A forest stand quality intelligent improvement system based on soil multi-source data

Abstract

The application relates to the field of forestry information processing and discloses a forest stand quality intelligent improvement system based on soil multi-source data, which comprises the following modules: a data preparation module, which collects and constructs a soil multi-source database, generates homogeneous forest stand plots and a restriction item set, a measure decision module, which queries positive evidence indexes and veto evidence indexes, generates a supplementary measurement task according to a gap formula when positive and negative evidence conflicts occur, and generates a measure combination after query recovery according to a frozen version, an execution control module, which allocates a delayed reference forest stand plot, establishes a delayed execution plan lock, a master token and a channel sub-token and issues a management plan, an evidence construction module, which calculates and processes a reference net response and writes an evidence pointer, an archiving update module, which completes atomic archiving, enables an index version and releases a delayed reference to re-query. The application can direct supplementary measurement according to positive and negative evidence conflicts, keep the query and index versions consistent, complete delayed reference allocation, channel evidence archiving and index backwriting at the same time.

Description

Technical Field

[0001] This invention relates to the field of forestry information processing, specifically to an intelligent forest stand quality improvement system based on multi-source soil data. Background Technology

[0002] Forest stand management-related data includes indicators such as soil pH, organic matter, nutrients, moisture, microorganisms, and tree growth. Data sources cover sampling and testing, sensor monitoring, and management records. Existing methods typically compare indicators with fixed thresholds, or weight multiple indicators to obtain a comprehensive score before providing management recommendations; alternatively, historical measures can be retrieved based on spatial proximity or case similarity, or changes in indicators before and after management of the same plot can be compared.

[0003] During the measure retrieval phase, valid and invalid results are often included in the same scoring or ranking process, and positive and negative historical information can easily cancel each other out. When positive and negative evidence coexist, there is a lack of processing to determine the minimum supplementary test field based on the conflict relationship and to restore the retrieval according to the original query generation and the original index version after supplementary testing.

[0004] Regarding the control organization, fixed control plots are often set up first for re-evaluation measures, but the correspondence between the reference objects and the currently proposed measures and their response cycles is insufficient. The system also lacks a mechanism to restrict the issuance of conflict plans while allowing entries into actual on-site management records. In terms of results archiving, observations from different response cycles are often evaluated together, and data from the same time period and response channel are repeatedly cited by multiple measures. Even when protection indicators have not been verified, results are still preserved as valid evidence, and there is a lack of connection between index updates and plot re-queries. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent forest stand quality improvement system based on multi-source soil data. This system solves the problems in existing technologies, such as the difficulty in completing targeted supplementary testing and peer-to-peer query recovery when there are conflicts between positive and negative historical evidence, the difficulty in dynamically organizing processing and comparison according to current measures, and the difficulty in controlling evidence attribution and completing archiving and write-back according to response channels.

[0006] To achieve the above objectives, the present invention provides the following technical solution: This invention provides an intelligent forest stand quality improvement system based on multi-source soil data, including a data preparation module, a measure decision-making module, an execution control module, an evidence construction module, and an archiving and updating module.

[0007] The data preparation module is used to collect and build a multi-source soil database, and generate homogeneous forest stand plots and a set of restrictions.

[0008] The measure decision module is used to query the positive evidence index and the negative evidence index based on the set of restrictions. When a conflict between positive and negative evidence is formed, the gap formula is called to determine the minimum discrimination gap and generate a supplementary test task. The frozen query generation and frozen index version are recorded. After the supplementary test task is completed, the query is restored according to the frozen query generation and frozen index version. After the conflict between positive and negative evidence is eliminated, a measure combination is generated, and the homogeneous forest stand is converted into a treated forest stand.

[0009] The execution control module is used to allocate delayed reference forest stands to the treated forest stands based on the combination of measures, establish delayed execution plan locks, master tokens, and channel sub-tokens, and issue management plans according to the master token and channel sub-tokens. The evidence construction module is used to calculate the net response of the treated forest stands and delayed reference forest stands, and write evidence pointers into the positive evidence index or the negative evidence index based on the net response of the treated reference forest stands.

[0010] The archive update module is used to perform atomic archiving of evidence pointers by channel sub-token, generate and enable indexed versions, release the master token, postpone the execution plan lock and postpone the reference forest stand plots and re-query.

[0011] Furthermore, the data preparation module includes a multi-source access unit, a homogeneous partitioning unit, and a fragment preparation unit.

[0012] The multi-source access unit is used to collect soil and stand monitoring data, management events, and environmental disturbance events to construct a multi-source soil database. The homogeneity partitioning unit reads valid sampling points and category fields, calls the homogeneity formula to calculate spatial heterogeneity, and identifies homogeneous stand plots when the spatial heterogeneity is no greater than 1 and the category fields are consistent; otherwise, it splits the stand and recalculates. The segment preparation unit forms root zone state segments with management events as boundaries, determines the baseline state before management and the response window after management, generates a set of constraints and a set of priority constraints, and registers homogeneous stand plots as queryable stand plots.

[0013] Furthermore, the evidence construction module includes an evidence initialization unit, a net response determination unit, and a dual index unit.

[0014] The evidence initialization unit is used to trigger evidence construction when the fields required for evidence access are available or when the response window after management ends and response collection is completed. When both the positive evidence index and the negative evidence index are empty, a processing and reference verification experiment is established. The net response determination unit is used to establish an evidence pair consisting of the treated forest stand and the delayed reference forest stand. Based on the evidence pair, the baseline comparability before management is calculated. When the baseline comparability before management is not greater than 1, the net response formula is called to calculate the processing reference net response, and the removal of restrictions and inverse response judgments are completed based on the processing reference net response. The dual-index unit is used to form the evidence index key and determine the archiving category. When the archiving category is positive, the evidence pointer is written to the positive evidence index by pressing the evidence index key; otherwise, it is written to the negative evidence index or the non-indexed evidence area.

[0015] Furthermore, the measure decision-making module includes an evidence retrieval unit.

[0016] The evidence retrieval unit is used to treat homogeneous forest stands as the current forest stand, generate query generations, read root zone state fragments and determine the post-management response window and response cycle category, query the positive evidence index and the negative evidence index, and read the measure codes from candidate historical evidence to form condition fields and calculate migration mismatch. When the migration mismatch is not greater than 1, it forms either transferable positive evidence or unresolved negative evidence according to the archive category. When transferable positive evidence and unresolved negative evidence simultaneously correspond to the same current forest stand, the same restriction, the same measure code, the same response cycle category, and the same query generation, it forms positive-negative evidence conflict and positive-negative evidence pairs.

[0017] Furthermore, the action decision module also includes a conflict retesting unit.

[0018] The conflict retesting unit is used to determine the separable fields of positive and negative evidence pairs. When a separable field is missing or invalid, it is included in the candidate field set for retesting. The gap formula is called to determine the minimum discriminative gap and generate a retesting task, recording the frozen query generation and frozen index version. After completing the retesting task and obtaining the retesting results, the results are written back to the condition fields. The query is then restored according to the frozen query generation and frozen index version, and the conflict between positive and negative evidence pairs is eliminated. When a new management event occurs, the data in the soil multi-source database associated with the current query context undergoes substantial correction, or the frozen query generation and frozen index version become unreadable, the original query is abandoned and a query re-trigger is executed. When no separable field exists, the process proceeds to processing and reference verification experiments.

[0019] Furthermore, the measure decision-making module also includes a measure generation unit.

[0020] The measure generation unit, when there is transferable positive evidence and no unresolved veto evidence, takes the intersection of the historical execution parameter range supported by positive evidence, the permitted range of technical specifications, and the executable range of the current project to form a parameter range, and forms a candidate measure set. When the priority restriction set is not empty, the measure selection formula is called to generate measure combinations from the candidate measure set and generate a management plan, transferring the current forest stand to a treatment forest stand. When the newly generated combination measure code has no historical evidence, it proceeds to the treatment and reference verification experiment.

[0021] Furthermore, the execution control module includes a reference locking unit.

[0022] The reference locking unit is used to determine the target response channel and necessary protection channel based on the combination of measures. It matches delayed reference forest stands as reference forest stands for the treated forest stands from the queryable forest stand plots with pre-management baseline status. A delayed execution plan lock is generated; during its effective period, no corresponding or conflicting measures are issued to the delayed reference forest stand. Actual management events are received from the site and written to the soil multi-source database. When the delayed reference forest stand actually implements corresponding measures or conflicting measures within the post-management response window, a reference failure record is generated, and the affected channel sub-token becomes invalid.

[0023] Furthermore, the execution control module also includes a token management unit.

[0024] The token management unit is used to establish master tokens and channel sub-tokens, and to write target response channels, necessary protection channels, and necessary response channels. It consists of a set of forest stands composed of treated stand plots and delayed reference stand plots, with the response time interval defined by the post-management response window. Effective occupancy is formed according to the forest stand set, response time interval, and response channel. When channel ownership conflicts occur, after eliminating the conflicts through delayed, phased, or combined measures, a conflict-free plan is generated. If channel ownership conflicts cannot be eliminated, the process transitions to a treatment and reference verification experiment.

