Soil organic carbon increment and erosion emission reduction collaborative evaluation method oriented to slope-to-terrace measure

Abstract

The invention discloses a soil organic carbon increment and erosion emission reduction collaborative evaluation method oriented to a slope-to-terrace measure, and relates to the technical field of water and soil conservation evaluation, and the method comprises the following steps: registering a terrace space distribution base map, a digital elevation model, a multi-temporal remote sensing image and rainfall data, and carrying out terrace unitization; forming a terrace unit account book, an observability mark and an abnormal unit; extracting a structural continuity component and a hydrodynamic overtopping risk component, and synthesizing a terrace structural integrity index and confidence; under double triggering of rainfall and change, dynamic support measure factors are obtained through confidence gating, hysteretic retention and segmented nonlinear mapping, and erosion emission reduction and transverse carbon retention are calculated; and reversely deducing a marginal repair income generation list, carrying out spot check, backfilling, updating a parameter version and archiving. The method can still update and output a recalculable evidence chain during cloud shielding and missing measurement, reduces the erosion reduction overestimation in the failure period, and supports post-flood repair order sending and performance verification.

Description

Technical Field

[0001] This invention relates to the field of soil and water conservation assessment technology, specifically to a method for synergistic assessment of soil organic carbon increment and erosion reduction for slope-to-terrace conversion measures. Background Technology

[0002] Terracing is an important project in soil and water conservation engineering for sloping farmland, often used in areas such as the Loess Plateau and the hilly areas of Southwest China. With the increasing demand for soil and water conservation performance evaluation and carbon sequestration accounting, it is usually necessary to examine the reduction of soil erosion and SOC retention / carbon emission reduction of terraces at the county / small watershed scale. Current technologies often use the method of first obtaining the distribution of terraces and then selecting support measure factors in models such as USLE / RUSLE to calculate the erosion differences under different scenarios.

[0003] Regarding terrace extraction, Chinese patent CN110415265B (application date: September 23, 2022, application number: 201910729929.0) discloses a method for extracting terraces based on high-precision DEM slope features from unmanned aerial vehicles (UAVs); regarding carbon sequestration accounting, Inner Mongolia Autonomous Region local standard dB15 / T4248-2025 (publication date: December 25, 2025, implementation date: January 25, 2026) specifies the accounting framework; Chinese patent CN116429 Patent 723B (application date: September 5, 2023, application number: 202310355004.0) discloses a method for calculating soil carbon sequestration based on parameters such as the erosion reduction benefits of measures and surface SOC; Chinese patent CN115438470B (application date: November 28, 2023, application number: 202210987716.X) proposes a carbon sequestration calculation model for soil and water conservation in production and construction projects, and proposes an approach to calculate carbon emission reduction based on the difference in erosion modulus before and after the implementation of measures.

[0004] The aforementioned technical approach, which relies on terraced field distribution, fixed parameters for erosion assessment, and further carbon sequestration calculation, faces the following challenges in county / small watershed flood season operation and maintenance scenarios: Terraced field projects suffer from degradation and damage due to tillage disturbance, rainfall erosion, and varying management practices. Features such as field ridges, drainage systems, and micro-topography deteriorate and break down. Heavy rainfall can also lead to overtopping and altered runoff paths, resulting in a time lag in the actual erosion reduction capacity. Current technologies rely on historical or one-off data on terrace distribution, using fixed support factor or erosion reduction modulus on annual or long-term timescales. SOC / carbon sequestration calculations then use erosion reduction as input. When the surface engineering condition deviates from static assumptions, the erosion reduction and the resulting extrapolated lateral soil conservation and carbon sequestration may not reflect reality. Furthermore, updating the terrace condition relies heavily on periodic interpretation or on-site verification. Given the vastness of the terrain, fragmented topography, and dispersed management units, timely and verifiable data is difficult to obtain, impacting the consistency of post-flood hazard mitigation, management decisions, multi-year performance accounting, and audit reviews. Summary of the Invention

[0005] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a method for the synergistic assessment of soil organic carbon increment and erosion reduction for slope-to-terrace conversion measures. By extracting structural continuity and hydrodynamic overtopping risk components, a terrace structure integrity index and confidence level are synthesized. Under dual triggering of rainfall and changes, confidence level gating and hysteresis maintenance are applied, and piecewise nonlinear mapping is used to obtain dynamic support measure factors and calculate erosion reduction and lateral carbon conservation. A list of marginal restoration benefits is generated by reverse calculation, and updated parameter versions are checked and archived. This method can still update and output a recalculated chain of evidence even under cloud cover and missing measurements, reducing overestimation of erosion reduction during the expiration period and solving the technical problems described in the background art.

[0006] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: A method for synergistic assessment of soil organic carbon increment and erosion reduction for slope-to-terrace conversion measures includes: registering the base map of terrace spatial distribution, digital elevation model and multi-temporal remote sensing images; unitizing the base map of terrace spatial distribution and generating observable markers; forming a terrace unit ledger and marking abnormal units. Structural continuity components and hydrodynamic overtopping risk components are extracted within the terraced unit, and the structural integrity index and confidence level of the terraced field are synthesized with positive weight constraints. Based on the dual triggering of rainfall events and integrity changes, dynamic support measures factors are obtained by confidence gating and combining hysteresis retention and piecewise nonlinear mapping with failure penalty zones. These factors are then substituted into the RUSLE / CSLE or InVEST sediment transport framework to obtain erosion emission reduction and estimate lateral carbon retention. Based on the piecewise nonlinear mapping, a priority list is generated by back-deriving marginal repair benefits. Sampling tasks are generated for low-confidence units, and the positive weight constraints, dual trigger conditions, and piecewise nonlinear mapping parameters are updated according to the sampling results.

[0007] Furthermore, the base map of the spatial distribution of terraced fields is registered with the digital elevation model and multi-temporal remote sensing images after unifying the coordinates and resolutions. The base map of the spatial distribution of terraced fields is then divided into terraced units according to the plots. Geometric boundaries, areas and adjacency relationships are written into the terraced unit ledger, and observability markers are also written.

[0008] Furthermore, for terraced units that cannot be registered, have missing images marked as true, or have abrupt changes in the base map of the terraced spatial distribution, abnormal unit markers are set, and the abnormal unit markers, abnormal causes, and corresponding times are written into the terraced unit ledger and simultaneously written into the queue to be verified, while keeping the geometric boundaries of the terraced units unchanged.

[0009] Furthermore, when multiple remote sensing images exist in the same period and at different times, the observability markers corresponding to each image are calculated separately. The image with the largest effective pixel ratio and the smallest cloud occlusion ratio and shadow occlusion ratio is selected as the remote sensing image used to update the observability markers in that period, and the observability markers are written into the terraced field unit ledger.

[0010] Furthermore, a strip buffer zone is set at the boundary of each terrace unit. The width of the strip buffer zone is determined by the resolution of the multi-temporal remote sensing image. Within the strip buffer zone, the breaks and gaps in the boundary line of the terrace are identified based on the multi-temporal remote sensing image. After merging the gap segments, the gap rate and continuity are statistically analyzed as structural continuity components, and the component interpretation of the structural continuity components is output.

[0011] Furthermore, based on the digital elevation model and its hydro-topographic derivation results, micro-topographic low points and runoff concentration are extracted. Potential overtopping low points and abnormal runoff paths are located within the terraced units, generating hydrodynamic overtopping risk components. The spatial location of potential overtopping low points and abnormal runoff paths are output as component interpretations.

[0012] Furthermore, the structural continuity component and the hydrodynamic overtopping risk component are combined into a terraced structure integrity index according to a preset positive weight constraint, and the confidence level is determined based on the observability marker and inter-period consistency. When the observability marker indicates that the cloud cover ratio exceeds the preset threshold, the structural continuity component of the previous period is maintained and only the hydrodynamic overtopping risk component is updated. When the image missing marker is true, it enters the sampling queue.

[0013] Furthermore, update candidates are triggered when rainfall events reach the local high quantile threshold, when structural continuity components show significant changes, and when hydrodynamic overtopping risk components show significant changes. After confidence gating, update candidates are updated using dynamic support measures factors. Terraced units that do not pass the gating remain in the previous period's state and enter the queue for sampling.

[0014] Furthermore, after determining that a terraced field unit has entered the degradation zone and includes the case of entering the failure zone, hysteresis maintenance is initiated. During the maintenance period, the dynamic support measure factor is restricted to not allowing rapid recovery, and the maintenance period status and triggering reasons are written into the terraced field unit ledger. When the terraced field structural integrity index returns to the target area and meets the exit conditions, the hysteresis maintenance is lifted.

