, linear interpolation can be used to generate a complete multi-dimensional data set from data at adjacent time points and spatial points. This method ensures data continuity and provides reliable support for subsequent agricultural monitoring or disaster warning.

[0114] It should be noted that the processing of each link described above is carried out around the agricultural monitoring scenario, from data acquisition to feature analysis and spatio-temporal completion, forming a complete technical chain. Through these methods, not only the data quality is improved, but also the visualization and prediction ability of environmental changes are enhanced, providing an important basis for precision agriculture management.

[0115] Further, the process of calculating the fusion weight coefficient of the multi-dimensional data set by the weighted average method to generate the integrated comprehensive data set includes:

[0116] According to the data dimensions of the multi-dimensional data set, the data set input is extracted, the data structure of each dimension is determined, and the structure data set is obtained;

[0117] Through the preset fusion algorithm, the weight coefficient is calculated for each data dimension of the structure data set, and the fusion weight is obtained;

[0118] Using the weighted average method, the weighted calculation is performed on each data dimension of the structure data set by combining the fusion weight, and the weighted data set is obtained;

[0119] If the dimension consistency of the weighted data set meets the preset threshold value, the weighted data set is normalized to obtain the standardized data set;

[0120] According to the standardized data set, a data fusion algorithm is applied to generate a comprehensive data set;

[0121] The structure of the comprehensive data set is analyzed to verify the integrity of the data fusion, and the output result is obtained;

[0122] Key features are extracted from the output result to generate the final comprehensive data set.

[0123] For example, in the scenario of acquiring a multi-dimensional dataset and performing data processing, assuming the data comes from remote sensing monitoring of a certain farmland area, covering an area of 50 square kilometers and a time span from March to September 2023. For the extraction of the multi-dimensional dataset, first separate the key information such as ground temperature, soil moisture and vegetation coverage from the data dimensions. The ground temperature data is stored in the form of daily average, the soil moisture is expressed in percentage, and the vegetation coverage is recorded in the range of 0 to 1. These data structures are divided by a uniform grid to ensure consistent spatial resolution, such as 10 meters x 10 meters per grid cell.

[0124] For example, after determining the data structure of each dimension, the weight calculation of the fusion algorithm for the initial dataset can use a statistical analysis method based on historical data. Assuming that ground temperature has a greater impact on crop growth, its weight coefficient can be set to 0.5, while the weights of soil moisture and vegetation coverage are 0.3 and 0.2 respectively. This weight distribution is based on the priority evaluation of the factors affecting the farmland environment. The weighted average method then linearly combines each dimension data according to the weight to form a weighted dataset. Assuming that the ground temperature of a certain grid cell is 30 degrees Celsius, the soil moisture is 40%, and the vegetation coverage is 0.6, the comprehensive value after weighted calculation is 30x0.5+40x0.3+0.6x0.2, resulting in a comprehensive index.

[0125] For example, in the normalization processing link, if the dimension consistency of the weighted dataset passes the preset threshold test, such as the fluctuation range of each dimension data being within ±10%, the data is mapped to the interval of 0 to 1 to form a standardized dataset. This processing facilitates the uniform input standard of subsequent algorithms. The data fusion algorithm can choose a multi-dimensional interpolation-based method to integrate the standardized dataset into a comprehensive dataset. Assuming that some grid data is missing at a certain time point, it can be filled by the average value of the adjacent grid to ensure data integrity.

[0126] For example, in verifying the integrity of data fusion, the spatial distribution of the comprehensive dataset can be analyzed to see if it is uniform and the time series is continuous. If an abnormal data distribution is found in a certain area, it can be traced back to the original data collection link to check for sensor errors. Finally, key features such as the distribution of high temperature and high humidity in the farmland area are extracted from the output results to generate the final comprehensive dataset. This helps to adjust irrigation or fertilization strategies in subsequent agricultural management, improving resource utilization efficiency.

