Agricultural space-time digital base data organization method, device and system and storage medium

Abstract

The invention discloses an agricultural space-time digital base data organization method, device and system, and a storage medium. The method comprises the following steps: collecting multi-source data; comprehensively considering landform characteristics, soil type distribution and climate condition differences, and carrying out geographical zoning; on the basis of partition constraint conditions, identifying edge-texture features of different ground features, and realizing refined layered extraction of spatio-temporal information based on the land parcels and integrating multi-source data for attribute assignment of multiple types of farmland land parcels; the method comprises the following steps: establishing a unified space-time reference by adopting a CGCS2000 geodetic coordinate system, and storing agricultural space-time digital base data by adopting a PostgreSQL (Structured Query Language) database management system; and carrying out three-dimensional visualization on the agricultural space-time digital base data in a Web application by utilizing Cesium and Sprg-Boot. By adopting the technical scheme of the invention, the technical problems that the agricultural spatio-temporal data is difficult to reasonably apply and the data is difficult to carry out standard unified management are solved.

Description

Technical Field

[0001] This invention belongs to the field of big data processing technology, and in particular relates to a method, device, system, and storage medium for organizing agricultural spatiotemporal digital base data. Background Technology

[0002] Today, data is gradually transforming from a simple information resource into an asset with real economic value, becoming a major driver of economic development. Among these, people's daily production activities are closely related to spatiotemporal big data.

[0003] With the deepening development of digital transformation in agriculture, traditional spatiotemporal data organization methods are facing severe challenges: multi-source agricultural data have significant differences in spatiotemporal scales, traditional administrative divisions or regular grids are difficult to accurately express the boundaries of actual farmland units, single-scale analysis methods cannot meet the multi-level decision-making needs from plots to regions, and there is a lack of effective systematic correlation mechanisms between agricultural expertise and spatiotemporal data.

[0004] To meet the urgent need for high-quality development to empower geographic spatiotemporal information, it is imperative to construct an efficient and intelligent data organization method for agricultural spatiotemporal digital foundations, enhance the integration, analysis, and application capabilities of agricultural data, and contribute to the construction of precision agriculture and smart villages. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method, device, system, and storage medium for organizing agricultural spatiotemporal digital base data.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for organizing agricultural spatiotemporal digital infrastructure data includes:

[0008] Step S1: Collect data from multiple sources;

[0009] Step S2: Within the administrative division framework, conduct geographical zoning based on topographic features, soil type distribution, and differences in climate conditions;

[0010] Step S3: Based on the partitioning constraints, the spatiotemporal information of various types of farmland plots is extracted by adaptively identifying the edge-texture features of different land features through a deep learning model.

[0011] Step S4: Assign attribute values ​​based on the spatiotemporal information and multi-source data of various types of farmland plots;

[0012] Step S5: Establish a unified spatiotemporal reference using the CGCS2000 geodetic coordinate system, and use the PostgreSQL database management system to store the agricultural spatiotemporal digital base data.

[0013] Step S6: Use Cesium and Spring-Boot to perform 3D visualization of agricultural spatiotemporal digital base data in a web application.

[0014] As a preferred option, in step S2, a clustering algorithm is used to divide the administrative division into several geographical units with high internal homogeneity and significant inter-class differences; the partitioning results meet the following constraints: maintaining the spatial integrity of the plots, highlighting the topographical differences, and ensuring the consistency of environmental conditions within the planting area.

[0015] As a preferred option, in step S3, based on the prior constraints established by geographical zoning, the visual differences between different farmland plots are considered, and the feature objects are extracted hierarchically by matching the corresponding deep neural network model, so as to achieve accurate hierarchical extraction of multiple types of farmland plots.

[0016] As a preferred option, in step S4, the unique identifier of the land cover is determined by matching spatiotemporal features, multi-source heterogeneous data is fused into the corresponding land cover unit, a unified data model with both raster and vector expression is generated, and finally an agricultural spatiotemporal digital base architecture containing five spatial scales of "wide area - region - local - object - signal" is constructed.

[0017] The present invention also provides an agricultural spatiotemporal digital base data organization device, comprising:

[0018] The first processing module is used to collect data from multiple sources;

[0019] The second processing module is used to perform geographical zoning within the administrative zoning framework, based on topographic features, soil type distribution, and differences in climate conditions.

[0020] The third processing module is used to adaptively identify the edge-texture features of different land features and extract the spatiotemporal information of multiple types of farmland plots based on partition constraints and deep learning models.

