Drought monitoring method and system based on ground feature detection

Abstract

The present invention provides a drought monitoring method and system based on ground feature detection, which relates to the field of ground feature detection technology, including: retrieving geological distribution data of a target monitoring area, performing preliminary division, and obtaining multiple initial division grid areas; performing sensor deployment, and obtaining multiple sensor deployment arrays; connecting a satellite remote sensing platform in a preset detection window, and obtaining multiple remote sensing images; performing ground feature data collection, and obtaining multiple ground feature data sets; performing secondary similar aggregation, and determining multiple aggregation division grid area sets; determining multiple first monitoring bandwidths and multiple second monitoring bandwidths, and constructing multiple first monitoring channels and multiple second monitoring channels of a drought monitor for drought monitoring. The present invention solves the technical problem that the drought monitoring method of the prior art usually relies on a static monitoring strategy, and cannot automatically adjust the monitoring strategy according to the real-time status of drought changes, resulting in low monitoring efficiency and poor accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of ground feature detection, and in particular to a drought monitoring method and system based on ground feature detection. Background Art

[0002] Drought monitoring is one of the most important tasks in agricultural management, especially in drought-prone areas. Accurate drought monitoring and prediction are of great significance to ensuring agricultural output, preventing soil degradation and rationally utilizing groundwater resources. However, traditional monitoring methods usually use fixed frequency or static strategies for data collection and analysis, lacking the ability to respond dynamically to changes in drought conditions. In practical applications, changes in drought conditions often have significant temporal and spatial heterogeneity, and the speed and trend of drought development in different regions may vary significantly. For example, in areas where water is rapidly lost, static monitoring may not be able to capture the rapid spread of drought in a timely manner, while in more stable areas, too frequent monitoring will cause data redundancy and waste of resources. Summary of the invention

[0003] The present application provides a drought monitoring method and system based on ground feature detection, aiming to solve the technical problems that the drought monitoring methods in the prior art usually rely on static monitoring strategies and cannot automatically adjust the monitoring strategies according to the real-time status of drought changes, resulting in low monitoring efficiency and poor accuracy.

[0004] The first aspect disclosed in the present application provides a drought monitoring method based on ground feature detection, the method comprising: retrieving geological distribution data of a target monitoring area, and preliminarily dividing the target monitoring area according to the geological distribution data to obtain a plurality of initially divided grid areas; deploying sensors based on the area sizes and geological features of the plurality of initially divided grid areas to obtain a plurality of deployed sensor deployment arrays; connecting a satellite remote sensing platform in a preset detection window to obtain a plurality of remote sensing images of the plurality of initially divided grid areas; collecting monitoring data of the plurality of sensor deployment arrays in the preset detection window to obtain a plurality of ground feature data sets; and monitoring the plurality of remote sensing images as background information and the plurality of ground feature data sets as foreground information. A plurality of initially divided grid areas are subjected to secondary similar aggregation to determine a plurality of aggregated divided grid area sets, wherein the plurality of aggregated divided grid area sets include a plurality of aggregated ground feature concentrated value sets; a plurality of first monitoring bandwidths and a plurality of second monitoring bandwidths are determined based on the plurality of aggregated ground feature concentrated value sets, and a plurality of first monitoring channels are constructed according to the plurality of first monitoring bandwidths and a plurality of second monitoring channels are constructed according to the plurality of second monitoring bandwidths, a drought monitor is constructed by one-to-one mapping the plurality of first monitoring channels and the plurality of second monitoring channels, and drought monitors are performed on the plurality of aggregated divided grid area sets using the drought monitor, wherein each aggregated divided grid area set is subjected to drought monitoring through the one-to-one mapped first monitoring channel and second monitoring channel.

[0005] The second aspect disclosed in the present application provides a drought monitoring system based on ground feature detection, the system is used for the above-mentioned drought monitoring method based on ground feature detection, the system includes: a region division module, the region division module is used to retrieve the geological distribution data of the target monitoring area, and preliminarily divide the target monitoring area according to the geological distribution data to obtain a plurality of initially divided grid areas; a sensor deployment module, the sensor deployment module is used to deploy sensors based on the area size and geological characteristics of the plurality of initially divided grid areas to obtain a plurality of sensor deployment arrays that have been deployed; a remote sensing image acquisition module, the remote sensing image acquisition module is used to connect to a satellite remote sensing platform in a preset detection window to obtain a plurality of remote sensing images of the plurality of initially divided grid areas; a data acquisition module, the data acquisition module is used to collect monitoring data of the plurality of sensor deployment arrays in the preset detection window to obtain a plurality of ground feature data sets; and a similar aggregation module. A module, wherein the same type aggregation module is used to perform secondary same type aggregation on the multiple initial divided grid areas with the multiple remote sensing images as background information and the multiple ground feature data sets as foreground information, and determine multiple aggregated divided grid area sets, wherein the multiple aggregated divided grid area sets include multiple aggregated ground feature concentrated value sets; a drought monitoring module, wherein the drought monitoring module is used to determine multiple first monitoring bandwidths and multiple second monitoring bandwidths based on the multiple aggregated ground feature concentrated value sets, and construct multiple first monitoring channels according to the multiple first monitoring bandwidths and construct multiple second monitoring channels according to the multiple second monitoring bandwidths, and perform one-to-one mapping of the multiple first monitoring channels and the multiple second monitoring channels to construct a drought monitor, and use the drought monitor to perform drought monitoring on the multiple aggregated divided grid area sets, wherein each aggregated divided grid area set performs drought monitoring through the one-to-one mapped first monitoring channel and second monitoring channel.