[0127] Further, the process of generating temperature anomaly distribution data includes:

[0128] Obtain ground temperature related records from the comprehensive dataset, use data filtering method to separate target temperature data, get preliminary extracted temperature dataset;

[0129] According to the preliminary extracted temperature data set, the temperature mean value data corresponding to the historical same period is obtained, and the data matching method is used for one-to-one correspondence to determine the correlation between the two sets of data;

[0130] If the correlation between the two sets of data meets the preset matching condition, the difference between the preliminary extracted temperature data set and the historical same period temperature mean value data is calculated to obtain the temperature deviation data set;

[0131] For the temperature deviation data set, a statistical analysis method is used to identify deviation values that exceed a preset threshold to determine abnormal temperature data points;

[0132] The spatial distribution of abnormal temperature data points is subjected to grid processing to obtain the regional characteristics of abnormal distribution and determine the temperature abnormal distribution data set;

[0133] According to the temperature abnormal distribution data set, a regression analysis method is used to predict the potential change trend of abnormal distribution to obtain the prediction result of abnormal distribution;

[0134] If the prediction result shows that the potential change trend of abnormal distribution exceeds the preset range, a dynamic mapping of abnormal distribution is generated through a data visualization tool to obtain the key attention area of abnormal distribution.

[0135] For example, when extracting surface temperature data from a comprehensive data set, the data processing system can automatically read remote sensing image data stored in the database. Assuming that the data set contains surface temperature records of each region in the study area in July 2023, the unit is Celsius, and the data resolution is 1 km x 1 km. The system uses a preset filtering algorithm to automatically filter the surface temperature field based on data tags, extracts the temperature value of the region at 30 degrees north latitude and 120 degrees east longitude as 35.5 degrees Celsius, and generates a structured data set containing latitude and longitude coordinates and corresponding temperature values. Then, the extracted data is compared with the historical same period surface temperature average value, assuming that the historical data comes from the average value database of the same period in the past 10 years, and the average temperature of the same position in July is 33.2 degrees Celsius. The system calculates the difference between the current temperature and the historical average value point by point through a batch calculation script, i.e. 35.5 minus 33.2, to obtain a temperature deviation value of 2.3 degrees Celsius.

[0136] Next, based on the deviation value, temperature anomaly distribution data is generated. The system uses a spatial interpolation algorithm (such as the Kriging interpolation method) to grid the deviation values ​​of all sampling points to generate a temperature anomaly distribution map covering the entire world. Areas with deviation values ​​greater than 2.0 degrees Celsius are marked as high temperature anomaly areas, and areas with deviation values ​​less than -2.0 degrees Celsius are marked as low temperature anomaly areas. At the same time, combined with geographic information system (GIS) technology, the anomaly distribution data is superimposed on the map to form a visual result. For example, it was found that the high temperature anomaly value in some areas of East Asia reached 3.5 degrees Celsius, indicating that there may be a risk of heat waves.

[0137] To ensure logical integrity, the system can also integrate with meteorological forecast models, using anomaly distribution data as input to predict temperature trends for the coming week. For example, it can predict that a high temperature anomaly may persist for three days, assisting in decision support. The entire process utilizes automated scripts and algorithms to extract, compare, calculate, and visualize data, ensuring efficiency and accuracy.

[0138] Furthermore, the disaster intensity index is calculated using the weighted comprehensive index method, and the process of generating the disaster intensity distribution number includes:

[0139] Obtain respective change trend data from temperature anomaly distribution data, soil moisture data and vegetation index data;

[0140] The temperature deviation value, soil moisture change value and vegetation index change value were calculated using the standardization processing method to obtain the normalized change trend data set;

[0141] Determine the weight coefficient of each parameter based on the normalized change trend data set;

[0142] The entropy method is used to calculate the weight coefficients of temperature deviation value, soil moisture change value and vegetation index change value to obtain the weight coefficient set;

[0143] Calculate the weighted comprehensive index based on the weight coefficient set and the normalized change trend data set to obtain the weighted comprehensive index value;

[0144] The weighted comprehensive index is calculated using the weighted comprehensive index formula: DSI = w1×TD+w2×SM+w3×VI, where DSI represents the disaster intensity index, w1, w2, and w3 represent the weight coefficients of temperature deviation value, soil moisture change value, and vegetation index change value, respectively, TD represents the temperature deviation value, SM represents the soil moisture change value, and VI represents the vegetation index change value, thereby obtaining the weighted comprehensive index value.