[0025] Furthermore, the execution control module also includes a job execution unit.

[0026] The operation execution unit is used to issue maintenance plans based on the master token, channel sub-tokens, and conflict-free plans. It receives actual operation records and reads actual execution parameters, collects response observations, and transmits them to the archiving and updating module. When the actual execution parameters are not within the parameter range or a reference failure record occurs, the affected channel sub-token is marked as invalid.

[0027] Furthermore, the archive update module includes a channel archive unit and a closed-loop update unit.

[0028] The channel archiving unit is used to complete atomic archiving and generate an index version based on response observations when the channel sub-token's channel status is archiveable. When the target response channel is completed but the necessary protection channel is not, the evidence to be verified is written to the protection verification area. After the necessary protection channel is completed, the protection verification area is read and then archived. The closed-loop update unit is used to release the valid occupancy after all the channel sub-tokens corresponding to the necessary response channel have entered the archived or expired state, release the main token and the delayed execution plan lock, enable the submitted index version, restore the delayed reference forest stand to a queryable forest stand, and query again.

[0029] This invention provides an intelligent forest stand quality improvement system based on multi-source soil data. It has the following beneficial effects: 1. This invention uses a measure decision module to simultaneously query the positive evidence index and the negative evidence index. When there is a conflict between positive and negative evidence, the minimum discriminative gap is determined according to the gap formula and a supplementary test task is generated. Then, the query is restored according to the frozen query generation and the frozen index version, so that the supplementary test content is given by the fields that are divisible by both conflicting parties and are currently missing or invalid, and the query generation and index version are kept consistent before and after restoration.

[0030] 2. After the combination of measures is determined, the execution control module allocates delayed reference forest stand plots to the forest stand plots to be processed and establishes a delayed execution plan lock for organizing the processing-reference concurrent control. During the effective period of the plan lock, no corresponding or conflicting measures are issued to the delayed reference forest stand plots. The actual on-site management events are still written into the soil multi-source database. When the reference conflict is implemented, only the affected channel sub-tokens are invalidated.

[0031] 3. In this invention, the token management unit forms an effective occupation based on the forest stand and plot set, response time interval and response channel. The archiving and updating module completes atomic archiving based on the channel sub-token. When the target response channel is completed but the necessary protection channel is not completed, the evidence to be verified is written into the protection verification area. After the necessary response channel ends, the submitted index version is activated and the delayed reference forest stand and plot is released for re-query. Attached Figure Description

[0032] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a flowchart of the method of the present invention; Figure 3 This is a working logic diagram of the data preparation module of the present invention; Figure 4 This is a working logic diagram of the evidence construction module of the present invention; Figure 5 This is a working logic diagram of the decision-making module of the present invention; Figure 6 This is a working logic diagram of the control module of the present invention; Figure 7 This is a working logic diagram of the archiving and updating module of the present invention; Figure 8 This is a bar chart showing the grouped verification results of the system group and the control group in the application embodiment of the present invention; Figure 9 The system group and control group, as examples of applications of this invention, are treated with reference to the net response box plot. Detailed Implementation

[0033] The technical solutions in 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.

[0034] Please see the appendix Figure 1 This invention provides an intelligent forest stand quality improvement system based on multi-source soil data, comprising: The data preparation module is used to access soil and forest stand monitoring data, divide homogeneous forest stand plots using a homogeneous formula, and form root zone state fragments and constraint sets.

[0035] The evidence construction module is used to create evidence pairs, calculate the net response of the reference using the net response formula, and determine the archiving category and evidence index key.

[0036] The measure decision module is used to retrieve the positive evidence index and the negative evidence index, use the gap formula to determine the minimum discriminant gap, and use the measure selection formula to generate a combination of measures.

[0037] The execution control module is used to perform dynamic reference allocation, obtain delayed reference forest stand plots, establish master tokens and channel sub-tokens, execute management plans, and collect response observations.

[0038] The archive update module is used to complete channel-level atomic archiving, enable new index versions, release deferred reference forest stand plots, and re-query.

[0039] See appendix Figure 2 This invention also provides a method for intelligently improving forest stand quality based on multi-source soil data, comprising the following steps: S1, the data preparation module accesses soil and forest stand monitoring data, uses the homogeneous formula to divide homogeneous forest stand plots, and forms root zone state fragments and constraint sets.

[0040] S2, the evidence construction module establishes evidence pairs, calculates the net response using the net response formula, and determines the archiving category and evidence index key.

[0041] S3, the measure decision module retrieves the positive evidence index and the negative evidence index, uses the gap formula to determine the minimum discriminant gap, and uses the measure selection formula to generate a combination of measures.

[0042] S4 executes the management module to allocate delayed reference forest stand plots, establishes main tokens and channel sub-tokens, executes the management plan, and collects response observations.

[0043] S5, the archive update module completes channel-level atomic archiving, enables the new index version, releases deferred reference forest stand plots, and re-queries.

[0044] The technical solutions in the embodiments of the present invention will be further explained below in conjunction with the specific processing procedures.

[0045] See appendix Figure 3 In this embodiment, step S1 is performed by the data preparation module, which includes a multi-source access unit, a homogeneous division unit, and a fragment preparation unit, used to generate a multi-source soil database, homogeneous forest stand plots, root zone state fragments, a set of restrictions, a set of priority restrictions, and a set of candidate protection indicators.

[0046] S11, the multi-source access unit accesses soil pH, organic matter, nitrogen, phosphorus, and potassium, soil temperature, moisture content, microbial community, functional microorganisms, altitude, slope, aspect, slope position, soil thickness, tree height, diameter at breast height (DBH), crown width, canopy closure, preservation status, root activity, management events, and environmental disturbance events to establish a multi-source soil database. Environmental disturbance events must include at least the disturbance type, occurrence time, affected area, duration, disturbance intensity, record source, and valid markers.

[0047] The multi-source access unit adds data annotation information to each data entry, including plot identifier, sampling location, soil layer, collection time, testing method, source number, quality mark and data version.

[0048] Forest stand plots are spatial objects to be processed with defined spatial boundaries; homogeneous forest stand plots are the minimum processing objects whose tree species, stand age class, soil texture category, and main site conditions meet the consistency requirements within the plot; initial forest stand plots are derived from project zoning boundaries or forest stand boundaries already registered in the soil multi-source database; site type is obtained by encoding the category field and the preset site classification field. The site classification field is set according to technical specifications, project procedures, and site constraints, and is written into the parameter version.

[0049] Data lacking spatial, temporal, or source identifiers enters the raw data area. After the corresponding identifiers are added, it re-enters the multi-source access unit. Data that cannot be supplemented is only used for data quality auditing. Valid markers are marked as invalid if monitoring data fails quality control, exceeds the instrument's range, or the sampling location does not belong to a forest stand. Valid markers are also marked as invalid if the occurrence time, affected area, or record source of an environmental disturbance event cannot be confirmed. Subsequent processing only uses data with valid markers and environmental disturbance events.

[0050] Routine data quality control can be completed according to the calibration, blank sample, and repeat sample requirements specified in the testing method. This is a well-known technique in the field and will not be elaborated here.

[0051] S12, the homogeneity partitioning unit reads existing sampling points and sensor points corresponding to existing sensor locations. Sensor points with plot identifiers, sampling locations, and observation values, and are validly marked, are included in the valid sampling points. If the number of valid sampling points is less than the preset lower limit, a supplementary sampling point task is generated. The supplementary sampling point task includes the initial forest stand plot, the indicators to be supplemented, the point range, soil layer, and collection period. The newly generated sampling points, observation values, and category fields are written into the soil multi-source database and homogeneity calculation is performed again. The lower limit of sampling points is set based on the sampling design and natural variations within the plot, for example, 3 to 8. The homogeneity partitioning unit uses a homogeneity formula to calculate the degree of spatial heterogeneity: ; In the formula: Forest stand plots Up indicators The degree of spatial heterogeneity; The number of the forest stand plot; The indicator number; Calculate the maximum value; The sampling point number; Forest stand plots Up indicators The set of sampling points consisting of valid sampling points; Sampling points Up indicators Observed values; The median represents the median state of the observations in the set of sampling points; As an indicator The difference function; As an indicator The pre-defined boundaries within the plot allow for differences, and 0.

[0052] For continuous indicators, the difference function is the absolute value of the difference between the observed value and the median state; for categorical indicators, the absolute difference is calculated after logarithmic transformation or standardization according to the parameter version; for categorical indicators, the difference is 0 when they are the same, and they are judged not to meet the homogeneity condition when they are different.

[0053] If the set of valid sampling points is empty, the number of valid sampling points is less than the lower limit of sampling points, or the allowable difference boundary within the plot is not configured, the degree of spatial heterogeneity will not be calculated, and the task will be transferred to supplementary sampling points or supplementary investigation. The allowable difference boundary within the plot is set according to the sampling design and natural variations within the plot based on indicators, such as soil pH value set to 0.2–0.5 and soil volumetric water content set to 0.03–0.08 m³. 3 / m 3 .

