Method and system for determining forestation area of fast-growing and high-yield forest tree species

Abstract

The invention provides a fast-growing high-yield forest tree species afforestation area establishment method and system, and relates to the field of afforestation. According to the method, ecological environment and social economic two-dimensional information are fused, an improved maximum entropy reinforcement learning model is adopted to construct an ecological environment suitable area prediction model, social economic suitability is judged based on rule modeling, and a final suitable afforestation area is obtained through space superposition. According to the method, intellectualization and precision of afforestation site selection are achieved, scientificity and actual operability of suitability judgment are improved, and the method is suitable for afforestation planning and optimization under large-scale and multi-constraint conditions.

Description

Technical Field

[0001] The present invention relates to the field of tree planting and afforestation, and in particular to a method and system for establishing an afforestation area for fast-growing and high-yield tree species. Background Art

[0002] Fast-growing, high-yield forests are typically densely planted in rows of a single tree species. Intensive management methods such as fertilization and irrigation promote rapid growth, resulting in extremely short rotation periods and clear-cutting within a short period of time. These forests are typically planted to produce pulp, paper, timber, and wood pellets for bioenergy production. They also play a significant role in reducing logging pressure on natural forests. With changing ecological environments, the importance of fast-growing, high-yield forests in ecological protection and carbon sequestration has become increasingly prominent. In stark contrast to the continued decline in global forest cover, the area of ​​planted forests has more than doubled over the past few decades, with fast-growing, high-yield forests accounting for approximately half of this area. Furthermore, with the growing demand for timber, the establishment of fast-growing, high-yield forests has become increasingly important. It is estimated that the area of ​​fast-growing, high-yield forests will expand by 2% annually, reaching a total area of ​​approximately 90 million hectares by mid-century. Furthermore, due to their high yields, fast-growing, high-yield forests are also likely to have a high carbon sequestration capacity, making them a cost-effective climate change mitigation measure. Furthermore, they can increase the land value of marginal agricultural land and bring higher resource utilization benefits.

[0003] Selecting appropriate afforestation areas has a crucial impact on the growth, yield, and implementation effectiveness of fast-growing, high-yield forest species. Traditional methods for establishing afforestation areas for fast-growing, high-yield forest species rely solely on empirical judgment, relying solely on ecological and environmental suitability. Suitable afforestation sites are determined by comparing the ecological and environmental characteristics of the tree species' distribution area with those of the target afforestation area. This traditional approach considers only ecological and environmental suitability, without factoring in the impact of regional feasibility, such as irrigation and management costs, and market accessibility. Consequently, the area of ​​suitable afforestation areas for fast-growing, high-yield forests is overestimated. In some areas, although fast-growing, high-yield forests meet expected yields, they fail to achieve the expected resource utilization returns. Alternatively, the post-afforestation growth of some fast-growing, high-yield forest species is significantly affected by human activities and fails to achieve the expected yields. This significantly impacts the feasibility of afforestation for fast-growing, high-yield forest species. Therefore, a scientific, rational, and precise method for establishing afforestation areas for fast-growing, high-yield forest species is urgently needed to synergistically optimize yield and resource utilization. Summary of the Invention

[0004] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide a method and system for establishing afforestation areas for fast-growing and high-yield forest tree species, which integrates reinforcement learning and rule modeling to achieve intelligent and precise delineation of fast-growing and high-yield forest afforestation areas under the dual-dimensional constraints of ecological and regional implementation conditions, significantly improving the scientific nature and feasibility of suitability judgment.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for establishing an afforestation area for fast-growing and high-yield tree species, comprising:

[0007] Obtain afforestation distribution data of target fast-growing and high-yield forest tree species;

[0008] Based on the afforestation distribution data, extracting ecological environment data of the first dimension and development basic data of the second dimension respectively;

[0009] Taking the afforestation distribution data as the response variable and the ecological environment data as the explanatory variable, an improved maximum entropy reinforcement learning model is used to establish an ecological environment suitable area prediction model, and the ecological environment suitable area of ​​the target area is obtained according to the ecological environment suitable area prediction model;

[0010] Performing an index interval statistical analysis on the basic development data, constructing a rule-based construction feasible area prediction model, and obtaining a construction feasible area of ​​the target area according to the construction feasible area prediction model;

[0011] The scope of the ecological environment suitable zone is revised so that the spatial intersection of the revised ecological environment suitable zone and the construction feasible zone is consistent, and the final suitable afforestation area for the target fast-growing and high-yield forest tree species is obtained.

[0012] Preferably, the afforestation distribution data is obtained through remote sensing or vegetation map interpretation, forest resource inventory, published literature, online database data and forest resource inventory data.

[0013] Preferably, based on the afforestation distribution data, extracting the ecological environment data of the first dimension and the development basic data of the second dimension respectively includes:

[0014] Downloading ecological environment spatial raster data by searching network shared data within the coverage of the afforestation distribution data; the ecological environment spatial raster data includes: climate data, soil data and topographic data;

[0015] Extract the corresponding indicator values ​​in the ecological environment spatial raster data to the afforestation distribution points of the afforestation distribution data through the extraction value to point tool of the geographic information system platform, and generate an ecological environment data table containing each afforestation distribution point in a field format;

[0016] Merging fields and arranging the extracted ecological environment data table in a standard format to form ecological environment data of the first dimension for modeling and analysis;

[0017] Obtain spatial vector location data of roads, settlements, water sources, cities, towns, and ports from OpenStreetMap and the National Basic Geographic Information Database, as well as spatial raster data of population density, gross domestic product, and night light index;

[0018] Utilizing the distance calculation and network analysis tools of the geographic information system platform, the spatial distance and access time from each of the afforestation distribution points to the nearest road, settlement, water source, and city are calculated based on the spatial vector position data, and a market accessibility index is generated based on the market accessibility formula;

[0019] Using the Extract Values ​​to Points tool of the geographic information system platform, extract corresponding index values ​​from the population density spatial raster data, the gross domestic product spatial raster data, and the nighttime light index spatial raster data to each of the afforestation distribution points, and combine the corresponding index values ​​with the spatial distance and access time, and the market accessibility index to generate a development basic data table;

[0020] The fields of the development basic data table are sorted and the format is unified to form the development basic data of the second dimension.