[0145] Generate initial disaster intensity distribution data based on the weighted comprehensive index value;

[0146] If the weighted comprehensive index value is greater than a preset threshold, it is determined to be a high-intensity disaster area, and the first disaster intensity distribution data is generated;

[0147] For the first disaster intensity distribution data, a spatial interpolation method is used to process the missing area;

[0148] Using the Kriging interpolation algorithm to process the data spatially, the second disaster intensity distribution data is obtained;

[0149] According to the second disaster intensity distribution data, the geographical information data is fused for spatial mapping;

[0150] Using the raster processing method, the disaster intensity index is associated with the geographic coordinates to generate the final disaster intensity distribution data;

[0151] For the final disaster intensity distribution data, a data compression method is used for storage optimization;

[0152] Using the Zstandard compression algorithm to compress the data, the compressed disaster intensity distribution data is obtained.

[0153] For example, for temperature anomaly distribution data combined with soil moisture data and vegetation index data trends, the disaster intensity index is calculated and the specific implementation method of generating disaster intensity distribution data by weighted comprehensive index method is as follows: Assuming that the temperature deviation value of a certain grid point in the temperature anomaly distribution data of a certain area is 3.5 degrees Celsius, indicating that the temperature of this point is 3.5 degrees higher than the historical average value; the soil moisture data of this point is-0.2, indicating that the soil moisture is 0.2 units lower than the historical average value; the change value of the vegetation index data of this point is-0.15, indicating that the health of the vegetation has decreased by 0.15 units. First, through historical disaster data analysis and expert model, the weighted coefficients are determined as follows: the weight of temperature deviation value is 0.5, the weight of soil moisture change value is 0.3,

[0154] The weight of vegetation index change value is 0.2, which reflects the influence degree of each factor on disaster intensity. Then, based on the weighted comprehensive index formula, i.e. disaster intensity index=(temperature deviation value x temperature weight)+(soil moisture change value x humidity weight)+(vegetation index change value x vegetation weight), the specific numerical value is calculated: disaster intensity index=(3.5 x 0.5)+(-0.2 x 0.3)+(-0.15 x 0.2)=1.75-0.06-0.03=1.66. The calculation result 1.66 indicates that the disaster intensity of this grid point is medium to high.

[0155] Subsequently, the calculation method is applied to all grid points in the entire region to generate a distribution data matrix containing the disaster intensity index of each point. Finally, the matrix is mapped into a disaster intensity distribution map by a geographic information system, in which the intensity values are divided into three levels of low, medium and high from 0 to 3, and 1.66 belongs to the medium level, providing data support for subsequent disaster warning. The above process is realized automatically by algorithm, and data processing and calculation are completed by the system, which is associated with the disaster warning platform to ensure real-time updating and analysis, forming a complete logical chain from data input to distribution map output.

[0156] Further, the process of generating disaster impact range and duration prediction data includes:

[0157] Converting and cleaning the disaster intensity distribution data to obtain a standardized input data set;

[0158] Extracting spatio-temporal features based on the standardized input data set, performing time series analysis using a long short-term memory network model to obtain the disaster intensity change trend;

[0159] According to the disaster intensity change trend, combine with meteorological factors for fusion, input into the pre-established meteorological driven simulation system, generate disaster evolution path;

[0160] If the intensity of the disaster evolution path exceeds the preset threshold, calculate the impact range by the spatial interpolation algorithm to obtain the disaster impact range data;

[0161] Through the time step analysis of the disaster evolution path, the duration is predicted by using the linear regression model to obtain the disaster duration data;

[0162] Based on the disaster impact range data and duration data, fuse the geographic information system for visual processing to generate a disaster prediction distribution map;

[0163] According to the disaster prediction distribution map, extract the high-risk area features to generate disaster warning information.