[0054] If all continuous indicators have a spatial heterogeneity of no more than 1 and the category fields are consistent, the current forest stand is determined to be a homogeneous forest stand. If there are indicators with a spatial heterogeneity greater than 1 or inconsistent category fields, the sampling points that do not meet the conditions are used as seeds, and the stands are expanded into non-overlapping continuous sub-regions according to the relationship between the grids where the sampling points are located or the spatial boundaries. The spatial heterogeneity is then recalculated.

[0055] Continuous sub-regions that still do not meet the conditions but whose number of valid sampling points is not less than the lower limit of sampling points will be further subdivided; if non-overlapping continuous sub-regions cannot be formed or the number of valid sampling points after subdivision is insufficient, they will be transferred to supplementary investigation. After the supplementary investigation data is written into the soil multi-source database and recalculated, areas that still do not meet the homogeneity condition will be transferred to manual processing and will not participate in subsequent evidence construction.

[0056] S13, the segment preparation unit arranges management events in homogeneous forest stands according to their execution time. Each management event includes at least the type of measure, execution time, actual application amount or intensity of work, area of ​​effect, and source of record. All management events that actually occur on site are recorded in the soil multi-source database.

[0057] The segment preparation unit converts the measure type into a measure code based on the preset measure code lookup; non-routine operations without a corresponding measure code retain the measure type, and their response data enters the non-indexed evidence area. The measure code lookup is set according to the project procedures and measure type catalog, and is written into the parameter version.

[0058] The segment preparation unit uses adjacent management events as mandatory time boundaries, classifying valid data from the same homogeneous forest stand and soil layer into root zone state segments. Data spanning management events are not classified into the same root zone state segment. The starting valid data time is taken as the acquisition time of the first valid data from the corresponding forest stand and soil layer.

[0059] For continuous monitoring indicators, the units of measurement should be standardized first, and then the scale conversion should be completed according to the logarithmic transformation or standardization transformation specified in the parameter version.

[0060] The effective duration of each observation is calculated from its acquisition time to the acquisition time of the next effective observation or the end time of the root region state segment, and is truncated at the segment boundary; the effective duration shall not exceed the acquisition period recorded in the preset data source requirements, and the time period exceeding the acquisition period shall not be included in the effective coverage duration; the time period before the first effective data shall not be included in the effective coverage duration.

[0061] The sum of the products of each observation and its corresponding effective duration, divided by the sum of the effective durations, yields the segment summary value. The effective coverage duration is the sum of all effective durations, and the effective coverage rate is the ratio of the effective coverage duration to the segment duration. If the sum of effective durations is 0 or the effective coverage rate is less than a preset lower limit, this indicator is recorded as missing in the current root region state segment. The lower limit of coverage is set based on the sampling design and data continuity requirements, for example, 70%–90%.

[0062] Periodic testing data is only used when the preset testing validity period covers the current root zone state segment. Sparse indices are selected from qualified test values ​​belonging to the current root zone state segment within the testing validity period, using a time-weighted average or the last qualified value as the segment summary value, according to preset index fixing rules. The testing validity period and index fixing rules are set based on the testing methods, technical specifications, and project procedures. For example, the testing validity period for pH value and effective nutrients is calculated from the sampling time of the corresponding data and is set to 15–90 days; the testing validity period for microbial indices is calculated from the sampling time of the corresponding data and is set to 7–30 days.

[0063] Where d represents the time unit day, 1d = 24h.

[0064] For any maintenance event, the root region state segment before the occurrence of the maintenance event and with the previous maintenance event as the lower bound is taken as the pre-maintenance baseline segment; if there is no previous maintenance event, the data start time is taken as the lower bound.

[0065] The response channels are determined by pre-defined response indicator categories, data source requirements, and response time scales, including hydrothermal channels, pH channels, organic matter channels, available nutrient channels, microbial channels, root vitality channels, stand growth channels, and stand health protection channels. Data whose detection methods do not meet the data source requirements are not included in the segment summary value of the corresponding response channel.

[0066] The post-management response window is determined by the response channel, the preset response window length, the execution time of the management event, and the boundaries of subsequent new management events. The response window length is set based on effective experiments, technical specifications, or project procedures. For example, the response window length for the hydrothermal channel is calculated from the execution time of the management event and is set to 1–7 days; the response window length for the effective nutrient channel is set to 15–90 days; and the response window length for the stand growth channel is set to 90–365 days. The summation value of the pre-management baseline segments constitutes the pre-management baseline state; within the post-management response window, the summation value of segments starting from the management event and not crossing subsequent new management events constitutes the post-management response state.

[0067] When a new maintenance event appears in the post-maintenance response window, the post-maintenance response window is truncated at the new maintenance event. If two measures belong to the same combined plan, have a common evidence attribution starting point, have already established a combined measure code before implementation, and the channel sub-token has been effectively occupied according to the combined measure code, it is processed according to the combined measure code; otherwise, the affected channel sub-token is invalidated.

[0068] The segment preparation unit generates constraint items for effective indicators in the pre-management baseline segment based on the preset suitable intervals corresponding to tree species, forest age class, and site type. When the indicator value is less than the lower boundary of the suitable interval, the deviation direction is considered too low; when it is greater than the upper boundary of the suitable interval, the deviation direction is considered too high. The width of the suitable interval must be greater than 0. If the width of the suitable interval is missing or not greater than 0, the standardized distance is not calculated, and the process is transferred to manual review. The standardized distance is calculated by dividing the absolute value of the difference between the indicator value and the nearest suitable interval boundary by the width of the suitable interval. The deviation level is determined based on the standardized distance and the preset deviation grading threshold.

[0069] The persistence level is determined by comparing the number of times the deviation recurs or the duration of the deviation in consecutive valid root region state segments with a preset persistence threshold. The deviation threshold and persistence threshold are set according to technical specifications, project procedures, and local verification tests. For example, the standardized distance is divided into levels according to less than 0.25, 0.25 to 0.50, and greater than 0.50; deviations that occur only once and are not confirmed by subsequent consecutive valid root region state segments are classified into the lowest persistence level.

[0070] In this embodiment, the priority restriction set consists of restrictions whose deviation level and persistence level both reach the second level or above, and which do not belong to the lowest persistence level. The segment preparation unit generates a candidate set of protection indicators based on tree species, stand age class, site type, and soil and stand health indicators related to the restrictions, according to the correspondence recorded in the parameter version records. Suitable intervals and correspondences are derived from technical specifications, healthy reference stands, project parameters, or local validation experiments, and the parameter source and parameter version are recorded.

[0071] When key fields required for generating restrictions or retrieving formal measures are missing, a supplementary survey task is generated. Key fields include category fields, pre-management baseline status indicators related to priority restrictions, and minimum data source markers formed according to the data source requirements of the response channel. The supplementary survey task records the fields to be surveyed, forest stand plots, sampling location range, soil layers, testing methods, and completion deadlines; after the supplementary survey results are written to the soil multi-source database, the preparation of the corresponding root zone status segment is re-executed.

[0072] See appendix Figure 4In this embodiment, step S2 is executed by the evidence construction module, which includes an evidence initialization unit, a net response determination unit, and a dual index unit. The module is used to calculate and process the reference net response, complete the evidence admission, restriction removal and inverse response judgment, and determine the archiving category and evidence index key.

[0073] S21, the evidence initialization unit reads homogeneous forest stand plots, root zone state fragments, constraint set, and protection indicator candidate set, and checks historical data. If the historical data simultaneously includes processed forest stand plots, reference forest stand plots, pre-management baseline status, post-management response status, management event records, post-management response windows, environmental disturbance events, and data sources, it enters the evidence admission process; if fields are missing, sources cannot be confirmed, or valid data is marked as invalid, it enters the non-indexed evidence area. The protection indicator candidate set is used to screen protection indicators and response channels participating in the inverse response judgment.

[0074] Evidence construction is triggered when historical data contains the fields required for evidence access, processing and reference verification tests are completed, or the post-management response window for new management is closed and response collection is completed. The data to be processed is submitted to the channel archiving unit, which sequentially calls the net response judgment unit and the dual-index unit to complete the archiving and writing process.

[0075] When both the positive evidence index and the negative evidence index are empty, the evidence initialization unit generates candidate verification measures and establishes processing and reference verification tests according to the technical specifications.

[0076] Candidate validation measures include the measures to be validated, target constraints, preset parameter ranges, and target response channels. Treatment and reference stands are selected from homogeneous stands, ensuring consistency in tree species, stand age class, soil texture type, and comparability with pre-management baselines. Parameter ranges are the intersection of the technical specification's allowable range and the current project's feasible range. Before the treatment and reference validation trials form acceptance evidence, supplementary testing tasks, validation trial tasks, insufficient evidence markers, or termination of formal recommendations are output; no formal management recommendations are output.

