Data Fusion Method of Forestry Data Management System

Abstract

The present application relates to the field of data fusion processing, and specifically discloses a data fusion method for a forestry data management system, which processes real-time collected data of the forestry data management system; divides the forest resource data of a target area into different independent areas, and establishes a special forest resource growth quantification model for the forest resource data of each independent area; detects whether there is a forest resource data conflict; when a conflict occurs in the forest resource data of a certain independent area, corrects the conflicting data through the forest resource growth quantification model of the independent area; when a conflict occurs in the forest resource data of multiple independent areas, corrects the conflicting data jointly through the forest resource growth quantification models of the multiple independent areas; and completes data fusion after correcting the conflicting data.

Description

Technical Field

[0001] The present application relates to the field of data fusion processing, and in particular to a data fusion method for a forestry data management system. Background Art

[0002] Forestry data has significant spatiotemporal characteristics. The growth and change of forest resources is a dynamic process, and spatially, forest ecosystems vary across geographic locations. During spatiotemporal data fusion, it is difficult to accurately match and integrate data from different temporal and spatial scales. This often results in conflicting data. For example, when combining satellite imagery from different time periods with ground-based data to analyze forest resource dynamics, it is difficult to accurately reconcile data from the same spatial location at different points in time. Summary of the Invention

[0003] In response to the deficiencies of the existing technology, this application provides a data fusion method for a forestry data management system, which effectively solves the existing problems.

[0004] To achieve the above objectives, this application is implemented through the following technical solutions:

[0005] The present application provides a data fusion method for a forestry data management system, comprising the following steps:

[0006] Unify the data format, purify the data quality, normalize the spatial coordinates, and standardize the time series of the real-time collected data of the forestry data management system;

[0007] Divide the forest resource data of the target area into different independent areas, and establish a special forest resource growth quantitative model for the forest resource data of each independent area;

[0008] The forestry data management system detects whether there is any conflict in forest resource data before data fusion; the conflict in forest resource data refers to the inconsistency of forest resource growth indicators collected in the same area at different time points;

[0009] When there is a conflict in the forest resource data of a certain independent area, the conflicting data will be corrected through the quantitative model of forest resource growth in that independent area;

[0010] When there is a conflict in the forest resource data of multiple independent regions, the conflicting data will be corrected through the quantitative forest resource growth model of the multiple independent regions;

[0011] Data fusion is completed after correcting the conflicting data.

[0012] On the other hand, the present application also discloses an electronic device, comprising:

[0013] at least one processor; and

[0014] a memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to execute the data fusion method of the forestry data management system.

[0016] This application provides a data fusion method for a forestry data management system, which has at least the following significant beneficial effects:

[0017] Conflict detection was performed before data fusion to identify inconsistent forest resource growth indicators in the same area collected at different time points, avoiding inaccurate analysis results due to data conflicts. For data conflicts in single independent areas and multiple independent areas, corresponding correction methods were adopted respectively, and the forest resource growth quantitative model was used to correct the data, which improved the accuracy and reliability of the data and ensured the overall consistency and rationality of multi-region data. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of the data fusion method for the forestry data management system in this application. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0020] This application discloses a data fusion method for a forestry data management system. Figure 1The method includes the following steps: S1. performing data format unification, data quality purification, spatial coordinate normalization, and time series standardization on the real-time collected data of the forestry data management system; S2. dividing the forest resource data of the target area into different independent areas; S3. establishing a special forest resource growth quantitative model for the forest resource data of each independent area; S4. The forestry data management system detects whether there is a forest resource data conflict before data fusion, and the forest resource data conflict refers to the inconsistency of forest resource growth indicators of the same area collected at different time points; S5. When a conflict occurs in the forest resource data of a certain independent area, the conflicting data is corrected by the forest resource growth quantitative model of the independent area; S6. When a conflict occurs in the forest resource data of multiple independent areas, the conflicting data is jointly corrected by the forest resource growth quantitative models of the multiple independent areas; S7. After the conflicting data is corrected, the data fusion is completed.

[0021] Specifically, for S1. The real-time data collected by the forestry data management system is unified in data format, data quality is purified, spatial coordinates are normalized, and time series are standardized; the real-time data collected by the forestry data management system is unified in data format, specifically including spatial data specifications and attribute data management. Among them, the spatial data specifications for vector data, whether it is forest resource monitoring points or forest area sample plot boundary data, are stored in Geo JSON format. This format can accurately describe various geometric types such as points, lines, and surfaces, and has good readability and interoperability. For raster data, satellite remote sensing images are unified into 16-bit TIFF format, and the WGS84 geographic coordinate system and UTM projection are embedded at the same time.