[0021] The fourth processing module is used to assign attribute values ​​based on the spatiotemporal information of multiple types of farmland plots and multi-source data;

[0022] The fifth processing module is used to establish a unified spatiotemporal reference using the CGCS2000 geodetic coordinate system and to store agricultural spatiotemporal digital base data using the PostgreSQL database management system.

[0023] The sixth processing module is used to perform 3D visualization of agricultural spatiotemporal digital base data in a web application using Cesium and Spring-Boot.

[0024] The present invention also provides a storage medium storing a computer program, which executes an agricultural spatiotemporal digital base data organization method when running.

[0025] This invention collects multi-source data; comprehensively considers topographic features, soil type distribution, and climatic differences to perform geographical zoning; based on zoning constraints, it identifies the edge-texture features of different land features, achieving refined stratification of various types of farmland plots, extracting spatiotemporal information based on plots, and integrating multi-source data for attribute assignment; it establishes a unified spatiotemporal benchmark using the CGCS2000 geodetic coordinate system, and uses a PostgreSQL database management system to store agricultural spatiotemporal digital base data; it utilizes Cesium and Spring-Boot to perform 3D visualization of agricultural spatiotemporal digital base data in a web application. The technical solution of this invention addresses the technical problems of the difficulty in rationally utilizing agricultural spatiotemporal data and standardizing its management. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0027] Figure 1 This is a flowchart of the agricultural spatiotemporal digital base data organization method according to an embodiment of the present invention;

[0028] Figure 2 This is a flowchart of the method for hierarchical extraction of agricultural spatiotemporal digital base data according to an embodiment of the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0031] Example 1:

[0032] like Figure 1 As shown, this embodiment of the invention provides a method for organizing agricultural spatiotemporal digital base data, including:

[0033] Step S1: Multi-source data acquisition;

[0034] Step S2: Within the framework of administrative division, geographical divisions are carried out by comprehensively considering topographic features, soil type distribution, and differences in climate conditions.

[0035] Step S3: Based on the partitioning constraints, the edge-texture features of different land features are adaptively identified through a deep learning model to achieve refined layered extraction of multiple types of farmland plots;

[0036] Step S4: Assign attribute values ​​based on the spatiotemporal information of the land parcel and by integrating multi-source data;

[0037] Step S5: Establish a unified spatiotemporal reference using the CGCS2000 geodetic coordinate system, and use the PostgreSQL database management system to store the agricultural spatiotemporal digital base data.

[0038] Step S6: Use Cesium and Spring-Boot to perform 3D visualization of agricultural spatiotemporal digital base data in a web application.

[0039] As one embodiment of the present invention, in step S1, the present invention obtains data through multi-source remote sensing data fusion and geospatial data integration, which is used to support geographic zoning and assign plot attributes. This includes: extracting vegetation phenological characteristic parameters based on the MCD12Q2 V6 dataset; calculating the Normalized Difference Vegetation Index (NDVI) using the near-infrared and red reflectance bands of the Sentinel-2 dataset; acquiring Gaofen-2 (GF-2) satellite remote sensing images, performing refined preprocessing such as radiometric correction, atmospheric correction, and noise reduction, followed by image stitching and cropping; using a digital elevation model (DEM) and its derived topographic parameters (elevation, slope, aspect), hydrological network data, and soil type distribution data; and integrating temperature-related parameters (mean diurnal range, seasonal variation coefficient), precipitation characteristics (seasonal precipitation distribution), real-time meteorological elements (current temperature, precipitation, evapotranspiration), and climate indicators such as the drought index.

[0040] As one embodiment of the present invention, in step S2, the present invention performs geographical zoning based on relevant factors within the framework of administrative zoning. Within the same administrative region, due to significant differences in natural environments such as topography, soil type, and climate factors, crop phenological characteristics (such as periodic biological activities like seedling emergence, growth, flowering, and maturity) also exhibit spatial heterogeneity. The final zoning result satisfies the following constraints: maintaining the spatial integrity of plots, highlighting topographical differences, and ensuring the consistency of environmental conditions within the planting area. Therefore, when there are significant differences in phenological characteristics in different areas within the same administrative region, by analyzing and comparing the ecological requirements of crops and the regional ecological environment conditions, the administrative vector surface is divided into most zoning surfaces with similar planting conditions, realizing "geographical zoning." This helps to better understand the natural environment of the administrative region and establish prior constraints for subsequent stratified extraction.

[0041] Specifically, in the implementation of this invention, representative vegetation cover indicators are first extracted based on remote sensing and phenological observation data, such as the mean normalized difference vegetation index (NDVI) for March, August, and the growing season (April–October), which are denoted as follows: This can be used as a response variable to reflect the growth status of crops in different regions.

