Urban ecosystem carbon reserve prediction method, equipment and medium

Abstract

The invention discloses an urban ecosystem carbon reserve prediction method and device and a medium, and relates to the technical field of carbon reserve prediction, and the method comprises the steps: carrying out the scene simulation through a carbon reserve prediction model based on a fusion ecological data set, evaluating the different influences of a climate change scene and a land utilization change scene on the carbon reserve, and obtaining a prediction result; generating a multi-scenario carbon reserve prediction result; performing spatial division on the multi-scenario carbon reserve prediction result, identifying the trend and difference of carbon reserve changes in different regions, and generating carbon reserve spatial distribution; according to the carbon reserve space distribution, optimizing weight factors and parameter setting in the carbon reserve prediction model, and generating an optimized carbon reserve prediction model; and simulating carbon reserve changes in different scenes through the optimized carbon reserve prediction model, extracting a key trend of the carbon reserve changes, and generating a carbon reserve prediction report. According to the method, adaptive parameter adjustment is realized, the prediction precision is improved, and differentiated carbon management decisions are supported.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of carbon storage prediction, in particular to a city ecosystem carbon storage prediction method, device and medium. BACKGROUND

[0002] With the intensification of global climate change, the dynamic changes of carbon storage in urban ecosystems, as the key interface between human activities and natural ecological processes, have become an important indicator of the sustainable development capacity of cities. In recent years, the rapid development of remote sensing observation, geographic information systems (GIS), Internet of Things sensing network and big data analysis technology has provided an unprecedented data basis and method support for urban-scale carbon sink monitoring. Researchers have gradually shifted from relying on limited ground sample points to a comprehensive modeling approach that integrates multi-source remote sensing images, land use maps, climate sequences and social and economic statistical data.

[0003] The existing urban carbon storage prediction technology generally has two major shortcomings: first, the data system relied on by most models lacks systematicness and timeliness, and cannot effectively integrate high-dimensional, heterogeneous spatio-temporal big data, resulting in a disconnection between input features and real urban ecological processes, which weakens the response ability of the model to the changes in carbon sinks in the process of rapid urbanization; second, the existing prediction framework usually adopts a fixed parameter structure, ignores the sensitivity differences of different urban functional areas to carbon storage driving factors, and lacks model self-optimization ability based on spatial feedback mechanism, so it is difficult to support fine and differentiated scenario simulation and policy evaluation. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a city ecosystem carbon storage prediction method to solve the problems of insufficient fusion of multi-source heterogeneous spatio-temporal data and lack of spatial optimization of model parameters, which leads to imperfect and inaccurate prediction methods.

[0006] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a city ecosystem carbon storage prediction method, which comprises, Collecting city ecological data and spatio-temporal big data and preprocessing to generate a fusion ecological data set; Based on the fusion ecological data set, scenario simulation is performed through a carbon storage prediction model to evaluate the different effects of climate change scenarios and land use change scenarios on carbon storage, and to generate multi-scenario carbon storage prediction results; Divide the multi-scenario carbon storage prediction results into different regions, and identify the trends and differences of carbon storage changes in different regions to generate a carbon storage spatial distribution; According to the spatial distribution of carbon storage, the weight factor and parameter setting in the carbon storage prediction model are optimized, and an optimized carbon storage prediction model is generated; The optimized carbon storage prediction model is used to simulate the change of carbon storage under different scenarios, and the key trends of the change of carbon storage are extracted, and a carbon storage prediction report is generated.

[0007] As a preferred scheme of the urban ecosystem carbon storage prediction method, the urban ecological data includes land use data, climate data, vegetation cover data, urbanization process data and population density data. The spatio-temporal big data includes historical ecological data, remote sensing data, social and economic data, urban infrastructure data and environmental pollution data.

[0008] As a preferred scheme of the urban ecosystem carbon storage prediction method, the generation of the fusion ecological data set is as follows, The urban ecological data and the spatio-temporal big data are spatially aligned and temporally aligned to generate a spatio-temporally aligned ecological data set; The spatio-temporally aligned ecological data set is cleaned, formatted and unified in dimension to generate a fusion ecological data set.

[0009] As a preferred scheme of the urban ecosystem carbon storage prediction method, based on the fusion ecological data set, the carbon storage prediction model is used for scenario simulation to evaluate the different effects of climate change scenarios and land use change scenarios on carbon storage, and a multi-scenario carbon storage prediction result is generated, and the steps are as follows, Based on the fusion ecological data set, the urban area is divided into scenarios, and scenario parameters and change values are set to generate climate change scenarios and land use change scenarios; The fusion ecological data set is input into the carbon storage prediction model, and scenario simulation is performed for different climate change scenarios and land use change scenarios, and the scenario simulation result is output; The scenario simulation results are summarized, and the key trends and influencing factors are extracted to generate a multi-scenario carbon storage prediction result.