[0164] For example, when importing disaster intensity distribution data into the meteorological driving simulation system, it can be achieved through an automated data interface. For example, suppose the intensity data of a typhoon disaster includes a wind speed of 80 kilometers per hour and a rainfall of 50 millimeters per hour. These data are stored in CSV format and automatically parsed and uploaded to the system database through a Python script. The script sets data validation rules to ensure that the wind speed value is within the range of 0 to 200 kilometers per hour and the rainfall is not less than 0 millimeters. After removing outliers, a standardized input data set is generated. Next, the disaster evolution process is simulated using time series analysis methods. The ARIMA model can be used for prediction. Suppose the historical data is the wind speed and rainfall sequence for the past 24 hours. Set the model parameters p = 2, d = 1, q = 1, and optimize the model by minimizing the AIC value. The calculated future 12-hour wind speed may increase to 90 kilometers per hour, and the rainfall may decrease to 30 millimeters per hour. A disaster evolution trend chart is generated.

[0165] Subsequently, based on the simulation results, disaster impact range and duration prediction data are generated. Using GIS technology combined with meteorological simulation output, suppose the typhoon center coordinates are 23.5 degrees north latitude and 118.3 degrees east longitude, the initial value of the influence radius is 50 kilometers, the radius expands to 70 kilometers over time through the wind speed decay formula, and the duration prediction is 48 hours. The system automatically generates an impact range heat map, superimposes population density data (assuming 1000 people per square kilometer) to estimate the number of people affected at about 300,000, and automatically calculates the required rescue supplies such as 5000 tents by associating with the emergency resource scheduling module. Form a complete logical chain from data input to prediction output to ensure the efficiency of disaster warning and response.

[0166] Further, the disaster impact range data and duration prediction data are compared with the preset disaster warning threshold. If the prediction data exceeds the threshold, the process of generating warning data including disaster category, impact area, and predicted time information includes:

[0167] The disaster impact range data and duration prediction data are again standardized and cleaned to obtain a standardized prediction data set.

[0168] The standardized prediction data set is compared with the preset warning threshold item by item. If any index in the prediction data set exceeds the threshold, it is marked as high-risk data, and a high-risk data set is determined.

[0169] According to the high-risk data set, the corresponding disaster category and impact area information are extracted, and the disaster type is divided to obtain classified disaster category information.

[0170] According to the classified disaster category information, combined with the prediction time data, the time series analysis tool is used for integration in the time dimension to determine the release time point of the warning information;

[0171] According to the release time point and the disaster category information, the impact area data is associated with the disaster category to generate structured warning information content;

[0172] According to the structured warning information content, the geographic information processing tool is used to perform spatial mapping on the impact area data to obtain a visual warning area distribution;

[0173] The boundary information of the high-risk area is extracted from the visual warning area distribution and recorded through the data storage module to determine the final warning information data set.

[0174] For example, for the comparison of disaster impact range and duration prediction data with the preset disaster warning threshold and the generation of warning data, the present embodiment can realize the whole process through the self-developed automatic system. Assuming that the system predicts that the impact range of a certain flood disaster is 80 kilometers in radius and the duration is 72 hours, first, the system compares these prediction data with the preset threshold, and the threshold is set to trigger the warning when the impact radius exceeds 60 kilometers or the duration exceeds 48 hours. The system automatically identifies that the current prediction value has exceeded the threshold range through the built-in logical judgment algorithm, and the specific calculation process is as follows: the impact radius of 80 kilometers is greater than the threshold of 60 kilometers, and the duration of 72 hours is greater than the threshold of 48 hours, which meets the double condition triggering rule.

[0175] Next, the system automatically extracts disaster-related information, including the disaster category as flood, the impact area center coordinates as north latitude 25.7 degrees and east longitude 116.4 degrees, and the prediction time as within the next three days. Based on these data, the system calls the warning generation module, combines the predefined warning template, and automatically generates warning data containing the disaster category, impact area, and prediction time. The warning information also includes the distribution data of key infrastructure in the impact area, such as 5 hydropower stations and 10 main roads in the area. The system calculates that the proportion of possible affected infrastructure is 60% through spatial analysis algorithm, and embeds this analysis result into the warning data to form a complete warning report.

[0176] At the same time, the system compares the warning data with the historical disaster database, and assumes that the historical data shows that a similar scale flood disaster has caused economic losses of about 200 million yuan. The system predicts that the loss range of this disaster may be between 180 million and 250 million yuan through regression analysis algorithm, further enriching the reference value of the warning data, forming a rigorous logic chain from prediction comparison to warning generation, and ensuring the integrity and accuracy of information transmission.