[0042] Subsequently, an explanatory variable matrix X was constructed for numerous environmental factors (including climate data, topographic data, and soil data) that may influence phenological characteristics. To avoid the risk of overfitting due to multicollinearity, redundancy analysis (RDA) was used to pre-screen the variables. RDA is a method that combines multiple regression analysis with principal component analysis (PCA), and its basic modeling form is as follows: Y = X·B + E.

[0043] Where Y is the response variable matrix (such as the NVDI index), X is the explanatory variable matrix (such as slope, temperature, soil type, etc.), B is the regression coefficient matrix, and E is the residual term.

[0044] The principal axes (RDA1, RDA2, etc.) obtained by RDA mapping reflect the direction of maximum explanatory power of the explanatory variables on the response variables. Based on this, the key environmental variables most relevant to the vegetation status are selected as the input index set X' for subsequent zoning analysis.

[0045] After variable optimization, to achieve automatic segmentation of areas with similar planting conditions within geographic space, this invention employs the Spatial Toeplitz Inverse Covariance Clustering (STICC) method. The STICC method introduces spatial constraints based on the traditional TICC method, comprehensively considering the conditional dependency structure and spatial continuity between variables by minimizing the following joint optimization objective function:

[0046]

[0047] Where, Θ k S is the inverse covariance matrix of the Toeplitz structure corresponding to the k-th partition. k λ is the sample covariance matrix, λ is the sparse regularization coefficient, Z is the spatial allocation variable, Φ(Z) is the spatial smoothing penalty term based on the Markov random field, and γ is the weight parameter that controls the intensity of spatial smoothing.

[0048] In practice, due to the complexity of topography, the spatial samples are divided into k=8 clusters, each corresponding to a geographical sub-region with similar planting conditions. The division fully considers the structural relationships and spatial proximity among the explanatory variables, forming clearly defined and continuously distributed natural geographical zoning surfaces. This zoning result serves as a spatial prior condition for the subsequent "hierarchical extraction" step, providing physically meaningful boundary support for parameter constraints and model construction.

[0049] In one embodiment of the present invention, in step S3, based on the planting zoning results formed in step S2, a plot-level extraction operation is further performed within each geographical zone. This step aims to adaptively identify the remote sensing image features presented by different farmland types through a deep learning model, achieve accurate contour extraction at the plot level, and generate high-quality agricultural spatiotemporal digital base object vector data.

[0050] Specifically, in high spatial resolution remote sensing imagery, farmland plots exhibit rich visual differences in edges, textures, and semantics. For example... Figure 2 As shown, this invention combines the morphological characteristics of cultivated land in different regions and uses heterogeneous deep network models to construct adaptive land parcel extraction processes.

[0051] In areas with concentrated arable land, such as plains, the fields are laid out regularly, with clear boundaries and square shapes. This invention preferably uses classic edge detection models such as HED and RCF to extract strong edge features of farmland boundaries, quickly identifying the boundaries of regular arable land.

[0052] In hilly areas with undulating terrain, sloping farmland and terraced fields are often scattered, with discontinuous edges but significant differences in surface texture. This invention employs an ENCNet-like texture semantic segmentation model to enhance the model's ability to learn local texture and morphological differences, thereby enabling the extraction of sloping farmland and terraced fields.

[0053] In complex ecological areas such as mountainous forest edges, cultivated land is often mixed with forest or shrubland, with blurred boundaries and indistinct color variations in images. This invention preferably uses the LinkNet model, which has a lightweight structure and strong boundary refinement capabilities, and leverages its residual connectivity and shallow feature fusion capabilities to perform precise identification of blurred cultivated land in forest areas.

[0054] The extracted land parcel boundary lines from the above model are integrated and processed. Combining mathematical morphology algorithms and the "line-to-surface" topology reconstruction method, the boundary lines are closed, topologically corrected, and reconstructed to generate topologically correct and structurally complete polygonal land parcel vector outlines. To address the issue of omissions or errors in the model's extraction in complex terrain areas, a small amount of manual checking and correction is performed to ensure boundary integrity and geometric accuracy. Ultimately, structured agricultural spatiotemporal digital base object vector data is formed.

[0055] In one embodiment of the present invention, step S4, after completing the extraction of land parcel boundaries, further assigns attribute values ​​based on the spatiotemporal positioning information of the land parcels and the integrated multi-source heterogeneous data. Agricultural planting land parcel data represents the granularity of the production-to-minimum information model in geographic spatiotemporal data. It carries a lot of signal spatial data and is an integration of spatiotemporal data, realizing the transformation of complex land surfaces into individual geographic entities.