[0021] Preferably, the calculation formula of the market reachability formula is:

[0022] ;

[0023] Among them, a ij represents the market accessibility from afforestation distribution point i to destination j, d ij is the distance between point i and point j; for each location i, the distance from it to the nearest first and second group destinations j is a ij Value; S j represents the importance of destination j, where the first group of destinations S j Assigned value 1, the second group of destinations S j The value is assigned as 0.5, v is a preset constant, and the assignment process is: ; Among them, d * It represents the distance from point i to the location where the accessibility drops the fastest.

[0024] Preferably, the afforestation distribution data is used as the response variable, the ecological environment data is used as the explanatory variable, and an improved maximum entropy reinforcement learning model is used to establish an ecological environment suitable area prediction model, and the ecological environment suitable area of ​​the target area is obtained according to the ecological environment suitable area prediction model, including:

[0025] The state vector is constructed using the ecological environment data of each afforestation distribution point. and uses the presence or absence of afforestation distribution records as labels ;

[0026] Will Randomly divided into training set and validation set;

[0027] In the improved maximum entropy reinforcement learning framework, the state vector Input graph convolution-transformer coupled hierarchical spatiotemporal state embedding network to obtain multi-scale embedding ;

[0028] Setting the action space ,in Indicates that the grid is judged to be an ecologically suitable area. It means it is judged as an unsuitable area;

[0029] Building a reward function , the reward function Including ecological prediction rewards and resource allocation cost items; ecological prediction rewards are the current action When , give positive rewards , otherwise the reward is 0; the resource allocation cost item is: from the development basic data of the second dimension, according to the rule model, determine whether the regional construction feasibility standard is met. If it is met, the cost item , otherwise the cost item ; The reward function The formula is: ;in, is the development constraint weight coefficient;

[0030] Construct a maximum entropy reinforcement learning model, which includes: a strategy network and value network The policy network Output in state Take action probability distribution of value network For estimating state-action pairs expected returns;

[0031] The Soft Actor-Critic strategy optimization method is used to train the model to maximize the objective function; the objective function is: ;in, For the The state of the time step; is the action selected by the strategy in this state; is the entropy weight coefficient, which is used to adjust the exploration and stability of the strategy;

[0032] After each update, the state-action-reward triplet Store the experience replay buffer and adaptively adjust the entropy weight coefficient based on the performance of the validation set ;

[0033] When the policy network After convergence, the model parameters are fixed to obtain the ecological environment suitable area prediction model;

[0034] Based on the ecological environment suitable zone prediction model, all locations to be determined in the target area Make predictions and calculate probability distributions ;

[0035] The probability distribution exceeds the set threshold All locations are aggregated and output as the ecological environment suitable area of ​​the target area.

[0036] Preferably, the development basic data is statistically analyzed for index intervals to construct a rule-based construction feasible area prediction model, and the construction feasible area of ​​the target area is obtained according to the construction feasible area prediction model, including:

[0037] Calculating the minimum value, maximum value, mean value and standard deviation of each indicator in the development basic data at the afforestation distribution points of the afforestation distribution data to form a statistical analysis table;

[0038] According to the statistical analysis table and the preset appropriate threshold reference table, determine the lower limit for each indicator With upper limit , get the index interval set ; k is the indicator index, ;

[0039] Construct a construction feasible area prediction model; the judgment rule of the construction feasible area prediction model is: if If both are established, the feasibility determination result of regional construction is ;otherwise ; For the grid to be evaluated Item index value, Indicates the construction feasible area, Indicates unsuitable area;

[0040] The judgment rule is called on all candidate grids in the target area one by one to generate a binary grid layer. ; Among them, the binary raster layer The value 1 indicates that the construction is feasible area pixel, binary raster layer The value of 0 indicates a pixel in an unsuitable area;

[0041] right Connected domain analysis is performed to retain connected patches whose area is not less than the pre-set first minimum threshold, and morphological opening operation is performed to smooth the boundaries to obtain the feasible construction area.

[0042] Preferably, the various indicators include: population density, night light index, gross domestic product, road distance, settlement distance, water source distance, city access time and market accessibility index.

[0043] Preferably, the ecological environment suitable zone is revised so that the spatial intersection of the revised ecological environment suitable zone and the construction feasible zone is consistent, and the final suitable afforestation area of ​​the target fast-growing and high-yield forest tree species is obtained, including:

[0044] In the geographic information system platform, the ecological environment suitable area and the construction feasible area are uniformly set to the same plane coordinate system;

[0045] Call the spatial overlay-intersection tool of the geographic information system platform to calculate the spatial intersection of the layers of the ecological environment suitable area and the construction feasible area, and generate the intersection vector layer ;

[0046] right Perform a dissolve operation to remove internal boundaries and form a single collection of spatial features;

[0047] Use the "area screening" tool to delete isolated patches in the single spatial feature set whose area is smaller than a preset second minimum threshold;

[0048] Perform morphological opening operation on the retained patches to smooth the boundaries and output the corrected ecological environment suitable area vector;

[0049] The revised ecological environment suitable area will be determined as the final suitable afforestation area.

[0050] A system for establishing afforestation areas for fast-growing and high-yield tree species, comprising:

[0051] Afforestation distribution data acquisition unit, used to acquire afforestation distribution data of target fast-growing and high-yield forest tree species;

[0052] A multi-dimensional environment-development data extraction unit, configured to extract first-dimensional ecological environment data and second-dimensional development basic data based on the afforestation distribution data;

[0053] an ecological environment suitable zone prediction unit, configured to use the afforestation distribution data as a response variable and the ecological environment data as an explanatory variable, establish an ecological environment suitable zone prediction model using an improved maximum entropy reinforcement learning model, and obtain an ecological environment suitable zone of a target area based on the ecological environment suitable zone prediction model;

[0054] A construction feasible area prediction unit is used to perform an index interval statistical analysis on the development basic data, construct a rule-based construction feasible area prediction model, and obtain a construction feasible area of ​​a target area according to the construction feasible area prediction model;

[0055] The range correction and result fusion unit is used to correct the range of the ecological environment suitable area so that the spatial intersection of the corrected ecological environment suitable area and the feasible construction area is consistent, thereby obtaining the final suitable afforestation area for the target fast-growing and high-yield forest tree species.