[0006] One or more technical solutions provided in this application have at least the following beneficial effects:

[0007] By retrieving the geological distribution data of the target monitoring area, a preliminary division of the area is carried out, and the grid areas with different geological characteristics are accurately identified, thus realizing a detailed division of the monitoring area. Since each grid area has different geological characteristics and areas, its water retention capacity, vegetation coverage and other responses to drought are also different. This precise division allows the optimization of sensor layout according to geological characteristics and area size, and obtains multiple sensor layout arrays, making the layout of monitoring equipment more targeted, improving monitoring efficiency and reducing costs. By connecting to the satellite remote sensing platform to obtain multiple remote sensing images of multiple initially divided grid areas, and combining multiple sensor layout arrays to collect ground feature data, the fusion of ground features and remote sensing data is realized. By using remote sensing images as the background and ground sensor data as the foreground, a secondary aggregation analysis of multiple information is performed, which can more accurately reflect the soil and vegetation status, groundwater changes, etc., and enhance the monitoring accuracy. ; By analyzing the concentrated values ​​of multiple aggregated ground features, the first monitoring bandwidth and the second monitoring bandwidth of each aggregated area are automatically determined to form different monitoring channels. The monitoring bandwidth is dynamically adjusted according to the deviation amount. A smaller bandwidth is used for areas with larger fluctuations, and a larger bandwidth is used for areas with smaller fluctuations, thereby realizing an adaptive monitoring strategy, which not only ensures high-frequency monitoring of abnormally changing areas, but also reduces the monitoring burden on stable areas, reduces unnecessary data collection and processing, and improves the efficiency and response sensitivity of drought monitoring; after constructing multiple first monitoring channels and second monitoring channels, each aggregated area can be continuously and dynamically monitored, and the real-time data inflow of the monitoring channels can be used to quickly respond to changes in drought conditions, especially for areas with rapid water loss or severe vegetation degradation. The high-frequency monitoring mode can be triggered in time to quickly discover potential drought crises, thereby helping disaster prevention and mitigation departments to better make early prevention and intervention measures.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic diagram of a drought monitoring method based on ground feature detection provided in an embodiment of the present application;

[0010] Figure 2 A schematic diagram of the structure of a drought monitoring system based on ground feature detection provided in an embodiment of the present application.

[0011] Explanation of reference numerals: region division module 10 , sensor deployment module 20 , remote sensing image acquisition module 30 , data collection module 40 , similarity aggregation module 50 , drought monitoring module 60 . DETAILED DESCRIPTION

[0012] The embodiments of the present application provide a drought monitoring method and system based on ground feature detection, thereby solving the technical problem that the drought monitoring methods in the prior art usually rely on static monitoring strategies and cannot automatically adjust the monitoring strategies according to the real-time status of drought changes, resulting in low monitoring efficiency and poor accuracy.

[0013] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0014] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a drought monitoring method based on ground feature detection, the method comprising:

[0015] The geological distribution data of the target monitoring area is retrieved, and the target monitoring area is preliminarily divided according to the geological distribution data to obtain a plurality of initial divided grid areas.

[0016] First, the location and scope of the target monitoring area are clarified, such as specific farmland in a province, city or region, land around a reservoir, etc., and the geological distribution data of the target monitoring area is obtained from the geological survey database, including but not limited to soil type, rock structure, groundwater layer distribution, surface height and surface slope. In the retrieved data, key features that can reflect changes in drought conditions are selected for analysis, including water retention capacity, surface water flow trends and vegetation coverage characteristics. The grid size is determined according to the total area of ​​the target monitoring area. The grid is the basic unit for dividing the target monitoring area into multiple small units. According to the feature analysis results, a grid divider is used to divide the target area into small unit grids to obtain multiple initial divided grid areas, each of which has similar geological characteristics.

[0017] Furthermore, the geological distribution data of the target monitoring area is retrieved, and the target monitoring area is preliminarily divided according to the geological distribution data to obtain a plurality of initial divided grid areas, including:

[0018] Connecting to a geological survey database to obtain geological distribution data of the target monitoring area, wherein the geological distribution data includes soil type, rock structure, groundwater layer distribution, surface height and surface slope; obtaining the total area of ​​the target monitoring area, and determining grid size constraints based on the total area; using water retention capacity, surface water flow trend and vegetation coverage characteristics as a preliminary division indicator set; using a grid divider to analyze and identify the geological distribution data, grid size constraints and the preliminary division indicator set, and obtaining the multiple initial division grid areas.

[0019] The geological survey database is a pre-built database for storing geological data of different regions. The geological distribution data of the target monitoring area is retrieved through the geological survey database, including soil type, rock layer structure, groundwater layer distribution, surface height and surface slope. Among them, the soil type reflects the composition, particle structure and chemical properties of the surface soil, such as clay, sandy soil, loam, etc.; the rock layer structure includes the type of underground rock, layered structure and its depth information; the groundwater layer distribution shows the depth, water content and distribution range of the groundwater layer; the surface height is the height data of the surface in the target area; the surface slope is calculated from the surface height data, reflecting the terrain undulation and water flow direction.

[0020] Determine the boundary range of the target monitoring area, such as using GIS tools to extract the boundary and obtain the total area of ​​the target area. The grid size is the spatial resolution of each grid unit. According to the total area of ​​the target monitoring area and the monitoring accuracy requirements, set the appropriate grid unit size. If the total area is large, use a larger grid unit to reduce the complexity of data processing; if the total area is small, use a smaller grid unit to improve the spatial resolution.

[0021] Water retention capacity, surface water flow trend, and vegetation coverage characteristics are used as a set of preliminary division indicators to comprehensively evaluate the drought resistance of each grid unit. Among them, soil type is an important factor in determining water retention capacity. Sandy soil usually has poor water retention capacity, while clay soil or loam has better water retention capacity. The flow direction and convergence point of surface water are calculated based on surface height and slope data. Areas with large slopes have fast water flow speeds and are prone to water loss. Flat areas have slow water flow and are more likely to retain water. Vegetation coverage is evaluated through remote sensing images or land cover data. Areas with high vegetation coverage often have better water retention capacity.

[0022] The grid divider is a tool for automatically dividing the target area. According to the input geological distribution data, grid size constraints and a set of preliminary division indicators, the target monitoring area is rasterized and divided into multiple grid cells with different geological characteristics. Each grid cell represents a small space in the target monitoring area. Each grid cell has a spatial position and area size, forming multiple initial divided grid areas for further monitoring and analysis.

[0023] Sensors are deployed based on the area sizes and geological features of the multiple initially divided grid areas to obtain multiple sensor deployment arrays.