[0056] Specifically, all acquired signals are divided into two sets according to their observability: a deterministic set S[i] = {S1, S2, ..., S...} m} and the set of uncertainties U[j]={U1,U2,...,U n ), where m and n represent the quantity of the two types of signals, respectively. The former is the collected measured data, which can be directly mapped and assigned to the specific plot units extracted in step S3; while the latter is problem data with information uncertainty due to observation scale, time gaps, or incomplete data. For processing uncertain data, this invention uses a five-level spatial architecture based on an agricultural spatiotemporal digital foundation for knowledge reasoning and attribute restoration. This architecture is divided into five levels from top to bottom: wide-area space, regional space, local space, object space, and signal space. During attribute assignment, for each item U in the uncertainty set... j By combining the unique temporal and spatial identifiers of the data, relevant attributes can be extracted from high-level spatial granularity. Attribute restoration is then performed through methods such as rule mapping, model fitting, or statistical interpolation. Based on the data coverage and stability during the calculation process, corresponding confidence intervals are output. Finally, the assignment results are stored in the "object space" level, forming land parcel object data that includes both structural boundaries and rich spatiotemporal attributes. This enables full-scale attribute modeling and deduction, from macro-level administrative division information to micro-level agricultural cultivated land parcel representation. Specific attributes are shown in Table 1.

[0057] Table 1

[0058]

[0059]

[0060] As one embodiment of the present invention, in step S5, regarding the spatial reference issues involved in geographic information data, the National Geodetic Coordinate System (CGCS2000, EPSG:4490) is uniformly adopted as the standard coordinate system for spatial positioning and data storage of raster data models and vector data models. The geographic coordinate system is used to uniquely determine any location on the Earth's surface, but different selections of the reference ellipsoid and prime meridian will lead to slight differences in the coordinate results. The CGCS2000 coordinate system, with the Earth's center of mass as the origin, is constructed based on a global reference frame and possesses characteristics such as high precision and high consistency, effectively improving the accuracy of the three-dimensional geographic representation of agricultural spatial objects.

[0061] To construct a stable and efficient data structure for the agricultural spatiotemporal digital foundation, this embodiment uses a PostgreSQL database as the main database platform and installs a PostGIS spatial extension module to achieve unified storage and spatial indexing support for vector data (points, lines, polygons, and their multiple geometric sets) and raster data. During data management, all spatial data includes coordinate system information to ensure consistency and compatibility. The PostgreSQL database supports various index types, such as B-tree, Hash, GiST, SP-GiST, GIN, and BRIN, which can be flexibly configured according to query type to meet the needs of complex spatial queries and analytical calculations, providing powerful support for data organization, index optimization, and spatial function calculations for the digital foundation.

[0062] As a further extension of this embodiment of the invention, in step S6, a service management system for spatiotemporal data is constructed based on the GeoServer platform, and integrated publishing between the database and map services is realized. GeoServer, as a geographic information service platform conforming to the Open Geospatial Consortium (OGC) standard, supports various spatial data service protocols, including Web Map Service (WMS), Web Feature Service (WFS), Web Overlay Service (WCS), and Web Map Tile Service (WMTS), and is an important intermediate layer for the publishing of agricultural spatiotemporal digital foundation networks.

[0063] In actual deployment, the connection between GeoServer and the PostgreSQL database is first established, and raster and vector data from the database are loaded through the layer management function to achieve layer tiling and service configuration. Subsequently, the data is standardized and published as various OGC services through GeoServer for unified access by clients. To enhance the visualization of the layers, QGIS software is used to configure and adjust the SLD style of the layers, including color mapping, transparency settings, and classification representation.

[0064] On the front-end system side, a web application interface is built based on the Vue framework, integrating the Cesium 3D visualization engine. The WebMapTileServiceImageryProvider method is used to access the WMTS layers published by GeoServer, enabling dynamic loading and interactive display of the digital base layers within the 3D scene. All layers are centrally managed in a LayerList array.

[0065] Meanwhile, the backend system uses the Spring Boot framework to build API services. Through spatial attribute statistics and structured interface calls, it transmits land feature attribute information from the database to the frontend application, enabling tabular statistics and chart-based display functions. Ultimately, it completes the integration, dynamic rendering, and attribute visualization of the agricultural spatiotemporal digital foundation on the web, supporting multi-scenario, multi-scale, and multi-dimensional agricultural business applications.