[0077] S22, the net impact assessment unit establishes an evidence pair consisting of the treated stand and the reference stand. When the tree species, age class, and soil texture category are consistent on both sides, the baseline indicator set is compared. The baseline indicator set consists of the indicators involved in the constraint terms and the preset site matching indicators. For continuous indicators, the difference in baseline state before management on both sides is calculated using the difference function in the homogeneous formula, and then divided by the preset baseline difference boundary; indicators that have undergone logarithmic transformation or standardization transformation are calculated on the transformed comparison scale, and categorical indicators must be completely consistent. The maximum value of each ratio is the comparability of the baselines before management.

[0078] If the baseline comparability before management is no greater than 1, the calculation proceeds to the net response calculation for the treatment reference; if it is greater than 1, the calculation proceeds to the non-indexed evidence area. The baseline difference boundary must be greater than 0; if it is missing or not greater than 0, the calculation for the current evidence pair is stopped, and the process proceeds to manual review, where abnormal baseline difference boundary information is recorded. The baseline difference boundary is set according to indicators based on detection error, natural baseline fluctuations, and the matching requirements between treatment and reference. For example, soil pH is set to 0.2–0.5, and soil volumetric moisture content is set to 0.03–0.08 m³. 3 / m 3 .

[0079] The net impact assessment unit reads evidence comparing the baseline state before and post-management response state on both sides under the same response channel and consistent comparison scale, and calculates the net soil change and net stand response using the net impact formula: ; ; In the formula: For evidence Soil Indicators Net change in soil; Evidence against Shanglin Sub-indicators Net response of the forest stand; For the number of the evidence pair; This refers to the numbering of soil indicators; The numbering of forest stand indicators; Indicates the handling of forest stand plots; Indicates a reference forest stand plot; A window indicating the response status after maintenance; A window representing the baseline status before maintenance; , Soil indicators on forest stand plots were treated respectively. Values ​​in both windows; , Soil indicators on reference forest stands Values ​​in both windows; , These are forest stand indicators for processing forest stand plots. Values ​​in both windows; , These are the forest stand indicators on the reference forest stand plots. Values ​​taken in both windows.

[0080] After subtracting the concurrent changes of the reference forest stand from the changes in the forest stand itself, the treatment reference net response is obtained, which consists of the net soil change and the net forest stand response. Each response channel is calculated separately.

[0081] For historical evidence-based paths, target response channels are determined based on the measure codes and restrictions in historical management events, and a set of protection indicators and necessary protection channels are generated from the candidate set of protection indicators.

[0082] For the proposed measures, after the measure generation unit generates a combination of measures, it calls the net response determination unit to determine the target response channel, protection index set, and necessary protection channels based on the measure code and restrictions corresponding to the measure combination, and then passes it to the token management unit. Historical authentication paths do not create channel sub-tokens for execution control.

[0083] Access to the response channel must simultaneously meet the following requirements: complete baseline, complete response data, continuously valid reference, complete management event records, no asymmetric interference, complete response window, and valid evidence attribution. Furthermore, the comparability of the baseline before management must not exceed 1. Complete baseline and complete response data refer to the absence of missing information on both sides of the baseline; complete management event records refer to the ability to confirm the measure code, execution time, area of ​​effect, and key actual application amount or workload.

[0084] In the new management implementation path, the continued validity of the reference forest stand requires that the reference forest stand maintain its reference status within the post-management response window, and that the channel sub-token has not expired; the validity of the evidence attribution requires that there are no other master tokens or channel sub-tokens effectively occupying the same response channel, or that a unique attribution has been established according to the combined measures code. In the historical evidence establishment path, the continued validity of the reference forest stand is determined based on the management event records of the reference forest stand, and the validity of the evidence attribution is determined based on the management event records of the processed forest stand and whether there are any additional management events within the post-management response window.

[0085] A complete response window requires that the post-management response window end after reaching the corresponding response window length, and not be prematurely truncated due to new unassigned management events. For continuous monitoring indicators, the number of valid observations must be no less than the preset minimum number of valid observations, and the effective coverage rate must be no less than the preset lower limit of coverage. For periodic monitoring data and sparse indicators, the validity period of the monitoring must cover the post-management response window, and the number of valid observations must be no less than the minimum number of valid observations. If the preset evaluation time point is reached but the post-management response window has not yet ended, only a preliminary judgment is made, and it is not included in the formal archiving.

[0086] If the preset evaluation time point is reached but the post-maintenance response window has not yet closed, only a preliminary judgment is made, and it is not formally archived. The evaluation time point and the minimum number of effective observations are set based on valid tests, technical specifications, or project procedures. For example, the evaluation time point for hydrothermal channels is set within 1 to 7 days after the maintenance event is executed. The specific evaluation time point is written into the parameter version, and the minimum number of effective observations is set to 3 to 10.

[0087] Asymmetric interference is not established when there are no effective environmental interference events affecting the target response channel on either side, the environmental interference events on both sides conform to the preset symmetric interference rules, or the environmental interference events do not affect the target response channel; otherwise, asymmetric interference is established. The symmetric interference rules compare the interference type, occurrence time, area of ​​effect, duration, and interference intensity in sequence. If any necessary field is missing or the comparison result exceeds the allowable difference, it is determined that the symmetric condition is not met.

[0088] The symmetrical interference rules are set according to technical specifications, project procedures and local verification tests. For example, the overlap ratio of the duration of events on both sides is set to 80% to 100%, and the allowable difference in interference intensity is set to 10% to 20%.

[0089] The net impact determination unit adds the baseline state of the treated forest stand before management to the concurrent change of the reference forest stand to obtain the presumed state without measures. For the corresponding indicators of the limiting items, the interval distance from the response state after management and the presumed state without measures to the suitable interval is calculated respectively; when the value is within the suitable interval, the interval distance is 0; when it is below the lower limit, it is the lower limit minus the value; when it is above the upper limit, it is the value minus the upper limit.

[0090] The restriction is lifted when the sum of the interval distance of the response state after management and the preset minimum distinguishable change is not greater than the interval distance of the presumed state without measures; otherwise, the restriction is not lifted. The minimum distinguishable change is set based on the repeatability of the detection method and the results of quality control, for example, the soil pH value is set to 0.05–0.20.

[0091] The inverse response is valid when the distance between the post-management response state intervals of any protection indicator is not less than the sum of the distance between the presumed state intervals without measures and the corresponding minimum identifiable change; otherwise, the inverse response is invalid. When the target response channel is completed but the necessary protection channel is not, the evidence enters the pending protection verification area, the channel status is set to pending protection verification, and it is not formally archived. After the necessary protection channel is completed, the channel archiving unit reads the evidence to be verified, calls the net response judgment unit again to complete the inverse response judgment, and submits it to the dual-index unit to determine the archiving category.

[0092] S23, the dual-index unit forms the evidence index key based on the site comparability category, restriction items, measure code, and response cycle category. The site comparability category is obtained by encoding the tree species, forest age class, soil texture category, and the preset site classification field; The response cycle category is determined according to the cycle file to which the post-management response window belongs. The cycle file is set based on the response timescale and project procedures. For example, if the duration of the post-management response window is no more than 30 days, the response cycle category is classified as the first cycle file; if the duration is greater than 30 days but no more than 180 days, it is classified as the second cycle file; and if the duration is greater than 180 days, it is classified as the third cycle file. When one piece of evidence corresponds to multiple restrictions, evidence index keys are generated separately for each restriction. After atomic archiving is completed, the evidence pointers corresponding to each evidence index key point to the same evidence master record.

[0093] When the response channel admission is established, the restriction removal is established, and the reverse response is not established, the archive category is positive; when the response channel admission is established but the restriction removal is not established, or when the reverse response is established, the archive category is negative; when the response channel admission is not established, the archive category is non-indexed.

[0094] The dual-index unit outputs the archive category and evidence index key, but does not write them directly to the database. Evidence index keys are created separately for combined measure codes and individual measure codes; evidence generated from combined measures is not split into individual measure codes. The channel archive unit, according to the archive category, writes the evidence pointer of positive evidence to the positive evidence index and the evidence pointer of negative evidence to the negative evidence index; when the archive category is non-indexed, the evidence master record, channel response record, and reason for failure are written to the non-indexed evidence area.

[0095] See appendix Figure 5 In this embodiment, step S3 is executed by the measure decision module, which includes an evidence retrieval unit, a conflict supplementation unit, and a measure generation unit. It is used to retrieve transferable positive evidence and unresolved veto evidence, process conflicts between positive and negative evidence, and generate a combination of measures.

[0096] S31, the evidence retrieval unit assembles the current forest stand plot, the set of restrictions, the response cycle category, the query generation, the index version, and the parameter version into a query context; the response cycle category is determined based on the response channel corresponding to the current restriction and the response window length of that response channel. The current forest stand plot is either a homogeneous forest stand plot or a queryable forest stand plot that has entered this round of query. Based on the site comparability category, restrictions, and response cycle category, the evidence retrieval unit reads candidate historical evidence from the positive evidence index and the negative evidence index, and obtains the evidence master record and channel response record through the evidence pointer.

[0097] The current condition fields are derived from the current root zone state fragment, category field, and site comparable category related fields. The historical condition fields are derived from the pre-management baseline state and site comparable category related fields in the evidence master record. The difference between the condition field directory and the preset allowed migration is recorded in the parameter version. The condition field directory also records whether each condition field is a critical condition field.