[0177] Further, the process of generating disaster risk distribution data by spatial interpolation method and dynamically monitoring the disaster risk distribution data comprises:

[0178] Fusing the standardized regional data set and the comprehensive data set, a fused data set is generated by using a weighted average method;

[0179] For the fused data set, a Kriging interpolation method is used to generate disaster risk distribution data, and an initial risk distribution map is obtained;

[0180] If there is data missing in the initial risk distribution map, the missing area is completed by adjacent point interpolation to obtain a complete risk distribution map;

[0181] The change trend of the complete risk distribution map is extracted by time series analysis to obtain the risk dynamic change characteristics;

[0182] According to the risk dynamic change characteristics, a sliding window method is used to calculate the risk fluctuation value to judge the stability of the risk change trend;

[0183] If the risk fluctuation value exceeds the preset threshold value, the dynamic monitoring system updates the risk distribution data to obtain the real-time risk distribution state.

[0184] For example, based on the impact area information in the early warning data, first, the key data such as rainfall, wind speed, terrain elevation and soil moisture is extracted from the comprehensive data set. Assuming that the rainfall data of a certain area is 50 mm per hour, the wind speed is 20 m / s, the elevation range is 100 to 500 meters, and the soil moisture is 30%. The Kriging spatial interpolation algorithm is used to generate disaster risk distribution data. In the specific implementation, the ordinary Kriging method is selected, the search radius is set to 5 kilometers, the weight function adopts the Gaussian model, and the risk value of each grid point (resolution 1 kilometer x 1 kilometer) is calculated. Taking rainfall as an example, the Kriging algorithm estimates the risk value of the unknown point by combining the rainfall data of the known point and the distance weight to obtain a risk distribution map, in which the high-risk area (rainfall greater than 45 mm) is concentrated in the low-lying terrain (elevation less than 200 meters). Subsequently, the disaster risk distribution data is dynamically monitored, and a time series analysis method is used in combination with real-time meteorological data stream to update the risk distribution map every 10 minutes. The monitoring system compares the current risk value with the historical threshold value.

[0185] For example, given the average rainfall in the past 24 hours is 40 mm, the system identifies the areas with increased risk, and if the risk value of a grid point exceeds 1.5 times of the threshold value (i.e. 60 mm), an early warning signal is triggered. The analysis process further combines terrain and soil data to assess the possibility of floods or landslides, for example, a grid point with an elevation of 100 meters and soil moisture exceeding 35% has an increased risk weight of 20%. Through superimposed analysis, a dynamic heat map is generated to show the distribution of risk from low (green, risk value <0.3) to high (red, risk value >0.7). The entire process can also be implemented through an automated script, with data processing using the GeoPandas library in Python and dynamic monitoring relying on the Kafka streaming processing platform to ensure real-time and accuracy.

[0186] Embodiment Two

[0187] As shown in the Figure 2 embodiment, based on the same inventive concept, the embodiment also provides a remote sensing technology-based agricultural meteorological disaster monitoring and early warning system, comprising:

[0188] a data acquisition and processing module, configured to acquire initial multi-source remote sensing data, process visible light, near-infrared, and thermal infrared band information of the initial multi-source remote sensing data, and generate a multi-dimensional data set containing time identifiers and spatial location information; wherein the multi-source remote sensing data includes surface temperature data, soil moisture data, and vegetation index data;

[0189] a data fusion module, configured to calculate fusion weight coefficients for the multi-dimensional data set through a weighted average method, and generate a fused comprehensive data set;

[0190] a temperature anomaly analysis module, configured to extract surface temperature data from the comprehensive data set, compare the surface temperature data with pre-established historical contemporaneous average surface temperature values, calculate temperature deviation values, and generate temperature anomaly distribution data;

[0191] a disaster intensity calculation module, configured to calculate disaster intensity indices through a weighted comprehensive index method for the temperature anomaly distribution data in combination with change trends of soil moisture data and vegetation index data in the comprehensive data set, and generate disaster intensity distribution data;