[0066] Example 2:

[0067] This invention also provides an agricultural spatiotemporal digital base data organization device, comprising:

[0068] The first processing module is used to collect data from multiple sources;

[0069] The second processing module is used to perform geographical zoning within the administrative zoning framework, based on topographic features, soil type distribution, and differences in climate conditions.

[0070] The third processing module is used to adaptively identify the edge-texture features of different land features and extract the spatiotemporal information of multiple types of farmland plots based on partition constraints and deep learning models.

[0071] The fourth processing module is used to assign attribute values ​​based on the spatiotemporal information of multiple types of farmland plots and multi-source data;

[0072] The fifth processing module is used to establish a unified spatiotemporal reference using the CGCS2000 geodetic coordinate system and to store agricultural spatiotemporal digital base data using the PostgreSQL database management system.

[0073] The sixth processing module is used to perform 3D visualization of agricultural spatiotemporal digital base data in a web application using Cesium and Spring-Boot.

[0074] Example 3:

[0075] The present invention also provides an agricultural spatiotemporal digital base data organization system, comprising: a memory and a processor, wherein the memory stores a computer program executed by the processor, and the computer program executes an agricultural spatiotemporal digital base data organization method when executed by the processor.

[0076] Example 4:

[0077] The present invention also provides a storage medium storing a computer program, which executes an agricultural spatiotemporal digital base data organization method during runtime.

[0078] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for organizing agricultural spatiotemporal digital base data, characterized in that, include: Step S1: Collect data from multiple sources; Step S2: Within the administrative division framework, conduct geographical zoning based on topographic features, soil type distribution, and differences in climate conditions; Step S3: Based on the partitioning constraints, the spatiotemporal information of various types of farmland plots is extracted by adaptively identifying the edge-texture features of different land features through a deep learning model. Step S4: Assign attribute values ​​based on the spatiotemporal information and multi-source data of various types of farmland plots; Step S5: Establish a unified spatiotemporal reference using the CGCS2000 geodetic coordinate system, and use the PostgreSQL database management system to store the agricultural spatiotemporal digital base data. Step S6: Use Cesium and Spring-Boot to perform 3D visualization of agricultural spatiotemporal digital base data in a web application.

2. The method for organizing agricultural spatiotemporal digital base data as described in claim 1, characterized in that, In step S2, a clustering algorithm is used to divide the administrative division into several geographical units with high internal homogeneity and significant inter-class differences. The partitioning results meet the following constraints: maintain the spatial integrity of the plots, highlight the topographic differences, and ensure the consistency of environmental conditions within the planting area.

3. The method for organizing agricultural spatiotemporal digital base data as described in claim 2, characterized in that, In step S3, based on the prior constraints established by geographical zoning, and considering the visual differences between different farmland plots, the feature objects are extracted hierarchically by matching the corresponding deep neural network model, thereby achieving accurate hierarchical extraction of multiple types of farmland plots.

4. The method for organizing agricultural spatiotemporal digital base data as described in claim 3, characterized in that, In step S4, the unique identifier of the land cover is determined by spatiotemporal feature matching, multi-source heterogeneous data is fused into the corresponding land cover unit, a unified data model with both raster and vector expression is generated, and finally an agricultural spatiotemporal digital base architecture containing five spatial scales of "wide area - region - local - object - signal" is constructed.

5. An agricultural spatiotemporal digital base data organization device, characterized in that, include: The first processing module is used to collect data from multiple sources; The second processing module is used to perform geographical zoning within the administrative zoning framework, based on topographic features, soil type distribution, and differences in climate conditions. The third processing module is used to adaptively identify the edge-texture features of different land features and extract the spatiotemporal information of multiple types of farmland plots based on partition constraints and deep learning models. The fourth processing module is used to assign attribute values ​​based on the spatiotemporal information of multiple types of farmland plots and multi-source data; The fifth processing module is used to establish a unified spatiotemporal reference using the CGCS2000 geodetic coordinate system and to store agricultural spatiotemporal digital base data using the PostgreSQL database management system. The sixth processing module is used to perform 3D visualization of agricultural spatiotemporal digital base data in a web application using Cesium and Spring-Boot.

6. An agricultural spatiotemporal digital base data organization system, characterized in that, include: A memory and a processor, wherein the memory stores a computer program executed by the processor, the computer program, when executed by the processor, performs the agricultural spatiotemporal digital base data organization method as described in any one of claims 1-4.

7. A storage medium, characterized in that, The storage medium stores a computer program that, when executed, performs the agricultural spatiotemporal digital base data organization method as described in any one of claims 1-4.

Images