[0098] For continuous indicators, the difference between the current condition field and the historical condition field is calculated using the difference function in the homogeneous formula, then divided by the allowable migration difference, and the maximum value among the ratios is taken as the migration mismatch. Categorical indicators must be completely consistent. When a key condition field is missing, a supplementary test task is generated; when a non-key condition field is missing, the corresponding candidate historical evidence is not included in the current query results.

[0099] Migration differences must be greater than 0; if a difference is missing or not greater than 0, the migration assessment of the corresponding candidate historical evidence is stopped. The allowable migration differences are set based on the applicable boundaries of historical evidence and local validation experiments; for example, soil pH is set at 0.3–0.8, and soil volumetric moisture content is set at 0.05–0.10 m³. 3 / m 3 When the comparable categories are consistent, the restrictions correspond, and the migration mismatch is no greater than 1, the candidate historical evidence is included in the current query results and is formed into transferable positive evidence or unremoved negative evidence according to the archive category.

[0100] When both transferable positive evidence and unresolved negative evidence correspond to the same current forest stand, the same restriction item, the same measure code, the same response cycle category, and the same query generation, a conflict of positive and negative evidence is formed, and they constitute a pair of positive and negative evidence.

[0101] The value range of continuous fields is centered on the evidence record value, extended outwards according to the detection uncertainty, and then intersected with the scope of application of the evidence; when the intersection is empty, the evidence does not participate in the discrimination of the corresponding field. Category fields form a category set according to the evidence record category or the category of application of the evidence.

[0102] S32, the conflict re-testing unit checks for positive and negative evidence pairs in a preset set of discrimination fields. When the value range or category set of transferable positive evidence and unresolved negative evidence on the same field has no intersection, that field is determined to be a separable field. If the current forest stand already has a valid discrimination field, and the field value belongs only to the value range or category set of one side of the evidence, the matching side is retained and the other side is removed; if the field value does not belong to either side, both sides of the evidence are removed from the current query results.

[0103] If no separable field exists for the positive or negative evidence, an irreparable conflict arises, and the case is transferred to a processing and reference verification test. If the processing and reference verification test fails to generate admission evidence, the new evidence still matches both positive and negative evidence, the test results still lack separable fields, or the reference becomes invalid or key data is missing, the case is transferred to manual review.

[0104] If a separable field exists, but the corresponding field is missing, has exceeded its validity period, is marked as invalid, or the current detection uncertainty interval intersects with the value intervals of both positive and negative evidence, then that field is added to the candidate field set for supplementary testing. The conflict supplementary testing unit uses a gap formula to determine the minimum discriminant gap: ; ; In the formula: To minimize the discriminative gap; This indicates that the minimum value has been obtained; This is the set of fields to be selected for supplementary testing. For the set of discriminant fields; For set The number of fields; Indicates constraints; This is transferable positive evidence; The evidence for rejection has not been eliminated; For the sake of evidence, whether it is positive or negative is correct; This is the set of positive and negative evidence pairs that have not yet been resolved and have separable fields in the current query generation; For or against evidence A set of separable fields; This is the set of candidate fields for supplementary testing that are currently missing, invalid, or insufficient for judgment. Represents subset relations; Indicates belonging to; Indicates intersection; Indicates any; It is an empty set.

[0105] The gap formula makes the set Each positive or negative evidence pair is covered by at least one supplementary test field, and the set with the fewest fields is selected. When multiple minimum discriminant gaps exist, the discriminant field sets are selected according to the fixed order in the parameter version; when no set satisfies the constraints, the process proceeds to the processing and reference verification test. The discriminant field set and fixed order are set based on the applicable boundaries of historical evidence, detection costs, and local verification tests.

[0106] The conflict re-testing unit generates re-testing tasks based on the minimum discriminant gap and records the frozen query generation, frozen index version, and frozen parameter version. If no new management events occur during the re-testing period and the frozen version is still readable, the re-testing results are written to the condition field of the current forest stand. The original positive and negative evidence are retrieved through a restored query, and a new judgment is made according to the frozen parameter version. When the conflict is resolved, the current query results are updated; if the conflict remains unresolved, the process proceeds to processing and reference verification experiments. If a new management event occurs, the data in the soil multi-source database associated with the current query context is substantially corrected, or the frozen version becomes unreadable, the original query is discarded and a query re-trigger is executed.

[0107] S33, the measure generation unit reads the current query results after conflict resolution. If both the positive evidence index and the negative evidence index are empty, no formal management recommendations are output. Specific measures are entered into candidate measure screening when there is transferable positive evidence, no outstanding negative evidence, a valid parameter range, and no preset disabling conditions. The parameter range is the intersection of the historical execution parameter range supported by positive evidence, the range allowed by technical specifications, and the executable range of the current project; if the intersection is empty, the corresponding measure is not included in the candidate measure set.

[0108] The measure generation unit reads the current availability status from the material ledger and equipment ledger, and calculates the availability flag according to the preset availability flag determination rules. If a measure belongs to the preset prohibited list, meets the prohibited conditions, its execution parameters are not within the parameter range, or the required materials or equipment are unavailable, the availability flag is 0; otherwise, it is 1. The availability flag determination rules are set based on project procedures, safety requirements, and project implementation constraints. Measures with an availability flag of 1 are included in the candidate measure set.

[0109] When the priority constraint set is empty, the measure selection formula is not executed, and a supplementary testing task is output or the formal recommendation is stopped based on the data status of the unconfirmed constraints; when the priority constraint set is not empty, the measure generation unit uses the measure selection formula to generate a combination of measures. ; ; In the formula: Minimize lexicographical order; For a set of candidate measures; and The candidate measure number; and For whether the corresponding measures should be selected, a binary decision variable; measures The cost of implementation; measures Priority restrictions Coverage coefficient; This is the number of the priority restriction item; For the set of priority restrictions; This is a set of conflicting measures that prohibit co-selection. measures Available tags; To achieve a sum; Indicates any; It represents the set of binary values.

[0110] A coverage coefficient of 1 indicates that the evidence index key of the corresponding measure covers the priority restriction item, while a coefficient of 0 indicates no coverage. Implementation costs are determined based on project quotas, labor requirements, and project implementation constraints; the implementation costs of different measures are converted to the same pricing unit. The set of conflicting measures is determined based on the target of the measures, the response channel, and on-site verification. The selection formula first reduces the number of measures, and then, when the number of measures is the same, reduces the total cost comprised of the sum of the implementation costs of all selected measures.

[0111] If no combination of measures satisfies the constraints, and the lack of transferable positive evidence or the inability of existing evidence to cover the priority restrictions results in the output of a verification test task, and the lack of executable measures due to the presence of prohibited conditions, available markers, or conflicting sets of measures results in the output of a halt to formal recommendations, the output of a halt to formal recommendations will be provided.

[0112] Measures with overlapping response channels but whose channel attribution conflicts can be eliminated by staggering response time intervals will be implemented in phases and will not be included in the set of measures prohibited from being selected together. When combined measures must be implemented concurrently and there is transferable positive evidence for the combined measure code and no unresolved veto evidence, a combined measure code will be generated. If the newly generated combined measure code lacks historical evidence, it will be transferred to the treatment and reference verification experiment and will not be used as a formal management recommendation. When there is an emergency response requirement, necessary measures will be implemented, and the corresponding response data will be entered into the non-indexed evidence area, or evidence will be formed according to the treatment and reference verification experiment. After the combined measure is approved for implementation, a management plan will be generated, and the current forest stand in the query context will be transferred to the treatment forest stand.

[0113] See appendix Figure 6 In this embodiment, step S4 is executed by the execution control module, which includes a reference locking unit, a token management unit, and an operation execution unit. It is used to dynamically allocate delayed reference forest stand plots, establish master tokens and channel sub-tokens, control evidence attribution, and collect actual operation records and response observations.

[0114] S41, referring to the combination of measures received by the locking unit, the processing of forest stand plots and the query context, and matching the delayed reference forest stand plots from the currently queryable forest stand plots.

[0115] The reference stand must be different from the treated stand, with consistent tree species, age class, and soil texture category. The pre-management baseline comparability must be no greater than 1. It must possess the pre-management baseline status required for the target response channel and necessary protection channels, have clear spatial boundaries, not be occupied by other valid channel sub-tokens, and have no registered conflict plans within the post-management response window. If multiple eligible stands exist, the stand with the lower pre-management baseline comparability should be selected; if the pre-management baseline comparability is the same, it should be determined according to the order of stand identification.

[0116] The reference locking unit associates the matching results with the combination of measures, the treated forest stand plots, the post-management response window, and the set of measures conflict relationships to generate a delayed execution plan lock, thus completing dynamic reference allocation. During the period when the delayed execution plan lock is in effect, no corresponding measures or conflict measures are issued to the delayed reference forest stand plots, but actual on-site management events are still written into the soil multi-source database.