[0015] Furthermore, the terrace structure integrity index is transformed into a dynamic support measure factor through piecewise nonlinear mapping, and when the terrace structure integrity index is lower than the failure threshold, it enters the failure penalty zone and is set with an upper limit saturation. Under the condition of keeping other factors unchanged, the dynamic scenario erosion and the counterfactual scenario erosion are calculated and the difference is obtained to obtain erosion reduction. At the same time, the horizontal carbon retention is estimated by combining the soil organic carbon content and the confidence label is output.

[0016] Furthermore, marginal restoration benefits are calculated for each terrace unit, including restoring the terrace structure integrity index to the target range without changing rainfall, soil and cover conditions, and improving the magnitude of improvement by using piecewise nonlinear mapping to reverse dynamic support measures factors. Then, the incremental erosion reduction and lateral carbon conservation are estimated and a priority list is generated.

[0017] Furthermore, for terraced fields with low confidence, high risk, and high marginal repair benefits, random inspection tasks are generated. The UAV elevation results and inspection records are used as calibration inputs to backfill the terraced field unit ledger. The parameter versions of positive weight constraints, dual trigger conditions, and piecewise nonlinear mapping are updated based on the backfill results, and archive records of parameter versions are established.

[0018] Furthermore, after sampling and backfilling, when the terrace structure integrity index returns to the target area for a consecutive preset period without triggering the double triggering condition, the hysteresis is released and the normal update frequency is restored. At the same time, a recalcible evidence chain containing the input version, trigger record, parameter version and result record is output. When the basic parameters of soil organic carbon are missing, only the relative ranking of horizontal carbon retention is output and written into the resampling plan.

[0019] (III) Beneficial Effects This invention provides a method for synergistic assessment of soil organic carbon increment and erosion reduction in response to slope terracing measures, which has the following beneficial effects: Terraced fields are modularized by registering the base map of the spatial distribution of terraced fields, digital elevation model, multi-temporal remote sensing images and rainfall data. A ledger of terraced field units with geometric boundaries, adjacency relationships and observable markers of the structure is generated. Abnormal units with registration failure, missing images or abrupt changes in the base map are enqueued to complete the unified and recalculated input under cloud cover and missing data conditions, which serves as the trigger gating input after trigger gating input.

[0020] Within terraced units, structural continuity and hydrodynamic overtopping risk components are extracted to synthesize a terraced structural integrity index, providing component interpretations and confidence levels. This allows maintenance degradation to be expressed through structural evidence, avoiding categorization based on single vegetation textures or single damage patterns, and aligning it with dynamic support measure factor updates. Simultaneously triggered by both rainfall events and integrity changes, and through confidence gating and hysteresis preservation, a piecewise nonlinear mapping with a failure penalty zone is used to update the dynamic support measure factors. This ensures a stable representation of the impact of event-driven degradation on model parameters, enabling conservative updates and state preservation even in the event of missing data.

[0021] Dynamic support measures factors are substituted into the erosion model to calculate erosion emission reduction. Horizontal carbon retention is calculated using the same ledger and confidence level for the same terrace unit, enabling collaborative accounting and traceability of erosion and carbon output across spatial units, triggering timelines, and confidence level labels. Marginal restoration benefits are inferred based on piecewise nonlinear mapping, and a priority list is constructed by combining risk level and confidence level. The results are then used to coordinate post-flood emergency repairs and annual maintenance assignments, prioritizing the maintenance of high-risk, high-return terrace units with limited maintenance resources.

[0022] For low-confidence or anomalous units, a sampling inspection task is generated. The sampling inspection backfill is used to update the positive weight constraints, dual trigger conditions and piecewise nonlinear mapping parameter versions, archive the input version, trigger record and parameter version, form a recalculated evidence chain, and complete the evaluation, verification and calibration closed loop. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the method for synergistic evaluation of soil organic carbon increment and erosion reduction for slope-to-terrace conversion measures according to the present invention. Figure 2 This is a schematic diagram illustrating the marginal revenue back-calculation and closed-loop optimization of the present invention; Figure 3 This is a flowchart illustrating the overall implementation of the present invention; Figure 4 This is a schematic diagram of the ledger data structure for the terraced field unit of the present invention; Figure 5 This is a block diagram of the observability and anomaly detection logic of the present invention; Figure 6 This is a schematic diagram illustrating the principle of structural integrity index extraction in this invention; Figure 7 This is a flowchart of the dynamic factor mapping and state preservation process of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please see Figures 1-7 This invention provides a method for synergistic assessment of soil organic carbon increment and erosion reduction for slope-to-terrace conversion measures. The method includes the following steps: Step 1: Organizing the base map of the spatial distribution of terraces, digital elevation model, multi-temporal remote sensing images and rainfall data into a recalcible ledger of terrace units, and simultaneously generating observable markers and a queue to be verified, thereby providing a unique data entry point for subsequent maintenance degradation index extraction and dynamic parameter updates.

[0026] Assessments of slope-to-terrace conversion in counties or small watersheds often utilize terrace base maps, digital elevation models, and pre- and post-flood imagery simultaneously. These data sources are scattered, with inconsistent coordinate benchmarks and resolutions, and the imagery is frequently affected by clouds and shadows. Direct overlay calculations can easily lead to problems such as misalignment of the same terrace unit in different datasets, unclear image usability, and inability to trace the cause of missing measurements. Therefore, it is necessary to pre-determine the reliability of spatial alignment and the usability of imagery in the terrace unit ledger.

[0027] First, align all data to the same coordinate reference system and the same cell grid, and convert the alignment quality into registration reliability coefficients. Then, divide the terraced base map into terraced units with unique numbers and adjacency relationships, and write the geometric attributes. Subsequently, generate observability scores using cloud / shade masks, missing measurement markers, and registration reliability coefficients. Finally, categorize terraced units with low observability, registration anomalies, and base map mutations into the verification queue and write the anomaly reason field.

[0028] Specifically, to ensure that the ledger can be reused for a long time, the terraced field unit ledger adopts a static field + period field organization method: the static field is only updated when the base map version changes or the unit division rules change, and the period field is written to multiple records under the same terraced field unit number with the timestamp as the index.

[0029] During project implementation, static fields can be stored in a geographic layer file, and period-specific fields can be written to the same file or an accompanying structured text file using an index table with the same name. Subsequent steps only need to read the terrace unit number and timestamp to locate the corresponding field. , This helps prevent unrecoverable errors caused by file renaming or path changes during multiple runs, thus avoiding the need for recalculation in abnormal states.

[0030] In areas where slopes are converted to terraces, the terrain is highly undulating. If the imagery and the digital elevation model are not aligned, the edge features within the boundary strip of the terraced fields will be misaligned and superimposed, affecting subsequent gap identification and the location of low points at the top of the terraces. To ensure the registration quality is traceable, a registration reliability coefficient is generated and written into the ledger simultaneously during the alignment process.

[0031] First, a baseline grid for the ledger is established, fixing the coordinate reference system, cell size, and cell origin. Then, the base map of the terraced field spatial distribution, remote sensing imagery, cloud / shadow quality masks, and digital elevation models are reprojected onto the baseline grid. Categorical data uses nearest-neighbor interpolation, while continuous elevation data uses bilinear interpolation to maintain the continuity of slope and runoff paths. If orthorectification discrepancies exist in the remote sensing imagery, translation and rotation parameters are solved through image matching and written back into the reprojection parameters to align image edges with terraced field boundaries within the same grid. To solidify the reliability of the alignment as a ledger field, a registration reliability coefficient is introduced. The average offset between the terrace boundary and the image edge within the boundary band area is mapped to an interval: Among them, the registration reliability coefficient : Value This is used to characterize the alignment reliability and serves as a multiplicative constraint for subsequent observability scores; registration error : Value , representing the average offset between the boundary position and the image edge position within the boundary strip region, in units of the reference grid cell length; attenuation coefficient : Value Used to control the intensity of punishment. The larger the value, the more sensitive it is to misalignment.

[0032] Optionally, the attenuation coefficient is selected such that the registration reliability coefficient drops to about half when the registration error reaches one pixel length, thereby directly converting a one-pixel misalignment into a clear penalty in the ledger.