[0024] The area size of each initially divided grid area is counted to obtain the area of ​​multiple grid areas; the geological feature data of each grid area is extracted, including soil type, rock structure, groundwater layer distribution, surface height and surface slope, to obtain a set of geological features of multiple grid areas. The strategy of sensor deployment is set according to the area and geological characteristics of the grid area. Different sensor types and deployment densities are optimized according to different geological characteristics and areas to ensure the effectiveness and accuracy of data acquisition. Specifically, first, the sensor type is determined. For example, soil moisture sensors are deployed in areas with weak water retention capacity or low vegetation coverage to monitor the water content in the soil; meteorological sensors are deployed in grid areas with large surface temperature differences to monitor meteorological data such as temperature, precipitation, wind speed and wind direction. Furthermore, the deployment density of sensors is determined according to the area size and geological characteristics of the grid area. The deployment density of grid areas with larger areas is lower to reduce resource consumption; the deployment density of grid areas with complex geological characteristics is higher to obtain higher resolution data. According to the analysis results, sensors are arranged in the target grid area to form multiple sensor deployment arrays. Through the layout of the array, comprehensive data coverage of each grid unit in the monitoring area can be achieved.

[0025] Furthermore, the sensors are deployed based on the area sizes and geological features of the multiple initially divided grid regions to obtain multiple sensor deployment arrays that have been deployed, including:

[0026] Based on the geological distribution data and the multiple initially divided grid areas, geological feature extraction is performed to obtain multiple grid area geological feature sets; the areas of the multiple initially divided grid areas are respectively counted to obtain multiple grid area areas; sensor deployment schemes are identified based on the multiple grid area areas and the multiple grid area geological feature sets to determine multiple sensor deployment schemes; based on the multiple sensor deployment schemes, sensors are deployed in the multiple initially divided grid areas to obtain multiple sensor deployment arrays.

[0027] Cross-match the geological distribution data with each grid area to extract the geological features corresponding to each grid unit. Specifically, based on the soil distribution data, identify the main soil types in each grid unit, such as sandy soil, clay soil, loam, etc., to obtain soil type features; extract the groundwater layer distribution information in each grid unit, including data such as groundwater depth, water content and phreatic layer thickness, to obtain groundwater layer features; extract the rock layer type, layer thickness distribution and its anti-permeability features in each grid unit to obtain rock layer structure features. Summarize all the geological features extracted from each grid unit to form multiple grid area geological feature sets, that is, the geological feature data set of each grid unit.

[0028] Use GIS tools or spatial analysis software to calculate the actual area of ​​each initially divided grid area. The area can be expressed in square meters or square kilometers, depending on the overall scale of the target monitoring area and the resolution of the grid. Integrate the area data of each grid area to form multiple grid area areas.

[0029] According to the geological characteristics and area size of each grid area, identify the appropriate sensor type, quantity and layout density. Specifically, match the geological characteristics with the sensor type. Specifically, in areas with low soil moisture or rapid water loss, soil moisture sensors are preferentially deployed; water flow monitoring sensors are deployed in areas with large slopes and strong water flow trends; groundwater monitoring sensors are deployed in areas with dense groundwater layers. Match the area with the sensor density. Specifically, the layout density is lower for grid areas with larger areas to reduce resource consumption; the layout density is higher for grid areas with complex geological characteristics to obtain higher resolution data. Generate a layout plan for each grid area based on the analysis results of sensor type, quantity and layout density, and determine multiple sensor layout plans.

[0030] According to the sensor type, quantity and density in the sensor deployment plan, sensors are deployed in the corresponding grid area to obtain multiple sensor deployment arrays. Each deployment array contains multiple sensor nodes, and data can be transmitted between nodes through a wireless network.

[0031] A satellite remote sensing platform is connected in a preset detection window to obtain a plurality of remote sensing images of the plurality of initially divided grid areas.

[0032] The detection window is a time period set in the drought monitoring process, which is used to synchronize satellite remote sensing and ground sensor data collection. The preset detection window can be defined according to the season, crop growth cycle, and potential time period of drought. The satellite remote sensing platform is a sensor and its control system carried on a remote sensing satellite, which can monitor and collect data from the earth from space. The remote sensing platform is divided into optical remote sensing, radar remote sensing, and thermal infrared remote sensing according to the sensor type. During the preset detection window, a data request is sent to retrieve multiple multi-phase images of the initially divided grid area. The multi-phase image is a remote sensing image sequence of the same monitoring area obtained in different time periods. This image sequence can reflect the changes in surface characteristics of the monitoring area over time, such as vegetation index, surface moisture distribution, surface temperature changes, etc. Multiple remote sensing images are obtained by integration, and each image corresponds to an initially divided grid area.

[0033] The monitoring data of the multiple sensor arrays within the preset detection window are collected to obtain multiple ground feature data sets.

[0034] During the preset detection window, multiple sensor arrays deployed in different grid areas are activated. Each sensor node collects real-time ground feature data in a specific grid area and transmits it to the central data platform to form multiple ground feature data sets. The ground feature data set is a collection of ground data collected by multiple sensor nodes within the preset detection window. The data set includes soil, meteorological and groundwater data at different times and in different regional locations, and is an important data source for drought monitoring.

[0035] The multiple remote sensing images are used as background information and the multiple ground feature data sets are used as foreground information to perform secondary homogeneous aggregation on the multiple initial divided grid areas to determine multiple aggregated divided grid area sets, wherein the multiple aggregated divided grid area sets include multiple aggregated ground feature concentration value sets.

[0036] Image semantic recognition technology is used to analyze multiple remote sensing images and extract multiple image background features, including vegetation index, surface reflectivity and surface temperature. Based on the above background feature indicators, the background vector of each initial divided grid area is constructed, and each background vector represents the overall characteristics of the area in the remote sensing image. The ground feature data set in each grid area is converted into a foreground vector, and each foreground vector represents the feature distribution of the grid area in the ground sensor data. A similarity clustering algorithm, such as K-means, clustering analysis algorithm DBSCAN or aggregation algorithm based on feature similarity, is used to regionally aggregate the background vector and foreground vector of each grid area. During the aggregation process, each grid area is represented as a point, and its corresponding background vector and foreground vector are used as the features of the point. The similarity of each grid area point is calculated in the feature space, and the grid areas with high similarity are merged using the aggregation algorithm to form a new aggregated divided grid area set. Among them, multiple aggregated divided grid area sets include multiple aggregated ground feature concentrated value sets. The aggregated ground feature concentrated value is the comprehensive statistical value of all ground feature data in the aggregated grid area, such as average soil moisture, average surface temperature, etc. The concentrated value reflects the overall ground feature status of the aggregated area and is an important reference data in drought monitoring.