[0022] For example, projection zones are automatically divided according to the longitude and latitude of different regions. For example, the UTM50N zone is used for the area between 30° and 40° north latitude. This ensures that spatial data can be accurately positioned and displayed regardless of its source and purpose. Structured data in attribute data management, such as key information such as tree diameter at breast height, tree height, and growth, is stored in a PostgreSQL database. Its structured data management capabilities facilitate data query, statistics, and analysis. Unstructured data, such as metadata from remote sensing images, is managed through the HDFS distributed file system. In the attribute data management process, the "Forestry Data Element Standard" (LY / T1662-2006) is strictly followed to ensure uniformity in the units and encoding rules of data fields.

[0023] The real-time collected data of the forestry data management system is purified from data quality aspects, including data integrity check, consistency processing, outlier and noise data processing, to ensure that the data quality meets the needs of subsequent analysis and fusion.

[0024] Specifically, for data integrity check, traverse the structured attribute data and check whether there are missing values in key fields. For missing data, if it is ground monitoring data, manual supplementation is preferred, and it is filled through field verification or calling historical data from similar areas; if the amount of missing data is large, it is estimated for data with time series characteristics; for unstructured data, check whether the metadata information is complete, such as the acquisition time, resolution, band and other key metadata of the remote sensing image. If there is any missing, obtain the supplement from the data source or related metadata repository. For spatial data integrity verification, for vector data, check whether the geometric elements of data such as forest resource monitoring points and forest plot boundaries are complete, such as whether there are problems such as broken lines and fragmented surfaces, perform topological verification on the vector data, and repair the topological errors found. For raster data, check whether there are areas with data loss or damage in the image, and compare the checksum of the image with the original storage record. If inconsistency is found, re-acquire the corresponding area data from the backup data source.

[0025] Specifically, for data consistency processing, including attribute data consistency adjustment, in the attribute data management process, the attribute fields of different data sources are standardized. For the case where there are multiple ways of expressing the same attribute, such as "tree variety" and "tree species name", they are standardized as "tree species" and standardized. For numerical attribute data, check whether the units of different data sources are unified, such as converting "forest growth" data of different units into "cubic meters / hectare". When there are multiple numerical records for the same attribute (such as the diameter at breast height of trees recorded by different monitoring equipment on the same plot), a weighted average is performed according to the data credibility weight. The credibility weight is determined according to the accuracy of the data acquisition equipment and the acquisition frequency. For example, the weight of high-precision sensor data is 0.8, and the weight of manual measurement data is 0.5, to generate a unique and accurate attribute value.

[0026] Specifically, this involves integrating spatial data consistency. For vector data, spatial topological relationships (such as intersection and inclusion) are used to accurately match forest resource boundaries and monitoring points from different sources. Overlap analysis is conducted between forest farm boundaries and forest boundaries interpreted from satellite remote sensing, and any deviated boundaries are corrected to ensure that the spatial error between the two does not exceed one grid cell.

[0027] For raster data, seamless mosaicking is used based on the UTM projection zone. During the mosaicking process, histogram matching is used to adjust the brightness and color differences between adjacent images to ensure consistent tones in the fused images. Resampling technology is used to unify images of varying resolutions to a 10-meter resolution, ensuring consistency in spatial detail.

[0028] Specifically, when processing outliers and noisy data, the median absolute deviation method is used to replace outliers to reduce their impact on data analysis results. For high-frequency data collected by ground sensors (such as minute-level data), methods such as sliding average filtering and median filtering are used to remove random noise. For soil moisture sensor data, for example, by setting an appropriate sliding window size (such as 15 minutes), the average or median of the data within the window is calculated and used as the filtered data value. For salt and pepper noise in remote sensing image data, algorithms such as median filtering and mean filtering are used to remove it, improving image quality while preserving image detail.

[0029] Specifically, the real-time data collected by the forestry data management system was spatially normalized, unifying all forestry data to the national geodetic coordinate system. Planar coordinates used the Gauss-Krüger projection (3° zoning), and the elevation datum was the 1985 National Elevation Datum. Historical data was corrected using georeferencing tools. Through precise control point matching and geometric transformation, the spatial error between ground monitoring points and remote sensing pixels was kept within 10 meters, or less than the size of a single 10-meter resolution pixel, achieving precise spatial alignment of data from different sources. The study area was divided into a regular 100m×100m grid, each assigned a unique ID. Through spatial overlay analysis, data of different resolutions, such as 30m-resolution remote sensing imagery and 5m-resolution ground monitoring point data, were linked to corresponding grids. This allows for grid-level spatial alignment during data processing and analysis, facilitating rapid retrieval and processing of multi-source data within the same spatial area.