[0033] When the registration reliability coefficient is in a low confidence state, the ledger generation is not directly stopped. Instead, the registration status is written into the ledger, so that subsequent steps will automatically switch to conservative processing when image edges are needed, and continue to advance when relying only on the digital elevation model, thereby avoiding the interruption of the post-flood process due to misalignment of individual images.

[0034] In practice, reprojection and resampling are first performed using the reference grid as the target, and nearest neighbor interpolation or bilinear interpolation is selected according to the data type; then, the registration error is calculated and the registration reliability coefficient is generated in the terrace boundary strip area. The registration status will be written into the terraced field unit ledger.

[0035] Calculating registration error First, an edge evidence layer is generated from the remote sensing image within the boundary strip area of ​​the terrace unit. This edge evidence layer can be a binary edge map obtained using the gradient operator or the Canny operator. Then, a set of boundary sampling points is obtained by sampling along the boundary of the terrace unit at a fixed arc length step. The number of boundary sampling points is denoted as […]. For each boundary sampling point, calculate its Euclidean distance to the nearest edge pixel in the edge evidence layer to obtain a distance sequence, and use the truncated mean or mean of the distance sequence as the registration error. .when If the number of samples is less than the preset minimum number, the terraced field unit is written into the abnormal status field and entered into the verification queue.

[0036] Furthermore, this ensures that the boundaries of terraced fields correspond to the imagery and elevation data within the same grid, reducing misjudgments caused by mismatches; it also explicitly records the registration reliability in the ledger, providing an executable basis for subsequent gating; and it makes registration issues traceable, facilitating on-site or manual verification of their origin.

[0037] Furthermore, in order to ensure that subsequent maintenance degradation indicators and management actions are applied to specific objects, the terrace base map needs to be divided into stable terrace units, and their numbering, geometric attributes, and adjacency relationships need to be fixed so that calculations at any time can refer back to the same terrace unit.

[0038] Prioritize using project construction zones or plot boundaries as dividing lines to ensure that terraced units align with actual management boundaries. If management boundaries are lacking, use regular grid division and the main confluence paths derived from the digital elevation model as constraints to split units crossing major confluence channels, preventing a single terraced unit from simultaneously covering two significantly different confluence backgrounds. After generating terraced units, assign a unit number to each unit and calculate its area, perimeter, centroid coordinates, and boundary strip geometry. This boundary strip serves as a fixed spatial window for subsequent gap extraction, registration error calculation, and overtopping risk location. Adjacency relationships are then determined based on the shared boundary length, and adjacent terraced unit numbers are written to the adjacency list field.

[0039] First, the terraced base map is divided according to the management boundary or rule grid, and cross-channel units are split using the main confluence path constraints; then, terraced unit numbers are generated, geometric attributes and adjacency relationships are calculated and written into the terraced unit ledger.

[0040] When in use, the terraced fields are transformed into location-locatable and dispatchable terraced field units to support subsequent maintenance and verification; the numbering and adjacency relationships are fixed to ensure that the same set of objects is maintained even when repeated calculations are performed across time phases; the confluence background is embedded into the unitization rules to make the subsequent explanation of overtopping risk more direct.

[0041] Furthermore, during the flood season, clouds and shadows make it difficult to reliably extract gaps and vegetation textures. Forcing calculations on low-quality imagery will produce unexplained fluctuations in results. To address this, cloud / shadow masks, missing markers, and registration reliability coefficients are coupled into an observability score, which is then written into a ledger as a subsequent gating input.

[0042] First, the cloud / shadow quality masks are converted into pixel-level cloud occlusion masks and shadow masks, ensuring consistency with the ledger reference grid. Cloud occlusion masks can directly use the remote sensing product quality bands, while shadow masks can be determined jointly by quality bands and terrain occlusion direction. Then, the cloud occlusion ratio and shadow ratio are statistically analyzed within each terrace unit, and it is determined whether there are missing image files at that moment to generate missing data markers. To compress multidimensional quality information into an executable scalar, an observability score is introduced. And by using the registration reliability coefficient as a multiplicative constraint, a consistent logic is formed to prevent misaligned images from being considered usable: Among them, the observability score : Value The registration reliability coefficient is used to characterize whether the terraced field unit can reliably extract image features at a certain moment and use them as subsequent gating input; : Range of values Missing test marker : Range of values , 1 indicates that the image is missing at that moment. This indicates that the image exists; Cloud cover ratio : Value This indicates the proportion of pixels obscured by clouds within a terraced field unit; the proportion of shadows. : Value , representing the proportion of shadow pixels within a terraced unit; cloud penalty coefficient : Value This is used to control the severity of the penalty imposed by cloud cover on the observability score; the shadow penalty coefficient. : Range of values The cloud penalty coefficient and the shadow penalty coefficient are used to control the intensity of the penalty imposed on the observability score. Both the cloud penalty coefficient and the shadow penalty coefficient reduce the exponential term to about half when the cloud occlusion ratio or the shadow ratio is half. The cloud and shadow penalties have interpretable quantities.

[0043] Cloud cover ratio With shadow ratio The statistics are performed using the effective pixels within the terrace unit as the denominator; effective pixels are defined as the set of pixels that are both located within the geometric boundaries of the terrace unit and are not without data (NoData) in either the image or the quality mask. Missing measurement markers. The value selection rule is as follows: when there is no image index at that moment, or when the number of valid pixels in all candidate images within the terrace unit is less than a preset threshold, it is recorded as... Otherwise, remember And write it to the selected image index field.

[0044] When multiple images exist at the same time, the ledger selects the image with the highest observability score as the representative image for that time and writes it into the image index; when only images with low observability scores exist, the observability score is still written and the low observability status is marked, ensuring that subsequent steps can automatically switch to a conservative strategy or sampling queue based on the ledger fields.

[0045] When in use, image availability is solidified into ledger fields, allowing subsequent steps to be directly gated without relying on experience-based judgments; by coupling spatial alignment quality into observability scores, misaligned images are prevented from misleading gap extraction; and the cause chain of low observability states is preserved, making it easier to explain the sources of post-flood updates.

[0046] Furthermore, construction, renovation, or natural damage can alter the boundaries of terraced fields, and missing imagery can create status gaps. If these gaps are not categorized and recorded, subsequent steps may misjudge base map differences as maintenance degradation or misinterpret missing data as a stable state. A verification queue is generated based on ledger fields, allowing abnormal objects to be explicitly marked before entering subsequent calculations.

[0047] Anomaly unit classification is mainly based on three trigger types: registration reliability coefficient is in a low confidence state, observability score is in a low observability state, or there is a missing marker. Significant boundary changes occur between the base map versions.

[0048] Base map changes are determined using the base map abrupt change index, which calculates the intersection and union differences of the same terrace unit across the set of terrace regions in adjacent base map versions. Among them, the base map mutation index : Value Used to characterize the spatial consistency differences between the old and new base maps; set of terraced areas in the new base map. : The spatial set of terraced areas within this terraced unit under the current base map version; the set of terraced areas under the old base map. : The spatial set of terraced areas within this terraced unit under the previous base map version; New base map terraced field area collection Combined with the old base map of the terraced fields A spatial set of terraced field pixels (or polygons) within the same terraced field unit; area operator. Implementation: If a raster representation is used, then It equals the number of pixels in the set multiplied by the area of ​​a single pixel; intersection and union are obtained by pixel logical AND / OR; if vector representation is used, then The area of ​​a polygon is given by the intersection and union of the polygons, which are obtained through polygon superposition operations; when (For example, if the unit has no terraced field pixels in either of the two base maps) Set as It also writes a marker for objects without terraces to avoid division by zero.

[0049] When the mutation index of the base map meets the mutation condition, the terrace unit is placed in the verification queue, and the verification priority is set in combination with the observability score: when the observability score is in the observable state, the verification action is mainly new construction, modification or collapse; if the observability score is in the low observability state, the verification action is mainly base map update error or gap.

[0050] The queue to be verified uses the terraced field unit number as the primary key and records the anomaly type, trigger time, and registration reliability coefficient. Observability score The base map mutation index and recommended verification actions can be completed by on-site inspection or UAV aerial survey, or by surveying equipment with equivalent functions.

[0051] When using it, registration anomalies, observation anomalies and base map mutations are recorded separately to reduce the risk of anomalies being misread as degradation; by binding the verification objects with the terrace unit numbers, the field results can be written back to the ledger and form a closed loop; even under conditions of missing image measurements or frequent base map updates, a complete ledger can still be output to ensure continuous input in subsequent steps.