[0010] As a preferred scheme of the urban ecosystem carbon storage prediction method, the multi-scenario carbon storage prediction result is spatially divided, and the trends and differences of carbon storage changes in different regions are identified to generate a carbon storage spatial distribution, and the steps are as follows, Based on the multi-scenario carbon storage prediction result, the spatial division data set is generated according to the urban planning type; The carbon storage change trends in different regions in the spatial division data set are extracted, and the differences and reasons of the carbon storage changes between regions are identified to generate a carbon storage change trend. Based on the carbon storage change trend, the relative position of different urban planning areas in the carbon storage change is evaluated, and the carbon storage spatial distribution is generated.

[0011] As a preferred scheme of the urban ecosystem carbon storage prediction method, wherein: according to the carbon storage spatial distribution, the weight factor and the parameter setting in the carbon storage prediction model are optimized, and an optimized carbon storage prediction model is generated, the steps are as follows, The carbon storage spatial distribution is identified in detail, the main driving factors of the carbon storage change in different regions are extracted, and a key influence factor set is generated; According to the key influence factor set, the actual influence degree of the carbon storage change is adjusted, the parameter setting is optimized through a regression analysis algorithm, and an optimized weight factor and parameter setting are generated; Based on the optimized weight factor and parameter setting, the influence factors in the carbon storage prediction model are reweighted, and the parameter setting of the carbon storage prediction model is adjusted, and an optimized carbon storage prediction model is generated.

[0012] As a preferred scheme of the urban ecosystem carbon storage prediction method, wherein: the carbon storage change under different scenarios is simulated through the optimized carbon storage prediction model, and the steps are as follows, Based on the optimized carbon storage prediction model, different climate change scenarios and land use change scenarios are selected, and detailed simulation parameters are set, and the set scenarios and simulation parameters are generated; The set scenarios and simulation parameters are input into the optimized carbon storage prediction model, the carbon storage change is simulated, and a carbon storage change simulation report is generated.

[0013] As a preferred scheme of the urban ecosystem carbon storage prediction method, wherein: the key trend of the carbon storage change is extracted, and a carbon storage prediction report is generated, and the steps are as follows, According to the carbon storage change simulation report, the time series data and the spatial distribution characteristics of the carbon storage change are extracted, and a carbon storage change key trend and difference data set is generated; The carbon storage change key trend and difference data set and the carbon storage change simulation results under different scenarios are integrated into a carbon storage prediction report.

[0014] In the second aspect, the present application provides a computer device, comprising a memory and a processor, the memory stores a computer program, wherein: the computer program is executed by the processor to realize any step of the urban ecosystem carbon storage prediction method of the first aspect of the present application.

[0015] In a third aspect, the present application provides a computer readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements any step of the urban ecosystem carbon storage prediction method according to the first aspect of the present application.

[0016] The present application has the beneficial effects that: through spatiotemporal alignment and fusion of multi-source data, high-consistency ecological data set construction is achieved, model input quality is improved, and the ability to depict urban dynamic carbon sinks is enhanced; through spatial feedback optimization of model weights, parameter adaptive adjustment is achieved, prediction accuracy is improved, and differentiated carbon management decisions are supported. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Fig. 1 The flowchart of the urban ecosystem carbon storage prediction method.

[0019] Fig. 2 The flowchart of the fusion ecological data set generation.

[0020] Fig. 3 The flowchart of generating multi-scenario carbon storage prediction results.

[0021] Fig. 4 The flowchart of generating carbon storage spatial distribution. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification.

[0023] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the concept of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0024] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. In this specification, "in one embodiment" appearing in different places does not mean the same embodiment, nor is it an independent or selective embodiment that excludes other embodiments.

[0025] REFERENCE Figs. 1-4For an embodiment of the present application, the embodiment provides a method for predicting carbon storage in urban ecosystems, comprising the following steps: S1: Collect urban ecological data and spatio-temporal big data and preprocess to generate a fusion ecological dataset.

[0026] S1.1: Urban ecological data includes land use data, climate data, vegetation cover data, urbanization process data, and population density data.

[0027] Further, land use data is obtained through remote sensing images or satellite data, recording and identifying the use of different types of land; climate data is usually obtained from weather stations, through the collection of data over a period of time for time series analysis, and according to different seasons, years, etc. Time dimension to dynamically track climate change; vegetation cover data is obtained through remote sensing images and satellite monitoring, using vegetation index to quantitatively analyze vegetation change; urbanization process data is usually combined with geographic information nodes and urban and rural planning data to record the development of different stages of the city; population density data comes from statistical departments, recording the number of permanent residents in each administrative district or grid node, reflecting the impact of population growth on land use and resource allocation by analyzing population distribution and density changes.

[0028] S1.2: Spatio-temporal big data includes historical ecological data, remote sensing data, social and economic data, urban infrastructure data, and environmental pollution data.

[0029] Further, historical ecological data is mainly obtained from long-term environmental monitoring records, usually provided by environmental monitoring agencies, and summarized annually or quarterly. Remote sensing data is collected through satellite images or unmanned aerial vehicles, using remote sensing image processing methods to extract relevant data for spatial analysis and time series analysis. Social and economic data is usually obtained from statistical bureaus or research institutions, with data collection periods generally being annual or quarterly. Urban infrastructure data is usually provided by planning departments, covering various types of urban construction and facility operation, and can reflect the potential impact of urban development on carbon storage. Environmental pollution data is mainly obtained from monitoring networks of environmental protection departments and meteorological agencies, usually recorded in hours or days, reflecting the impact of environmental pollution on the ecosystem.