[0030] Time series standardization is performed on real-time data collected by the forestry data management system. Specifically, all data timestamps are standardized using UTC time, accurate to the second. Satellite remote sensing data (temporal resolution 5-16 days), ground sensor data (minute-level), and forest resource inventory data (annual) are uniformly converted to a daily scale (YYYY-MM-DD) to ensure consistency in the temporal dimension of data of different time frequencies, facilitating time series analysis and data fusion. For high-frequency ground sensor data, aggregation and downsampling are performed by calculating the mean, converting minute-level data to daily-scale data. For example, data collected every 15 minutes is used to generate a daily average. For low-frequency forest resource inventory data, cubic spline interpolation is used to expand the annual data into a continuous time series, ensuring that the time axis error does not exceed 24 hours, meeting the needs of subsequent time series-based forest resource growth model analysis and dynamic monitoring.

[0031] Specifically, for S2. Divide the target area's forest resource data into distinct independent regions. In forestry data management systems, this division is fundamental to precise data management and conflict resolution. Specifically, spatial data such as remote sensing imagery, topographic data (DEM), soil type maps, and vegetation distribution maps are integrated, and the overlay analysis capabilities of geographic information systems (GIS) are utilized to identify areas with similar characteristics. For example, vegetation type maps can be overlaid with soil type maps to identify distribution areas with specific vegetation and soil combinations. Combined with vector data such as administrative boundaries and forest farm boundaries, the initially divided regions can be adjusted to better meet actual management needs.

[0032] Specifically, the division method involves using a clustering algorithm to automatically divide the target area into clusters based on the spatial distribution characteristics of forest resources (such as tree density, average diameter at breast height, and tree height) and ecological and environmental factors (altitude, slope, and aspect). The clustering results are manually verified and optimized to ensure that the forest resources within each independent area are homogeneous and the boundaries are clearly distinguishable.

[0033] The division method also includes grid optimization. This involves merging adjacent grids with similar attributes into independent regions based on the established 100m×100m regular grid. For example, multiple consecutive grids dominated by pine trees and sharing the same soil type are combined into a single independent pine planting region. Smaller grids or those with unique attributes are assigned to adjacent independent regions based on their similarity to surrounding areas, thus avoiding the creation of overly small or isolated division units.

[0034] After demarcating the independent regions, each was assigned a unique identification code (e.g., "RZ-001") and annotated with key attribute information, including the region name (e.g., "XX Mountain Coniferous Forest Region"), administrative division, primary vegetation type, soil type, and average altitude, to form a regional attribute profile. Forest resource statistics were then calculated for each region, such as total area, total forest volume, and average canopy density, providing foundational data for subsequent growth model development and data conflict analysis.

[0035] The forest resource data (including spatial data, attribute data, monitoring data, etc.) in each independent area are associated with the identification code of the area to ensure the integrity and traceability of the data.

[0036] Specifically, S3. For each independent region's forest resource data, a dedicated forest resource growth quantification model is established. The forest resource growth quantification model is specifically as follows: ;

[0037] y(t) is the forest resource growth index, and y(t) specifically represents the diameter at breast height or tree height or biomass;

[0038] K is the limit value of forest resource growth, which indicates the maximum value that the forest resource growth indicator y(t) (such as breast diameter or tree height or biomass) can reach under the current site conditions, reflecting the environmental limitations on forest growth.

[0039] r is the intrinsic growth rate, which indicates the inherent growth rate of forest resource growth indicators under no environmental restrictions, and reflects the growth potential of the tree species itself.

[0040] The average growth rate of forest resource growth indicators over different time periods is calculated. Through multiple fitting and adjustments, the r value that best describes the growth trend of a particular tree species is determined. For example, the intrinsic growth rate r of poplar trees in a certain region is calculated to be 0.2 per year through fitting, meaning that under ideal conditions, the growth indicator for poplar trees increases by 20% annually.

[0041] t0 is the growth inflection point time, which is the time point when the growth rate of forest resources reaches its maximum value. Before t0, the growth rate gradually increases; after t0, the growth rate gradually slows down.