[0052] If the target area is mainly affected by wind erosion or there is no daily rainfall data, the ledger can still be completed with unitization and observable marking. Subsequent event-triggered updates will be based solely on imagery and on-site verification. Rainfall data gaps can be marked in the ledger so that corresponding models can be used or data can be supplemented in subsequent steps.

[0053] Step 2: Forming within each terrace unit - - - The single-chain output enables maintenance degradation to be interpreted, recalculated, and subsequently gated.

[0054] Heavy rainfall can cause variations in the erosion reduction capacity of terraced fields due to gaps in the field ridges, changes in drainage outlets, and alterations in surface runoff paths. However, when imagery is affected by clouds and shadows, relying solely on texture analysis can easily lead to unexplained abrupt changes in results. Step one has already compressed image availability into an observability score and written it into the ledger. Therefore, Step two uses gating to introduce usable terraced units from the imagery into the extraction of structural continuity components, while transferring unusable terraced units from the imagery to the extraction of hydrodynamic overtopping risk components relying solely on the digital elevation model. This ensures consistent structural indicators even under conditions of missing data. To guarantee subsequent inter-period comparisons, Step two also requires all intermediate objects to be written back to the ledger's period-by-period field area with terraced unit number + timestamp. Subsequent steps must not skip the ledger and directly read the original imagery or elevation file.

[0055] Maintenance degradation is constrained by a combination of gap-related structural continuity and overtopping-related hydrodynamic risks. Image quality is explicitly injected into subsequent gating through confidence levels, thereby reducing the occurrence of numerical values ​​without source interpretation in post-flood assessments under cloud-covered scenarios. At the same time, the evidence object write-back ledger allows the state changes of the same terrace unit to be recalculated and verified in different personnel and different batches of operation.

[0056] Terrace ridges often appear as linear abrupt changes in brightness on imagery and as abrupt changes in slope broken lines on digital elevation models; however, seasonal farming alters the texture, so observability scores must be used to determine whether an image should be included in the computation chain. To avoid inconsistencies in boundary orientation leading to filtering failure, arc length resampling of terrace unit boundaries is also required, ensuring that the boundary orientation is determined by the tangent vectors of adjacent sampling points and written into the ledger parameter field.

[0057] Specifically, the boundary strip area, registration reliability coefficient, and observability score are first read from the terraced field unit ledger. And set observable thresholds. The observable threshold range is as follows: When the observability score falls below the observability threshold, the ledger record imagery becomes unavailable and the process moves to the next step, where it continues working based on the digital elevation model. When the threshold is not lower than the specified value, the image within the boundary strip area is sequentially denoised, contrast stretched, and edge extracted. Bilateral filtering is selected for denoising, and quantile truncation is selected for contrast stretching. The Canny operator is selected for edge extraction; other edge operators must output a single-pixel wide edge with a replayable threshold parameter. After edge extraction, morphological thinning is performed to remove burrs and ensure line continuity. Skeletonization is selected as the thinning operator; other thinning methods with the same function can also be used. Then, the edge line direction is selected. Only a set of pixels whose boundary distance between terrace units is not less than the boundary strip width and whose orientation is consistent is retained. The boundary strip width is selected as 1-3 reference pixels, preferably 2 reference pixels, to reduce the amount of field surface texture entering the evidence layer and avoid losing slightly recessed field ridge lines.

[0058] The extracted results are not directly used in the final index, but are written back to the terraced unit ledger periodically as evidence. These fields include at least the edge line cell set index, image index, denoising parameters, edge threshold, thinning method identifier, orientation threshold, and boundary band width. This enables subsequent verification to reproduce the same layer of evidence and determine whether the difference stems from image quality or parameter settings.

[0059] When in use, observability score gating prevents missing images from entering gap identification; boundary band width and direction filtering reduces crop row texture interference; evidence layer write-back ledger enables subsequent verification to replay the same processing chain and locate the source of differences.

[0060] The evidence layer is transformed into a verifiable gap description, and slope polygon evidence from the digital elevation model is introduced as a consistency check, ensuring that image fractures are not necessarily considered field ridge gaps. In terracing scenarios, true gaps are often adjacent to slope polygons or micro-topographical abrupt changes, while fractures caused by tillage textures usually do not correspond to slope polygon abrupt changes. Therefore, consistency checks can distinguish between the two types of cases. To ensure that inter-period gap comparisons are not affected by sampling density, the boundary arc length sampling step size is fixed to an integer multiple of the reference pixel, and the sampling step size is written into the ledger parameter field.

[0061] First, the edge line pixel set is projected into a one-dimensional sequence along the arc length direction of the terrace unit boundary. Connectivity and gap segment markings are then applied to this sequence to obtain a continuous segment set and a gap segment set. The gap segment length and number are written into the ledger's periodic field. The gap segment length is expressed in boundary arc length units to naturally align with the geometric boundary of the terrace unit, facilitating on-site positioning by arc length. Subsequently, slope change rate and curvature are derived from the digital elevation model to extract the slope polygon pixel set. Slope polygon extraction preferably uses a joint determination of slope threshold and curvature threshold; extreme value discrimination of terrain profiles with similar functions is also acceptable. For each gap segment, the existence of a slope polygon pixel set within its neighborhood is checked. If it exists, it is marked as a high-confidence gap; otherwise, it is marked as a low-confidence gap, and the confidence marker is written into the ledger.

[0062] If the registration reliability coefficient is low confidence, the terrace unit is marked as structural evidence verification. This provides an upper limit constraint in subsequent confidence calculations to prevent misalignment from amplifying gaps in the system. The structural continuity component consists of the continuous segment coverage ratio and the high-confidence gap penalty, ranging from 0 to 1. It forms a one-to-one interpretation chain with the gap length, number of gaps, and confidence markers. Subsequent personnel can directly consult the gap field to explain the cause of the structural continuity component.

[0063] When in use, the gap segment can directly point to the on-site verification location, making the gap a visible object instead of an abstract indicator; the false gaps caused by cultivation texture are reduced by slope break line consistency verification and enter the high-confidence set; the structural continuity component coexists with the intermediate field, which facilitates the subsequent explanation of the abrupt change by the confidence level.

[0064] Furthermore, local low points on the terrace ridges can become overflow outlets, and the concentrated upstream flow can amplify the scouring risk of these outlets. Therefore, a digital elevation model is used as the primary method to locate low points on the boundaries of terrace units and calculate their upstream flow accumulation. To reduce the impact of elevation noise on low point location, the boundary elevation sequence is locally smoothed after depression processing. The smoothing window length is bound to the reference cell length and written into the ledger parameter field.

[0065] First, depression processing is performed on the digital elevation model to obtain continuous flow paths. Depression filling is the preferred method, but trenching is also acceptable. Then, boundary elevation sequences are sampled at the boundaries of terrace units to locate local minima as candidate overtopping low points. The relative depression level is represented by the difference in elevation between neighboring boundaries. A fixed arc-length window is used for the neighborhood length to avoid incomparability in depression levels between terrace units of different areas. Next, the confluence direction and confluence accumulation are calculated within the terrace units. The D8 method is preferred for the confluence direction, but the D∞ method is also acceptable, as long as the upstream confluence accumulation reaching the candidate overtopping low point can be obtained. In engineering implementation, confluence calculations can be performed using common hydro-topographic tools, and the depression processing method, confluence direction algorithm identifier, and raster resolution are written into the ledger parameter field to ensure consistency during recalculation. For each candidate overtopping low point, the upstream confluence accumulation and relative depression level are jointly evaluated to obtain the hydrodynamic overtopping risk component. The coordinates of the candidate flood peak and low point, the relative degree of depression, and the upstream confluence cumulative index are written into the ledger period-by-period fields.

[0066] When the base map mutation index is in a mutation state or the registration reliability coefficient is in a low confidence state, the terrace unit should be written into the hydrodynamic evidence verification mark and the upper limit constraint should be reflected in the confidence level to avoid the candidate low point being mistakenly regarded as stable risk due to coarse elevation or mismatch.

[0067] First, the digital elevation model is processed to remove depressions, and candidate overtopping low points are located at the boundaries of terraced units to calculate the relative degree of depression. Then, the accumulation of runoff within the unit is calculated to generate the hydrodynamic overtopping risk component. At the same time, the candidate top and bottom points and their attributes are written into the terraced field unit ledger.