[0117] When a delayed reference forest stand actually implements corresponding or conflict-related measures within the post-management response window, a reference failure record is generated based on the actual management event's area of ​​effect, implementation time, and affected response channels. The reference failure record includes the delayed reference forest stand, conflict-related measures, affected response channels, and response time intervals. Only sub-tokens for channels where both the area of ​​effect and the response time interval intersect will become invalid; unaffected response channels will retain their reference eligibility.

[0118] If there are no eligible delayed reference stands and no emergency response requirements, the issuance of formal management recommendations will be stopped; if a match cannot be made due to missing baseline status before management, a supplementary measurement task will be output, and a new match will be made after the supplementary measurement; if the data is complete but there are still no eligible delayed reference stands, the output of formal recommendations will be stopped.

[0119] When an emergency response is required, necessary measures may be implemented, and actual operation records and response observations may be included in the non-indexed evidence area. Emergency response requirements must originate from on-site warnings, maintenance procedures, or manual confirmation, and the source, confirmation time, and affected area must be recorded.

[0120] S42, the token management unit establishes a master token and channel sub-tokens based on the combination of measures, the forest stand plots to be processed, the reference forest stand plots to be postponed, the response channel, the query generation, and the index version.

[0121] Each management and protection execution corresponds to a master token, which records the measure code, the forest stand to be treated, the reference forest stand to be postponed, the query generation, the index version, and the execution status. Each response channel corresponds to a channel sub-token, which records the response channel, the response window after management and protection, the data source requirements, the protection index requirements, and the channel status.

[0122] The token management unit reads the set of protection indicators, target response channels, and necessary protection channels determined by the net response judgment unit based on the combination of measures, and writes them into the corresponding channel sub-tokens. Necessary response channels are determined based on the measure code, the preset correspondence between measure codes and response channels, and the protection indicator requirements, and are written into the main token. The correspondence between measure codes and response channels is set based on the target of the measure, technical specifications, project procedures, and local verification tests, and is bound to the parameter version.

[0123] The forest stand set consists of the processed forest stand set and the delayed reference forest stand set. The response time interval is taken as the post-management response window of the corresponding channel. The response channel set consists of the response channels occupied by the sub-tokens of each channel under the main token. When the channel sub-token takes effect, it forms a valid occupation according to the forest stand set, response time interval and its response channel; the main token records the union of the valid occupations of each channel sub-token.

[0124] The channel remains in valid use when its status is pending collection, collection in progress, pending protection verification, or archiveable. The corresponding valid use is released when the channel status is archived, expired, or the channel sub-token is revoked.

[0125] After the maintenance plan is issued, the status changes from pending data collection to data collection in progress; when the target response channel is completed and protection verification is not required, the status changes from data collection in progress to archiveable; when the necessary protection channel is not yet completed, the status changes from data collection in progress to pending protection verification; after the necessary protection channel is completed and the inverse response judgment is completed, the status changes to archiveable; when reference failure, key data missing, asymmetric interference, or invalid evidence attribution occurs, the corresponding channel status changes to invalid; after atomic archiving is completed, the status changes from archiveable to archived.

[0126] When the intersection of the forest stand sets of two channel sub-tokens is not empty, the intersection of their response time intervals is not empty, and the response channels are the same, a channel attribution conflict is formed; if a unique attribution has already been established according to the same combination of measures, no duplicate determination is made.

[0127] The token management unit staggers response time intervals by extending them, and arranges phased execution if conflicts still exist; if conflicts cannot be eliminated by extension or phased execution, it checks the evidence status of the combined measures code.

[0128] If there is transferable positive evidence for the combined measure code and no unresolved veto evidence, a new master token and channel sub-token are created according to the combined measure code; if there is no corresponding evidence for the combined measure code, the process proceeds to the processing and reference verification test. A conflict-free plan is generated after the channel ownership conflict is resolved; if it is not yet resolved, the corresponding maintenance plan is not issued.

[0129] When transitioning to processing and reference verification testing, the token management unit creates a new master token and channel sub-tokens, and sets the original channel sub-tokens to invalid. Response data associated with the original channel sub-tokens must not be entered into the positive evidence index or the negative evidence index according to the original measure encoding.

[0130] S43, the work execution unit issues maintenance plans according to the effective master token, channel sub-token, and conflict-free plan, and receives actual work records. The maintenance plan includes the measure code, area of ​​effect, execution time, parameter range, materials, tools, and record requirements; the actual work records include the actual execution parameters, execution time, area of ​​effect, material usage, tool usage, and record source.

[0131] All on-site maintenance events and actual operation records are written into the soil multi-source database. For the routine issuance of maintenance plans and the receipt of operation records, those skilled in the art can use existing project management terminals or database interfaces; the data transmission and storage are well-known technologies in the field and will not be elaborated upon here.

[0132] When the actual execution parameters can be confirmed but are not within the parameter range, the corresponding channel sub-token is set to invalid, and the current response data enters the non-indexed evidence area; when it is necessary to verify the actual execution parameters, a separate main token and channel sub-token are created for processing and reference verification tests.

[0133] When the actual execution parameters cannot be confirmed, the current response data enters the non-indexed evidence area, the original channel sub-token is set to expire and transferred to manual review; the manual review result does not change the non-indexed status of the current response data, and if further verification is required, a separate processing and reference verification test will be established.

[0134] If, within the post-management response window, it is detected that a delayed reference stand has been processed, key data is missing, asymmetric interference occurs, or a response channel is effectively occupied by other measures and a unique attribution is not established according to the combined measures coding, the operation execution unit will only set the sub-token of the affected channel to invalid and form a token invalidation. Unaffected response channels will continue to collect data as long as the baseline is complete, the response data is complete, the reference remains valid, and the evidence attribution is valid.

[0135] The response observation includes additional plot identifier, acquisition time, response channel, data source, master token identifier, channel sub-token identifier, quality flag, and data version. The job execution unit transmits the actual job record, response observation, and channel sub-token status to the channel archiving unit as input for atomic archiving.

[0136] See appendix Figure 7 In this embodiment, step S5 is executed by the archive update module, which includes a channel archive unit and a closed-loop update unit, used to complete atomic archiving, enable a new index version, release delayed reference forest stand plots and re-query.

[0137] S51, the channel archiving unit receives historical data, or receives actual operation records, response observations, and channel sub-token status for new maintenance execution paths. Historical evidence-based paths are not archived based on channel sub-token status; new maintenance execution paths are archived according to channel sub-token status.

[0138] When a channel sub-token has expired, the collected data, the reason for the token's expiration, and the data version are written to the non-indexed evidence area. If the channel sub-token has not expired and the post-maintenance response window has not yet closed, the channel sub-token remains in the process of data collection or awaits protection verification. If the post-maintenance response window closes but the actual work record or key response observation is missing, the channel sub-token is set to expired, and the available data and the reason for the expiration are written to the non-indexed evidence area. When the actual work record and response observation are complete and the channel sub-token is in an archiveable state, evidence access review is performed.

[0139] The evidence access review includes checking whether the baseline status before management has been corrected; recalculating the baseline comparability before management using the corrected baseline status before management, and confirming that the baseline comparability before management is not greater than 1; checking whether the delayed reference forest stand plots have maintained their reference status; and checking whether the conditions of no asymmetric interference and valid evidence attribution are met.

[0140] When the pre-management baseline status is corrected, the corrected data must include the data source, quality markers, and a new data version. The archiving of pathway units does not use the similarity of the post-management response status of the treated stand to that of the delayed reference stand as an archiving criterion.

[0141] If the baseline comparability before maintenance is greater than 1, the reference is invalid, there is asymmetric interference, or the evidence attribution is invalid, the archiving category is determined to be non-indexed, and the evidence admission conditions that failed are recorded. After the review is passed, the channel archiving unit calls the net response judgment unit to calculate and process the reference net response, completes the removal of restrictions and inverse response judgment, and then calls the dual index unit to determine the archiving category and evidence index key.

[0142] When the target response channel is completed but the necessary protection channel is not yet completed, the corresponding evidence enters the protection verification area and retains the evidence pair, the target response channel, the necessary protection channel, the master token identifier, and the channel sub-token identifier, without performing formal archiving.

[0143] After the necessary protection channel is completed, the channel archiving unit reads the evidence to be verified based on the above-mentioned related information, and re-performs the inverse response judgment and archiving category determination; when the token of the necessary protection channel expires or is revoked, the channel sub-token corresponding to the target response channel is set to expired, and the evidence to be verified and the reason for expiration enter the non-indexed evidence area.

[0144] Within the same database transaction, the channel archiving unit writes the evidence master record, channel response record, evidence pointer, master token and channel sub-token status, and the new index version. The evidence master record includes evidence pairs, maintenance events, measure codes, pre-maintenance baseline status, post-maintenance response status, processing reference net response, evidence admission results, restriction removal results, reverse response results, data version, and parameter version. The evidence master record for new maintenance execution paths also includes the master token identifier and channel sub-token identifier; the master token identifier and channel sub-token identifier are not written to historical evidence creation paths.