[0192] a disaster simulation module, configured to import the disaster intensity distribution data into a pre-established meteorological driving simulation system, simulate disaster evolution processes through a time series analysis method, and generate disaster impact range and duration prediction data;

[0193] an early warning generation module, configured to compare the disaster impact range data and the duration prediction data with pre-set disaster early warning threshold values, and if the prediction data exceeds the threshold values, generate early warning data containing disaster categories, impact areas, and prediction time information;

[0194] The risk monitoring module is configured to generate disaster risk distribution data by a spatial interpolation method in combination with data in the comprehensive data set for the impact area information in the early warning data, and to dynamically monitor the disaster risk distribution data.

[0195] The agricultural meteorological disaster monitoring and early warning system based on remote sensing technology provided by the embodiment has all the advantages of the agricultural meteorological disaster monitoring and early warning method based on remote sensing technology provided by the first embodiment.

[0196] Embodiment Three

[0197] The embodiment also discloses a computer device, which comprises a memory, a processor and a computer program stored in the memory, and the processor executes the computer program to implement the steps of the method in the first embodiment.

[0198] Embodiment Four

[0199] The embodiment also discloses a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the method in the first embodiment.

[0200] Embodiment Five

[0201] The embodiment also discloses a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the steps of the method in the first embodiment.

[0202] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited to this, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for monitoring and early warning of agricultural meteorological disasters based on remote sensing technology, characterized in that: include: Acquiring initial multi-source remote sensing data, processing visible light, near infrared, and thermal infrared band information of the initial multi-source remote sensing data to generate a multidimensional dataset containing time stamps and spatial location information; wherein the multi-source remote sensing data includes surface temperature data, soil moisture data, and vegetation index data; Calculating a fusion weight coefficient for the multidimensional data set by a weighted average method to generate a fused comprehensive data set; Extracting surface temperature data from the comprehensive data set, comparing it with the pre-established historical surface temperature average value for the same period, calculating the temperature deviation value, and generating temperature anomaly distribution data; For the temperature anomaly distribution data, combined with the change trends of the soil moisture data and vegetation index data in the comprehensive data set, a disaster intensity index is calculated by a weighted comprehensive index method to generate disaster intensity distribution data; Importing the disaster intensity distribution data into a pre-established meteorological-driven simulation system, combining it with a time series analysis method to simulate the disaster evolution process and generate disaster impact range and duration prediction data; Compare the disaster impact range data and duration prediction data with the preset disaster warning threshold. If the prediction data exceeds the threshold, generate warning data containing disaster category, impact area and prediction time information; Based on the impact area information in the early warning data and combined with the data in the comprehensive data set, disaster risk distribution data is generated through a spatial interpolation method, and the disaster risk distribution data is dynamically monitored.

2. The method according to claim 1, characterized in that The process of obtaining initial multi-source remote sensing data, processing visible light, near infrared, and thermal infrared band information of the initial multi-source remote sensing data, and generating a multidimensional dataset containing time stamps and spatial location information includes: Acquire multi-source remote sensing data, extract raw band information from visible light, near infrared, and thermal infrared bands, fuse surface temperature data, soil moisture data, and vegetation index data to generate an initial dataset; Correcting the initial data set through preprocessing, and using a radiation correction method to eliminate sensor bias in the band information to obtain a corrected band data set; If there is noise in the corrected band dataset, the Gaussian filtering algorithm is used to denoise the visible light, near infrared, and thermal infrared band information to obtain the denoised band dataset; Extracting time stamps and spatial location information from the denoised band data set, and associating the time stamps and spatial location information with the band information using a spatiotemporal indexing method to generate a spatiotemporal association data set; Extracting features from the spatiotemporal correlation dataset using a principal component analysis algorithm, fusing features of surface temperature, soil moisture, and vegetation index to obtain a feature-enhanced dataset; If the time series of the feature enhancement dataset is complete, the long short-term memory network algorithm is used to analyze the temporal variation trends of surface temperature, soil moisture and vegetation index to obtain dynamic variation characteristics; According to the dynamic change characteristics, an interpolation method is used to perform spatiotemporal completion on the feature-enhanced dataset to generate a multidimensional dataset containing time identification and spatial location information.

