A vegetation restoration threshold value calculation method and system

Abstract

The application discloses a vegetation recovery threshold calculation method and system, and belongs to the technical field of ecological restoration. The method comprises the following steps: obtaining long-time sequence vegetation coverage data; determining whether the time continuity condition is met; if the time continuity condition is met, further determining whether the fluctuation stability condition is met; if the above conditions are met, determining that the pixel has reached the vegetation recovery threshold; calculating the average value of the vegetation coverage of the pixel that has reached the recovery threshold in the Tmax to Tend period; extracting the pixel that has reached the recovery threshold as a sample data set; extracting the multi-year average rainfall and other environmental factors corresponding to each sample; constructing a nonlinear relationship model of the environmental factors and the vegetation recovery threshold based on a random forest model; and applying the random forest model to predict the vegetation recovery threshold of all pixels in the study area, realizing spatial expansion, and thus providing a scientific and universal decision basis for macro ecological restoration engineering.

Description

Technical Field

[0001] This invention relates to the field of ecological restoration technology, and in particular to a method and system for calculating vegetation restoration threshold. Background Technology

[0002] With the continuous advancement of ecological civilization construction, my country has implemented numerous large-scale vegetation construction projects in areas such as land greening and ecological restoration. Scientifically assessing the vegetation restoration status and rationally determining the vegetation carrying capacity of different regions have become key scientific issues for achieving the systematic governance and sustainable development of "mountains, rivers, forests, fields, lakes, grasslands, and deserts." The vegetation restoration threshold, as a critical state indicator characterizing the transition of an ecosystem from the restoration stage to the stable stage, has significant practical value for guiding the spatial layout, species selection, and density control of vegetation construction.

[0003] Currently, methods for determining vegetation restoration thresholds mainly rely on ground observations at the micro- or slope scales, and local calculations based on theories such as water balance. However, in actual large-scale ecological construction, site conditions such as topography, climate, and soil exhibit significant spatial heterogeneity, making it difficult to directly apply these thresholds established at small scales to large areas, and thus failing to scientifically guide the layout of vegetation construction and resource allocation at the regional scale. Summary of the Invention

[0004] This invention provides a method and system for calculating vegetation restoration thresholds, which overcomes the limitations of traditional small-scale methods. By utilizing remote sensing data with the advantage of large-scale observation and fully considering the impact of environmental factors on vegetation restoration, a method and system are established that can effectively determine the vegetation restoration threshold and its spatial distribution over a large area, thereby providing a scientific and universal decision-making basis for macro-ecological restoration projects.

[0005] To achieve the aforementioned objective, the technical solution adopted by the present invention is as follows:

[0006] This invention provides a method for calculating vegetation restoration threshold, comprising:

[0007] S1: Obtain long-term vegetation cover data;

[0008] S2: Determine if the time duration condition is met: T end - T max ≥ 5 years; of which, This represents the year in which FVC reaches its maximum value throughout the entire time series. This represents the end year of the time series.

[0009] S3: If satisfied, further determine whether the fluctuation stability condition is met: (FVC) _local-max - FVC _local-min )≤ 0.05 × (FVC_global-max - FVC _global-min ); where FVC _local-max With FVC _local-min These represent the maximum and minimum values ​​in the local time series "after reaching the peak"; FVC _global-max With FVC _global-min These represent the maximum and minimum values ​​of the entire time series over the entire study period, respectively.

[0010] S4: If the time persistence condition and the fluctuation stability condition are met at the same time, it is determined that the pixel has reached the vegetation restoration threshold.

[0011] S5: For pixels that have reached the recovery threshold, calculate their value in T. max To T end The average vegetation cover over a given period is used as the vegetation restoration threshold for that pixel.

[0012] S6: Extract the pixels that have reached the recovery threshold as the sample dataset;

[0013] S7: Extract the environmental factors corresponding to each sample, such as multi-year average rainfall, temperature, elevation, slope, aspect, soil type, soil bulk density, sandy soil ratio, loam soil ratio, and clay soil ratio.

[0014] S8: Construct a nonlinear relationship model between the environmental factors and the vegetation restoration threshold based on the random forest model;

[0015] S9: The random forest model is applied to predict the vegetation restoration threshold of all pixels in the study area to achieve spatial expansion.

[0016] Furthermore, the calculation formula for S5 is as follows:

[0017]

[0018] in, For pixels The vegetation restoration threshold; For pixels In time vegetation coverage; The time when vegetation cover reaches its peak; This is the end time of the observation period.

[0019] Furthermore, the formula for constructing the sample dataset is:

[0020]

[0021] in, Indicates the first One pixel that has reached the recovery threshold. The total number of samples.