[0042] Furthermore, it is preferred to further incorporate site factors (such as altitude, slope, aspect, soil type, and soil fertility) to modify the parameters of the quantitative forest resource growth model. A regression relationship between site factors and model parameters is established. For example, the growth limit K of a particular tree species decreases with increasing altitude, and the higher the soil organic matter content, the greater the K value. Through extensive data statistics and analysis, the specific regression equation parameters are determined. For the intrinsic growth rate r, slope and aspect affect the tree's access to light and water, and thus its growth rate. By comparing and analyzing data from plots with different slopes and aspects, a correction formula for r and slope and aspect is established, enabling the model to more accurately reflect the growth of forest resources under different site conditions.

[0043] Specifically, the growth limit value K is corrected: with K as the dependent variable, altitude (x1), soil fertility index (x2), and vegetation type (x3, dummy variable) as independent variables, a linear regression equation is constructed, and the parameters of the linear regression equation are fitted by the least squares method to determine the relationship for the correction of the growth limit value K. For example, K (the maximum tree height that can be reached by forest resources) = 60-0.04x1+15x2+x3, (x1 is in meters, x2 is the fertility index, and x3 takes a fixed value), which means that for every 100m increase in altitude, the K value decreases by 4%; for every 0.1 increase in the fertility index, the K value increases by 1.5%.

[0044] Specifically, the intrinsic growth rate r is corrected: for example, in view of the influence of slope (y1) and slope direction (y2, sunny slope = 1, shady slope = 0), a nonlinear correction model is constructed: r = r0 * (1-0.015y1) * (1 + 0.1y2), where r0 is the basic growth rate. After calibration with sample plot data, the r value decreases by 15% for every 10° increase in slope; the r value in the sunny slope area is 10% higher than that in the shady slope.

[0045] Specifically, for S4. The forestry data management system detects whether there are forest resource data conflicts before data fusion. This forest resource data conflict refers to the inconsistency of forest resource growth indicators collected in the same area at different time points. For the same monitoring point, if the growth increase in adjacent years exceeds 1.5 times the maximum growth rate predicted by the forest resource growth quantification model, it is determined that there is a data conflict in the time dimension. For example, if the model predicts that the maximum annual growth rate of a certain tree species, poplar, is 5cm, when the actual growth increase in adjacent years exceeds 7.5cm, conflict detection is triggered. Although tree growth is affected by many factors, it will not show drastic abnormal growth in a short period of time. Data exceeding this threshold is very likely to have problems.

[0046] Specifically, based on a pre-built 100m x 100m regular grid, each grid is assigned a unique GridID. Data detection will use the grid as the basic unit, and the forest resource data within the grid will be centrally analyzed.

[0047] Specifically, when performing conflict detection, the system uses the current grid and its ±1 adjacent grid as the spatial detection range. This comprehensively considers the influence of adjacent areas and avoids misjudgments due to boundary errors or local anomalies. In terms of time, the detection window extends 30 days forward and backward, centered on the current time point. This is because forest resource growth changes relatively continuously over a short period of time (30 days). Data inconsistencies within this time window are more likely to indicate a data conflict than a true growth mutation.

[0048] For the selected data, the system compares and calculates the actual measured forest resource growth indicators (such as diameter at breast height, tree height, and biomass) with the predicted values based on the forest resource growth quantification model. During the calculation process, the system fully considers the impact of model parameters (such as the growth limit value K, the intrinsic growth rate r, and the growth inflection point time t0) and site factors (such as altitude, slope, and soil type) on the predicted values, thereby determining the deviation between the measured and predicted values.

[0049] The calculated deviation is compared with the pre-set test standard. If the deviation exceeds the standard range, the data is considered to be in conflict and the conflicting data is marked.

[0050] Specifically, S5. When a conflict occurs in the forest resource data of an independent area, the conflicting data is corrected through the forest resource growth quantification model of the independent area; after detecting the data conflict, the system first determines the specific type of the conflicting data, such as the tree diameter at breast height, tree height, and biomass growth index data. At the same time, based on the previous grid division and detection area setting, the grid range where the conflicting data is located is accurately located, its spatial position in the independent area is clarified, the forest resource growth quantification model parameters are called, and the conflicting data correction operation is performed.