[0056] The present invention discloses the following technical effects:

[0057] This paper, by combining an improved maximum entropy reinforcement learning model with a rule-based prediction model for feasible construction zones, dynamically couples ecological and environmental suitability with development feasibility at the same spatial scale for the first time. On the one hand, maximum entropy reinforcement learning utilizes multi-scale ecological and environmental characteristics to adaptively optimize decision-making strategies, significantly improving the accuracy of ecologically suitable zone determination while maintaining exploratory nature. On the other hand, the regional implementation rule model uses eight hard constraints, including population density, transportation accessibility, and resource utilization return output, to ensure that the selected areas are realistically feasible in terms of market, infrastructure, and cost investment. Finally, through post-processing such as spatial intersection and patch removal, a contiguous, feasible, and feasible "final suitable afforestation area" is output, addressing both ecological and implementation requirements. This method avoids both the ecological risks associated with relying solely on empirical thresholds and the wasteful afforestation costs associated with ignoring implementation constraints. It achieves precise layout of fast-growing and high-yield forests, efficient resource utilization, and long-term sustainable management, significantly improving afforestation survival rates and resource utilization returns. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0059] Figure 1 A flow chart of a method provided by an embodiment of the present invention;

[0060] Figure 2 A schematic diagram of a technical route provided by an embodiment of the present invention;

[0061] Figure 3 A schematic diagram of the system structure provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0063] The purpose of the present invention is to provide a method and system for establishing afforestation areas for fast-growing and high-yield forest tree species, which integrates reinforcement learning and rule modeling, realizes the intelligent and precise delineation of fast-growing and high-yield forest afforestation areas under the dual-dimensional constraints of ecological and regional implementation conditions, and significantly improves the scientific nature and feasibility of suitability judgment.

[0064] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0065] Figure 1 A flow chart of the method provided in the embodiment of the present invention is shown in FIG. Figure 1 As shown, the present invention provides 1. A method for establishing an afforestation area for fast-growing and high-yield tree species, characterized by comprising:

[0066] Step 100: Obtaining afforestation distribution data of target fast-growing and high-yield forest tree species;

[0067] Step 200: extracting ecological environment data of the first dimension and development basic data of the second dimension based on the afforestation distribution data;

[0068] Step 300: Using the afforestation distribution data as the response variable and the ecological environment data as the explanatory variable, an improved maximum entropy reinforcement learning model is used to establish an ecological environment suitable area prediction model, and the ecological environment suitable area of ​​the target area is obtained based on the ecological environment suitable area prediction model;

[0069] Step 400: Performing an index interval statistical analysis on the development basic data, constructing a rule-based construction feasible area prediction model, and obtaining the construction feasible area of ​​the target area based on the construction feasible area prediction model;

[0070] Step 500: Correct the range of the ecological environment suitable area so that the spatial intersection of the corrected ecological environment suitable area and the construction feasible area is consistent, and obtain the final suitable afforestation area for the target fast-growing and high-yield forest tree species.

[0071] Preferably, the afforestation distribution data is obtained through remote sensing or vegetation map interpretation, forest resource inventory, published literature, online database data and forest resource inventory data.

[0072] like Figure 2As shown, the afforestation distribution data of fast-growing and high-yield forest tree species in this embodiment must come from commercial artificial afforestation sites, and the sources include: (1) remote sensing interpretation maps; (2) vegetation maps or forest resource maps; (3) published literature (monographs, papers) data; (4) online database data; and (5) forest resource inventory data.

[0073] Exemplary online database data include: Global Biodiversity Platform GBIF, China Digital Herbarium CVH, etc.

[0074] Preferably, based on the afforestation distribution data, extracting the ecological environment data of the first dimension and the development basic data of the second dimension respectively includes:

[0075] Downloading ecological environment spatial raster data by searching network shared data within the coverage of the afforestation distribution data; the ecological environment spatial raster data includes: climate data, soil data and topographic data;

[0076] Extract the corresponding indicator values ​​in the ecological environment spatial raster data to the afforestation distribution points of the afforestation distribution data through the extraction value to point tool of the geographic information system platform, and generate an ecological environment data table containing each afforestation distribution point in a field format;

[0077] Merging fields and arranging the extracted ecological environment data table in a standard format to form ecological environment data of the first dimension for modeling and analysis;

[0078] Obtain spatial vector location data of roads, settlements, water sources, cities, towns, and ports from OpenStreetMap and the National Basic Geographic Information Database, as well as spatial raster data of population density, gross domestic product, and night light index;

[0079] Utilizing the distance calculation and network analysis tools of the geographic information system platform, the spatial distance and access time from each of the afforestation distribution points to the nearest road, settlement, water source, and city are calculated based on the spatial vector position data, and a market accessibility index is generated based on the market accessibility formula;

[0080] Using the Extract Values ​​to Points tool of the geographic information system platform, extract corresponding index values ​​from the population density spatial raster data, the gross domestic product spatial raster data, and the nighttime light index spatial raster data to each of the afforestation distribution points, and combine the corresponding index values ​​with the spatial distance and access time, and the market accessibility index to generate a development basic data table;

[0081] The fields of the development basic data table are sorted and the format is unified to form the development basic data of the second dimension.

[0082] Specifically, the steps for acquiring ecological environment data of the distribution locations of fast-growing and high-yield forest tree species in this embodiment are as follows:

[0083] Ecological and environmental data are obtained by retrieving shared data on the Internet. Ecological and environmental spatial raster data include climate, soil, and topography data. Then, the "extract value to point" function module in geographic information systems or R language software is used to extract the ecological and environmental data values ​​of the corresponding locations based on the geographic coordinates of the tree species distribution points.