[0030] S1.3: Spatially align and temporally align the urban ecological data with the spatio-temporal big data to generate a spatio-temporally aligned ecological dataset.

[0031] Further, a unified spatial coordinate is selected for urban ecological data and spatio-temporal big data from different sources to unify the spatial positions of various types of data. The geographic coordinates of land use data, climate data, remote sensing data, and data sets are converted to a unified coordinate system to ensure that the geographic positions of all data can be matched with each other. During spatial alignment, if the data sources are different, interpolation or resampling operations are performed to ensure that the spatial accuracy of each data set is consistent. For each type of data, time matching is performed according to a unified time scale. Different data sets may have different time intervals or time formats, and the time ranges of all data are uniformly processed to ensure that all data have corresponding relationships in the same time dimension. Time format conversion and time point matching ensure that each data point corresponds to the correct time dimension. Through spatial alignment and time matching operations, all data sets are consistently processed in space and time to form a spatio-temporal aligned ecological data set.

[0032] S1.4: Data cleaning, formatting, and unified dimension of the spatio-temporal aligned ecological data set to generate a fused ecological data set.

[0033] Further, through the data cleaning process, missing values, outliers, and repeated values existing in the spatio-temporal aligned ecological data set are removed; missing values are processed by interpolation method or deletion of incomplete data records for filling or rejection; outliers are detected and corrected by statistical methods to ensure the accuracy and consistency of the data. Data formatting operations include unifying various types of data into a standard format to ensure consistency in field names and data types of data from different sources, unifying different source date formats into the same standard time format to ensure consistency of data in the time dimension. When unifying the dimensions of data from different sources, standardization methods are used to scale different characteristics such as air temperature, precipitation, and land use type to a comparable numerical range, eliminating scale differences between different data sources, and ensuring the comparability and consistency of the data. After completing data cleaning, formatting, and dimension unification, a fused ecological data set is generated.

[0034] S2: Based on the fused ecological data set, scenario simulation is performed through a carbon storage prediction model to evaluate the different effects of climate change scenarios and land use change scenarios on carbon storage, and to generate multi-scenario carbon storage prediction results.

[0035] S2.1: Based on the fused ecological data set, scenarios are divided for the urban area, and scenario parameters and change values are set to generate climate change scenarios and land use change scenarios.

[0036] Further, according to the geographical characteristics, land use types and future development trends of urban areas, the areas are divided into multiple different sub-areas, each sub-area has different ecological characteristics (such as vegetation coverage and soil type and health) and carbon storage change characteristics (such as carbon storage change rate and carbon storage accumulation and fluctuation). When the scenario is divided, the spatial distribution characteristics of the city (such as the carbon storage difference between the core area and the surrounding area of the city and the ecological fracture zone in the region) need to be considered, and the climate conditions, land use status and future planning factors of different regions are considered to determine the division boundary and category of each sub-region. The scenario parameter setting includes selecting the key driving factors of climate change, such as temperature change amplitude, precipitation change trend and extreme weather event frequency, and setting reasonable change value range for each driving factor, for example: temperature rise change (1-4 degrees) and precipitation change (increase by 10%, decrease by 15%). The land use change scenario is set according to the current and future land use development planning, and the scenarios of urban expansion, farmland development and forest restoration are determined, and different change rates and spatial distribution are set for each scenario to generate climate change scenarios and land use change scenarios.

[0037] S2.2: Input the fusion ecological data set into the carbon storage prediction model, and simulate the scenario for different climate change scenarios and land use change scenarios, and output the scenario simulation results.

[0038] Further, read the land use type, climate index, vegetation coverage, urbanization level, population density and other multi-dimensional ecological factor information in the fusion ecological data set, and standardize the data format according to the input layer structure of the carbon storage prediction model. For different types of data, use a unified dimension conversion method to convert them into a format compatible with other data sets, so that each ecological factor corresponds to the model input node under the unified dimension and spatio-temporal scale; after input, according to the climate change scenario and land use change scenario, a plurality of scenario simulation running tasks are constructed respectively, each task corresponds to a combination of climate parameters and land parameters, and through the feature extraction layer of the carbon storage prediction model, the spatio-temporal features are extracted through multi-dimensional convolution operation, the time series data are processed through recurrent neural network (RNN), and the nonlinear response relationship between ecological factors and carbon storage change is obtained; during the simulation process, the carbon storage prediction model iterates the dynamic evolution value of carbon storage under each scenario according to the time sequence, and outputs the carbon storage estimation result set of different time nodes through the time sequence recursive structure of the correlation calculation layer; in the output layer, the estimation results of each time node are aggregated into the spatio-temporal distribution and change trend of the overall carbon storage through weighted aggregation; the simulation results of all climate change scenarios and land use change scenarios are uniformly recorded and time indexed, and the scenario simulation simulation results are integrated.