[0037] Based on the multiple aggregated ground feature concentrated value sets, multiple first monitoring bandwidths and multiple second monitoring bandwidths are determined, and multiple first monitoring channels are constructed according to the multiple first monitoring bandwidths and multiple second monitoring channels are constructed according to the multiple second monitoring bandwidths. The multiple first monitoring channels and the multiple second monitoring channels are mapped one-to-one to construct a drought monitor, and the drought monitor is used to monitor the multiple aggregated divided grid area sets, wherein each aggregated divided grid area set is monitored for drought through the one-to-one mapped first monitoring channel and second monitoring channel.

[0038] Monitoring bandwidth refers to the monitoring threshold range set for different monitoring indicators, such as soil moisture, surface temperature, vegetation index, etc. The bandwidth defines the sensitivity and monitoring interval of a certain monitoring channel to specific ground features. The first monitoring bandwidth is used to reflect the baseline value or normal interval value of the target feature. The setting of the first monitoring bandwidth is based on the overall mean set of aggregated ground feature concentration values, that is, the baseline bandwidth is defined based on the feature distribution under normal conditions; the second monitoring bandwidth is used to reflect abnormal conditions or feature intervals that deviate from the normal state, and is calculated based on the deviation of the feature data, that is, it is defined based on the deviation between the ground feature concentration value set and the overall mean set.

[0039] The monitoring channel is a drought monitoring path set for different features. Each channel corresponds to the monitoring of a specific feature and its bandwidth interval. Through different monitoring channels, the impact of different features on drought can be evaluated separately. Multiple first monitoring channels are set according to the first monitoring bandwidth to detect whether the aggregated grid area is in a normal state; multiple second monitoring channels are set according to the second monitoring bandwidth to detect the deviation and abnormality of the feature data. By integrating multiple first monitoring channels and multiple second monitoring channels, a drought monitor is obtained. The drought monitor can output areas with abnormal characteristics, such as too low soil moisture, according to the output results of different monitoring channels, to achieve drought monitoring.

[0040] Further, determining a plurality of first monitoring bandwidths and a plurality of second monitoring bandwidths based on the plurality of aggregated ground feature concentrated value sets includes:

[0041] Taking ground features as indexes, the overall mean of each ground feature in the target monitoring area is calculated in combination with multiple aggregated ground feature concentrated value sets to obtain an aggregated ground feature concentrated value set; deviations are calculated based on the multiple aggregated ground feature concentrated value sets and the aggregated ground feature concentrated value set to determine multiple aggregated ground feature concentrated value deviation sets; deviation sets in the multiple aggregated ground feature sets are traversed to calculate deviation means to obtain multiple deviation means; and the multiple first monitoring bandwidths and the multiple second monitoring bandwidths are determined based on the sizes of the multiple deviation means.

[0042] Taking ground features as indexes, multiple aggregated ground feature concentrated value sets are combined. Each set contains ground feature concentrated values ​​in different aggregated grid areas, such as the average soil moisture and temperature of each aggregated area. For each ground feature, the overall mean of all aggregated grid areas in the target monitoring area is calculated to form an aggregated ground feature concentrated mean set.

[0043] The deviation is the difference between the feature concentration value of each aggregated grid area and the overall mean value of the feature in the entire target monitoring area. By calculating the deviation, the abnormal distribution of features in different grid areas can be identified. A deviation greater than 0 indicates that the feature value of the area is higher than the overall mean, and a deviation less than 0 indicates that the feature value of the area is lower than the overall mean. For each feature in each aggregated grid area, the deviation between it and the overall mean value of the feature in the target monitoring area is calculated to obtain multiple aggregated ground feature concentration value deviation sets, which can intuitively reflect the deviation of each ground feature in each aggregated grid area.

[0044] The mean deviation is used to measure the average deviation of any feature from the overall mean in all aggregated grid areas. For any feature, the sum of the absolute values ​​of the deviations of the feature in each aggregated grid area is calculated, and then divided by the number of aggregated grid areas to obtain the mean deviation of the feature. The mean deviation of each ground feature in all aggregated grid areas is calculated in turn to obtain multiple mean deviations, which represent the average deviation of all features in the monitoring area.

[0045] Calculate the deviation mean ratio of each ground feature. The ratio is used to measure the deviation between different ground features to determine the importance and priority of their monitoring. That is, calculate the sum of the deviation mean of any feature divided by the deviation mean of the ground features to obtain the deviation mean ratio of the feature. The larger the ratio, the higher the proportion of the deviation of the feature in all features, indicating that the feature has a strong volatility in the monitoring area and there are large abnormalities. Therefore, it requires closer monitoring, and the corresponding monitoring bandwidth is relatively small, and the time interval between two adjacent monitorings is short; conversely, the smaller the ratio, the lower the volatility of the feature and the relatively uniform distribution. Therefore, the corresponding monitoring bandwidth is relatively large, and the time interval between two adjacent monitorings is longer. In this way, multiple first monitoring bandwidths and multiple second monitoring bandwidths are set. The first monitoring bandwidth is used to reflect the normal monitoring frequency under the baseline state, and the second monitoring bandwidth is used to reflect the high-frequency monitoring under abnormal conditions.

[0046] Further, the multiple remote sensing images are used as background information, and the multiple ground feature data sets are used as foreground information to perform secondary homogeneous aggregation on the multiple initial divided grid areas to determine multiple aggregated divided grid area sets, including:

[0047] The multiple remote sensing images are analyzed by using an image semantic recognizer to determine multiple image background features, and the multiple image background features are used as multiple background vectors, wherein each image background feature includes a vegetation index, a surface reflectivity, and a surface temperature; using ground features as vectors, foreground vectors are constructed for the multiple ground feature data sets to obtain multiple foreground vectors; based on the multiple background vectors and the multiple foreground vectors, the multiple initially divided grid areas are secondary aggregated in the same manner to obtain the multiple aggregated divided grid area sets, wherein each background vector corresponds to a foreground vector.