[0084] Climate data include indicators such as annual mean temperature, annual temperature range, annual mean precipitation, and summer precipitation. Data can be obtained through the following methods: (1) spatial interpolation of meteorological station data using methods such as kriging interpolation to obtain climate spatial raster data; (2) spatial raster data freely available in literature or on the Internet. Literature data includes spatial raster climate data published in data papers, and online data includes climate surface spatial raster data provided by the WorldClim and CHELSA websites.

[0085] Soil data includes soil organic carbon content, total nitrogen, total phosphorus, and total potassium content, and soil pH. Data can be contributed free of charge through literature or online. Online data includes spatial raster data from the Global Soil Grid Dataset SoilGRID250m, the HWSD World Soil Database, and the National Tibetan Plateau Data Center.

[0086] Global terrain data sources include elevation, slope, and aspect. Spatial raster terrain data can be generated from topographic maps or digital elevation models (DEMs) using a geographic information system (GIS). Digital elevation models can be obtained from shared data sets such as ASTER DEM, SRTM DEM, and ALOS DEM.

[0087] Furthermore, the spatial data of development condition indicators in this embodiment include but are not limited to: market accessibility, distance to water sources, population density, travel time to towns, distance to roads, distance to settlements, night light index, gross domestic product, etc.

[0088] The spatial vector location data of roads, water sources (rivers, lakes, swamps and wetlands), settlements, cities, towns and ports are derived from online shared data, including: OpenStreetMap, the world's open street map, 1:1 million public version of basic geographic information data, 1:250,000 national basic geographic database, etc.

[0089] The population density spatial raster data comes from free shared data on the Internet, including: National Earth System Science Data Center population density data, NASA population density data GWPv4, Oak Ridge National Laboratory population density data Landscan, global population density data WorldPop, global human settlement raster data GHS_POP, global population raster data GPW, etc.

[0090] The GDP spatial raster data comes from free shared data on the Internet, including: GDP data from the National Earth System Science Data Center, China's GDP raster paper data, etc.

[0091] The spatial raster data of the night light index comes from free shared data on the Internet, including: night light data from the National Earth System Science Data Center, VIIRS data from the National Oceanic and Atmospheric Administration of the United States, Suomi NPP satellite data, DMSP satellite data, etc.

[0092] Specifically, the market accessibility calculation method is as follows:

[0093] Market accessibility refers to the distance, time, and monetary cost incurred when traveling to a series of market destinations. Due to the lack of market location data, two groups of locations are used as destinations. The first group uses cities and ports as market destinations instead, and the second group uses towns as market destinations instead. The market accessibility to the first and second groups of destinations is calculated according to the following formula. Finally, the market accessibility of each spatial location is determined by the maximum value of the two. The calculation formula is as follows:

[0094]

[0095] Among them, a ij represents the market accessibility from point i to destination j, d ij is the distance between point i and point j. For each location i, we need to calculate the distance to the nearest first and second group destinations j. ij value.

[0096] S j represents the importance of destination j, where the first group of destinations S j Assigned value 1, the second group of destinations S j Assign a value of 0.5.

[0097] v is a preset constant, and the assignment process is as follows:

[0098]

[0099] Among them, d *represents the distance from point i to the point where accessibility decreases most rapidly. Different constants were calculated for two groups of destinations: the inflection point corresponding to the fastest decline in distance for the first group of destination markets was set at 2 hours, and the inflection point for the second group of destination markets was set at 45 minutes. The distance d corresponding to the inflection point was calculated based on the assumed speed for calculating travel time provided in Table 1 below. * .

[0100] Table 1. Assumed speeds for calculating travel times

[0101] Through the "network analysis" module of the geographic information system platform, d ij and d * To calculate the market accessibility, use the above formula in the GIS's "Raster Calculator" module to create market accessibility spatial raster data. Distance to water sources, roads, and settlements is calculated using the GIS platform's "Distance Analysis" module, generating spatial raster surface data. Travel time to towns is calculated using the GIS platform's "Network Analysis" module.

[0102] Preferably, the afforestation distribution data is used as the response variable, the ecological environment data is used as the explanatory variable, and an improved maximum entropy reinforcement learning model is used to establish an ecological environment suitable area prediction model, and the ecological environment suitable area of ​​the target area is obtained according to the ecological environment suitable area prediction model, including:

[0103] The state vector is constructed using the ecological environment data of each afforestation distribution point. and uses the presence or absence of afforestation distribution records as labels ;

[0104] Will Randomly divided into training set and validation set;

[0105] In the improved maximum entropy reinforcement learning framework, the state vector Input graph convolution-transformer coupled hierarchical spatiotemporal state embedding network to obtain multi-scale embedding ;

[0106] Setting the action space ,in Indicates that the grid is judged to be an ecologically suitable area. It means it is judged as an unsuitable area;

[0107] Building a reward function , the reward function Including ecological prediction rewards and resource allocation cost items; ecological prediction rewards are the current action When , give positive rewards , otherwise the reward is 0; the resource allocation cost item is: from the development basic data of the second dimension, according to the rule model, determine whether the regional construction feasibility standard is met. If it is met, the cost item , otherwise the cost item ; The reward function The formula is: ;in, is the development constraint weight coefficient;

[0108] Construct a maximum entropy reinforcement learning model, which includes: a strategy network and value network The policy network Output in state Take action probability distribution of value network For estimating state-action pairs expected returns;

[0109] The Soft Actor-Critic strategy optimization method is used to train the model to maximize the objective function; the objective function is: ;in, For the The state of the time step; is the action selected by the strategy in this state; is the entropy weight coefficient, which is used to adjust the exploration and stability of the strategy;

[0110] After each update, the state-action-reward triplet Store the experience replay buffer and adaptively adjust the entropy weight coefficient based on the performance of the validation set ;

[0111] When the policy network After convergence, the model parameters are fixed to obtain the ecological environment suitable area prediction model;

[0112] Based on the ecological environment suitable zone prediction model, all locations to be determined in the target area Make predictions and calculate probability distributions ;

[0113] The probability distribution exceeds the set threshold All locations are aggregated and output as the ecological environment suitable area of ​​the target area.