[0022] Furthermore, S7~S9 include:

[0023] For each sample cell Extract the corresponding environmental factors to form an environmental feature vector. :

[0024]

[0025] In the formula, This represents the average annual rainfall over many years. The average temperature over many years For elevation, For slope, Slope direction, Soil type For soil bulk density, The proportion of sandy soil, The proportion of soil is loam. The clay ratio;

[0026] Establish vegetation restoration threshold Nonlinear mapping relationship model between environmental factors:

[0027]

[0028] in, This represents a complex nonlinear function fitted by a random forest. This refers to model error;

[0029] Using the random forest regression method, For input features, Using the target variable, train the model, optimize the hyperparameters through cross-validation, and obtain the fitted model. .

[0030] The environmental factors corresponding to all pixels in the study area are input into the trained random forest model. Per-pixel simulated vegetation restoration threshold:

[0031]

[0032] in, The total number of pixels in the study area. For the first An environmental factor vector of a pixel This is an estimated vegetation restoration threshold for that pixel;

[0033] Generate a spatial distribution map of vegetation restoration thresholds in the study area, realize refined spatial expansion of the thresholds, and provide spatial decision support for regional ecological restoration projects.

[0034] This invention also provides a vegetation restoration threshold calculation system, comprising:

[0035] Acquisition module: used to acquire long-term vegetation cover data;

[0036] The first judgment module is used to determine whether the time duration condition is met: Tend - Tmax ≥ 5 years; where, This represents the year in which FVC reaches its maximum value throughout the entire time series. This represents the end year of the time series.

[0037] The second judgment module is used to further determine whether the fluctuation stability condition (FVC) is met if it is satisfied. _local-max - FVC _local-min ) ≤ 0.05 × (FVC _global-max - FVC _global-min ); where FVC _local-max With FVC _local-min These represent the maximum and minimum values ​​in the local time series "after reaching the peak"; FVC _global-max With FVC _global-min These represent the maximum and minimum values ​​of the entire time series over the entire study period, respectively.

[0038] Determination module: used to determine that the pixel has reached the vegetation restoration threshold if the time duration condition and the fluctuation stability condition are met simultaneously.

[0039] Calculation module: Used to calculate the value of pixels that have reached the recovery threshold in T. max To T end The average vegetation cover over a given period is used as the vegetation restoration threshold for that pixel.

[0040] First extraction module: used to extract pixels that have reached the recovery threshold as sample dataset;

[0041] The second extraction module is used to extract environmental factors corresponding to each sample, such as multi-year average rainfall, temperature, elevation, slope, aspect, soil type, soil bulk density, sandy soil ratio, loamy soil ratio, and clay soil ratio.

[0042] Module: Used to construct a nonlinear relationship model between the environmental factors and the vegetation restoration threshold based on the random forest model;

[0043] Prediction module: used to apply the random forest model to predict the vegetation restoration threshold of all pixels in the study area, thereby achieving spatial expansion.

[0044] Compared with the prior art, the technical solution disclosed in this invention has the following beneficial effects:

[0045] Compared with existing technologies, this invention utilizes long-term remote sensing observation data to obtain pixels where vegetation has reached the recovery threshold and simultaneously calculates the vegetation recovery threshold corresponding to each pixel. Subsequently, a random forest model is used to construct a nonlinear complex response relationship between environmental factors such as rainfall, temperature, topography, and soil and the determined vegetation recovery threshold. Based on this model, the threshold space is extended to areas not directly observed, ultimately obtaining the complete spatial distribution of vegetation recovery thresholds within the study area. This invention overcomes the limitations of traditional small-scale methods, utilizes remote sensing data with the advantage of large-scale observation, and fully considers the impact of environmental factors on vegetation recovery. It establishes a method and system that can effectively determine vegetation recovery thresholds and their spatial distribution over a large area, thereby providing a scientific and universal decision-making basis for macro-ecological restoration projects. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a schematic diagram of the vegetation restoration threshold calculation method provided in an embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram illustrating the principle of the vegetation restoration threshold calculation method provided in this embodiment of the invention;

[0049] Figure 3 This is a schematic diagram of the vegetation restoration threshold judgment rule provided in an embodiment of the present invention. Detailed Implementation

[0050] 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.

[0051] To make the objectives, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0052] With the continuous advancement of ecological civilization construction, my country has implemented numerous large-scale vegetation construction projects in areas such as land greening and ecological restoration. Scientifically assessing the vegetation restoration status and rationally determining the vegetation carrying capacity of different regions have become key scientific issues for achieving the systematic governance and sustainable development of "mountains, rivers, forests, fields, lakes, grasslands, and deserts." The vegetation restoration threshold, as a critical state indicator characterizing the transition of an ecosystem from the restoration stage to the stable stage, has significant practical value for guiding the spatial layout, species selection, and density control of vegetation construction.