[0039] It should be noted that the carbon storage prediction model is constructed through a multi-source ecological data driven modeling process, and the structure is based on a multi-layer nonlinear feature mapping architecture, which consists of an input layer, a feature extraction layer, a correlation calculation layer, and an output layer. The input layer receives the multi-dimensional ecological factor information in the fused ecological data set and standardizes the coding under the unified dimension and spatial scale, ensuring the consistency and comparability of the data; the feature extraction layer extracts the coupling characteristics of ecological factors in time and space through the combination of multi-dimensional convolution and time series recursive structure, capturing the nonlinear response of climate and land use change to carbon storage dynamics, and these extracted features provide spatio-temporal information for the calculation of the next layer; the correlation calculation layer uses a combination algorithm based on multivariate regression and attention mechanism to learn the interaction weight relationship between each ecological factor, analyze the influence of different features on carbon storage changes, and assign weights to each factor; the output layer obtains the spatio-temporal distribution and change trend estimates of carbon storage through weighted aggregation, and the output results of the previous layers are integrated to generate the final prediction results. During the construction process, the number of layers, the number of nodes, and the regularization parameters are determined through cross-validation to balance the complexity and prediction accuracy, and to ensure the generalization performance under multiple scenario inputs.

[0040] In the training phase, first, normalize and synchronize the time series of various input features, and perform local interpolation repair on missing values and add data quality labels; then, optimize the parameters of the carbon storage prediction model through a supervised learning strategy, and use the weighted mean squared error between the measured and predicted values of carbon storage as the optimization objective; in the training process, input dynamic learning rate adjustment and early stopping mechanism are used to prevent overfitting, and hierarchical sampling is used to ensure the balanced sample proportion of different urban functional areas; through multiple rounds of iterative training, the prediction error and stability indicators are obtained on the training set and validation set respectively, until the error converges below the weighted mean squared error threshold (example range: 0.50%-2.00%, defined as the allowable error range of the predicted value of carbon storage relative to the reference value, where 0.50% is set according to the resolvable lower bound of data measurement uncertainty, spatio-temporal matching error, and numerical noise, and 2.00% is set according to the marginal improvement inflection point of the error convergence curve, model generalization performance, and computational cost constraints), the training is stopped, and finally a carbon storage prediction model that can stably represent the dynamic rules of urban ecological system carbon storage is obtained.

[0041] S2.3: Summarize the results of scenario simulation, extract key trends and influencing factors, and generate multi-scenario carbon storage prediction results.

[0042] Further, the simulation results generated under different climate change scenarios and land use change scenarios are aligned and integrated according to time series and spatial region indexes, a scenario simulation result summary table is established, and the corresponding carbon storage evolution sequence is extracted according to the simulation task identifier; in the scenario simulation result summary table, the carbon storage change values of the same urban area under different scenarios are counted to form a carbon storage change rate matrix, the long-term trend and phase fluctuation characteristics are extracted by time series analysis method, and the trend slope and fluctuation intensity are obtained, combined with the climate change scenario parameter set and the land use change scenario parameter set, the carbon storage change rate matrix is subjected to correlation analysis and regression analysis to identify the driving factor with the strongest impact on carbon storage change; according to the driving factor analysis result, the growth or decay trend of carbon storage under each scenario is marked, and a key trend feature table and a main influencing factor table are generated; after all the results are integrated, the key trend feature table and the main influencing factor table are fused, and output as multi-scenario carbon storage prediction results according to time index and spatial region index.

[0043] S3: Spatially divide the multi-scenario carbon storage prediction results and identify the trends and differences in carbon storage changes in different regions to generate a carbon storage spatial distribution.

[0044] S3.1: Based on the multi-scenario carbon storage prediction results, spatially divide according to city planning types to generate a spatial division dataset.

[0045] Further, read the spatial coordinate information, city land attribute information, and planning type information in the multi-scenario carbon storage prediction results, make type attribution judgments for the spatial nodes within the city range according to the city overall planning map and the land use classification standard, and map different city planning types (such as residential areas, industrial areas, green areas, water areas, and transportation areas) to regional grids in a unified spatial coordinate system; during spatial division, perform coordinate matching between city planning types and the spatial distribution of multi-scenario carbon storage prediction results through spatial overlay algorithms, perform boundary correction and topology correction on overlapping regions to ensure seamless connection of division region boundaries and no spatial overlap; according to the city planning boundary parameter set, set accurate division region boundaries for each city planning type region, the resolution and accuracy of the boundary line are determined according to the city data scale, and the division region number and the corresponding carbon storage prediction value are indexed and bound to establish a one-to-one correspondence between city planning types and carbon storage prediction data. Finally, all division results are summarized into a spatial division dataset.

[0046] It should be noted that city planning types refer to functional regional categories divided according to city overall planning and land use classification standards, including residential areas, industrial areas, green areas, water areas, and transportation areas, etc., which are used to reflect the ecological function differences of different land properties in the city spatial structure.

[0047] S3.2: Extract the carbon storage change trends in different regions of the spatial partition dataset, identify the differences and reasons for carbon storage changes between regions, and generate carbon storage change trends.