[0145] The master evidence record is saved only once. The channel response record includes the response channel, post-maintenance response window, response observation, and archiving category; the evidence pointer points to the master evidence record.

[0146] When the archive category is positive, the evidence pointer is written to the positive evidence index according to the evidence index key; when the archive category is negative, the evidence pointer is written to the negative evidence index; when the archive category is non-indexed, the evidence master record, channel response record, and reason for failure are written to the non-indexed evidence area.

[0147] Before executing a database transaction, the channel archiving unit verifies whether the same master evidence record already exists based on the evidence pair, response channel, and channel sub-token identifier. Historical evidence creation paths are verified based on the evidence pair, response channel, and data version; if the same master evidence record exists, it is not written repeatedly. When the evidence retrieval unit reads the master evidence record through the evidence pointer, it also reads the corresponding channel response record.

[0148] Atomic archiving ensures that the evidence master record, channel response record, evidence pointer, token state, and index version remain in the same committed state. If any write fails, the entire database transaction is rolled back, and no partial index results are retained. After a successful database transaction commit, the current channel sub-token changes from archiveable to archived, the master token continues to maintain the execution state corresponding to other unterminated channel sub-tokens, and the new index version is marked as committed and ready for activation.

[0149] For the commit and rollback of database transactions, those skilled in the art can use the transaction mechanism of relational databases to implement it. The specific operation is a well-known technology in this field and will not be described in detail here.

[0150] The new index version is generated sequentially by overlaying the current channel archiving results on top of the previous committed index version. When multiple response channels are archived sequentially, the later index version includes the previously committed channel archiving results. After a transaction fails, the channel sub-token remains archiveable, and the transaction is re-initiated if the evidence admission conditions are still met.

[0151] If the number of retries reaches the preset retry limit or cannot be completed after manual confirmation, an archiving failure is recorded, and the channel sub-token is set to expired. After the database becomes writable again, the corresponding data and the reason for the archiving failure are written to the non-indexed evidence area, and an index version containing the non-indexed archiving result and the termination status of the channel sub-token is generated. Then, the query is retried or manually reviewed. The maximum number of retries is set according to the database operating conditions and project management requirements, for example, 2 to 5 times.

[0152] Parameter sources and versions are saved along with the master evidence record. The appropriate range, detection validity period, evaluation time point, measure code comparison, and the correspondence between measure codes and response channels are written into the parameter version. Permissible difference boundaries within the plot, lower limits for sampling points and coverage are derived from sampling design and natural variations within the plot. Baseline difference boundaries are derived from detection errors, natural baseline fluctuations, and processing and reference matching requirements. Permissible migration differences, discriminant field sets, and site matching indicators are derived from historical evidence application boundaries and local validation experiments. Minimum discernible variation is derived from the repeatability and quality control of the detection method. Response window length, minimum number of valid observations, response indicator categories, response time scale, and data source requirements are derived from valid experiments, technical specifications, or project procedures.

[0153] Deviation threshold, continuous threshold, site classification field, symmetric interference rule, indicator fixed rule, prohibited list and prohibited conditions are derived from technical specifications, project procedures, safety requirements and site constraints; the set of conflicting relationships of measures, coverage coefficient, implementation cost and available marker judgment rules are derived from the target of the measures, response channel, site verification and project implementation constraints; the parameter range is the intersection of the historical execution parameter range supported by positive evidence, the range allowed by technical specifications and the current project executable range.

[0154] Data version records data changes, parameter version records the basis for calculations and judgments, and index version records the status of evidence indexes. When parameters are updated, a new parameter version is generated, which does not overwrite the old version parameters bound to archived historical data.

[0155] S52, the closed-loop update unit releases the effective occupation of the corresponding channel after all channel sub-tokens corresponding to the necessary response channel have entered the archived or expired state, and the corresponding evidence has been archived in a positive, negative, or non-indexed manner. If there are still other valid channel sub-tokens under the main token, the main token and the delayed execution plan lock are retained; after all channel sub-tokens have entered the archived, expired, or revoked state, the main token and the delayed execution plan lock are released, completing the reference release.

[0156] The closed-loop update unit selects and enables the latest committed index version that includes the archived results of all necessary response channels. If all necessary response channels are non-indexed archives, the latest committed index version that includes the corresponding non-indexed archived results and token termination status is enabled. Index versions that are not committed, have been rolled back, or do not contain the archived results of all necessary response channels are not included in the current query. After the index version is enabled, the original delayed reference forest stand plots are restored to queryable forest stand plots.

[0157] The closed-loop update unit regenerates the root zone state fragment of the queryable forest stand, re-identifies the set of constraints, and re-queries using the enabled positive evidence index and negative evidence index. The queryable forest stand can be reallocated in the new query generation as a processing forest stand or a deferred reference forest stand, but it must not conflict with response channels that still have valid occupancy.

[0158] After a re-query, the closed-loop update unit outputs a maintenance plan, supplementary testing tasks, and verification test tasks. Alternatively, if evidence is insufficient or conflicts between positive and negative evidence are not resolved, it outputs a halt to formal recommendations. When a maintenance plan is approved for execution, it becomes a formal maintenance recommendation. If a supplementary testing task is in a frozen query generation, frozen index version, or frozen parameter version, the restored query still reads evidence and interprets supplementary testing fields according to the frozen version. If a new maintenance event occurs, data in the soil multi-source database associated with the current query context is substantially corrected, or the frozen version becomes unreadable, the original query is abandoned and a query re-trigger is executed. The above outputs constitute the implementation results of the current query generation.

[0159] To further aid in understanding the technical solutions of the embodiments of the present invention, a specific application example is given below.

[0160] Masson pine stands aged 15–20 years were selected, and 36 initial stand plots of 20m × 20m were connected to a multi-source soil database. Soil pH, available nutrients, water content, root activity and stand growth data were collected, and management events and environmental disturbance events were recorded.

[0161] Homogeneous partitioning units were used to homogeneously divide each initial forest stand plot. Taking four soil pH values ​​(4.28, 4.34, 4.37, and 4.72) of an initial forest stand plot as an example, the median pH is 4.355, and the allowable difference boundary within the plot is 0.30. Based on the homogeneity formula, the following is obtained: ; If the spatial heterogeneity is greater than 1, the initial forest stand is split. After recalculation, the spatial heterogeneity of the continuous sub-region formed by the first three sampling points is 0.20, and it is determined to be a homogeneous forest stand. The segment preparation unit divides the root zone state segments according to the management events, and forms a set of restrictions, a set of priority restrictions, and a set of candidate protection indicators based on suitable intervals.

[0162] A homogeneous forest stand exhibits both low soil pH and low available nutrients. The retrieved transferable positive evidence and unresolved negative evidence form two positive-negative evidence pairs. Both pairs include soil pH as a separable field, and the current soil pH value has exceeded its validity period. Based on the gap formula, we obtain: ; The conflict retesting unit generates a retesting task based on this, and the soil pH value obtained from the retest is 4.34. After the query is restored, any unresolved veto evidence that does not conform to this value range is removed from the current query results.

[0163] The candidate measure set includes individual measures that cover both priority restrictions, and combined measures that cover both priority restrictions and have transferable positive evidence. The corresponding coverage coefficients are (1,0), (0,1), and (1,1), with implementation costs of 420 yuan, 360 yuan, and 740 yuan, respectively. The selection formula reduces the number of measures in lexicographical order and lowers the total cost when the number of measures is the same; therefore, the combined measure with an implementation cost of 740 yuan is selected.

[0164] The execution control module matches the processed forest stand plots with delayed reference forest stand plots and establishes a master token and channel sub-tokens. The soil pH values ​​in the pre-management baseline state and post-management response state of the processed forest stand plots are 4.34 and 4.76, respectively, while the corresponding values ​​for the delayed reference forest stand plots are 4.36 and 4.44, respectively. According to the net response formula, the net change in soil pH value is: ; The suitable pH range for soil is 5.00–6.00, with a presumed pH of 4.42 without intervention. The intervals between the response state after management and the presumed state without intervention are 0.24 and 0.58, respectively, with a minimum distinguishable change of 0.05. Since 0.24 + 0.05 ≤ 0.58, the restriction is lifted; none of the protection indicators meet the conditions for a reverse response, and the corresponding archiving category is determined to be positive.

[0165] To verify the effectiveness of the technology, 18 pairs of evidence were divided into a system group and a control group, with 9 groups in each group. The system group completed homogeneous partitioning, net response calculation, conflict resolution of positive and negative evidence, and atomic archiving according to the embodiments of this invention. The control group selected measures based on the initial average value of forest stand plots and changes before and after management. Both groups used the same detection methods, sampling cycles, and management conditions, and were continuously monitored for 365 days.

[0166] The system group achieved a restriction removal rate of 77.8%, a reverse response rate of 0%, and a formally archiveable evidence ratio of 88.9%. The control group achieved a restriction removal rate of 44.4%, a reverse response rate of 22.2%, and a formally archiveable evidence ratio of 55.6%. (Appendix) Figure 8 Plotting percentages on the vertical axis, the rates of restriction removal, adverse response occurrence, and the proportion of formally archived evidence are presented for both the systemic and control groups. The systemic group is represented by filled bars with diagonal lines, while the control group is represented by unfilled bars. (See attached...) Figure 8 It is evident that, under the conditions of this verification, the system group had a higher rate of restriction removal and a higher proportion of formally archived evidence than the control group, while the incidence of adverse responses was lower than that of the control group.