[0114] Specifically, the core approach of this embodiment is to collect and stitch together ecological indicators such as temperature, precipitation, soil nutrients, and topography for each afforestation sample site, then feed them into a "hierarchical spatiotemporal embedding network." This network first uses graph convolution to interpret the spatial correlations between adjacent samples, then employs multi-head attention to capture climate-topography coupling at different scales. Ultimately, it outputs an ecological feature vector that simultaneously expresses local continuity and macroscopic gradients. This deep embedding eliminates the need for manual feature selection in subsequent models, avoiding the fragility and lack of context inherent in traditional pixel-by-pixel approaches.

[0115] During the reinforcement learning phase, the model only needs to determine whether a location is suitable for afforestation. Reward design adheres to the dual objectives of ecological accuracy and development feasibility: positive incentives are given when predictions align with historical afforestation records; if development criteria such as a location's market accessibility and transportation distance fall outside a preset acceptable range, a portion of the score is deducted as a penalty. The policy and value networks employ a soft actor-critic architecture, accelerating convergence through experience replay and early stopping. Exploration intensity is dynamically adjusted during training to ensure both expansion into new areas and robust adherence to prior knowledge.

[0116] Once the strategy network is fixed, a grid-by-grid inference is performed across the entire study area to generate an "ecological suitability probability map." Connected domains are merged and boundaries smoothed for locations with probabilities above a threshold, and patches below the minimum final patch threshold are removed to identify ecologically suitable areas. Subsequently, the implementation rule model is used to generate a "construction feasibility zone" grid. The two are then overlaid using an intersection operation, retaining only contiguous areas that meet both ecological and feasibility criteria. This ultimately creates a map of suitable areas for fast-growing, high-yield forestation that can be directly used for planning.

[0117] As an example, in the reinforcement learning training process of the ecological environment suitable area prediction model in this embodiment, in order to enable the model to not only have ecological judgment capabilities but also actively avoid areas with higher resource allocation costs, a "rule model" module is introduced when constructing the reward function. The rule model does not directly participate in spatial zoning, but serves as a penalty basis in the reward function. Its operation method is as follows: in each training iteration, the system reads the development condition index value of the current training sample and compares it with the defined intervals of the feasibility indicators of each regional construction item by item; if any indicator does not meet the preset interval conditions, the sample is marked as "not meeting the implementation constraint conditions", and this result is input into the reward function of the reinforcement learning algorithm as a deduction signal. As an auxiliary module embedded in the maximum entropy reinforcement learning framework, the rule model is an important mechanism for achieving the goal of "ecological-implementation dual constraints".

[0118] Optionally, in addition to using reinforcement learning, this embodiment can also use statistical analysis methods or other machine learning / artificial intelligence technologies to realize the association between tree species distribution data and the obtained ecological environment data of the distribution points, that is, using tree species distribution data as the response variable (the existence of the tree species is assigned a value of 1, and the non-existence of the tree species is assigned a value of 0), and using the environmental data value corresponding to the distribution point as the explanatory variable to establish a prediction model, and then use the established model to predict the ecological environment suitable for afforestation in the target afforestation area.

[0119] Statistical methods include, but are not limited to, the following types: generalized linear models, generalized additive models, multivariate adaptive regression splines, flexible discriminant analysis, and the like.

[0120] Machine learning / artificial intelligence algorithms include but are not limited to the following types: various algorithms based on decision trees and ensemble learning, various algorithms based on artificial neural networks, support vector machines, Bayesian, maximum entropy model MAXENT, etc.

[0121] Within software platforms such as R or Python, use multiple statistical methods or machine learning / artificial intelligence algorithms to build models. Then, compare and analyze the prediction accuracy of these models to identify the optimal model. This optimal model is then used to predict the ecologically suitable afforestation areas for the target tree species. Alternatively, use one of these methods to build and evaluate a model. If the model has high prediction accuracy, use this model directly to predict the ecologically suitable afforestation areas for the target tree species.

[0122] Feature selection during model training includes but is not limited to the following methods: variance inflation factor method, correlation coefficient method, information gain method based on information entropy theory, feature importance method, forward recursive addition, backward recursive deletion, genetic algorithm, etc.

[0123] Use cross-validation or random data splitting to separate model building data from model evaluation data.

[0124] Model evaluation indicators include but are not limited to the following indicators: root mean square error (RMSE), mean absolute prediction error (MAE), coefficient of determination (R 2 ) and cross entropy (meancrossentropy, MXE), etc.

[0125]

[0126]

[0127]

[0128]

[0129] Among them, p i and o i are the predicted and observed values ​​at position i (1, tree species appears; 0, tree species does not appear), The mean of the observed values; n is the dataset size, and p is the occurrence rate of the tree species in the model validation data.

[0130] Preferably, the development basic data is statistically analyzed for index intervals to construct a rule-based construction feasible area prediction model, and the construction feasible area of ​​the target area is obtained according to the construction feasible area prediction model, including:

[0131] Calculating the minimum value, maximum value, mean value and standard deviation of each indicator in the development basic data at the afforestation distribution points of the afforestation distribution data to form a statistical analysis table;

[0132] According to the statistical analysis table and the preset appropriate threshold reference table, determine the lower limit for each indicator With upper limit , get the index interval set ; k is the indicator index, ;

[0133] Construct a construction feasible area prediction model; the judgment rule of the construction feasible area prediction model is: if If both are established, the feasibility determination result of regional construction is ;otherwise ; For the grid to be evaluated Item index value, Indicates the construction feasible area, Indicates unsuitable area;

[0134] The judgment rule is called on all candidate grids in the target area one by one to generate a binary grid layer. ; Among them, the binary raster layer The value 1 indicates that the construction is feasible area pixel, binary raster layer The value of 0 indicates a pixel in an unsuitable area;

[0135] right Connected domain analysis is performed to retain connected patches whose area is not less than the pre-set first minimum threshold, and morphological opening operation is performed to smooth the boundaries to obtain the feasible construction area.