[0048] Further, read the carbon storage prediction values and time series information corresponding to each city planning type region in the spatial partition dataset, sort the carbon storage change values of each region at different time nodes by time index, and construct a regional carbon storage change sequence set. According to the time series, perform trend fitting and fluctuation feature extraction on the carbon storage change sequence of each region: obtain the carbon storage change rate, growth or decay interval length, and fluctuation frequency, perform smoothing and difference calculation through sliding window regression and moving average algorithm, determine the local linear trend of the data using sliding window regression and perform weighted average, thereby removing short-term fluctuations and identifying long-term trends. Subsequently, use the moving average algorithm to smooth the time series data, reduce the influence of random noise, and identify the short-term fluctuation characteristics of carbon storage; perform difference analysis on the carbon storage change trends of different regions, obtain the difference matrix of carbon storage changes between regions through comparative analysis, to determine the distribution relationship between carbon storage growth and decay regions, and combine the city planning type information and environmental characteristic parameter set in the spatial partition dataset to attribute the driving reasons for carbon storage change differences and identify the main factor set affecting the differences (such as land use intensity, vegetation coverage, or climate change rate); unify and label the trend fitting results, difference matrix, and driving factor set to generate carbon storage change trends.

[0049] It should be noted that the environmental characteristic parameter set refers to a set of parameters including but not limited to climate change rate, land use intensity, vegetation coverage, urbanization degree, social and economic activities, etc. The environmental characteristic parameter set is derived from the integrated ecological dataset, integrating climate indicators, land use indicators, vegetation coverage data, and other related environmental characteristic indicators. Environmental pollution indicators and social and economic indicators can also be part of this set to more comprehensively analyze the driving factors of carbon storage changes.

[0050] S3.3: Based on the carbon storage change trends, evaluate the relative positions of different city planning regions in carbon storage changes, and generate carbon storage spatial distribution.

[0051] Further, read the carbon storage change rate, trend direction and driving factor weight information of each city planning area in the carbon storage change trend, and establish a region index table according to the region number in the spatial division data set; according to the numerical distribution of the carbon storage change rate of each region, obtain the mean and standard deviation of all regions, and grade the regions, define the regions with carbon storage change rate higher than the mean value by one standard deviation as carbon storage growth area, define the regions with carbon storage change rate lower than the mean value by one standard deviation as carbon storage decay area, and define the remaining regions as carbon storage stable area; after the grade division is completed, the distance matrix and carbon storage change gradient between adjacent regions are obtained based on the spatial index relationship of the regions, and the relative difference of carbon storage change between different regions is analyzed; through the joint analysis of the distance matrix and the carbon storage change gradient, the mutual influence of the carbon storage change of adjacent regions is analyzed, the relative position of the carbon storage change in the carbon storage distribution is determined, the concentration area, the transition area and the edge area of the carbon storage change are identified, and the relative position of different city planning areas in the overall carbon storage change is determined; the region grade division result, the spatial adjacency relationship and the mutual influence pattern are unified and integrated to generate the carbon storage spatial distribution.

[0052] S4: According to the carbon storage spatial distribution, the weight factor and parameter setting in the carbon storage prediction model are optimized to generate an optimized carbon storage prediction model.

[0053] S4.1: Detailed influence factor identification is performed on the carbon storage spatial distribution, the main driving factors of carbon storage change in different regions are extracted, and a key influence factor set is generated.

[0054] Further, read the carbon storage change rate and the carbon storage change gradient between adjacent regions of different city planning areas in the carbon storage spatial distribution, and combine the climate index set, the land use index set, the vegetation coverage index set and the social and economic index set in the ecological data set to establish a corresponding relationship table of regional attributes and carbon storage change indicators; in the corresponding relationship table, the influence coefficients of each regional attribute on the carbon storage change rate are obtained by multiple regression analysis and partial least squares regression method, and an influence coefficient matrix is constructed; according to the numerical size order in the influence coefficient matrix, the driving factor set with dominant effect on carbon storage change is identified, the factors with influence coefficient higher than the upper limit of the influence coefficient threshold interval are marked as main driving factors, and the factors with influence coefficient within the influence coefficient threshold interval are marked as secondary driving factors; after the driving factor identification, the interaction effect and independent contribution rate between each driving factor are obtained by correlation analysis and variance decomposition, so as to determine the difference of different driving factors in each city planning area; combine the city planning type and the regional attribute characteristics to classify, arrange and merge the main driving factors, and generate a key influence factor set.

[0055] It should be noted that the impact coefficient threshold interval example ranges from 0.30 to 0.60, which is defined as the value range for distinguishing the main and secondary driving factors in the overall value distribution of the impact coefficient matrix, wherein 0.30 is set according to the robust lower side quantile of the distribution and the false positive rate control requirement, and 0.60 is set according to the robust upper side quantile of the distribution and the marginal inflection point of the cross-validation performance improvement.

[0056] S4.2: Adjust the weight factor according to the actual influence degree of the carbon storage change set by the key influencing factors, and optimize the parameter setting by the regression analysis algorithm to generate the optimized weight factor and parameter setting.