[0053] Currently, most related studies focus on experimental plots or slope scales, comprehensively using fixed-point observations, hydrothermal models, and water balance formulas to estimate local vegetation restoration thresholds. For example, the research by Wang Yanping et al. focused on alfalfa artificial grasslands in Mizhi County, Shaanxi Province. Their core implementation plan was to construct a soil moisture-vegetation carrying capacity model based on water balance theory. Through long-term monitoring of local meteorological, soil moisture, and vegetation growth data, the model simulated soil moisture dynamics under different alfalfa biomass levels, ultimately determining the maximum biomass (2600–3500 kg / hm²) for maintaining long-term water balance as the vegetation restoration threshold for this region. Fu et al.'s research, using *Caragana sinica* and *Salix psammophila* in the Liudaogou watershed as examples, employed the SWCCV model to quantify the critical soil moisture conditions required for vegetation survival. Their implementation plan analyzed the relationship between soil moisture characteristic curves and vegetation transpiration water consumption, transforming the soil moisture threshold into a more intuitive leaf area index (LAI) threshold (1.27 and 0.7). This method focuses on defining the survival baseline of vegetation from the perspective of soil moisture physical limitations, providing clear physiological and ecological indicators for the restoration and management of shrub vegetation in semi-arid areas. Liu Bingxia et al.'s research also focused on the Liudaogou watershed in Shenmu, but selected two plant species, *Rhizophora stylosa* and *Sophora japonica*, and employed the more complex SHAW model. This model couples water and heat transfer processes, and by simulating water transport in the soil-plant-atmosphere continuum under different vegetation biomass scenarios, it identifies the maximum biomass (1980 kg / hm² and 5050 kg / hm², respectively) that can achieve system water balance.

[0054] In addition, the rapid development of remote sensing technology has provided new technical methods for determining the vegetation restoration threshold at the regional scale. Therefore, many researchers have also conducted research on methods for determining the regional scale vegetation restoration threshold. For example, the research by Zhao Guangju, Gao Haidong, and others adopted a spatial statistical method based on the "similarity habitat principle." This method first divides the study area into several relatively homogeneous habitat patches based on environmental factors such as topography, soil, and climate. Within each patch, a histogram distribution is constructed using long-term remote sensing vegetation indices (such as NDVI), and the upper boundary value of the distribution curve is used as the vegetation restoration threshold for that habitat type. Addressing the spatial heterogeneity of habitat division, Zhang et al. proposed an improved calculation method based on a sliding window. This method no longer pre-fixes habitat patches but uses a dynamic sliding window to traverse each pixel, selecting neighboring pixels with similar environmental conditions within a set spatial range centered on that pixel to form a temporary sample set. Then, a local threshold is determined through statistical methods. This scheme better captures the spatial gradation characteristics of environmental factors through dynamic adaptation, improving the spatial resolution of the threshold, but its core still relies on the empirical statistical extreme values ​​of vegetation parameters. The research by Feng et al. and later Liang et al. shifted towards a path combining eco-hydrological mechanisms with remote sensing. Based on the principle of ecosystem water balance, they used remotely sensed evapotranspiration (ET) and gross primary productivity (GPP) data to construct a quantitative response relationship between the two in different ecosystem types, and calculated the GPP threshold by setting water supply scenarios. Liang et al. further introduced site factors such as topography and soil to correct this relationship. This approach is more rigorous in mechanism, directly linking the threshold to the carbon-water coupling process of the ecosystem. However, as a core indicator of the carbon cycle, the threshold of GPP is difficult to directly convert into operational parameters (such as density and cover) to guide vegetation construction, and the model has high requirements for data quality and parameterization, which limits its operational application. Wang Kaili et al. used existing ET and vegetation parameter remote sensing products, considering the linear and nonlinear response relationship of ET to vegetation and the interaction between vegetation and meteorological elements, and used a stepwise regression method to construct the optimal response relationship between ET and meteorological elements and vegetation indicators pixel by pixel. Based on the principle of water balance, they calculated the forest and grassland vegetation cover restoration threshold under different climatic conditions.

[0055] The existing technology has the following main drawbacks:

[0056] (1) While methods that calculate local vegetation restoration thresholds by combining fixed-point observations, hydrothermal models, and water balance formulas at the plot or slope scale have theoretical value, they are highly dependent on detailed observation data from specific sites. They can only calculate the restoration thresholds of a certain type of typical shrub and grass species at the plot or small watershed scale, making it difficult to reflect the spatial heterogeneity under complex environmental gradients in large areas. In addition, the internal parameters of the mechanistic models used in related studies are complex and it is difficult to obtain true values, thus limiting the feasibility of such methods for widespread application. In actual engineering, the "site conditions" formed by different combinations of topography, climate, and soil vary greatly. Simply applying local thresholds can easily lead to unreasonable vegetation configuration, resulting in problems such as "over-restoration" causing soil drying or "under-restoration" failing to perform ecological functions, affecting the long-term effectiveness of the project and ecological security.