[0051] Conflicting data are growth indicators such as DBH, tree height, and biomass. The t value at that moment, along with the model parameters K, r, and t0, are directly substituted into the forest resource growth quantification model to calculate the predicted value. For example, the measured DBH growth of a 20-year-old pine tree is 6 cm per year, exceeding the model's predicted upper limit of 5 cm. Substituting t = 20 and the corresponding K = 50, r = 0.15, and t0 = 15 for pine trees in this area into the formula, we obtain y(20) = 45.5 cm. The DBH growth was 40 cm in the previous year, and the predicted growth for the current year was 5.5 cm. This predicted value replaces the measured conflict value to complete the data correction.

[0052] Specifically, S6. When conflicting forest resource data for multiple independent regions is detected, the conflicting data is corrected by combining the forest resource growth quantification models for the multiple independent regions. When conflicting forest resource data for multiple independent regions is detected, the system integrates the forest resource growth quantification models for the multiple regions to construct a joint correction. The specific implementation steps are as follows:

[0053] Collect all parameters for quantitative forest resource growth models in conflict areas, including the growth limit value K (representing the maximum value that forest resource growth indicators such as diameter at breast height, tree height, or biomass can reach under current site conditions, reflecting environmental constraints on forest growth), the intrinsic growth rate r (indicating the inherent growth rate of forest resource growth indicators in the absence of environmental constraints, reflecting the growth potential of the tree species itself), and the growth inflection point time t0 (the time point when the forest resource growth rate reaches its maximum; before t0, the growth rate gradually increases and then gradually slows down). Simultaneously, summarize site factor parameters for each region, such as altitude, slope, aspect, soil type, and soil fertility. These site factors will be used to adjust model parameters through regression relationships to adapt to the growth environment of different regions.

[0054] Each region participating in the joint correction is weighted based on factors such as its spatial distance from the conflicting data area, ecological similarity, and data reliability. For example, regions adjacent to the conflicting area and with similar ecological environments receive higher weights; regions with high data collection frequency and high-precision monitoring equipment also receive higher weights. Weights range from 0 to 1, with the sum of all region weights equal to 1. This is used to balance the influence of each region's model during subsequent joint calculations.

[0055] Based on the 100m×100m regular grid and regional division results established in the system, the specific grids and multiple independent regions containing conflicting data are determined. Combined with the spatial detection range set during conflict detection (the current grid and its ±1 adjacent grid), the affected regional boundaries and data range are clearly defined to ensure comprehensive coverage of potentially related forest resource data.

[0056] For each independent region participating in the joint correction, the time t corresponding to the conflicting data is substituted into its forest resource growth quantification model to calculate its respective predicted value. Then, based on pre-set regional weights, the predicted values for each region are weighted and summed to produce a joint predicted value. This joint predicted value comprehensively considers the growth patterns and site conditions of multiple regions, better reflecting the true growth trends in the region where the conflicting data resides.

[0057] The combined predictions are used as the correction for conflicting data, replacing the original conflicting data. For example, if the conflicting data is an abnormal annual growth rate of the measured diameter at breast height (DBH) of a particular cross-regional forest, and the predicted growth rate obtained through the combined calculation is more reasonable, this prediction is used to update the original data.

[0058] In addition, preferably, due to the spatial correlation of data in multiple regions, after completing the correction of conflicting data, the data in other regions associated with the corrected data are automatically checked. If it is found that the data in other regions may have new conflicts or logical contradictions due to this correction (such as unreasonable differences in tree growth in adjacent regions), local secondary detection and adjustment will be initiated to ensure the overall consistency and rationality of the multi-region data. Specifically:

[0059] A threshold for the difference in growth indicators between adjacent areas is set. Detection is triggered when the difference in similar growth indicators (such as diameter at breast height and tree height) between the corrected area and the adjacent independent area exceeds 15% of the model predicted value for the area.

[0060] For example, if a grid in region A shows a corrected annual growth rate of 5.5 cm in diameter at breast height (DBH), while the model-predicted value for the same type of tree in adjacent region B is 4.5 cm, a spatial correlation conflict is identified if the absolute value of the difference exceeds 0.675 cm (4.5 cm x 15%). For adjacent grids with similar site factors (elevation difference <50 m, slope difference <10°, soil fertility index difference <10%), if the difference in measured growth exceeds 20% of the corrected intrinsic growth rate (e.g., if the r-value for both regions is 0.15, the permissible difference threshold is 0.03), a potential logical contradiction is flagged. The detection area is automatically expanded to ±2 adjacent grids (forming a 3×3 grid detection area), centered around the grid containing the conflicting data, to cover any area potentially affected by spatial transmission effects. If the conflicting area crosses an independent regional boundary (e.g., a forest farm boundary or a vegetation type boundary), all associated grids in the adjacent region are simultaneously included to ensure cross-regional data consistency.