[0068] When in use, it can still output the hydrodynamic overtopping risk component and locate candidate low points even when the image is missing, so that the post-flood process is not interrupted; the candidate low point attributes are written into the ledger to facilitate direct verification along the boundary during inspection; the linkage mark between the base map mutation index and the registration reliability coefficient makes the risk output under boundary conditions constrained and interpretable.

[0069] Structural continuity components Combined with the hydrodynamic overtopping risk component, a single terrace structure integrity index is formed. Confidence levels are generated using observable scores and intertemporal differences. This allows step three to complete gating and conservative processing within the same ledger. To avoid difficulties in interpretation due to the opposing directions of the two components, the hydrodynamic overtopping risk component is incorporated into the synthesis in a formal way, so that the increase in the terrace structure integrity index uniformly represents a more complete structure.

[0070] The terraced field structural integrity index is calculated using the following formula: Where: Terraced field structural integrity index : Range of values Used to characterize the overall structural state of terraced units; structural continuity components : Range of values A larger value indicates fewer boundary gaps and higher coverage of continuous segments; hydrodynamic overtopping risk component : Value A larger value indicates that the candidate flood peak is lower and the upstream confluence is more concentrated; Structural weights : Value Used to adjust the contribution of the hydrodynamic weight. : Value This is used to adjust the contribution of the pair and satisfy the following: the weighted version is written into the parameter field of the terraced unit ledger, and the same version is directly called during recalculation. The confidence level is calculated using the following formula: Confidence : Range of values The reliability score is used to characterize the integrity index of terraced field structures; observability score. : Value Reflecting missing test markers Cloud cover ratio The combined effect of shading ratio and registration reliability coefficient; consistency penalty coefficient. : Value This is used to control the penalty strength of intertemporal differences on confidence levels; intertemporal differences : Value , defined as the absolute difference between the current period and the previous period, the terrace structure integrity index of the previous period is read from the previous timestamp field of the same terrace unit ledger; If the missing data is marked as 1, the observability score is 0, and the confidence level is 0. At this time, the terrace structure integrity index is still written, but the low confidence level is marked in the ledger. In step three, conservative update or transfer to the sampling queue is selected based on the confidence level.

[0071] The set of gap segments identified within the boundary band region is denoted as The total arc length of the gap segment is denoted as The total arc length of the boundary is denoted as For each gap segment, a confidence weight is assigned based on the consistency verification results of the slope polyline in the digital elevation model, thus obtaining the total arc length of the high-confidence gap segment. Total arc length of low-confidence gap segment and satisfy .

[0072] Structural continuity components It can be calculated and truncated as follows: : Among them, the high-confidence gap weighting coefficient : Range of values This is used to impose a stronger penalty on high-confidence gaps; the weighting coefficient for low-confidence gaps... : Range of values This is used to apply a weaker penalty to low-confidence gaps; Total arc length of the boundary : Range of values , where is the arc length of the terrace unit boundary; and is the total arc length of the high-confidence gap segment. Total arc length of low-confidence gap segment : Range of values To ensure consistency in physical meaning, the preferred option is to satisfy... , and when When .

[0073] When used, the terraced structure integrity index synthesizes gap information and overtopping risk on the same scale, which facilitates direct mapping of dynamic parameters in step three; after coupling observability score and inter-period differences, missing and abrupt states are marked and gated; the parameter version and period-by-period fields are stored at the same time, and the same input is recalculated to obtain the same output.

[0074] Step 3: Map the terrace structure integrity index into dynamic support measure factors in a gated, maintainable, and penalized manner, and output interpretable erosion reduction and lateral carbon conservation results when post-flood heavy rainfall and missing image data coexist.

[0075] Terraced terraces may experience gaps and overtopping after the flood season, but the annual assessment still uses fixed support factor calculations, resulting in the same terrace unit being considered valid in the model even though it has degraded on site. Step two has already output the structural continuity component. Hydrodynamic overtopping risk Terraced field structural integrity index and confidence level Step one has already output the observability score. Missing markers and basemap mutation index Therefore, step three directly reads the above ledger fields to complete the triggering, gating, maintenance and mapping, and writes the dynamic support measures factors and collaborative results back to the ledger, providing the only entry point for the marginal benefit back-calculation in step four.

[0076] Specifically, first, rainfall triggers and change triggers are used to generate update candidates, and then confidence levels are used to determine the update candidates. The system first uses missing markers and base map mutation index gating to divert water flow; then it uses a maintenance window to constrain the recovery path of degradation and failure states. Next, it uses piecewise nonlinear mapping to transform the terrace structure integrity index into a dynamic support measure factor, introducing a steeper penalty and saturation upper bound in the failure penalty zone. Then, it substitutes the dynamic support measure factor into the modified general soil loss equation (RUSLE) or the Chinese soil loss equation (CSLE) and sediment transport ratio framework to calculate the dynamic scenario erosion, and obtains erosion reduction by difference with the counterfactual scenario. Then, it calculates the lateral carbon conservation according to the soil conservation and carbon conservation caliber, and finally writes it back into the terrace unit ledger along with the triggering cause and gating state.

[0077] As a supplement: Rainfall triggering: Calculate rainfall events as an event erosivity index. and the event erosion threshold Compare; when The event was triggered by rainfall. (Event erosivity index) It can be calculated from hourly rainfall intensity sequences; when only daily rainfall is available, it can be constructed using daily rainfall and empirical coefficients. Alternative indicators are provided, and the substitution method and coefficient version used are recorded in the parameter field. Change trigger: Read the terraced field structure integrity index of the current period and the previous period. Calculate intertemporal differences ;when The time-change trigger is established, where the change trigger threshold is... It is embedded in the parameter field; Among them, rainfall-triggered scenarios cover post-flood scenes with increased risk but no imagery, while change-triggered scenarios cover scenes with less extreme rainfall but significant local degradation. Both types of triggers are generated in parallel to avoid missed triggers caused by relying on a single signal. To make the triggers engineering-feasible, the source of rainfall data, spatial matching, and gap handling are written into the ledger parameter fields, so that subsequent recalculations can reproduce the same rainfall-triggered conclusions.

[0078] The rainfall event sequence is read from the terrace unit ledger, and a rainfall trigger threshold is established based on local multi-year statistics. The rainfall trigger threshold is preferably taken from the historical high quantile, and even better from the 90th to 97th percentile of daily rainfall or event erosivity. The rainfall event sequence can be derived from automatic rain gauges, weather radar grids, or reanalysis rainfall products. When the source data is point data, the point rainfall is first assigned to the grid containing the centroid of the terrace unit using inverse distance weighting or Thiessen polygons, and then written into the rainfall event sequence field of the terrace unit ledger to avoid the same event being assigned to different units in different runs. When the threshold is met, the rainfall trigger establishment and trigger time are written, and the terrace unit is marked as an update candidate. Then, the structural continuity components of the previous and current periods for the same terrace unit are read. Hydrodynamic overtopping risk component and terrace structure integrity index The system determines changes based on inter-period differences. The threshold for triggering changes is preferably determined based on natural fluctuations during the non-flood season, but even better, it is determined based on the inter-period difference distribution within the same terrace unit, allowing the threshold to adapt to unit differences. When a change trigger is established, the "Change Trigger Establishment" and "Trigger Reason" fields are written, and the terrace unit is marked as an update candidate. If a gap exists in the rainfall data, the rainfall trigger is marked as a gap state instead of the default "Not Triggered," and the "Gap Reason" field is also written, enabling subsequent gating to identify gaps and switch to a conservative path.

[0079] When in use, rainfall triggers can still generate candidates even when images are missing, avoiding interruption of post-flood update entry; by using change triggers, evidence of local degradation is solidified into the ledger, reducing underreporting based solely on rainfall; and the sources of rainfall data and matching methods are written back into the ledger, making trigger conclusions recalculated and traceable to the data chain.

[0080] Furthermore, triggering only indicates a need for attention, therefore, based on confidence level... Missing data markers and base map mutation index gating are used to divert traffic, and a holding window is used to limit state recovery to an interpretable pace. To prevent gating anomalies caused by missing ledger fields, a field integrity check is performed before gating: if the terrace structure integrity index or confidence field is missing in the current period, the conservative branch is directly entered and the missing reason field is written.