[0136] Preferably, the various indicators include: population density, night light index, gross domestic product, road distance, settlement distance, water source distance, city access time and market accessibility index.

[0137] Specifically, this embodiment first calculates the minimum, maximum, mean, and standard deviation of the eight indicators in the second-dimensional development basic data—population density, night light index, regional GDP, road distance, settlement distance, water source distance, urban access time, and market accessibility—at the afforestation distribution points, forming a statistical analysis table. On this basis, combined with the preset appropriate threshold reference standard, the upper and lower limit intervals are determined for each indicator, thus forming a complete set of indicator intervals. This interval not only reflects the distribution range of regional support conditions for existing afforestation activities, but also provides a basis for adjustable parameters, providing clear boundary conditions for subsequent suitability judgment rules.

[0138] Optionally, this embodiment uses a serial Boolean decision rule based on indicator intervals to construct a construction feasible zone prediction model. Specifically, for each grid to be evaluated within the target area, each of its development condition indicators is checked item by item to see if they fall within the corresponding interval. If all indicators meet the preset upper and lower limits, the area is determined to be construction feasible; otherwise, it is determined to be unsuitable. This rule model has clear logic and transparent parameters, requiring no model training, facilitating rapid adjustment of decision criteria to different regional conditions, and exhibiting good adaptability and interpretability.

[0139] For example, when constructing a feasible construction area, this embodiment uses a rule-based judgment method to independently judge each candidate location in the target area. Specifically, the system compares the eight development condition indicators of each location with the upper and lower limits of the interval obtained in the early stage based on afforestation sample statistics. Only when all indicators are within the corresponding interval range, the location is judged as a feasible construction area; if any indicator exceeds the range, it is judged as an unsuitable area. The rule uses explicit "all-satisfaction" logic as the judgment criterion to form a Boolean logic model with strong interpretability and flexible adjustment. This model is directly used for raster judgment and spatial zoning, and is the core method for constructing a feasible construction area layer.

[0140] Furthermore, this embodiment expresses the results of the above judgment in the form of a binary grid, where a value of one indicates that the area meets the implementation conditions, and a value of 0 indicates that it does not meet the conditions. In order to enhance the spatial operability of the region, a connected domain analysis is further performed to remove scattered patches with an area less than the first minimum threshold. Subsequently, the boundaries of the remaining patches are smoothed by morphological opening operations to remove burrs and holes, ensuring that the spatial pattern structure is regular and the boundaries are clear. The final output of the construction feasible area layer can be used as a basic map for subsequent overlapping analysis with the ecologically suitable area, providing a precise implementation constraint boundary for the formation of the final suitable afforestation area.

[0141] Furthermore, this embodiment extracts tree species distribution point data (i.e., the above-mentioned afforestation distribution data), and then statistically analyzes the interval ranges corresponding to these distribution points in each development condition indicator spatial raster data, and establishes a rule-based construction feasible area prediction model based on these interval ranges; finally, this prediction model is used to predict the afforestation areas in the target afforestation area that have the implementation conditions.

[0142]

[0143] In the formula, y is the output of the rule-based regional construction feasibility model, 1 represents an afforestation area with implementation conditions, and 0 represents an unsuitable afforestation area. Indicators a, b, ..., c are all the development condition indicators used. is the interval range of indicator a. Represents logical AND, meaning all conditions must be met at the same time.

[0144] Preferably, the ecological environment suitable zone is revised so that the spatial intersection of the revised ecological environment suitable zone and the construction feasible zone is consistent, and the final suitable afforestation area of ​​the target fast-growing and high-yield forest tree species is obtained, including:

[0145] In the geographic information system platform, the ecological environment suitable area and the construction feasible area are uniformly set to the same plane coordinate system;

[0146] Call the spatial overlay-intersection tool of the geographic information system platform to calculate the spatial intersection of the layers of the ecological environment suitable area and the construction feasible area, and generate the intersection vector layer ;

[0147] right Perform a dissolve operation to remove internal boundaries and form a single collection of spatial features;

[0148] Use the "area screening" tool to delete isolated patches in the single spatial feature set whose area is smaller than a preset second minimum threshold;

[0149] Perform morphological opening operation on the retained patches to smooth the boundaries and output the corrected ecological environment suitable area vector;

[0150] The revised ecological environment suitable area will be determined as the final suitable afforestation area.

[0151] Specifically, this example first sets the ecologically suitable area layer and the construction feasible area layer to a consistent projected coordinate system within the geographic information system platform to eliminate scale distortion and offset. The platform's built-in "Spatial Overlay-Intersection" function then intersects the two layers pixel by pixel, generating new intersection vector data. This intersection result retains only spatial units that meet both ecological and implementation requirements, laying the foundation for subsequent patch optimization.

[0152] To avoid duplicate boundaries within the intersecting layers, this example uses the "Merge" tool to consolidate the fragmented polygons into a single polygonal set. Then, using the "Area Filter" function, it removes isolated patches below a second minimum threshold. This threshold, customized for the final afforestation area, is larger than the first threshold used in the aforementioned implementation criteria screening phase, ensuring that the remaining patches are practical in terms of ecological connectivity, management, and mechanical operation.

[0153] For contiguous areas that remain after area screening, this embodiment uses morphological opening to smooth their boundaries, removing jagged edges and narrow corners and filling local voids, resulting in a final image with regular morphology and smooth edges. The resulting corrected image layer identifies the final suitable afforestation area for the target fast-growing, high-yield tree species and can be directly used in afforestation planning and subsequent bidding and tendering drawings, significantly improving the spatial accuracy and feasibility of afforestation site selection decisions.