[0057] (2) At the regional scale, based on remote sensing observation data, the spatial statistical method of "similar habitat principle" can be used to estimate the vegetation restoration threshold. This method extends from the station scale to the regional scale and simplifies the complex environmental background with the help of spatial clustering, but it still has several limitations: on the one hand, the scientificity and rationality of habitat patch division are difficult to guarantee completely; on the other hand, the extracted threshold is essentially the maximum value of vegetation parameters during the observation period, which fails to fully incorporate the long-term dynamic change law of vegetation, and therefore cannot scientifically judge whether the statistical extreme values ​​in the patch truly represent the vegetation restoration threshold, which introduces uncertainty error into the threshold extraction process. Even if the relationship between evapotranspiration and vegetation response is included in the method, the drought resistance characteristics of the vegetation itself and the actual available water are difficult to quantify accurately, which will still bring significant uncertainty to the estimation of vegetation restoration threshold.

[0058] Therefore, developing a technical method applicable to regional scales and capable of objectively characterizing the spatial distribution of vegetation restoration thresholds has become a key issue urgently needing breakthroughs in the fields of ecology and remote sensing applications. Addressing the shortcomings of existing methods, this invention proposes an automatic method and system for determining vegetation restoration thresholds based on long-term remote sensing observation data. This method can accurately identify pixels within the study area that have reached the vegetation restoration threshold and simultaneously calculate the corresponding vegetation restoration threshold for each pixel. By fully utilizing the advantages of long-term remote sensing observations, it can clearly determine whether vegetation has entered a stable state after reaching its peak, thus determining the restoration threshold more scientifically and accurately. Subsequently, a nonlinear complex response relationship between environmental factors such as rainfall, temperature, topography, and soil and the determined vegetation restoration threshold is constructed using a random forest model. Based on this model, the threshold space is extended to areas not directly observed, ultimately obtaining the complete spatial distribution of vegetation restoration thresholds within the study area. Due to the spatial resolution advantage of remote sensing data, such as the ability of pixels at a 30-meter scale to finely characterize the spatial differentiation features of vegetation restoration thresholds, this method has higher application value in engineering practice and can provide direct and precise guidance for vegetation construction projects in specific areas such as slopes.

[0059] In short, this invention provides a method and system for calculating vegetation restoration thresholds, overcoming the limitations of traditional small-scale methods. Utilizing remote sensing data with its large-scale observation advantages and fully considering the impact of environmental factors on vegetation restoration, it establishes a method and system capable of effectively determining vegetation restoration thresholds and their spatial distribution over large areas, thereby providing a scientific and universally applicable decision-making basis for macro-ecological restoration projects. The specific technical terms and explanations involved are as follows:

[0060] the term explain Vegetation cover (FVC) The vertical projection area of ​​the vegetation canopy is the proportion of the total area of ​​the statistical area. It is a key remote sensing indicator for measuring the vegetation growth status and is usually obtained by inversion through remote sensing spectral indices (such as NDVI). Vegetation restoration threshold In this technology, it specifically refers to the maximum vegetation cover level that vegetation can achieve and sustainably maintain during natural or artificial restoration, corresponding to the "carrying capacity threshold" or "steady-state upper limit" in ecology. Time Duration Condition To ensure that peak vegetation cover is not a random fluctuation, the observation period must be maintained for at least 5 years after the vegetation cover reaches its peak, in order to eliminate the influence of interannual climate fluctuations. Wave stability condition To quantify "entering a stable state", the following rule is set: the fluctuation range (local range) of FVC in the period after the peak shall not exceed 5% of the total fluctuation range of FVC in the entire study period. Environmental factors This technique constructs a multivariate set of variables, including environmental variables such as climate, topography, and soil, to predict vegetation restoration thresholds, which serves as input to the random forest model. Time-series remote sensing This refers to a technique for dynamic process monitoring and analysis based on a series of continuously acquired remote sensing images of the same region across multiple time periods. This technique relies on time-series remote sensing data to extract the interannual variation process of vegetation cover, providing a data foundation for identifying the status of vegetation restoration.