[0061] Specifically, secondary inspection and adjustment include:

[0062] Using GIS overlay analysis, the conflicting grids were compared with adjacent grids for forest resource boundaries, vegetation type distribution, and site factor gradients. Changes in growth indicators across regional boundaries were required to conform to ecological transition patterns (e.g., for every 100-meter increase in altitude, the predicted diameter at breast height growth should decrease by 3%-5%). Alerts were triggered when the magnitude of the sudden change exceeded this pattern by more than twice. Growth data from the affected area over the past three years was extracted, and the standard deviation (σ) of the annual growth rate was calculated. If the corrected data caused the current growth rate to fluctuate by more than ±2σ of the historical mean (e.g., if the historical standard deviation was 0.5 cm and the current fluctuation was greater than 1 cm), a temporal anomaly was identified.

[0063] Retrieve site factor data (such as altitude, slope, and soil fertility) from the conflicting area and adjacent areas, and use established regression relationships to infer model parameters. For example, if the corrected K value (growth limit) in area A is 50 cm, and the K value for the same tree species in adjacent area B is 40 cm, but the soil fertility difference between the two sites is only 5% and the altitude difference is less than 30 meters, an abnormal K value alarm will be triggered, indicating a possible model parameter conflict.

[0064] It is also possible to quantify the weight of the impact of site factors on growth indicators (for example, the weight of the impact of altitude on K value is 0.4, and soil fertility is 0.3). If the corrected data leads to a contradiction in the weight contribution (for example, the K value in the low-altitude area is lower than that in the high-altitude area), it is determined to be a model fitting deviation.

[0065] For adjacent areas where conflicts were detected, corrected data were combined with historical monitoring data to refit the parameters of the forest resource growth quantitative model. For example, if the r-value deviation in area B due to the correction in area A exceeded 10%, data for the same tree species from the past five years in both areas were extracted and recalculated using a nonlinear least squares method to ensure that the cross-regional model parameters were consistent with the site factor gradient (e.g., for every 10° increase in slope, the r-value decrease should be within a range of 10%-20%). The regional weighting of the joint correction model was adjusted, increasing the weight of adjacent areas to 0.6-0.8 (from a previous weight of ≤0.5), minimizing the impact of distant regions and ensuring that the correction results were consistent with local ecological consistency.

[0066] For grids that cross regional boundaries, an inverse distance weighted (IDW) algorithm is used to generate transitional values. The weighting factor combines spatial distance (weight 0.7 for areas <50 m, 0.3 for areas 50-100 m) and ecological similarity (weight 0.2 for identical vegetation types, 0.1 for identical soil types). For example, if the corrected DBH value in area A is 45 cm, and the uncorrected value in adjacent area B is 40 cm, the boundary grid will be interpolated to 43 cm (45 × 0.7 + 40 × 0.3 at a distance of 50 m). To address time series fluctuations caused by the corrected data, cubic spline interpolation is used to supplement intermediate values to ensure continuity of the growth curve. For example, if the corrected growth rate in area A in 2023 is 5 cm, and the uncorrected growth rate in area B is 4 cm, the boundary grid will be interpolated to 4.5 cm, keeping the annual growth rate fluctuation within ±10% of the historical mean.

[0067] S7. Complete data fusion after correcting conflicting data; specifically, the data fusion levels include attribute data fusion, spatial data fusion, and time series fusion.

[0068] Attribute data fusion involves standardizing the mapping of attribute fields from different data sources according to the Forestry Data Metadata Standard (LY / T1662-2006). For example, fields such as "tree species" and "tree species name" are standardized as "tree species" and coded using the GB / T14396-2008 standard to ensure consistent semantics and coding for the same attribute.

[0069] For multiple values of the same attribute (such as the diameter at breast height of trees recorded by different monitoring devices on the same plot), a weighted average is performed based on the data's credibility weight. The credibility weight is determined based on the accuracy and frequency of the data collection device, for example, a weight of 0.8 for high-precision sensor data and 0.5 for manually measured data. This eliminates data discrepancies and generates a single, accurate attribute value.

[0070] Spatial data fusion involves accurately matching forest resource boundaries and monitoring points from different sources using spatial topological relationships (such as intersection and inclusion) within vector data. For example, forestry boundaries are overlapped with forest boundaries interpreted from satellite remote sensing, and any discrepancies between them are corrected to ensure that the spatial error between the two does not exceed one grid cell (100m x 100m).