[0081] Set a confidence level threshold; the confidence level threshold is preferably 0.5 to 0.7, more preferably 0.6. If the confidence level is not lower than the confidence level threshold, the missing data is marked as 0, and the base map mutation index has not triggered a mutation state, then proceed to the updatable branch. If the confidence level is lower than the confidence level threshold, or the missing data is marked as 1, or the base map mutation index has triggered a mutation state, then proceed to the conservative branch, maintaining the previous period's dynamic support measures factor or allowing only limited adjustments, and adding the terraced unit to the queue for sampling; the conservative branch also writes the low confidence level reason field, which at least includes the observability score. The status of missing test markers and base map mutation index is recorded and written to the value set version of the gated branch field to avoid ambiguity for the same branch name in different batches.

[0082] The preferred maintenance window is seven to thirty days, more preferably one field inspection cycle. The ledger maintains a status field and a maintenance count field for each terrace unit. When the terrace structural integrity index enters the degradation or failure zone, the corresponding status is written and the maintenance count is initiated. Before the maintenance count ends, even if the terrace structural integrity index recovers in the subsequent period, the status field is only allowed to remain in its original state or recover by a maximum of one level, thus incorporating the constraint that recovery must be accompanied by verification or continuous stability into the timeline. The preferred exit condition for the maintenance count is the backfilling of on-site verification records or continuous stable confidence levels over multiple periods, with the terrace structural integrity index exceeding the integrity threshold. When the above exit conditions are met, the exit reason field is written for subsequent audit review.

[0083] When in use, gating and routing avoid unexplained fluctuations caused by forced updates under low evidence quality; windows prevent short-term rebounds caused by seasonal phases and cloud cover; dynamic support measures and factor changes are closer to repair; field integrity checks and branch version write-back to the ledger reduce the caliber drift caused by multiple batch runs.

[0084] Furthermore, the terraced field structural integrity index is converted into a dynamic support measure factor, satisfying the requirements of slow changes in intact areas, steep penalties in failed areas, and the existence of an upper limit saturation. When the terrace ridges are continuous and drainage is unobstructed, the erosion reduction capacity is relatively stable. However, once overflows or gaps form, local confluences will be re-concentrated. Therefore, the mapping must allow for stronger penalties in failed areas while avoiding out-of-bounds values. To ensure practical implementation, the mapping calculation is required to read the parameter version field before assignment. If a parameter version is missing, the default version is used and a version backfilling flag is written, avoiding implicit parameter tuning.

[0085] The terrace structure integrity index and status field are read, and values ​​are assigned in three segments: intact zone, degraded zone, and failure penalty zone. The threshold for the intact zone is preferably 0.75 to 0.90, more preferably 0.80; the threshold for failure is preferably 0.40 to 0.60, more preferably 0.50. For intact zone units, the dynamic support measure factor is taken from the previous period or the effective terrace baseline value, with only small fluctuations allowed. For degraded zone units, the dynamic support measure factor is gradually increased with the terrace structure integrity index and is constrained by a maintenance window. For failure penalty zone units, the dynamic support measure factor is increased according to a steeper slope and a saturation upper bound is set, preferably the support measure factor value under conditions without terraces. To avoid false penalties based solely on the structural continuity component, the hydrodynamic overtopping risk component and confidence level are also considered in the failure penalty zone. When the hydrodynamic overtopping risk component increases and the status field is in a failed state, the penalty segment assignment is given priority; when the confidence level is low, the conservative branch is switched and the system waits for random checks and backfilling. When writing dynamic support measure factors back to the ledger, append writing is used instead of overwrite writing to ensure that the assignment history of the same terrace unit can be replayed to the window and parameter version.

[0086] When used, segmented mapping explicitly expresses the sharp drop in erosion reduction capability caused by failure at the parameter level, reducing the long-term overestimation of fixed factors; the saturation of the upper bound of the penalty segment makes the worst-case scenario fall back to no measures without generating out-of-bounds values, which is beneficial to model stability; the parameter version and append write-back mechanism make the same assignment process recalculated and traceable to the verification record.

[0087] Furthermore, the dynamic support measures factors are translated into verifiable outputs, ensuring that the results, triggers, gating, and traceability are maintained. Under the same terrain and cover management conditions, dynamic scenario erosion and counterfactual scenario erosion are calculated separately, and the difference is used as the erosion reduction. Subsequently, the basic parameters of soil organic carbon content are read to calculate the lateral carbon retention. Finally, the synergistic results and confidence labels are written back to the terraced unit ledger for use in step four.

[0088] As a supplement: horizontal carbon retention According to the erosion reduction modulus With soil organic carbon mass fraction Conversion: Among them, soil organic carbon mass fraction : Range of values The carbon retention coefficient can be determined from soil samples or assigned by soil type from a soil database; : Range of values This is used to express the retention ratio of organic carbon in erosion-reducing soils at the research scale; when enrichment / mineralization correction information is lacking, the linear caliber can be taken as... And write the linear caliber marker in the parameter field.

[0089] To avoid outputting only numerical values ​​without specifying the terms, the selected model route identifier and counterfactual scenario term identifier are written simultaneously during the write-back process, allowing different regions to use different terms under the same ledger structure without confusion.

[0090] The erosion model can use either the Modified Universal Soil Loss Equation (RUSLE) or the Chinese Soil Loss Equation (CSLE). If sediment transport along slopes needs to be considered, a sediment transport ratio framework can be overlaid, and the confluence path can be used to determine the outlet. The counterfactual scenario uses the baseline support factor, whose value can be either the support factor under conditions without terraces or the project initiation baseline value, but it must be written into the terrace unit ledger parameter field for recalculation. If there is a gap in the basic parameter of soil organic carbon content, the lateral carbon retention field should be written as a gap state, and the erosion reduction results should be retained, while keeping the confidence level and cause fields unchanged to prevent low evidence quality and parameter gaps from being masked. The output results file should preferably include the terrace unit number, timestamp, erosion reduction, lateral carbon retention, and confidence level. The trigger reason field, gated branch field, status field, and model route identifier are written into the terraced unit ledger with the same field name to avoid field mapping deviations during subsequent step four reading.

[0091] When used, the difference calculation caliber is the same as the previous assessment caliber, and the interpretation of erosion and emission reduction results can be recalculated; the caliber identifier and field name alignment ensures that cross-regional deployment does not confuse the model route and baseline setting; all fields are written back to the ledger throughout the process, low confidence results are displayed and trigger spot checks to avoid misuse.

[0092] Step 4: Based on the written-back fields, complete the marginal repair benefit back-calculation, order dispatch closure, spot check backfilling, parameter version archiving, and exit determination. Transform the above fields into an executable maintenance closure and a recalculated performance output.

[0093] Step 3 has written the erosion reduction, horizontal carbon conservation, triggering causes, gating branches, status fields and confidence levels of each terrace unit back into the terrace unit ledger. However, on-site management still faces limitations such as limited restoration resources, limited verification capabilities and tight time after the flood season. Therefore, it is necessary to back-infer the incremental value that can be obtained if restored to a good state from the ledger fields, convert low-confidence objects into sampling tasks, and then use the sampling backfill for weight and threshold calibration. Finally, the replayable evidence chain and exit conditions are output in the annual dimension.

[0094] The restoration process does not alter the rainfall event sequence, soil factors, topographic factors, or cover management factors, nor does it change the erosion model route selected in step three. Instead, it only performs a controlled replacement of the terrace structure integrity index, attributing the differences to a single source: structural condition recovery. This facilitates verification and review. To ensure the replacement values ​​are applicable, the lower limit of the integrity threshold is written as a ledger parameter field, preferably between 0.75 and 0.90, with 0.80 being even better. When the sample size is insufficient, the default value is used initially, and a new parameter version is generated after subsequent backfilling.

[0095] First, read the terraced field structure integrity index at the same timestamp. Confidence level Structural continuity components Hydrodynamic overtopping risk Base map mutation index Missing test marker The system checks the gating branch field and status field, and reads the erosion reduction and lateral carbon conservation result fields written back in step three. If the base map mutation index triggers a mutation state, or the missing data is marked as 1, resulting in a low confidence value, the marginal repair benefit is not directly inferred. Instead, a placeholder record to be verified is generated. The placeholder record is still included in the order list, but its suggested action field is fixed as random inspection to avoid mistaking the base map update error as repair benefit.

[0096] The set of candidate low points of the ridge top is located on the boundary of the terrace unit and denoted as . The relative depression level of each candidate low point is denoted as . The upstream confluence accumulation is recorded as and will and Linear normalization to Interval.