[0061] The method flow of a specific embodiment is as follows: Figures 1-2 As shown, it includes:

[0062] S1: Obtain long-term vegetation cover data;

[0063] Whether vegetation restoration has reached the ecological threshold (i.e., entered a sustainable stable state) is objectively determined by analyzing the spatiotemporal characteristics of interannual vegetation cover (FVC). This determination must simultaneously meet the following two rigid conditions with a specific order: Figure 3 As shown, namely S2 and S3, specifically:

[0064] After reaching its peak canopy cover during its recovery process, vegetation must undergo a sufficiently long period of stable observation to exclude accidental peaks caused by interannual climate fluctuations. That is, S2: Determine if the time persistence condition is met: T end - T max ≥5 years; among which, This represents the year in which FVC reaches its maximum value throughout the entire time series. This is the end year of the time series. This condition ensures a "post-peak" observation window of at least 5 years for assessing the stability of the state. If "No", it is determined that "the recovery threshold has not been reached"; if "Yes", proceed to the next step.

[0065] S3: If the above conditions are met, the vegetation growth status after reaching its peak must remain within a very narrow fluctuation range, indicating that it has left the recovery and upward phase and entered a stable phase. Further assessment is needed to determine whether the fluctuation stability condition is met: (FVC) _local-max - FVC _local-min ) ≤ 0.05 × (FVC _global-max - FVC _global-min ); where FVC _local-max With FVC _local-min These represent the maximum and minimum values ​​in the local time series "after reaching the peak"; FVC _global-max With FVC _global-minThese represent the maximum and minimum values ​​of the entire time series over the entire study period, respectively. The 5% threshold means that the interannual fluctuation amplitude in the post-peak period must not exceed one-twentieth of the total fluctuation amplitude of the entire recovery process, thus mathematically defining the concept of "fluctuations tending to level off." If "No," it is determined that "the recovery threshold has not been reached"; if "Yes," it is ultimately determined that "the recovery threshold has been reached."

[0066] S4: If the time persistence condition and the fluctuation stability condition are met at the same time, it is determined that the pixel has reached the vegetation restoration threshold.

[0067] S5: For pixels that have reached the recovery threshold, calculate their value in T. max To T end The average vegetation cover over a given period is used as the vegetation restoration threshold for that pixel.

[0068] S6: Extract the pixels that have reached the recovery threshold as the sample dataset;

[0069] S7: Extract the environmental factors corresponding to each sample, such as multi-year average rainfall, temperature, elevation, slope, aspect, soil type, soil bulk density, sandy soil ratio, loam soil ratio, and clay soil ratio.

[0070] S8: Construct a nonlinear relationship model between the environmental factors and the vegetation restoration threshold based on the random forest model;

[0071] S9: The random forest model is applied to predict the vegetation restoration threshold of all pixels in the study area to achieve spatial expansion.

[0072] Furthermore, the calculation formula for S5 is as follows:

[0073]

[0074] in, For pixels The vegetation restoration threshold; For pixels In time vegetation coverage; The time when vegetation cover reaches its peak; This is the end time of the observation period.

[0075] Furthermore, a set of pixels within the study area that meet the vegetation restoration threshold conditions is extracted from long-term remote sensing data to construct a vegetation restoration threshold sample dataset. The formula for constructing the sample dataset is as follows:

[0076]

[0077] in, Indicates the first One pixel that has reached the recovery threshold. The total number of samples.

[0078] Furthermore, S7~S9 include:

[0079] For each sample cell Extract the corresponding environmental factors to form an environmental feature vector. :

[0080]

[0081] In the formula, This represents the average annual rainfall over many years. The average temperature over many years For elevation, For slope, Slope direction, Soil type For soil bulk density, The proportion of sandy soil, The proportion of soil is loam. The clay ratio;

[0082] Establish vegetation restoration threshold Nonlinear mapping relationship model between environmental factors:

[0083]

[0084] in, This represents a complex nonlinear function fitted by a random forest. This refers to model error;

[0085] Using the random forest regression method, For input features, Using the target variable, train the model, optimize the hyperparameters through cross-validation, and obtain the fitted model. .

[0086] The environmental factors corresponding to all pixels in the study area are input into the trained random forest model. Per-pixel simulated vegetation restoration threshold:

[0087]

[0088] in, The total number of pixels in the study area. For the first An environmental factor vector of a pixel This is an estimated threshold for vegetation restoration in that pixel.

[0089] Generate a spatial distribution map of vegetation restoration thresholds in the study area, realize refined spatial expansion of the thresholds, and provide spatial decision support for regional ecological restoration projects.

[0090] In the spatial expansion modeling stage, random forest models can be replaced by various machine learning methods. For example, neural network models (such as multilayer perceptrons and convolutional neural networks) can capture higher-dimensional nonlinear relationships and are suitable for multi-source remote sensing feature fusion; gradient boosting models (such as XGBoost and LightGBM) often perform well in small sample scenarios; and Gaussian process regression can provide quantification of prediction uncertainty. In addition, geostatistical methods (such as co-kriging interpolation) or geographically weighted regression can also be used for spatial modeling. The former can explicitly utilize spatial autocorrelation, while the latter can better handle spatial nonstationarity issues.