[0071] For raster data such as satellite remote sensing images, seamless mosaicking is used based on the UTM projection band. During the mosaicking process, the brightness and color differences of adjacent images are adjusted through histogram matching to ensure the consistency of the tones of the fused images. At the same time, resampling technology is used to unify images of different resolutions to 10 meters to ensure the consistency of spatial details.

[0072] Time series fusion involves aligning data with different temporal resolutions (such as minute-level sensor data and annual inventory data) to a daily scale, using a unified UTC time (accurate to the second) as the benchmark. Data with time offsets is corrected based on its collection cycle and patterns. For example, if a sensor's data is two hours behind due to clock error, by calculating its fixed collection interval, all data can be shifted forward by two hours, achieving precise time series alignment.

[0073] Based on a quantitative model of forest resource growth, data from different time points are concatenated into a continuous growth curve. For missing time periods, the model is used to interpolate and predict, ensuring the completeness of the time series. For example, if a region's tree growth data is missing for six months between 2022 and 2023, the model inputs the region's site factors (elevation, slope, etc.) and existing growth parameters (K, r, t0) to predict growth indicators for the missing period, ensuring that the entire time series accurately reflects the dynamic changes in forest resources.

[0074] It is apparent that one or more steps of the method of the present invention can be implemented by a computer program. The computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flowchart and / or block diagram are implemented. The computer program can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0075] Therefore, it can be understood that the present invention discloses an electronic device, including:

[0076] at least one processor; and

[0077] a memory communicatively connected to the at least one processor; wherein,

[0078] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform one or more steps of the above method.

[0079] Finally, it should be noted that the above is only a preferred embodiment of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or make equivalent replacements for some of the technical features therein. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A data fusion method for a forestry data management system, characterized in that: The method includes the following steps: unifying the data format, purifying the data quality, normalizing the spatial coordinates, and standardizing the time series of the real-time collected data of the forestry data management system; Divide the forest resource data of the target area into different independent areas, and establish a special forest resource growth quantitative model for the forest resource data of each independent area; The forestry data management system detects whether there is any conflict in forest resource data before data fusion; the conflict in forest resource data refers to the inconsistency of forest resource growth indicators collected in the same area at different time points; When there is a conflict in the forest resource data of a certain independent area, the conflicting data will be corrected through the quantitative model of forest resource growth in that independent area; When there is a conflict in the forest resource data of multiple independent regions, the conflicting data will be corrected through the quantitative forest resource growth model of the multiple independent regions; Correct conflicting data before completing data fusion; After completing the conflicting data correction, the system automatically checks the data in other regions that are associated with the corrected data. If it is found that the data in other regions may have new conflicts or logical contradictions due to this correction, it will initiate a local secondary check and adjustment to ensure the overall consistency and rationality of the multi-region data. Secondary testing and adjustments include: using GIS overlay analysis to compare forest resource boundaries, vegetation type distribution, and site factor gradient changes between conflicting grids and adjacent grids. When the magnitude of a sudden change in growth indicators across regional boundaries exceeds the threshold of ecological transition laws, an early warning is triggered. Growth data for the affected area over the past three years is extracted, and the standard deviation of the annual growth rate is calculated. If the corrected data causes the current growth rate to fluctuate by more than the standard deviation of the historical mean, it is determined to be a temporal anomaly. Retrieve site factor data of the conflict area and adjacent areas, and infer model parameters through the established regression relationship; Quantify the weight of the influence of site factors on growth indicators. If the modified data leads to a contradiction in the weight contribution, it is considered a model fitting deviation. For adjacent areas where conflicts were detected, the revised data were combined with historical monitoring data to refit the parameters of the forest resource growth quantitative model; Adjust regional weight distribution of joint correction; Interpolation was performed on grids that crossed regional boundaries, and transition values were generated using an inverse distance weighted interpolation algorithm, with the weight factor combining spatial distance and ecological similarity; In order to address the time series mutations caused by the corrected data, cubic spline interpolation is used to supplement the intermediate values to ensure the continuity of the growth curve; By integrating the quantitative growth models of forest resources in multiple regions, a joint correction is constructed, specifically including: collecting the quantitative growth model parameters of all forest resources involved in the conflict area, summarizing the site factor parameters of each region, and correcting the model parameters through regression relationship to adapt to the growth environment of different regions; setting weights for each region participating in the joint correction based on the spatial distance between each region and the area where the conflict data is located, the similarity of the ecological environment, and the data reliability factor; determining the specific grid where the conflict data is located and the multiple independent regions to which it belongs, and determining the affected regional boundaries and data range based on the spatial detection range set during conflict detection; for each independent region participating in the joint correction, substituting the time t corresponding to the conflict data into its forest resource growth quantitative model to calculate its respective predicted values; then, according to the pre-set regional weights, the predicted values of each region are weighted and summed to obtain the joint predicted value, which is used as the correction result of the conflict data to replace the original conflict data.