[0048] Image semantic recognizer is an analysis tool based on deep learning or traditional image processing technology. It can identify and extract different features in remote sensing images, including vegetation, surface reflectivity, and surface temperature. Image semantic recognition tools include convolutional neural networks and support vector machines. Use image semantic recognizer to extract background feature data in remote sensing images, including vegetation index, surface reflectivity, and surface temperature. Vegetation index is a remote sensing feature parameter used to measure the coverage and health of surface vegetation. The commonly used vegetation index is NDVI, and the calculation formula is: , NIR represents the reflectivity of the near-infrared band, RED represents the reflectivity of the red light band, the higher the NDVI value is, the better the vegetation coverage and health; the surface reflectivity represents the reflection characteristics of surface materials, such as soil, vegetation, and water bodies, to solar radiation. The reflectivity of different materials varies significantly in different bands and can be used to distinguish surface types; surface temperature is an important indicator reflecting the energy balance of the surface and is used to evaluate surface evaporation, vegetation status, and drought conditions. Surface temperature information can be accurately extracted through thermal infrared band data.

[0049] Each image background feature is converted into a numerical vector form to represent the value of each pixel on each feature, where each component of the vector represents the average value of the vegetation index, surface reflectivity and surface temperature in the grid area. A background vector is generated for each grid area, and the background vectors of all grid areas in the target monitoring area are aggregated to form multiple background vectors. The background vector represents the overall feature state of all pixels in the grid area.

[0050] In the aforementioned steps, multiple ground feature data sets are collected through the ground sensor array, each set contains sensor data from different grid areas, the ground feature data in each initially divided grid area are statistically analyzed, the average value of each feature is calculated, and the statistical values ​​of each ground feature are combined into a vector form to form a foreground vector for the area, wherein each component represents the mean of the soil moisture, surface temperature and groundwater level in the grid area, and the foreground vectors of each grid area are aggregated to form multiple foreground vectors, which represent the comprehensive characteristics of the ground feature data in each grid area in the feature space.

[0051] Each grid area has a background vector (representing the background features in the remote sensing image) and a foreground vector (representing the ground feature data). The similarity between the background vector and the foreground vector of each grid area is calculated. Based on the calculated similarity, the initial divided grid areas with similar features are merged to form multiple aggregated grid area sets. Each set represents a group of grid areas that are similar in both background and foreground features.

[0052] Further, performing secondary homogeneous aggregation on the multiple initial divided grid regions based on the multiple background vectors and the multiple foreground vectors to obtain the multiple aggregated divided grid region sets includes:

[0053] Constructing an aggregation space, inputting the multiple background vectors, the multiple foreground vectors and the multiple initial divided grid regions into the aggregation space, obtaining multiple spatial points and multiple spatial point coordinates, wherein each spatial point corresponds to an initial divided grid region; selecting K spatial points from the multiple spatial points according to a preset screening rule, wherein the preset screening rule is that the distance between any two spatial points of the screened K spatial points is greater than or equal to a preset distance threshold, and K is an integer greater than or equal to 1; taking the K spatial points as the centers of the K grid regions, performing a pre-set iteration step in the aggregation space. Perform an iterative search to determine K initial aggregated divided grid area sets; obtain multiple discrete space points in the aggregated space that are not divided into grid areas, perform neighbor authentication on the multiple discrete space points, and if the authentication is successful, add them into a neighbor aggregated divided grid area set to obtain multiple neighbor aggregated divided grid area sets; if the authentication is unsuccessful, add the multiple discrete space points into the initial aggregated divided grid area set with the closest distance; obtain the multiple aggregated divided grid area sets based on the K initial aggregated divided grid area sets and multiple neighbor aggregated divided grid area sets.

[0054] Constructing an aggregation space, the aggregation space refers to mapping multiple grid regions with multidimensional feature data into a mathematical space, usually a feature space or a similarity space, which is used to analyze the similarity of these grid regions. In this aggregation space, each grid region is represented by its background vector and foreground vector to form a multidimensional space point.

[0055] Multiple background vectors, multiple foreground vectors, and multiple initial grid regions are input into the aggregation space. For each grid region, its background vector and foreground vector are combined to form a composite vector. Each composite vector represents the position of a spatial point in the aggregation space. Each spatial point corresponds to an initial grid region and is matched one by one with its background vector and foreground vector. Therefore, the feature data (background and foreground) of each grid region can be represented by a unique spatial point. The composite vectors of all grid regions are input into the aggregation space to form multiple spatial points and multiple spatial point coordinates.

[0056] The preset distance threshold is the standard for screening spatial points, which indicates the minimum distance between any two screened spatial points in the aggregated space. K indicates the number of spatial points that need to be screened from all spatial points. K is an integer greater than or equal to 1. The K value can be determined based on the size of the monitoring area, the complexity of the feature distribution, and the computing resources.

[0057] In the multidimensional aggregation space, the distance between any two spatial points represents the degree of difference between the two grid areas in all features. The distance can be calculated using Euclidean distance, Manhattan distance or other similarity measurement methods. Two spatial points are randomly selected from all spatial points and the distance between them is calculated. If the distance is greater than or equal to the preset distance threshold, the two points are added to the screening set. When the number of screened spatial points reaches the preset K value, or no new points that meet the distance condition can be found, the screening process is terminated and K spatial points are output.

[0058] K spatial points are taken as the centers of K grid regions and as the core points of subsequent aggregation analysis. The preset iteration step size represents the distance expanded outward from the aggregation center during each search. The size of the step size determines the speed and accuracy of the search.

[0059] Iterative search is performed. Specifically, starting from the center of each grid area, the area search is performed outward according to the preset step size to determine whether the surrounding non-aggregated spatial points meet the aggregation conditions. For example, when the similarity between the spatial point and the aggregation center is higher than a preset value, the point is included in the aggregation area. In each iteration, the newly added spatial points in each aggregation area are checked. If the newly added points can significantly change the feature mean of the aggregation area, the aggregation center of the aggregation area is updated. When all the neighboring spatial points of a certain aggregation area cannot meet the aggregation conditions, the expansion of the area is stopped and it is marked as a stable aggregation partition area. Finally, K initial aggregation partition grid area sets are formed, each set consisting of an initial aggregation center and its neighboring spatial points.