[0091] At the data input level, in addition to basic climate, topography and soil factors, micro-topographic feature data such as terraces and silt-retaining dams can also be introduced to more accurately characterize the site conditions of vegetation.

[0092] Based on the same idea, embodiments of the present invention also provide a vegetation restoration threshold calculation system, including:

[0093] Acquisition module: used to acquire long-term vegetation cover data;

[0094] First judgment module: used to determine whether the time duration condition is met: T end - T max ≥ 5 years; of which, This represents the year in which FVC reaches its maximum value throughout the entire time series. This represents the end year of the time series.

[0095] The second judgment module is used to further determine whether the fluctuation stability condition (FVC) is met if it is satisfied. _local-max - FVC _local-min ) ≤ 0.05 × (FVC _global-max - FVC _global-min ); where FVC _local-max With FVC _local-min These represent the maximum and minimum values ​​in the local time series "after reaching the peak"; FVC _global-max With FVC _global-min These represent the maximum and minimum values ​​of the entire time series over the entire study period, respectively.

[0096] Determination module: used to determine that the pixel has reached the vegetation restoration threshold if the time duration condition and the fluctuation stability condition are met simultaneously.

[0097] Calculation module: Used to calculate the value of pixels that have reached the recovery threshold in T.max To T end The average vegetation cover over a given period is used as the vegetation restoration threshold for that pixel.

[0098] First extraction module: used to extract pixels that have reached the recovery threshold as sample dataset;

[0099] The second extraction module is used to extract environmental factors corresponding to each sample, such as multi-year average rainfall, temperature, elevation, slope, aspect, soil type, soil bulk density, sandy soil ratio, loamy soil ratio, and clay soil ratio.

[0100] Module: Used to construct a nonlinear relationship model between the environmental factors and the vegetation restoration threshold based on the random forest model;

[0101] Prediction module: used to apply the random forest model to predict the vegetation restoration threshold of all pixels in the study area, thereby achieving spatial expansion.

[0102] In summary, this invention, by integrating long-term remote sensing observations with multi-source environmental factors, constructs a large-area vegetation restoration threshold determination model with both physical mechanisms and statistical reliability, achieving the following three objectives:

[0103] ① Scientific level: Promote the transformation of vegetation restoration research from "site characterization" to "spatial pattern", deepen the understanding of the regional differentiation law of ecosystem resilience and stability threshold, and provide new methods and parameter basis for large-scale ecological process simulation.

[0104] ② Technical aspect: Establish a set of operational threshold determination technology processes to achieve automated generation of threshold distribution map products from remote sensing data, thereby improving the spatiotemporal coverage and decision support efficiency of ecological monitoring and assessment.

[0105] ③ Practical level: It directly serves the planning, implementation and post-evaluation of national and local ecological construction projects, provides quantitative basis for "where to plant, what to plant and how much to plant", promotes the transformation of vegetation construction from "scale expansion" to "precision quality improvement", enhances the sustainability and climate adaptability of ecological projects, and helps to achieve my country's carbon neutrality and ecological security goals.

[0106] The advantages of the embodiments of the present invention compared to the core defects of the prior art are shown in the following table:

[0107] Existing technology types Core defects of existing technology Advantages of the embodiments of the present invention (1) Plot / slope scale method (based on fixed-point observation and mechanism model) ① Insufficient spatial representativeness: It relies heavily on site data and is difficult to reflect the spatial heterogeneity of complex environments in large areas. ② Complex parameters and difficulty in generalization: The mechanistic model has many parameters and the true value is difficult to obtain, which limits its wide application. ③ Poor engineering applicability: Simply applying local thresholds to diverse site conditions can easily lead to "over-repair" or "under-repair". ① Comprehensive Spatialization Capability: Based on long-term remote sensing data, it can objectively and automatically extract pixels that have reached the threshold across the entire region, forming a sample system covering complex environmental gradients and fully reflecting spatial heterogeneity. ② Data-Driven and Parameter Simplified: Employing machine learning methods such as neural networks, it does not rely on complex mechanistic parameters, requiring only readily available environmental factor data from multiple sources, and possesses strong generalizability. ③ Refined Engineering Guidance: The generated threshold spatial distribution map has a fine resolution (e.g., 30 meters), providing differentiated threshold references for different slopes and site conditions, supporting precise "one-site-one-policy" restoration. (2) Regional scale method (based on spatial statistics of "similar habitats") ① High subjectivity in patch division: The scientific validity and rationality of habitat clustering are difficult to guarantee. ② Static threshold definition: The extracted thresholds are actually the maximum values ​​of vegetation parameters during the observation period, without considering long-term dynamic stability. ③ Ignoring vegetation physiological and water constraints: The drought resistance characteristics of vegetation and the actual available water are not fully considered, introducing significant uncertainty. ① Sample selection based on objective time-series judgment: A rigid "dual-condition" rule (temporal persistence + fluctuation stability) directly determines whether a threshold has been reached from the pixel time-series features, avoiding subjective patch division. ② Dynamic stabilization of threshold definition: The threshold is defined as the average vegetation cover during the "post-peak stabilization period," reflecting the true carrying capacity of vegetation after it enters a steady state, thus demonstrating greater ecological rationality. ③ Multi-factor collaborative modeling: Multiple environmental factors, such as climate, topography, and soil, are incorporated into the neural network model, implicitly expressing key constraints such as water availability, and more systematically characterizing the comprehensive environmental boundary of vegetation restoration.