[0154] Corresponding to the above method, this embodiment further provides a system for establishing afforestation areas for fast-growing and high-yield tree species, comprising:

[0155] Afforestation distribution data acquisition unit, used to acquire afforestation distribution data of target fast-growing and high-yield forest tree species;

[0156] A multi-dimensional environment-development data extraction unit, configured to extract first-dimensional ecological environment data and second-dimensional development basic data based on the afforestation distribution data;

[0157] an ecological environment suitable zone prediction unit, configured to use the afforestation distribution data as a response variable and the ecological environment data as an explanatory variable, establish an ecological environment suitable zone prediction model using an improved maximum entropy reinforcement learning model, and obtain an ecological environment suitable zone of a target area based on the ecological environment suitable zone prediction model;

[0158] A construction feasible area prediction unit is used to perform an index interval statistical analysis on the development basic data, construct a rule-based construction feasible area prediction model, and obtain a construction feasible area of ​​a target area according to the construction feasible area prediction model;

[0159] The range correction and result fusion unit is used to correct the range of the ecological environment suitable area so that the spatial intersection of the corrected ecological environment suitable area and the feasible construction area is consistent, thereby obtaining the final suitable afforestation area for the target fast-growing and high-yield forest tree species.

[0160] The beneficial effects of the present invention are as follows:

[0161] This invention is the first to introduce two dimensions, ecological environment data and development basic data, into the delineation of fast-growing and high-yield forest afforestation areas. The spatial intersection of the ecological environment suitable area and the construction feasible area is used as the final recommended afforestation area, which effectively overcomes the one-sided site selection problem caused by the existing technology that only uses single ecological factors such as climate and soil as the criteria, and significantly improves the scientificity and practicality of afforestation area delineation.

[0162] This invention innovatively adopts an improved maximum entropy reinforcement learning model for ecological suitability modeling. By constructing a reward function consisting of ecological prediction rewards and implementation constraint costs, the model automatically avoids areas with high implementation constraint costs while learning historical afforestation distribution patterns. Combined with an adaptive entropy adjustment strategy, the model can achieve migration generalization and robust decision-making in different regions, enhancing the flexibility and accuracy of suitability prediction.

[0163] This paper constructs a rule-based decision model based on development condition index intervals, which serves as an adjustable external constraint logic to ensure that the delineation results meet practical afforestation conditions such as practical accessibility, land use rationality, and infrastructure supportability. This model offers good interpretability and flexibility, allowing for rapid adjustments based on different regions and planning policies, significantly improving the policy compliance and operational feasibility of afforestation plans.

[0164] During the post-processing phase of determining suitable areas, this method incorporates spatial optimization steps such as connected domain analysis, minimum patch area screening, and morphological opening operations. These effectively eliminate isolated and fragmented patches, improving the coherence, regularity, and manageability of the patches. The resulting output of the suitable afforestation area layer exhibits smooth boundaries and a rational structure, making it directly applicable to practical forestry management applications such as planting planning, investment assessment, and land allocation.

[0165] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0166] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for establishing a fast-growing and high-yield forest tree species afforestation area, characterized in that: include: Obtain afforestation distribution data of target fast-growing and high-yield forest tree species; Based on the afforestation distribution data, extracting ecological environment data of the first dimension and development basic data of the second dimension respectively; Taking the afforestation distribution data as the response variable and the ecological environment data as the explanatory variable, an improved maximum entropy reinforcement learning model is used to establish an ecological environment suitable area prediction model, and the ecological environment suitable area of ​​the target area is obtained according to the ecological environment suitable area prediction model; Performing an index interval statistical analysis on the basic development data, constructing a rule-based construction feasible area prediction model, and obtaining a construction feasible area of ​​the target area according to the construction feasible area prediction model; The scope of the ecological environment suitable zone is revised so that the spatial intersection of the revised ecological environment suitable zone and the construction feasible zone is consistent, and the final suitable afforestation area for the target fast-growing and high-yield forest tree species is obtained.

2. The method for establishing a fast-growing and high-yield forest tree species afforestation area according to claim 1, characterized in that: The afforestation distribution data are obtained through remote sensing or vegetation map interpretation, forest resource inventory, published literature, online database data and forest resource inventory data.

3. The method for establishing a fast-growing and high-yield forest tree species afforestation area according to claim 1, characterized in that: Based on the afforestation distribution data, the first dimension of ecological environment data and the second dimension of development basic data are extracted respectively, including: Downloading ecological environment spatial raster data by searching network shared data within the coverage of the afforestation distribution data; the ecological environment spatial raster data includes: climate data, soil data and topographic data; Extract the corresponding indicator values ​​in the ecological environment spatial raster data to the afforestation distribution points of the afforestation distribution data through the extraction value to point tool of the geographic information system platform, and generate an ecological environment data table containing each afforestation distribution point in a field format; Merging fields and arranging the extracted ecological environment data table in a standard format to form ecological environment data of the first dimension for modeling and analysis; Obtain spatial vector location data of roads, settlements, water sources, cities, towns, and ports from OpenStreetMap and the National Basic Geographic Information Database, as well as spatial raster data of population density, gross domestic product, and night light index; Utilizing the distance calculation and network analysis tools of the geographic information system platform, the spatial distance and access time from each of the afforestation distribution points to the nearest road, settlement, water source, and city are calculated based on the spatial vector position data, and a market accessibility index is generated based on the market accessibility formula; Using the Extract Values ​​to Points tool of the geographic information system platform, extract corresponding index values ​​from the population density spatial raster data, the gross domestic product spatial raster data, and the nighttime light index spatial raster data to each of the afforestation distribution points, and combine the corresponding index values ​​with the spatial distance and access time, and the market accessibility index to generate a development basic data table; The fields of the development basic data table are sorted and the format is unified to form the development basic data of the second dimension.

4. The method for establishing a fast-growing and high-yield forest tree species afforestation area according to claim 3, characterized in that: The calculation formula of the market reachability formula is: ； Among them, a ij represents the market accessibility from afforestation distribution point i to destination j, d ij is the distance between point i and point j; for each location i, the distance from it to the nearest first and second group destinations j is ij Value; S j represents the importance of destination j, where the first group of destinations S j Assigned value 1, the second group of destinations S j The value is assigned as 0.5, v is a preset constant, and the assignment process is: ; Among them, d * It represents the distance from point i to the location where the accessibility drops the fastest.