[0060] During the iterative aggregation process, some spatial points will not be included in any of the initial aggregation areas. These spatial points are discrete spatial points, which are usually located at the border of the initial aggregation area or in an area far from the initial aggregation center. Nearest neighbor authentication is to determine whether multiple discrete spatial points can be included in the same aggregation area. If the similarity of a pair of discrete spatial points is greater than the preset threshold, the pair of points is determined to have passed the nearest neighbor authentication.

[0061] All the multiple discrete space points that have not been aggregated are authenticated one by one by the nearest neighbor. Specifically, for each discrete space point, its similarity with the remaining discrete space points is calculated. If a discrete space point has a high similarity with this discrete space point, it means that the two discrete space points have passed the nearest neighbor authentication. Then they are added to the same aggregation area to form a new nearest neighbor aggregation division area. Finally, all the discrete space points that have passed the nearest neighbor authentication are divided into one or more new nearest neighbor aggregation areas to form multiple nearest neighbor aggregation division grid area sets.

[0062] For each discrete spatial point, its similarity with the remaining discrete spatial points is calculated. If the similarity between the discrete spatial point and any discrete spatial point is lower than the preset threshold, the authentication is unsuccessful. In this case, all discrete points that failed to be authenticated are merged one by one into the corresponding nearest initial aggregation area according to the principle of minimum distance. By incorporating unauthenticated points into the nearest aggregation area, the formation of isolated spatial points can be avoided.

[0063] K initial aggregated partitioned grid region sets and a plurality of neighboring aggregated partitioned grid region sets are integrated to obtain a plurality of aggregated partitioned grid region sets, each of which has similar features.

[0064] Furthermore, the mean value of each ground feature within the plurality of aggregated divided grid area sets is calculated according to the plurality of ground feature data sets to obtain the plurality of aggregated ground feature concentrated value sets.

[0065] The ground feature data set is the ground feature data of each grid unit in the area collected by the ground sensor array. The data types include soil moisture, surface temperature, groundwater level, etc. For each aggregated grid area set, the ground feature data of all grid units in the set are extracted. For each ground feature, the data mean of all grid units in the aggregated grid area set is calculated to form an aggregated ground feature concentrated value set. The mean of all ground features in each aggregated grid area is calculated one by one to obtain multiple aggregated ground feature concentrated value sets.

[0066] In summary, the drought monitoring method based on ground feature detection provided by the embodiments of the present application has the following technical effects:

[0067] By retrieving the geological distribution data of the target monitoring area, a preliminary division of the area is carried out, and the grid areas with different geological characteristics are accurately identified, thus realizing a detailed division of the monitoring area. Since each grid area has different geological characteristics and areas, its water retention capacity, vegetation coverage and other responses to drought are also different. This precise division allows the optimization of sensor layout according to geological characteristics and area size, and obtains multiple sensor layout arrays, making the layout of monitoring equipment more targeted, improving monitoring efficiency and reducing costs. By connecting to the satellite remote sensing platform to obtain multiple remote sensing images of multiple initially divided grid areas, and combining multiple sensor layout arrays to collect ground feature data, the fusion of ground features and remote sensing data is realized. By using remote sensing images as the background and ground sensor data as the foreground, a secondary aggregation analysis of multiple information is performed, which can more accurately reflect the soil and vegetation status, groundwater changes, etc., and enhance the monitoring accuracy. ; By analyzing the concentrated values ​​of multiple aggregated ground features, the first monitoring bandwidth and the second monitoring bandwidth of each aggregated area are automatically determined to form different monitoring channels. The monitoring bandwidth is dynamically adjusted according to the deviation amount. A smaller bandwidth is used for areas with larger fluctuations, and a larger bandwidth is used for areas with smaller fluctuations, thereby realizing an adaptive monitoring strategy, which not only ensures high-frequency monitoring of abnormally changing areas, but also reduces the monitoring burden on stable areas, reduces unnecessary data collection and processing, and improves the efficiency and response sensitivity of drought monitoring; after constructing multiple first monitoring channels and second monitoring channels, each aggregated area can be continuously and dynamically monitored, and the real-time data inflow of the monitoring channels can be used to quickly respond to changes in drought conditions, especially for areas with rapid water loss or severe vegetation degradation. The high-frequency monitoring mode can be triggered in time to quickly discover potential drought crises, thereby helping disaster prevention and mitigation departments to better make early prevention and intervention measures.

[0068] Embodiment 2 is based on the same inventive concept as the drought monitoring method based on ground feature detection in the above embodiment. Figure 2 As shown, the embodiment of the present application provides a drought monitoring system based on ground feature detection, the system comprising:

[0069] A region division module 10, the region division module 10 is used to retrieve the geological distribution data of the target monitoring area, and preliminarily divide the target monitoring area according to the geological distribution data to obtain a plurality of initially divided grid areas; a sensor deployment module 20, the sensor deployment module 20 is used to deploy sensors based on the area size and geological characteristics of the plurality of initially divided grid areas, and obtain a plurality of sensor deployment arrays that have been deployed; a remote sensing image acquisition module 30, the remote sensing image acquisition module 30 is used to connect to a satellite remote sensing platform in a preset detection window to obtain a plurality of remote sensing images of the plurality of initially divided grid areas; a data acquisition module 40, the data acquisition module 40 is used to collect monitoring data of the plurality of sensor deployment arrays in the preset detection window to obtain a plurality of ground feature data sets; a same type aggregation module 50, the same type aggregation module 50 is used to use the plurality of remote sensing images as Background information, taking the multiple ground feature data sets as foreground information, performing secondary similar aggregation on the multiple initial divided grid areas to determine multiple aggregated divided grid area sets, wherein the multiple aggregated divided grid area sets include multiple aggregated ground feature concentrated value sets; a drought monitoring module 60, the drought monitoring module 60 is used to determine multiple first monitoring bandwidths and multiple second monitoring bandwidths based on the multiple aggregated ground feature concentrated value sets, and construct multiple first monitoring channels according to the multiple first monitoring bandwidths and construct multiple second monitoring channels according to the multiple second monitoring bandwidths, one-to-one mapping the multiple first monitoring channels and the multiple second monitoring channels to construct a drought monitor, and use the drought monitor to perform drought monitoring on the multiple aggregated divided grid area sets, wherein each aggregated divided grid area set performs drought monitoring through the one-to-one mapped first monitoring channel and second monitoring channel.