[0108] The comprehensive advantages of this invention are summarized as follows: (1) High objectivity and automation: By rigidly determining temporal features and modeling with machine learning, human intervention is reduced, and the objectivity of threshold identification and spatialization processes is improved. (2) Strong spatiotemporal universality: Based on long-term remote sensing data and environmental factors, the method has the potential for cross-regional and cross-ecosystem transplantation. (3) Clear engineering guidance significance: It provides a high spatial resolution threshold distribution map, which can directly support the vegetation configuration and target setting of specific ecological restoration projects such as slopes and small watersheds, and improve the scientific nature and long-term effectiveness of the project.

[0109] In addition, the following should be noted: 1. Recommendations for data and tools during technology implementation. project illustrate Remote sensing data source It is recommended to use MODIS NDVI (250m / 500m, interannual sequence) or Landsat NDVI (30m, requires fusion and reconstruction), with a time span of ≥15 years to cover the entire vegetation restoration process. Environmental factor data Climate data can be obtained from WorldClim, ERA5, etc.; topographic factors are extracted based on DEM; soil data can be obtained from global or regional soil databases such as HWSD and SoilGrids. Tools and Platforms The algorithm can be implemented in Python (scikit-learn, TensorFlow), R, or IDL / ENVI platforms. Spatial output format The final vegetation restoration threshold distribution map is recommended to be output in GeoTIFF format with a spatial resolution of 30 meters, which is convenient for use in conjunction with engineering planning and design drawings.

[0110] II. Engineering Manifestations of Technological Advantages Scene How this invention provides support Slope ecological restoration design It directly provides the spatial distribution of vegetation restoration thresholds at different locations on the slope, guiding precise configuration of "planting grass where appropriate and irrigating where appropriate," and avoiding excessive restoration that leads to soil drying. Watershed ecological restoration assessment It can generate a distribution map of threshold achievement within the watershed, identify areas that have been restored and areas that still need repair, and provide a basis for subsequent management and key treatment. Ecological Red Line Management Assistance The spatial distribution of thresholds can serve as an input for evaluating regional ecological carrying capacity, providing a quantitative basis for delineating ecological protection red lines and determining vegetation cover management targets. Climate change adaptation planning By combining future climate scenarios, the potential spatial evolution of vegetation restoration thresholds can be simulated, ecologically vulnerable areas can be identified in advance, and adaptive restoration strategies can be formulated.

[0111] III. Adjustable parameters in the technical process

[0112] Flexible adjustment of the recovery threshold conditions: "5-year observation period" and "5% fluctuation ratio" can be used as adjustable parameters, which are reflected in the claims as "preset time window" and "preset fluctuation tolerance".

[0113] Flexibility in the combination of environmental factors: The claims may not limit the specific number and type of factors, but may use the expression "including but not limited to multiple factors in climate, topography, and soil" to cover implementation methods under different data conditions.

[0114] The basic principles of the present invention have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in the present invention are merely examples and not limitations, and should not be considered as essential features of each embodiment of the present invention. Furthermore, the specific details disclosed are for illustrative and facilitative purposes only, and are not limitations; these details do not restrict the present invention from being implemented solely using these specific details.

[0115] The block diagrams of devices, apparatuses, devices, and systems involved in this invention are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0116] It should also be noted that in the apparatus, device, and method of the present invention, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of the present invention.

[0117] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the invention. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the invention. Therefore, the invention is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0118] It should be understood that the qualifying terms "first", "second", "third", "fourth", "fifth" and "sixth" used in the description of the embodiments of the present invention are only used to more clearly illustrate the technical solutions and are not intended to limit the scope of protection of the present invention.