2. The data fusion method of a forestry data management system according to claim 1, characterized in that: Perform data quality purification on real-time collected data of forestry data management system, including: data integrity check, data consistency processing, and outlier and noise data processing; Normalize the spatial coordinates of the real-time data collected by the forestry data management system. This includes unifying all forestry data into the national geodetic coordinate system. Plane coordinates use the Gauss-Krüger projection. The study area is divided into regular grids, each grid is assigned a unique ID, and data of different resolutions are associated with the corresponding grids through spatial overlay analysis. The real-time data collected by the forestry data management system is standardized in time series, including: unifying the timestamps of all data to UTC time, converting satellite remote sensing data, ground sensor data, and forest resource inventory data to daily scale, and aggregating and downsampling high-frequency ground sensor data by calculating the mean to convert minute-level data into daily scale data. For low-frequency forest resource inventory data, cubic spline interpolation is used to expand the annual data into a continuous time series.

3. The data fusion method of a forestry data management system according to claim 1, characterized in that: In the forestry data management system, the forest resource data of the target area is divided into different independent areas, including: integrating remote sensing images, terrain data, soil type maps, and vegetation distribution map spatial data to identify areas with similar characteristics, and adjusting the initially divided areas based on administrative boundaries and forest farm boundary vector data; the division method includes clustering algorithm assistance and grid division optimization, merging adjacent grids with similar attributes into independent areas; after dividing the independent areas, assigning a unique identification code to each independent area and marking key attribute information, which includes the area name, administrative division, main vegetation type, soil type, and average altitude to form a regional attribute file; and associating the forest resource data in each independent area with the identification code of the area.

4. The data fusion method of a forestry data management system according to claim 1, characterized in that: The specific quantitative model for forest resource growth is as follows: , y(t) is the forest resource growth index; When y(t) specifically represents the DBH, K is the DBH limit of forest resources, which means the maximum value of DBH that forest resources can reach under the current site conditions, r is the intrinsic growth rate of DBH, which means the inherent growth rate of DBH of forest resources under no environmental restrictions, t0 is the inflection point of DBH growth. Before t0, the DBH growth rate gradually increases; after t0, the DBH growth rate gradually slows down. When y(t) specifically represents tree height, K is the limit value of tree height growth of forest resources, which means the maximum value that tree height can reach under the current site conditions. r is the intrinsic growth rate of tree height, which means the inherent growth rate of tree height of forest resources under no environmental restrictions. t0 is the inflection point of tree height growth. Before t0, the tree height growth rate gradually increases; after t0, the tree height growth rate gradually slows down. When y(t) specifically represents biomass, K is the limit value of forest resource biomass growth, which means the maximum value that forest resource biomass can reach under current site conditions; r is the intrinsic growth rate of biomass, which means the inherent growth rate of forest resource biomass under no environmental restrictions; t0 is the inflection point time of biomass growth. Before t0, the biomass growth rate gradually increases; after t0, the biomass growth rate gradually slows down.

5. The data fusion method of a forestry data management system according to claim 1, characterized in that: Detecting whether there is a forest resource data conflict specifically includes: for the same monitoring point, if the growth increase in adjacent years exceeds the threshold of the maximum growth rate predicted by the forest resource growth quantitative model, it is determined that there is a data conflict in the time dimension.

6. The data fusion method of a forestry data management system according to claim 5, characterized in that: When a conflict occurs in the forest resource data of a certain independent area, the conflict data is corrected through the forest resource growth quantification model of the independent area, including: first determining the specific type of conflict data, which includes: tree diameter at breast height, tree height, and biomass growth index data; at the same time, based on grid division and detection area setting, locating the grid range where the conflict data is located, clarifying its spatial position in the independent area, and calling the forest resource growth quantification model parameters; executing conflict data correction includes calculating the predicted value through the forest resource growth quantification model, replacing the measured conflict value with this predicted value, and completing the data correction.

7. The data fusion method of a forestry data management system according to claim 1, characterized in that: The data fusion levels include attribute data fusion level, spatial data fusion level, and time series fusion level.

Images