[0070] Furthermore, the area division module 10 further includes the following operation steps:

[0071] Connecting to a geological survey database to obtain geological distribution data of the target monitoring area, wherein the geological distribution data includes soil type, rock structure, groundwater layer distribution, surface height and surface slope; obtaining the total area of ​​the target monitoring area, and determining grid size constraints based on the total area; using water retention capacity, surface water flow trend and vegetation coverage characteristics as a preliminary division indicator set; using a grid divider to analyze and identify the geological distribution data, grid size constraints and the preliminary division indicator set, and obtaining the multiple initial division grid areas.

[0072] Furthermore, the same type aggregation module 50 further includes the following operation steps:

[0073] The multiple remote sensing images are analyzed by using an image semantic recognizer to determine multiple image background features, and the multiple image background features are used as multiple background vectors, wherein each image background feature includes a vegetation index, a surface reflectivity, and a surface temperature; using ground features as vectors, foreground vectors are constructed for the multiple ground feature data sets to obtain multiple foreground vectors; based on the multiple background vectors and the multiple foreground vectors, the multiple initially divided grid areas are secondary aggregated in the same manner to obtain the multiple aggregated divided grid area sets, wherein each background vector corresponds to a foreground vector.

[0074] Furthermore, the same type aggregation module 50 further includes the following operation steps:

[0075] Constructing an aggregation space, inputting the multiple background vectors, the multiple foreground vectors and the multiple initial divided grid regions into the aggregation space, obtaining multiple spatial points and multiple spatial point coordinates, wherein each spatial point corresponds to an initial divided grid region; selecting K spatial points from the multiple spatial points according to a preset screening rule, wherein the preset screening rule is that the distance between any two spatial points of the screened K spatial points is greater than or equal to a preset distance threshold, and K is an integer greater than or equal to 1; taking the K spatial points as the centers of the K grid regions, performing a pre-set iteration step in the aggregation space. Perform an iterative search to determine K initial aggregated divided grid area sets; obtain multiple discrete space points in the aggregated space that are not divided into grid areas, perform neighbor authentication on the multiple discrete space points, and if the authentication is successful, add them into a neighbor aggregated divided grid area set to obtain multiple neighbor aggregated divided grid area sets; if the authentication is unsuccessful, add the multiple discrete space points into the initial aggregated divided grid area set with the closest distance; obtain the multiple aggregated divided grid area sets based on the K initial aggregated divided grid area sets and multiple neighbor aggregated divided grid area sets.

[0076] Furthermore, the mean value of each ground feature within the plurality of aggregated divided grid area sets is calculated according to the plurality of ground feature data sets to obtain the plurality of aggregated ground feature concentrated value sets.

[0077] Furthermore, the drought monitoring module 60 further includes the following operation steps:

[0078] Taking ground features as indexes, the overall mean of each ground feature in the target monitoring area is calculated in combination with multiple aggregated ground feature concentrated value sets to obtain an aggregated ground feature concentrated value set; deviations are calculated based on the multiple aggregated ground feature concentrated value sets and the aggregated ground feature concentrated value set to determine multiple aggregated ground feature concentrated value deviation sets; deviation sets in the multiple aggregated ground feature sets are traversed to calculate deviation means to obtain multiple deviation means; and the multiple first monitoring bandwidths and the multiple second monitoring bandwidths are determined based on the sizes of the multiple deviation means.

[0079] Furthermore, the sensor deployment module 20 also includes the following operation steps:

[0080] Based on the geological distribution data and the multiple initially divided grid areas, geological feature extraction is performed to obtain multiple grid area geological feature sets; the areas of the multiple initially divided grid areas are respectively counted to obtain multiple grid area areas; sensor deployment schemes are identified based on the multiple grid area areas and the multiple grid area geological feature sets to determine multiple sensor deployment schemes; based on the multiple sensor deployment schemes, sensors are deployed in the multiple initially divided grid areas to obtain multiple sensor deployment arrays.

[0081] Through the above detailed description of the drought monitoring method based on ground feature detection in this specification, those skilled in the art can clearly understand the drought monitoring system based on ground feature detection in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0082] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A drought monitoring method based on ground feature detection, characterized in that: The method comprises: Retrieving geological distribution data of the target monitoring area, and preliminarily dividing the target monitoring area according to the geological distribution data to obtain a plurality of initial divided grid areas; Deploy sensors based on the area sizes and geological features of the multiple initially divided grid regions to obtain multiple deployed sensor arrays; Connecting to a satellite remote sensing platform in a preset detection window to obtain a plurality of remote sensing images of the plurality of initially divided grid areas; Collecting monitoring data of the plurality of sensor arrays within the preset detection window to obtain a plurality of ground feature data sets; The plurality of initially divided grid regions are secondary aggregated by taking the plurality of remote sensing images as background information and the plurality of ground feature data sets as foreground information to determine a plurality of aggregated divided grid region sets, wherein the plurality of aggregated divided grid region sets include a plurality of aggregated ground feature concentrated value sets; Determine a plurality of first monitoring bandwidths and a plurality of second monitoring bandwidths based on the plurality of aggregated ground feature concentrated value sets, construct a plurality of first monitoring channels according to the plurality of first monitoring bandwidths and construct a plurality of second monitoring channels according to the plurality of second monitoring bandwidths, perform one-to-one mapping of the plurality of first monitoring channels and the plurality of second monitoring channels to construct a drought monitor, and use the drought monitor to perform drought monitoring on the plurality of aggregated divided grid area sets, wherein each aggregated divided grid area set performs drought monitoring through the one-to-one mapped first monitoring channel and second monitoring channel; The ground characteristic data include soil, meteorological and groundwater data; Determining a plurality of first monitoring bandwidths and a plurality of second monitoring bandwidths based on the plurality of aggregated ground feature concentrated value sets includes: Taking the ground feature as an index, combining multiple aggregated ground feature concentrated value sets to calculate the overall mean of each ground feature in the target monitoring area, and obtaining an aggregated ground feature concentrated mean value set; Calculate the deviation amount according to the plurality of aggregated ground feature concentrated value sets and the aggregated ground feature concentrated mean value set to determine a plurality of aggregated ground feature concentrated value deviation sets; Traversing the deviation sets in the plurality of aggregated ground feature concentrations to calculate the deviation mean, and obtaining a plurality of deviation means; The plurality of first monitoring bandwidths and the plurality of second monitoring bandwidths are determined based on the magnitudes of the plurality of deviation means.