[0057] Further, read the impact coefficients of the main driving factors and the secondary driving factors in the key influencing factor set, and preliminarily allocate the weight proportions of different driving factors according to the impact coefficient threshold interval to establish an impact factor weight mapping table; in the impact factor weight mapping table, by comparing the regional carbon storage change rate in the carbon storage change trend analysis result with the impact coefficients of the driving factors, a weight adjustment coefficient matrix is calculated to proportionally correct the driving factors with high or low weights; after the weight adjustment is completed, the parameter set of the carbon storage prediction model is optimized by the regression analysis algorithm, the sensitivity values of each parameter in the output error of the carbon storage prediction model are obtained by the multivariate nonlinear regression and stepwise regression method, and the parameter setting is updated according to the sensitivity values; the optimized weight factor and parameter setting are subjected to consistency check to ensure that the weight allocation sum satisfies the normalization constraint condition, and finally the optimized weight factor and parameter setting are generated.

[0058] It should be noted that the regression analysis algorithm here refers to analyzing the influence of each driving factor on the carbon storage change by the multivariate nonlinear regression and stepwise regression method, and adjusting the model parameters according to the sensitivity values to optimize the prediction results.

[0059] The formula for calculating the weight adjustment coefficient matrix is:

[0060] wherein, represents the weight adjustment coefficient matrix, represents a nonlinear correction function for smoothing extreme deviation and limiting adjustment amplitude, represents the regional carbon storage change rate matrix, represents the mean value of the regional carbon storage change rate, represents the standard deviation of the regional carbon storage change rate, represents the element multiplication operation for calculating the interaction deviation of the regional change rate and the driving factor impact coefficient, represents the driving factor impact coefficient matrix, represents the mean value of the driving factor impact coefficient, a standard deviation of the driving factor influence coefficient.

[0061] It should be noted that the regional carbon storage rate of change matrix represents the rate of change of carbon storage in different regions, reflects the speed of carbon storage change in each region at different time nodes, and is derived from the carbon storage trend analysis result. The driving factor influence coefficient matrix represents the influence degree of each driving factor (such as climate, land use, etc.) on the change of carbon storage, which is derived from the weight analysis of each main and secondary driving factor in the key influence factor set.

[0062] S4.3: Based on the optimized weight factor and parameter setting, the influence factors in the carbon storage prediction model are reweighted, and the parameter setting of the carbon storage prediction model is adjusted to generate an optimized carbon storage prediction model.

[0063] Further, read the weight distribution ratio, sensitivity correction value and parameter adjustment coefficient in the optimized weight factor and parameter setting, replace the original influence factor weight mapping table in the carbon storage prediction model with the optimized weight mapping table; according to the optimized weight factor, perform weight redistribution operation on each driving factor of the input layer of the carbon storage prediction model, obtain the weighted contribution value of each driving factor and update the input layer feature matrix; after the weight redistribution is completed, based on the parameter adjustment set in the optimized parameter setting, update the regression parameter, regularization parameter and iteration control parameter of the carbon storage prediction model, and through iterative correction residual minimization in the updating process, make the fitting error of the carbon storage prediction model output and the historical carbon storage sample lower than the precision threshold; after the parameter adjustment is completed, perform consistency test and convergence verification on the optimized carbon storage prediction model, ensure that the carbon storage prediction results under different scenarios are stable and consistent with the trend logic, and finally generate the optimized carbon storage prediction model.

[0064] It should be noted that the precision threshold example range (for example, 0.50%–2.00%) is defined as the upper limit of the relative error of the carbon storage prediction result relative to the historical sample in the verification stage, 0.50% is set according to the lower bound of the reachable error of data measurement uncertainty and scenario noise, and 2.00% is set according to the marginal improvement inflection point of the error convergence curve and the calculation cost constraint.

[0065] S5: Simulate the change of carbon storage under different scenarios through the optimized carbon storage prediction model, and extract the key trends of carbon storage change to generate a carbon storage prediction report.

[0066] S5.1: Based on the optimized carbon storage prediction model, select different climate change scenarios and land use change scenarios, and set detailed simulation parameters to generate the set scenarios and simulation parameters.

[0067] Further, read the input interface information and parameter dependent structure of the optimized carbon storage prediction model, extract the climate change scenario parameter set and land use change scenario parameter set from the scenario parameter library, and check the rationality and boundary consistency of the value range of each scenario parameter set; In the scenario combination stage, construct the combination matrix of climate change scenarios and land use change scenarios, each combination contains a set of corresponding climate parameters and land parameters; After the generation of the combination matrix, set the detailed simulation parameter set according to the operation requirements of the carbon storage prediction model, including time step, iteration number and convergence judgment threshold, and verify the consistency of the parameter set to ensure the logical compatibility between parameters; The verified climate change scenario, land use change scenario and simulation parameter set are bound into a unified scenario simulation configuration record to generate the set scenario and simulation parameters.

[0068] It should be noted that the scenario parameter library is a standardized parameter set warehouse provided for scenario combination and simulation configuration, which compiles climate change scenario parameter sets and land use change scenario parameter sets based on historical observation data, remote sensing inversion results and urban planning constraints, and sets through benchmark period calibration, unit and dimension unification, value range and boundary consistency rules, and time and space resolution specifications.