[0097] Hydrodynamic overtopping risk It can be calculated as follows: Among them, the number of candidate low points : Range of values Candidate low point weights : Range of values This is used to express the relative importance of different candidate low points; when prior knowledge is lacking, the same value can be used and written into the parameter field; relative depression degree : Range of values It is obtained by normalizing the difference between the local elevation of the boundary and the elevation of the candidate low point; upstream confluence accumulation : Range of values It is obtained by sampling and normalizing the candidate low point positions of the confluence accumulation grid; Exponential function :satisfy This can thus ensure Falling And when all candidate low points are not low-lying or have no confluence concentration, Approaching 0; when there is a significant low-lying area and concentrated flow, Increase; If the updateable branch is satisfied and the confidence level is not lower than the gating threshold, a repair scenario record is constructed: all fields of the current record are copied, the terrace structure integrity index is replaced with the lower limit of the integrity threshold, and the status field is set to the intact state; then, the piecewise nonlinear mapping and erosion calculation process with the same parameter version in step three is called to generate erosion reduction and lateral carbon conservation results under the repair scenario. The marginal repair benefit is obtained by subtracting the repair scenario result from the current scenario result, and is written back to the marginal repair benefit field of the terrace unit ledger, and simultaneously written to the repair scenario record index and parameter version index. If the gap in the basic parameter of soil organic carbon content causes the lateral carbon conservation field to be in a gap state, only the erosion reduction increment is reversed, and the lateral carbon conservation increment field is written to the gap state and the gap reason field is written. When the current terrace structure integrity index is not lower than the lower limit of the integrity threshold and the status field is in the intact state, the repair scenario record index is still retained for recalculation, but the marginal repair benefit field is written to 0 and the reason field is written to avoid intact units from entering the front line of emergency repair. This writing rule is used to unify the dispatching caliber. As a supplement: when selecting the model route as the modified general soil loss equation, the dynamic erosion modulus... Compared with baseline erosion modulus Press respectively: Among them, rainfall erosivity factor : Range of values Calculated from rainfall data and written into the parameter field; Soil erodibility factor : Range of values The value is assigned by soil type or soil survey data and written into the parameter field; slope length and slope factor. : Range of values Derived from the digital elevation model and written into periodic or static fields; covering management factors. : Range of values The value is assigned by the remote sensing vegetation index or land use type and written into the period-by-period field; dynamic support measure factor Baseline support factor : Range of values .

[0098] Erosion reduction modulus Defined as: Among them, the maximum function This indicates that the larger of the two values ​​is taken to ensure that the erosion emission reduction is non-negative; To prevent implementation from remaining at the conceptual level, it is required to reuse the same entry point of the calculation engine in step three: the input should only replace the terraced structure integrity index. All other inputs come from the same timestamp field in the terraced unit ledger; after the calculation is completed, the results of the repair scenario and the current results are written back to the ledger side by side, so that auditors can verify the source of the difference.

[0099] When in use, the difference in marginal repair benefits is only triggered by the replacement of the terrace structure integrity index, making the attribution clear; by using placeholder records to advance the sampling of mutations and missing objects, the misassignment of orders is reduced; the repair scenario record and the current record are written back to the ledger in parallel, so that the recalculation can be completed in the same index chain.

[0100] The restoration scenario is defined as: the structural integrity index of the terraced fields in this period. Replace with intact threshold And recalculate the dynamic support measures factor under the same parameter version. Dynamic erosion modulus Erosion reduction modulus Compared with horizontal carbon retention Marginal erosion and emission reduction increments Marginal carbon increment They are defined as follows: And , The repair scenario parameter version identifier is written back to the terraced field unit ledger; Furthermore, the marginal repair benefit field written back is converted into an executable dispatch list, and confidence level is used as a necessary constraint during sorting, thereby preventing objects with high benefits but weak evidence from crowding out emergency repair resources.

[0101] Risk level assessment incorporates both the terrace structure integrity index and the hydrodynamic overtopping risk component, ensuring that no site with a relatively acceptable structure but high overtopping risk is overlooked. A stable ranking strategy is employed: first, ranking by risk level; then by marginal repair benefit; and finally by confidence level. When all three are the same, ranking is based on terrace unit number, guaranteeing consistent output order for the same input across different batches. Risk level determination uses a three-stage division: high risk is defined as a terrace structure integrity index below the failure threshold and a hydrodynamic overtopping risk component above the median threshold; medium risk is defined as a terrace structure integrity index in the degradation zone and a hydrodynamic overtopping risk component not below the median threshold; all others are classified as low risk.

[0102] The failure threshold and median threshold are given by the parameter version field and updated during subsequent backfilling. To avoid gated conservative branch objects being incorrectly sorted, when the gated branch field is identified as a conservative branch or placeholder record, it is recommended that the action field be fixed as sampling and the marginal repair benefit field be set to unavailable, thereby creating a hard isolation between updatable objects and objects to be verified.

[0103] The order list fields are fixed in the ledger as follows: terrace unit number, timestamp, risk level, marginal repair benefit, and confidence level. The system includes fields for triggering cause, gating branch, status, gap segment index, candidate overtopping low point index, suggested action, location coordinates, and verification backfill entry. The suggested action field lists actions visible on-site: backfilling and compaction at the ridge gap, dredging of the drainage outlet, reinforcement of the candidate overtopping low point, and spot-checking and photographing backfilling. The location coordinates field consists of the centroid of the terrace unit and the coordinates of the evidence object, which can be directly selected by on-site personnel on the mobile map.

[0104] When in use, the three-dimensional sorting incorporates both risk and return, reducing the bias caused by dispatching orders based on a single indicator; by using confidence level as a necessary constraint, objects with weak evidence are automatically transferred to random inspection; the order dispatch field set contains evidence object indexes and backfilling entry points, forming a closed loop between on-site actions and ledger fields.

[0105] Furthermore, the uncertainty is concentrated in the sampling inspection task, and the sampling inspection feedback is transformed into parameter version updates of weights and thresholds to avoid caliber drift. The sampling inspection task is triggered by confidence level. The base map mutation index and dispatch status jointly restrict: when the confidence level is below the gating threshold and the risk level is high risk, or the base map mutation index triggers a mutation state, or a high-risk placeholder record appears in the dispatch list, a sampling inspection task is generated. The sampling inspection task fields must include a gap segment index and a candidate top-low point index, so that the on-site verification revolves around the specific object.

[0106] On-site backfilling employed both ground patrol routes and UAV aerial survey routes. On the ground patrol routes, patrol personnel, carrying mobile terminals, reached the locations of the gap sections and candidate low points of the overtopping, taking photos and recording whether there were overflow channels extending over the top, whether new gaps existed, and whether drainage outlets were blocked. On the UAV aerial survey routes, surveyors completed flight paths within the boundaries of the terrace unit and generated elevation data, which was then reviewed and annotated by office staff at the candidate low points of the overtopping and the gap sections. Both routes recorded the backfilling results in the terrace unit ledger verification status field using the terrace unit number as the primary key, along with the backfilling timestamp and backfilling method fields. Surveying equipment and recording terminals with similar functions can replace the aforementioned tools.

[0107] When updating the parameter version, a batch of terraced units with backfilled verification status are read, and an alignment table of verification labels and prediction fields is constructed: the verification labels come from the verification status field, and the prediction fields come from the structural continuity component. Hydrodynamic overtopping risk Terraced field structure integrity index and gating branch fields. The update objective is to adjust the structural weights and hydrodynamic weights. The integrity threshold, failure threshold, and risk segmentation threshold are adjusted to ensure consistency of the prediction fields in the verification labels. The solution employs a constrained nonlinear minimization method. The constraints include structural weights ranging from 0 to 1 (not zero), hydrodynamic weights ranging from 0 to 1 (not zero), and the sum of the structural and hydrodynamic weights being 1, while maintaining a monotonic relationship with the thresholds. A sequential quadratic programming solver or an interior-point solver can be used to complete the solution. The new parameter version identifier is output and written back to the ledger parameter version field, and simultaneously written to the updated backfill sample index set.

[0108] When in use, the sampling task consists of the gap segment and the candidate top and bottom points. The backfilling result is the same as the source. The parameter version update adds the backfilled sample to the new version, and the threshold and weight are reviewed. The constrained solution keeps the relationship between the weight and the threshold unchanged to prevent the caliber from drifting due to multiple backfills.

[0109] Furthermore, the closed-loop results are converted into annual performance metrics, and the conditions for exiting the maintenance window are written into executable rules, thereby restoring the normal update frequency after the repair is completed, while ensuring that any annual report can be reproduced based on the terraced unit ledger.