2. The drought monitoring method based on ground feature detection according to claim 1, characterized in that: The geological distribution data of the target monitoring area is retrieved, and the target monitoring area is preliminarily divided according to the geological distribution data to obtain a plurality of initial division grid areas, including: Connecting to a geological survey database to obtain geological distribution data of the target monitoring area, wherein the geological distribution data includes soil type, rock layer structure, groundwater layer distribution, surface height and surface slope; Obtaining the total area of ​​the target monitoring area, and determining the grid size constraint condition according to the total area; Water retention capacity, surface water flow trends and vegetation cover characteristics were used as a preliminary set of classification indicators; The geological distribution data, grid size constraints and the preliminary division indicator set are analyzed and identified using a grid divider to obtain the multiple initial division grid areas.

3. The drought monitoring method based on ground feature detection according to claim 1, characterized in that: The multiple remote sensing images are used as background information and the multiple ground feature data sets are used as foreground information to perform secondary homogeneous aggregation on the multiple initial divided grid areas to determine multiple aggregated divided grid area sets, including: Analyzing the multiple remote sensing images using an image semantic recognizer to determine multiple image background features, and using the multiple image background features as multiple background vectors, wherein each image background feature includes a vegetation index, a surface reflectivity, and a surface temperature; Taking ground features as vectors, constructing foreground vectors for the plurality of ground feature data sets to obtain a plurality of foreground vectors; The multiple initially divided grid regions are secondary aggregated in the same manner based on the multiple background vectors and the multiple foreground vectors to obtain the multiple aggregated divided grid region sets, wherein each background vector corresponds to a foreground vector.

4. The drought monitoring method based on ground feature detection according to claim 3, characterized in that: The multiple initial divided grid regions are secondary aggregated into the same type based on the multiple background vectors and the multiple foreground vectors to obtain the multiple aggregated divided grid region sets, including: Constructing an aggregation space, inputting the multiple background vectors, the multiple foreground vectors and the multiple initial divided grid regions into the aggregation space, obtaining multiple space points and multiple space point coordinates, wherein each space point corresponds to an initial divided grid region; Selecting K spatial points from the multiple spatial points according to a preset screening rule, wherein the preset screening rule is that the distance between any two spatial points of the screened K spatial points is greater than or equal to a preset distance threshold, and K is an integer greater than or equal to 1; Taking the K spatial points as the centers of the K grid regions, performing an iterative search in the aggregation space according to a preset iteration step length, and determining K initial aggregation division grid region sets; Acquire multiple discrete space points in the aggregation space that are not divided into grid areas, perform neighbor authentication on the multiple discrete space points, and if the authentication is successful, add them into a neighbor aggregation division grid area set to obtain multiple neighbor aggregation division grid area sets; If the authentication fails, multiple discrete spatial points are added to the nearest initial aggregated partition grid area set; The multiple aggregated partitioned grid area sets are obtained according to the K initial aggregated partitioned grid area sets and multiple neighboring aggregated partitioned grid area sets.

5. The drought monitoring method based on ground feature detection according to claim 4, characterized in that: The mean value of each ground feature within the plurality of aggregated divided grid area sets is calculated according to the plurality of ground feature data sets to obtain the plurality of aggregated ground feature concentrated value sets.

6. The drought monitoring method based on ground feature detection according to claim 1, characterized in that: The sensors are deployed based on the area sizes and geological features of the multiple initially divided grid regions to obtain multiple sensor deployment arrays that have been deployed, including: Extracting geological features based on the geological distribution data and the multiple initially divided grid regions to obtain a plurality of grid region geological feature sets; Counting the areas of the plurality of initially divided grid regions respectively to obtain the areas of a plurality of grid regions; Identify sensor deployment plans based on the areas of the multiple grid regions and the geological feature sets of the multiple grid regions, and determine multiple sensor deployment plans; Sensors are arranged in the plurality of initially divided grid areas based on the plurality of sensor arrangement schemes to obtain a plurality of sensor arrangement arrays.

7. A drought monitoring system based on ground feature detection, characterized in that: A drought monitoring method based on ground feature detection for implementing any one of claims 1 to 6, the system comprising: A region division module, the region division module is used to retrieve geological distribution data of a target monitoring area, and preliminarily divide the target monitoring area according to the geological distribution data to obtain a plurality of initial division grid regions; A sensor deployment module, the sensor deployment module is used to deploy sensors based on the area sizes and geological characteristics of the multiple initially divided grid areas to obtain multiple sensor deployment arrays that have been deployed; A remote sensing image acquisition module, the remote sensing image acquisition module is used to connect to a satellite remote sensing platform in a preset detection window to acquire a plurality of remote sensing images of the plurality of initially divided grid areas; A data acquisition module, the data acquisition module is used to collect monitoring data of the multiple sensor arrays within the preset detection window to obtain multiple ground feature data sets; A similar aggregation module, the similar aggregation module is used to perform secondary similar aggregation on the multiple initial divided grid areas with the multiple remote sensing images as background information and the multiple ground feature data sets as foreground information to determine multiple aggregated divided grid area sets, wherein the multiple aggregated divided grid area sets include multiple aggregated ground feature concentration value sets; A drought monitoring module, the drought monitoring module is used to determine multiple first monitoring bandwidths and multiple second monitoring bandwidths based on the multiple aggregated ground feature concentrated value sets, and construct multiple first monitoring channels according to the multiple first monitoring bandwidths and construct multiple second monitoring channels according to the multiple second monitoring bandwidths, and one-to-one map the multiple first monitoring channels and the multiple second monitoring channels to construct a drought monitor, and use the drought monitor to monitor the multiple aggregated divided grid area sets, wherein each aggregated divided grid area set is drought monitored through the one-to-one mapped first monitoring channel and second monitoring channel.

Images