3. The method according to claim 1, characterized in that The process of calculating the fusion weight coefficient of the multidimensional data set by the weighted average method to generate the fused comprehensive data set includes: Extracting a data set input according to the data dimensions of the multidimensional data set, determining the data structure of each dimension, and obtaining a structured data set; By using a preset fusion algorithm, a weight coefficient is calculated for each data dimension of the structured data set to obtain a fusion weight; Using a weighted average method and combining the fusion weights, weighted calculation is performed on each data dimension of the structured data set to obtain a weighted data set; If the dimensional consistency of the weighted dataset meets a preset threshold, normalizing the weighted dataset to obtain a standardized dataset; applying a data fusion algorithm based on the standardized data set to generate a comprehensive data set; Analyze the structure of the comprehensive data set, verify the integrity of data fusion, and obtain output results; Key features are extracted from the output results to generate the final comprehensive dataset.

4. The method according to claim 1, wherein The process of generating temperature anomaly distribution data includes: Obtaining surface temperature related records from the comprehensive data set, separating target temperature data using a data screening method, and obtaining a preliminary extracted temperature data set; Based on the initially extracted temperature data set, the temperature mean data corresponding to the same period in history is obtained, and a data matching method is used to make a one-to-one correspondence to determine the correlation between the two sets of data; If the correlation between the two sets of data meets the preset matching conditions, the difference between the initially extracted temperature data set and the historical temperature mean data for the same period is calculated to obtain the temperature deviation data set; For the temperature deviation data set, a statistical analysis method is used to identify deviation values ​​exceeding a preset threshold value, and abnormal temperature data points are determined; Gridding the spatial distribution of the abnormal temperature data points to obtain regional characteristics of the abnormal distribution and determine a temperature abnormal distribution data set; Based on the temperature anomaly distribution data set, a regression analysis method is used to predict the potential change trend of the anomaly distribution to obtain a prediction result of the anomaly distribution; If the prediction result shows that the potential change trend of the abnormal distribution exceeds the preset range, a dynamic map of the abnormal distribution is generated by a data visualization tool to obtain the key focus area of ​​the abnormal distribution.

5. The method according to claim 1, wherein The disaster intensity index is calculated using the weighted comprehensive index method. The process of generating the disaster intensity distribution number includes: Obtain respective change trend data from temperature anomaly distribution data, soil moisture data and vegetation index data; The temperature deviation value, soil moisture change value and vegetation index change value were calculated using the standardization processing method to obtain the normalized change trend data set; Determining the weight coefficient of each parameter according to the normalized change trend data set; The entropy method is used to calculate the weight coefficients of temperature deviation value, soil moisture change value and vegetation index change value to obtain the weight coefficient set; Calculating a weighted comprehensive index based on the weight coefficient set and the normalized change trend data set to obtain a weighted comprehensive index value; generating initial disaster intensity distribution data according to the weighted comprehensive index value; If the weighted comprehensive index value is greater than a preset threshold, it is determined to be a high-intensity disaster area, and first disaster intensity distribution data is generated; For the first disaster intensity distribution data, a spatial interpolation method is used to process missing areas; The Kriging interpolation algorithm is used to perform spatial smoothing on the data to obtain the second disaster intensity distribution data; According to the second disaster intensity distribution data, the geographic information data is integrated to perform spatial mapping; The rasterization processing method is used to associate the disaster intensity index with the geographic coordinates to generate the final disaster intensity distribution data; For the final disaster intensity distribution data, data compression method is used for storage optimization; The data is compressed using the Zstandard compression algorithm to obtain compressed disaster intensity distribution data.