[0069] The convergence judgment threshold example range is (for example 0.0001-0.01), which is defined as the relative change allowed upper limit of the carbon storage prediction objective function and parameter update norm in the adjacent two iterations, 0.0001 is set according to the numerical noise level and double-precision distinguishability, and 0.01 is set according to the accuracy requirement and the marginal improvement inflection point of the benefit-cost curve.

[0070] S5.2: Input the set scenario and simulation parameters into the optimized carbon storage prediction model to simulate the carbon storage change and generate a carbon storage change simulation report.

[0071] Further, read the climate change scenario parameter set, land use change scenario parameter set and time control parameter set in the scenario and simulation parameter setting completed, and map each parameter to the input layer of the optimized carbon storage prediction model according to the input specification; in the input stage, perform normalization and scale correction on the parameter set to ensure that the climate factors, land factors and time steps are iterated at a unified scale; in the simulation execution stage, the optimized carbon storage prediction model solves the carbon storage change equation through step-by-step iteration according to the weight factor and parameter setting, obtains the regional carbon storage value and total carbon storage change at each time step, and judges whether the residual error is less than the convergence judgment threshold at the end of each iteration; if the convergence judgment threshold is not reached, continue to update the parameter set and repeat the iteration until the convergence judgment threshold is met; after the simulation is completed, match the time series carbon storage results under each scenario combination with the spatial index information, and summarize them into a complete carbon storage change sequence set to generate a carbon storage change simulation report.

[0072] It should be noted that the convergence judgment threshold is in the range of (i.e. 0.0001-0.01), which is defined as the relative change rate of the objective function or parameter update allowed in the adjacent two iterations of the carbon storage prediction model, wherein 0.0001 is set according to the minimum distinguishable error and floating point stability of numerical calculation, and 0.01 is determined according to the marginal improvement inflection point of the model error convergence curve and the balance condition of calculation efficiency.

[0073] S5.3: According to the carbon storage change simulation report, extract the time series data and spatial distribution characteristics of carbon storage change, and generate a carbon storage change key trend and difference data set.

[0074] Further, read the time series carbon storage data corresponding to different climate change scenarios and land use change scenarios in the carbon storage change simulation report, align the data according to the time index and spatial index, and establish a carbon storage change comparison table; in the carbon storage change comparison table, obtain the annual change rate, growth rate and cumulative change of carbon storage under each scenario, and identify the long-term trend and phase fluctuation characteristics of carbon storage change through time series difference analysis method; after the trend analysis is completed, difference comparison is performed on the carbon storage change amplitude between scenarios to generate a scenario difference matrix, and the contribution degree of climate change scenarios and land use change scenarios to carbon storage difference is judged by variance analysis method; according to the difference matrix and trend analysis result, the region and time interval most sensitive to carbon storage change are selected, and the key trend index set and difference feature set are extracted; combine the key trend index set and difference feature set to analyze the change direction, consistency degree and difference source of carbon storage change under different scenarios, and generate a carbon storage change key trend and difference data set.

[0075] S5.4: integrate the carbon storage change key trend and difference dataset with the carbon storage change simulation results under different scenarios into a carbon storage prediction report.

[0076] Further, the key trend indicators set and the difference feature set in the carbon storage change key trend and difference dataset are read, and time series carbon storage data, regional distribution information, and a scenario parameter set in the carbon storage change simulation results under different climate change scenarios and land use change scenarios are extracted; the data of different sources are uniformly indexed and matched according to the scenario identifier, the carbon storage change results under the same scenario are associated and integrated with the corresponding trend analysis data, and a scenario result summary table is formed; in the scenario result summary table, the carbon storage change curve characteristics, growth rate, and change period of each scenario are summarized and statistically analyzed, and the carbon storage sustained growth scenario, the stage fluctuation scenario, and the sustained decay scenario are distinguished according to the classification results of the key trend indicators set; the average difference, the extreme difference, and the trend consistency ratio of the carbon storage change between different scenarios are obtained through the difference feature set, so as to quantify the difference degree between scenarios; after the data integration is completed, the scenario summary results are logically verified and integrity checked to ensure that the time series is continuous, the region is correctly corresponding, and the parameter consistency meets the requirements; after the verification is passed, the carbon storage change characteristics, key trends, and difference analysis results of each scenario are arranged in order of time index and space index, and are uniformly compiled into a carbon storage prediction report.

[0077] The embodiment also provides a computer device suitable for the urban ecosystem carbon storage prediction method, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the urban ecosystem carbon storage prediction method proposed in the above embodiment.

[0078] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved through WIFI, an operator network, NFC (near field communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or can be a key, trackball, or touchpad arranged on the shell of the computer device. In addition, the input device can be an external keyboard, touchpad, or mouse, etc.

[0079] The embodiment also provides a storage medium on which a computer program is stored, the program being executed by a processor to implement the method for predicting carbon storage of urban ecosystems as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk, or an optical disk.