5. The method for establishing a fast-growing and high-yield forest tree species afforestation area according to claim 3, characterized in that: The afforestation distribution data is used as a response variable, and the ecological environment data is used as an explanatory variable. An improved maximum entropy reinforcement learning model is used to establish an ecological environment suitable area prediction model, and the ecological environment suitable area of ​​the target area is obtained according to the ecological environment suitable area prediction model, including: The state vector is constructed using the ecological environment data of each afforestation distribution point. and uses the presence or absence of afforestation distribution records as labels ; Will Randomly divided into training set and validation set; In the improved maximum entropy reinforcement learning framework, the state vector Input graph convolution-transformer coupled hierarchical spatiotemporal state embedding network to obtain multi-scale embedding ; Setting the action space ,in Indicates that the grid is judged to be an ecologically suitable area. It means it is judged as an unsuitable area; Building a reward function , the reward function Including ecological prediction rewards and resource allocation cost items; ecological prediction rewards are the current action When , give positive rewards , otherwise the reward is 0; the resource allocation cost item is: from the development basic data of the second dimension, according to the rule model, determine whether the regional construction feasibility standard is met. If it is met, the cost item , otherwise the cost item The reward function The formula is: ;in, is the development constraint weight coefficient; Construct a maximum entropy reinforcement learning model, which includes: a strategy network and value network The policy network Output in state Take action probability distribution of value network For estimating state-action pairs expected returns; The Soft Actor-Critic strategy optimization method is used to maximize the objective function for model training; the objective function is: ;in, For the The state of the time step; is the action selected by the strategy in this state; is the entropy weight coefficient, which is used to adjust the exploration and stability of the strategy; After each update, the state-action-reward triplet Store the experience replay buffer and adaptively adjust the entropy weight coefficient based on the performance of the validation set ; When the policy network After convergence, the model parameters are fixed to obtain the ecological environment suitable area prediction model; Based on the ecological environment suitable zone prediction model, all locations to be determined in the target area are Make predictions and calculate probability distributions ; The probability distribution exceeds the set threshold All locations are aggregated and output as the ecological environment suitable area of ​​the target area.

6. The method for establishing a fast-growing and high-yield forest tree species afforestation area according to claim 1, characterized in that: Performing an index interval statistical analysis on the basic development data, constructing a rule-based construction feasible area prediction model, and obtaining the construction feasible area of ​​the target area according to the construction feasible area prediction model, including: Calculating the minimum value, maximum value, mean value and standard deviation of each indicator in the development basic data at the afforestation distribution points of the afforestation distribution data to form a statistical analysis table; According to the statistical analysis table and the preset appropriate threshold reference table, determine the lower limit for each indicator With upper limit , get the index interval set ; k is the indicator index, ; Construct a construction feasible area prediction model; the judgment rule of the construction feasible area prediction model is: if If both are established, the feasibility determination result of regional construction is ;otherwise ; For the grid to be evaluated Item index value, Indicates the construction feasible area, Indicates unsuitable area; The judgment rule is called on all candidate grids in the target area one by one to generate a binary grid layer. ; Among them, the binary raster layer The value 1 indicates that the construction is feasible area pixel, binary raster layer The value of 0 indicates a pixel in an unsuitable area; right Connected domain analysis is performed to retain connected patches whose area is not less than the pre-set first minimum threshold, and morphological opening operation is performed to smooth the boundaries to obtain the feasible construction area.

7. The method for establishing a fast-growing and high-yield forestry area according to claim 6, wherein: The indicators include: population density, night light index, gross domestic product, road distance, distance to settlements, distance to water sources, urban access time and market accessibility index.

8. The method for establishing a fast-growing and high-yield forestry area according to claim 1, wherein: The scope of the ecological environment suitable area is revised so that the spatial intersection of the revised ecological environment suitable area and the construction feasible area is consistent, and the final suitable afforestation area for the target fast-growing and high-yield forest tree species is obtained, including: In the geographic information system platform, the ecological environment suitable area and the construction feasible area are uniformly set to the same plane coordinate system; Call the spatial overlay-intersection tool of the geographic information system platform to calculate the spatial intersection of the layers of the ecological environment suitable area and the construction feasible area, and generate the intersection vector layer ; right Perform a dissolve operation to remove internal boundaries and form a single collection of spatial features; Use the "Area Filter" tool to delete isolated patches in the single spatial feature set whose area is smaller than a preset second minimum threshold; Perform morphological opening operation on the retained patches to smooth the boundaries and output the corrected ecological environment suitable area vector; The revised ecological environment suitable area will be determined as the final suitable afforestation area.

9. A system for establishing afforestation areas for fast-growing and high-yield tree species, characterized in that: include: Afforestation distribution data acquisition unit, used to acquire afforestation distribution data of target fast-growing and high-yield forest tree species; A multi-dimensional environment-development data extraction unit, configured to extract first-dimensional ecological environment data and second-dimensional development basic data based on the afforestation distribution data; an ecological environment suitable zone prediction unit, configured to use the afforestation distribution data as a response variable and the ecological environment data as an explanatory variable, establish an ecological environment suitable zone prediction model using an improved maximum entropy reinforcement learning model, and obtain an ecological environment suitable zone of a target area based on the ecological environment suitable zone prediction model; A construction feasible area prediction unit is used to perform an index interval statistical analysis on the development basic data, construct a rule-based construction feasible area prediction model, and obtain a construction feasible area of ​​a target area according to the construction feasible area prediction model; The range correction and result fusion unit is used to correct the range of the ecological environment suitable area so that the spatial intersection of the corrected ecological environment suitable area and the feasible construction area is consistent, thereby obtaining the final suitable afforestation area for the target fast-growing and high-yield forest tree species.

Images