[0167] The average net response of soil pH in the system group and the control group was 0.31±0.08 and 0.18±0.12, respectively; the average net response of available nutrients was 0.42±0.11 and 0.21±0.15, respectively; and the average net response of root activity was 0.16±0.07 and 0.04±0.10, respectively.

[0168] Appendix Figure 9 Using the treatment reference net response as the vertical axis, box plots are presented for soil pH, available nutrients, and root activity. The system group is represented by gray-filled boxes and solid circles, while the control group is represented by unfilled boxes and hollow circles. The upper and lower boundaries of the boxes represent the interquartile ranges, the horizontal lines inside the boxes represent the median, the outer lines of the boxes represent the range of non-outliers, and the rhombus and its upper and lower endpoints represent the mean and standard deviation. (See attached...) Figure 9 It is evident that the overall net response of soil pH, available nutrients, and root activity in the system group was higher than that in the control group, and the observed values ​​were relatively concentrated.

[0169] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result using substantially the same method falls within the scope of the present invention.

Claims

1. A smart forest stand quality improvement system based on multi-source soil data, characterized in that, include: The data preparation module is used to collect and build a multi-source soil database and generate a set of homogeneous forest stand plots and restrictions. The measure decision module is used to query the positive evidence index and the negative evidence index according to the set of restrictions. When a conflict between positive and negative evidence is formed, the gap formula is called to determine the minimum discrimination gap and generate a supplementary test task. The frozen query generation and the frozen index version are recorded. After the supplementary test task is completed, the query is restored according to the frozen query generation and the frozen index version. After the conflict between positive and negative evidence is eliminated, a measure combination is generated, and the homogeneous forest stand is converted into a treated forest stand. The execution control module is used to allocate delayed reference forest stand plots to the treated forest stand plots according to the combination of measures, establish delayed execution plan locks, master tokens and channel sub-tokens, and issue management plans according to the master tokens and channel sub-tokens; The evidence construction module is used to calculate the net response of the treatment reference based on the treated forest stand and the delayed reference forest stand, and write the evidence pointer into the positive evidence index or the negative evidence index according to the net response of the treatment reference. The archive update module is used to perform atomic archiving of the evidence pointer according to the channel sub-token, generate and enable the index version, release the master token, the delayed execution plan lock and the delayed reference forest stand, and re-query.

2. The intelligent forest stand quality improvement system based on multi-source soil data according to claim 1, characterized in that, The data preparation module includes: A multi-source access unit is used to collect soil and forest stand monitoring data, management events and environmental disturbance events, and to construct the soil multi-source database. The homogeneous division unit is used to read valid sampling points and category fields, call the homogeneous formula to calculate the degree of spatial heterogeneity, and determine the homogeneous forest stand plot when the degree of spatial heterogeneity is not greater than 1 and the category fields are consistent; otherwise, it is split and recalculated. The segment preparation unit is used to form root zone state segments with management events as boundaries, determine the baseline state before management and the response window after management, generate the set of restrictions and the set of priority restrictions, and register the homogeneous forest stand plots as queryable forest stand plots.

3. The intelligent forest stand quality improvement system based on multi-source soil data according to claim 2, characterized in that, The evidence construction module includes: The evidence initialization unit is used to trigger evidence construction when the fields required for evidence access are available or when the post-protection response window ends and response collection is completed. When both the positive evidence index and the negative evidence index are empty, a processing and reference verification test is established. The net impact determination unit is used to establish an evidence pair consisting of the treated forest stand and the delayed reference forest stand, calculate the baseline comparability before management based on the evidence pair, and when the baseline comparability before management is not greater than 1, call the net impact formula to calculate the net response of the treated reference forest stand, and complete the removal of restrictions and inverse response determination based on the net response of the treated reference forest stand. A dual-index unit is used to form an evidence index key and determine the archive category. When the archive category is positive, the evidence pointer is written to the positive evidence index according to the evidence index key; otherwise, it is written to the denial evidence index or the non-indexed evidence area.

4. The intelligent forest stand quality improvement system based on multi-source soil data according to claim 3, characterized in that, The decision-making module for the measures includes an evidence retrieval unit; The evidence retrieval unit is used to take the homogeneous forest stand as the current forest stand, generate a query generation, read the root zone state fragment and determine the post-management response window and response cycle category, query the positive evidence index and the negative evidence index, and read the measure code in the candidate historical evidence to form a condition field and calculate the migration mismatch degree. When the migration mismatch is not greater than 1, transferable positive evidence or unresolved veto evidence is formed according to the archiving category; When the transferable positive evidence and the unremoved negative evidence simultaneously correspond to the same current forest stand, the same restriction item, the same measure code, the same response cycle category, and the same query generation, a positive-negative evidence conflict and a positive-negative evidence pair are formed.

5. The intelligent forest stand quality improvement system based on multi-source soil data according to claim 4, characterized in that, The measure decision-making module also includes a conflict supplementary testing unit; The conflict supplementation unit is used to determine the separable fields of the positive and negative evidence pair, and to include the candidate fields for supplementation when the separable fields are missing or invalid. It calls the gap formula to determine the minimum discrimination gap and generates the supplementation task, and records the frozen query generation and the frozen index version. After completing the supplementary testing task and obtaining the supplementary testing results, the supplementary testing results are written back to the condition field. The query is then restored according to the frozen query generation and the frozen index version, and the conflict between the positive and negative evidence is eliminated. When a new management event occurs, the data in the soil multi-source database associated with the current query context is substantially corrected, or the frozen query generation and the frozen index version become unreadable, the original query is discarded and the query is retried. If the separable field does not exist, proceed to processing and reference verification experiments.

6. The intelligent forest stand quality improvement system based on multi-source soil data according to claim 5, characterized in that, The measure decision-making module also includes a measure generation unit; The measure generation unit is used to form a parameter range by taking the intersection of the historical execution parameter range supported by the positive evidence, the range allowed by the technical specifications, and the executable range of the current project when there is the transferable positive evidence and there is no unresolved veto evidence. A set of candidate measures is formed. When the set of priority restrictions is not empty, the measure selection formula is called to generate the measure combination from the set of candidate measures and generate a management plan. The current forest stand is transferred to the forest stand to be treated. When there is no historical evidence for newly generated combined measures codes, the process is transferred to a processing and reference verification experiment.

7. The intelligent forest stand quality improvement system based on multi-source soil data according to claim 6, characterized in that, The execution control module includes a reference locking unit; The reference locking unit is used to determine the target response channel and the necessary protection channel according to the combination of measures, and to match the delayed reference forest stand as the reference forest stand for the treated forest stand from the queryable forest stand plots that have the baseline status before management; Generate the delayed execution plan lock, during which no corresponding or conflicting measures are issued to the delayed reference forest stand during the effective period, receive actual on-site management events and write them into the soil multi-source database; When the delayed reference forest stand actually implements corresponding measures or conflict measures within the post-management response window, a reference failure record is generated and the affected channel sub-token becomes invalid.

8. The intelligent forest stand quality improvement system based on multi-source soil data according to claim 7, characterized in that, The execution control module also includes a token management unit; The token management unit is used to establish the master token and the channel sub-token, write the target response channel, the necessary protection channel and the necessary response channel, and form a forest stand set by the processed forest stand and the delayed reference forest stand, with the post-management response window as the response time interval, and form an effective occupation according to the forest stand set, the response time interval and the response channel; When a channel ownership conflict occurs, after the channel ownership conflict is eliminated by coding through extension, phased or combined measures, a conflict-free plan is generated. If the channel ownership conflict cannot be eliminated, the process proceeds to processing and reference verification test.

9. The intelligent forest stand quality improvement system based on multi-source soil data according to claim 8, characterized in that, The execution control module also includes a job execution unit; The operation execution unit is used to issue the maintenance plan according to the main token, the channel sub-token and the conflict-free plan, receive actual operation records and read actual execution parameters, collect response observations and transmit them to the archive update module; When the actual execution parameters are not within the parameter range or when the reference failure record occurs, the affected channel sub-token will be set to invalid.

10. The intelligent forest stand quality improvement system based on multi-source soil data according to claim 9, characterized in that, The archive update module includes: A channel archiving unit is used to complete the atomic archiving and generate the index version based on the response observation when the channel status of the channel sub-token is archiveable; When the target response channel is completed but the necessary protection channel is not completed, the evidence to be verified is written into the verification area to be protected. After the necessary protection channel is completed, the verification area to be protected is read and then archived. The closed-loop update unit is used to release the effective occupation after all the channel sub-tokens corresponding to the necessary response channel have entered the archived or expired state, release the main token and the delayed execution plan lock, enable the submitted index version, restore the delayed reference forest stand to the queryable forest stand and query again.

Images