[0080] To sum up, the present application achieves high-consistency ecological data set construction by spatiotemporal alignment and fusion of multi-source data, improves model input quality, enhances the ability to depict urban dynamic carbon sinks, optimizes model weights through spatial feedback, realizes adaptive adjustment of parameters, improves prediction accuracy, and supports differentiated carbon management decisions.

[0081] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and all of them should be covered in the scope of the claims of the present application.

Claims

1. A method for predicting carbon storage in urban ecosystems, the method comprising: The application relates to a carbon storage prediction method and system. ​ The method comprises the following steps: Collecting urban ecological data and spatiotemporal big data and preprocessing the same to generate a fusion ecological data set; Based on the fusion ecological data set, a carbon storage prediction model is used for scenario simulation and simulation to evaluate the different influences of climate change scenarios and land use change scenarios on carbon storage, and to generate multi-scenario carbon storage prediction results; The multi-scenario carbon storage prediction results are spatially divided, and the trends and differences of carbon storage changes in different regions are identified to generate a carbon storage spatial distribution; According to the carbon storage spatial distribution, the weight factors and parameter settings in the carbon storage prediction model are optimized to generate an optimized carbon storage prediction model; 2. The urban ecosystem carbon stock prediction method of claim 1, wherein: The optimized carbon storage prediction model is used to simulate carbon storage changes under different scenarios, and the key trends of carbon storage changes are extracted to generate a carbon storage prediction report. The urban ecological data includes land use data, climate data, vegetation cover data, urbanization process data and population density data; 3. The urban ecosystem carbon stock prediction method of claim 2, wherein: The spatiotemporal big data includes historical ecological data, remote sensing data, social and economic data, urban infrastructure data and environmental pollution data. The fusion ecological data set is generated by the following steps: The urban ecological data and the spatiotemporal big data are spatially and temporally aligned to generate a spatiotemporal alignment ecological data set; 4. The urban ecosystem carbon stock prediction method of claim 3, wherein: The spatiotemporal alignment ecological data set is cleaned, formatted and unified in dimension to generate the fusion ecological data set. Based on the fusion ecological data set, a carbon storage prediction model is used for scenario simulation and simulation to evaluate the different influences of climate change scenarios and land use change scenarios on carbon storage, and to generate multi-scenario carbon storage prediction results, which comprises the following steps: Based on the fusion ecological data set, the urban area is divided into scenarios, and scenario parameters and change values are set to generate climate change scenarios and land use change scenarios; The fusion ecological data set is input into the carbon storage prediction model, and scenario simulation is carried out for different climate change scenarios and land use change scenarios, and the scenario simulation results are output; 5. The urban ecosystem carbon stock prediction method of claim 4, wherein: The scenario simulation results are summarized, and the key trends and influencing factors are extracted to generate multi-scenario carbon storage prediction results. The multi-scenario carbon storage prediction results are spatially divided, and the trends and differences of carbon storage changes in different regions are identified to generate a carbon storage spatial distribution, which comprises the following steps: Based on the multi-scenario carbon storage prediction results, the urban planning types are spatially divided to generate a spatial division data set; The carbon storage change trends of different regions in the spatial division data set are extracted, and the differences and reasons of carbon storage changes between regions are identified to generate carbon storage change trends; 6. The urban ecosystem carbon stock prediction method of claim 5, wherein: Based on the carbon storage change trends, the relative positions of different urban planning regions in carbon storage change are evaluated to generate a carbon storage spatial distribution. According to the carbon storage spatial distribution, the weight factors and parameter settings in the carbon storage prediction model are optimized to generate an optimized carbon storage prediction model, which comprises the following steps: The carbon storage spatial distribution is subjected to detailed influence factor identification, and the main driving factors of carbon storage changes in different regions are extracted to generate a key influence factor set; The weight factors are adjusted according to the actual influence degree of the key influence factor set on carbon storage changes, and the parameter settings are optimized through a regression analysis algorithm to generate optimized weight factors and parameter settings; Based on the optimized weight factor and parameter setting, the influencing factors in the carbon storage prediction model are reweighted, and the parameter setting of the carbon storage prediction model is adjusted to generate an optimized carbon storage prediction model.

7. The urban ecosystem carbon stock prediction method of claim 6, wherein: The carbon storage changes under different scenarios are simulated by the optimized carbon storage prediction model, and the steps are as follows, Based on the optimized carbon storage prediction model, different climate change scenarios and land use change scenarios are selected, and detailed simulation parameters are set to generate the set scenarios and simulation parameters; The set scenarios and simulation parameters are input into the optimized carbon storage prediction model to simulate the carbon storage changes, and a carbon storage change simulation report is generated.

8. The urban ecosystem carbon stock prediction method of claim 7, wherein: The key trends of carbon storage changes are extracted to generate a carbon storage prediction report, and the steps are as follows, According to the carbon storage change simulation report, the time series data and spatial distribution characteristics of the carbon storage changes are extracted to generate the carbon storage change key trends and difference data set; The carbon storage change key trends and difference data set are integrated with the carbon storage change simulation results under different scenarios to generate the carbon storage prediction report. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to realize the steps of the urban ecosystem carbon storage prediction method of any one of claims 1-8.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the urban ecosystem carbon storage prediction method of any one of claims 1-8.

Images