[0110] The exit decision for the hold window is based on the ledger fields: when the verification status field is marked as repaired, and the terrace structure integrity index is not lower than the lower limit of the intact threshold for several consecutive periods, and the confidence level is not lower than the gate threshold for several consecutive periods, and the concurrent rainfall trigger and change trigger are not established, then the status field is set to normal and the hold count field is cleared to zero; if the base map mutation index triggers a mutation status or the verification status field is empty, the hold window will not exit and will continue to generate sampling tasks to avoid false rebounds caused by releasing hold during the base map drift period.

[0111] The annual performance record is written back to the terraced unit ledger in the form of a recalculation evidence chain, and an archived copy is output. The recalculation evidence chain field set must include at least: input base map version identifier, rainfall data source identifier, and registration reliability coefficient. Observability score Missing test marker Base map mutation index Structural continuity components Hydrodynamic overtopping risk Terraced Field Structure Integrity Index Confidence level The system includes gating branch fields, status fields, dynamic support measure factor fields, erosion and emission reduction fields, horizontal carbon conservation fields, marginal repair benefit fields, dispatch list index, sampling task index, verification and backfilling index, and parameter version identifiers. Archived copies are sorted by timestamp and saved using a non-overwrite strategy, allowing subsequent recalculations to directly reference the same copy.

[0112] Evaluation metrics include: a recalculation consistency metric, verified by repeatedly running the program under the same input and parameter versions and comparing the order of tasks and key fields; a closed-loop completion rate metric, verified by calculating the percentage of verification backfill entry fields in the task list that have changed from empty to filled; and a sampling hit rate metric, verified by comparing the consistency between the backfill results of sampling tasks and the reason field of placeholder records. The control path involves replacing dynamic support measure factors with fixed support measure factors and running the program annually, then comparing the trends in the number of sampling tasks and the number of placeholder records to demonstrate the contribution of gating and maintenance chains to uncertainty convergence.

[0113] First, verify the status field and the terraced field structure integrity index. The confidence level and trigger flag are executed to exit the window and the status field and the count field are written back. Then, the evidence chain field set is sealed and recalculated by year and an archived copy is generated. At the same time, the calculation methods of three evaluation indicators, namely recalculation consistency, closed loop completion rate and sampling hit rate, are recorded.

[0114] When in use, the exit hold window determination is based on the ledger fields, making the status recovery path interpretable and recalculated; the annual report is bound to the full-process fields through the sealing of the recalculation evidence chain, which facilitates auditing and dispute resolution; the evaluation indicators and comparison paths provide executable verification methods and do not rely on unpublished data.

[0115] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0116] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0117] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0118] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0119] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for synergistic assessment of soil organic carbon increment and erosion reduction in response to slope-to-terrace conversion measures, characterized by: include, The base map of the spatial distribution of terraced fields, the digital elevation model and multi-temporal remote sensing images are registered. The base map of the spatial distribution of terraced fields is divided into terraced units and observable markers are generated. A ledger of terraced units is formed and abnormal units are marked. Structural continuity components and hydrodynamic overtopping risk components are extracted within the terraced unit, and the structural integrity index and confidence level of the terraced field are synthesized with positive weight constraints. Based on the dual triggering of rainfall events and integrity changes, dynamic support measures factors are obtained by confidence gating and combining hysteresis retention and piecewise nonlinear mapping with failure penalty zones. These factors are then substituted into the RUSLE / CSLE or InVEST sediment transport framework to obtain erosion emission reduction and estimate lateral carbon retention. Based on the piecewise nonlinear mapping, a priority list is generated by back-deriving marginal repair benefits. Sampling tasks are generated for low-confidence units, and the positive weight constraints, dual trigger conditions, and piecewise nonlinear mapping parameters are updated according to the sampling results.

2. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 1, characterized in that: After unifying the coordinates and resolutions of the base map of the spatial distribution of terraced fields with the digital elevation model and multi-temporal remote sensing images, the base map of the spatial distribution of terraced fields is divided into terraced units according to the plots. Geometric boundaries, areas and adjacency relationships are written into the terraced unit ledger, and observability markers are written.

3. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 2, characterized in that: For terraced units that cannot be registered, have missing images marked as true, or have abrupt changes in the base map of the spatial distribution of terraced fields, set abnormal unit marks, write the abnormal unit marks, the cause of the abnormality and the corresponding time into the terraced unit ledger and simultaneously write them into the queue to be verified, while keeping the geometric boundaries of the terraced units unchanged.

4. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 3, characterized in that: When multiple remote sensing images exist in the same period and at different times, the observability markers corresponding to each image are calculated separately. The image with the largest effective pixel ratio and the smallest cloud occlusion ratio and shadow occlusion ratio is selected as the remote sensing image used to update the observability markers in that period, and the observability markers are written into the terraced field unit ledger.

5. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 1, characterized in that: A strip buffer zone is set at the boundary of each terrace unit. The width of the strip buffer zone is determined by the resolution of the multi-temporal remote sensing image. Within the strip buffer zone, the breaks and gaps in the boundary line of the terrace are identified based on the multi-temporal remote sensing image. After merging the gap segments, the gap rate and continuity are statistically analyzed as structural continuity components, and the component interpretation of the structural continuity components is output.

6. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 5, characterized in that: Based on the digital elevation model and its hydro-topographic derivation results, micro-topographic low points and runoff concentration are extracted. Potential overtopping low points and abnormal runoff paths are located within the terraced units, generating hydrodynamic overtopping risk components. The spatial location of potential overtopping low points and abnormal runoff paths are output as component interpretations.

7. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 6, characterized in that: The structural continuity component and the hydrodynamic overtopping risk component are combined into a terraced structure integrity index according to a preset positive weight constraint, and the confidence level is determined based on the observability marker and inter-period consistency. When the observability marker indicates that the cloud cover ratio exceeds the preset threshold, the structural continuity component of the previous period is maintained and only the hydrodynamic overtopping risk component is updated. When the image missing marker is true, it is put into the sampling queue.

8. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 7, characterized in that: The update candidate is triggered when the rainfall event reaches the local high quantile threshold, when the structural continuity component changes significantly, and when the hydrodynamic overtopping risk component changes significantly. After the candidate is updated and gated with confidence, dynamic support measures factors are updated. Terraced units that do not pass the gate remain in the previous period and enter the queue for sampling.

9. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 8, characterized in that: After determining that a terraced field unit has entered the degradation zone and includes the case of entering the failure zone, hysteresis maintenance is initiated. During the maintenance period, the dynamic support measure factor is restricted to not allowing rapid recovery, and the status of the maintenance period and the triggering reason are written into the terraced field unit ledger. Hysteresis maintenance is lifted when the terraced field structural integrity index returns to the target area and meets the exit conditions.

10. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 9, characterized in that: The terrace structure integrity index is converted into a dynamic support measure factor through piecewise nonlinear mapping. When the terrace structure integrity index is lower than the failure threshold, it enters the failure penalty zone and is set with an upper limit saturation. Under the condition of keeping other factors unchanged, the dynamic scenario erosion and the counterfactual scenario erosion are calculated and the difference is obtained to obtain erosion reduction. At the same time, the horizontal carbon retention is estimated by combining the soil organic carbon content and the confidence label is output.

11. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 1, characterized in that: Marginal restoration benefits are calculated for each terrace unit, including restoring the terrace structure integrity index to the target range without changing rainfall, soil and cover conditions, and improving the magnitude of improvement by using piecewise nonlinear mapping to reverse dynamic support measures factors. Then, the incremental erosion reduction and lateral carbon conservation are estimated and a priority list is generated.

12. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 11, characterized in that: For terraced fields with low confidence, high risk and high marginal repair benefits, random inspection tasks are generated. The UAV elevation results and inspection records are used as calibration inputs to backfill the terraced field unit ledger. The parameter versions of positive weight constraints, dual trigger conditions and piecewise nonlinear mapping are updated according to the backfill results, and archive records of parameter versions are established.

13. The method for synergistic assessment of soil organic carbon increment and erosion reduction according to claim 12, characterized in that: After sampling and backfilling, when the terrace structure integrity index returns to the target area for a consecutive preset period and does not trigger the double trigger condition, the hysteresis is released and the normal update frequency is restored. At the same time, a recalculated evidence chain containing input version, trigger record, parameter version and result record is output. When the basic parameters of soil organic carbon are missing, only the relative ranking of horizontal carbon retention is output and written into the resampling plan.

Images