6. The method according to claim 1, characterized in that The process of generating data on the extent and duration of a disaster's impact involves: Performing format conversion and cleaning on the disaster intensity distribution data to obtain a standardized input data set; Extracting spatiotemporal features based on the standardized input data set, and using a long short-term memory network model to perform time series analysis to obtain the trend of disaster intensity changes; Based on the disaster intensity change trend, combined with meteorological factors, the data is integrated and input into a pre-established meteorological-driven simulation system to generate a disaster evolution path; If the intensity of the disaster evolution path exceeds a preset threshold, the impact range is calculated using a spatial interpolation algorithm to obtain disaster impact range data; Through the time step analysis of the disaster evolution path, the linear regression model is used to predict the duration and obtain the disaster duration data; Based on the disaster impact range data and duration data, the geographic information system is integrated for visualization processing to generate a disaster prediction distribution map; Based on the disaster prediction distribution map, the characteristics of high-risk areas are extracted and disaster warning information is generated.

7. The method according to claim 1, characterized in that The disaster impact range data and duration prediction data are compared with a preset disaster warning threshold. If the prediction data exceeds the threshold, a process of generating warning data containing disaster category, impact area, and prediction time information includes: The disaster impact range data and duration prediction data are further formatted and cleaned to obtain a standardized prediction data set; Comparing the normalized prediction data set with the preset warning thresholds item by item, if any indicator in the prediction data set exceeds the threshold, it is marked as high-risk data, and a high-risk data set is determined; Extracting corresponding disaster categories and affected area information based on the high-risk data set, and classifying disaster types to obtain classified disaster category information; Based on the classified disaster category information, combined with the predicted time data, a time series analysis tool is used to integrate the time dimension to determine the release time point of the warning information; According to the release time point and disaster category information, the affected area data is associated with the disaster category to generate structured warning information content; Based on the structured warning information content, geographic information processing tools are used to spatially map the affected area data to obtain a visual distribution of warning areas; The boundary information of high-risk areas is extracted from the visual warning area distribution, recorded through the data storage module, and the final warning information data set is determined.

8. The method according to claim 1, characterized in that The process of generating disaster risk distribution data by using a spatial interpolation method and dynamically monitoring the disaster risk distribution data includes: The standardized regional dataset is fused with the comprehensive dataset and a weighted average method is used to generate a fused dataset. For the fused data set, a Kriging interpolation method is used to generate disaster risk distribution data to obtain an initial risk distribution map; If there is data missing in the initial risk distribution map, the missing area is filled by interpolating adjacent points to obtain a complete risk distribution map; Through time series analysis, the changing trend of the complete risk distribution map is extracted to obtain the dynamic change characteristics of the risk; Based on the dynamic change characteristics of the risk, the sliding window method is used to calculate the risk fluctuation value and determine the stability of the risk change trend; If the risk fluctuation value exceeds the preset threshold, the dynamic monitoring system is triggered to update the risk distribution data and obtain the real-time risk distribution status.

9. An agricultural meteorological disaster monitoring and early warning system based on remote sensing technology, characterized in that: include: A data acquisition and processing module is used to obtain initial multi-source remote sensing data, process the visible light, near infrared, and thermal infrared band information of the initial multi-source remote sensing data, and generate a multi-dimensional dataset containing time stamps and spatial location information; wherein the multi-source remote sensing data includes surface temperature data, soil moisture data, and vegetation index data; A data fusion module, configured to calculate a fusion weight coefficient for the multidimensional data set by a weighted average method to generate a fused comprehensive data set; a temperature anomaly analysis module, configured to extract surface temperature data from the comprehensive data set, compare the data with a pre-established historical average surface temperature value for the same period, calculate a temperature deviation value, and generate temperature anomaly distribution data; a disaster intensity calculation module, configured to calculate a disaster intensity index by a weighted comprehensive index method based on the temperature anomaly distribution data and the change trends of the soil moisture data and vegetation index data in the comprehensive data set, thereby generating disaster intensity distribution data; A disaster simulation module is used to import the disaster intensity distribution data into a pre-established meteorological-driven simulation system, combine it with a time series analysis method, simulate the disaster evolution process, and generate disaster impact range and duration prediction data; An early warning generation module is used to compare the disaster impact range data and duration prediction data with a preset disaster early warning threshold value. If the prediction data exceeds the threshold value, early warning data containing disaster category, impact area and prediction time information is generated; The risk monitoring module is used to generate disaster risk distribution data through a spatial interpolation method based on the impact area information in the early warning data and the data in the comprehensive data set, and dynamically monitor the disaster risk distribution data.

Images