[0119] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the invention to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A method for calculating vegetation restoration threshold, characterized in that, include: S1: Obtain long-term vegetation cover data; S2: Determine if the time duration condition is met: T end - T max ≥ 5 years; of which, This represents the year in which FVC reaches its maximum value throughout the entire time series. This represents the end year of the time series. S3: If satisfied, further determine whether the fluctuation stability condition is met: (FVC) _local-max - FVC _local-min ) ≤0.05 × (FVC _global-max - FVC _global-min ); where FVC _local-max With FVC _local-min These represent the maximum and minimum values ​​in the local time series "after reaching the peak"; FVC _global-max With FVC _global-min These represent the maximum and minimum values ​​of the entire time series over the entire study period, respectively. S4: If the time persistence condition and the fluctuation stability condition are met at the same time, it is determined that the pixel has reached the vegetation restoration threshold. S5: For pixels that have reached the recovery threshold, calculate their value in T. max To T end The average vegetation cover over the time period is used as the vegetation restoration threshold for that pixel. S6: Extract the pixels that have reached the recovery threshold as the sample dataset; S7: Extract the environmental factors corresponding to each sample, such as multi-year average rainfall, temperature, elevation, slope, aspect, soil type, soil bulk density, sandy soil ratio, loam soil ratio, and clay soil ratio. S8: Construct a nonlinear relationship model between the environmental factors and the vegetation restoration threshold based on the random forest model; S9: The random forest model is applied to predict the vegetation restoration threshold of all pixels in the study area to achieve spatial expansion.

2. The vegetation restoration threshold calculation method according to claim 1, characterized in that, The calculation formula for S5 is as follows: in, For pixels The vegetation restoration threshold; For pixels In time vegetation coverage; The time when vegetation cover reaches its peak; This is the end time of the observation period.

3. The vegetation restoration threshold calculation method according to claim 1, characterized in that, The formula for constructing the sample dataset is: in, Indicates the first One pixel that has reached the recovery threshold. The total number of samples.

4. The vegetation restoration threshold calculation method according to claim 1, characterized in that, S7~S9 include: For each sample cell Extract the corresponding environmental factors to form an environmental feature vector. : In the formula, This represents the average annual rainfall over many years. The average temperature over many years For elevation, For slope, Slope direction, Soil type For soil bulk density, The proportion of sandy soil, The proportion of soil is loam. The clay ratio; Establish vegetation restoration threshold Nonlinear mapping relationship model between environmental factors: in, This represents a complex nonlinear function fitted by a random forest. This refers to model error; Using the random forest regression method, For input features, Using the target variable, train the model, optimize the hyperparameters through cross-validation, and obtain the fitted model. . The environmental factors corresponding to all pixels in the study area are input into the trained random forest model. Per-pixel simulated vegetation restoration threshold: in, The total number of pixels in the study area. For the first An environmental factor vector of a pixel This is an estimated threshold for vegetation restoration in that pixel. Generate a spatial distribution map of vegetation restoration thresholds in the study area, realize refined spatial expansion of the thresholds, and provide spatial decision support for regional ecological restoration projects.

5. A vegetation restoration threshold calculation system, characterized in that, include: Acquisition module: used to acquire long-term vegetation cover data; The first judgment module is used to determine whether the time duration condition is met: Tend - Tmax ≥ 5 years; where, This represents the year in which FVC reaches its maximum value throughout the entire time series. This represents the end year of the time series. The second judgment module is used to further determine whether the fluctuation stability condition (FVC) is met if it is satisfied. _local-max -FVC _local-min ) ≤ 0.05 × (FVC _global-max - FVC _global-min ); where FVC _local-max With FVC _local-min These represent the maximum and minimum values ​​in the local time series "after reaching the peak"; FVC _global-max With FVC _global-min These represent the maximum and minimum values ​​of the entire time series over the entire study period, respectively. Determination module: used to determine that the pixel has reached the vegetation restoration threshold if both the time duration condition and the fluctuation stability condition are met simultaneously; Calculation module: Used to calculate the value of pixels that have reached the recovery threshold in T. max To T end The average vegetation cover over the time period is used as the vegetation restoration threshold for that pixel. First extraction module: used to extract pixels that have reached the recovery threshold as sample dataset; The second extraction module is used to extract environmental factors corresponding to each sample, such as multi-year average rainfall, temperature, elevation, slope, aspect, soil type, soil bulk density, sandy soil ratio, loamy soil ratio, and clay soil ratio. Module: Used to construct a nonlinear relationship model between the environmental factors and the vegetation restoration threshold based on the random forest model; Prediction module: used to apply the random forest model to predict the vegetation restoration threshold of all pixels in the study area, thereby achieving spatial expansion.

Images