An lcz-based urban composite high temperature-pm2.5 pollution risk assessment method and system

Abstract

The application belongs to the field of smart city environment monitoring and regulation, and specifically discloses a city composite high temperature-PM2.5 pollution risk assessment method and system based on LCZ, which comprises an LCZ space unit construction module, a multi-source data acquisition and fusion module, a high temperature and PM2.5 pollution independent risk assessment module, a composite high temperature-PM2.5 pollution risk index assessment module, a city form influence mechanism analysis module, a composite high temperature-PM2.5 pollution risk control unit generation module and a control threshold reverse deduction and parameter output module. The application takes LCZ as a basic space technology unit, constructs a joint probability distribution model of composite high temperature-PM2.5 pollution risk through multi-source environmental data fusion, analyzes the influence mechanism of city form and reversely deduces the control parameters, so as to realize fine assessment and implementable control of composite high temperature-PM2.5 pollution risk.

Description

Technical Field

[0001] This invention relates to the field of smart city environmental monitoring and control, and in particular to a method and system for assessing the risk of urban composite high temperature-PM2.5 pollution based on LCZ. Background Technology

[0002] In the field of risk assessment and control of urban high temperature and PM2.5 pollution, although existing technologies have made some progress in the analysis of single environmental elements and macro-risk identification, they still have the following shortcomings from the perspective of refined assessment and spatial control of complex high temperature-PM2.5 pollution risks: 1. Existing technologies are insufficient to accurately characterize the combined hazards of high temperatures and PM2.5 pollution. Most existing technical solutions either assess high temperature and PM2.5 pollution independently or use simple weighting and linear superposition to construct comprehensive indicators. Neither of these solutions can effectively characterize the spatiotemporal joint occurrence of the two, ignore their statistical dependence and synergistic amplification effect, and are prone to resulting in low risk assessment results for combined high temperature and PM2.5 pollution. These results cannot accurately reflect the true degree of danger and are difficult to meet the precision needs of urban environmental governance.

[0003] 2. Existing technologies using spatial analysis units are insufficient to reflect the differences in the internal physical structure of cities. Existing technologies mostly use administrative divisions or coarse-scale grids as spatial analysis units. These units exhibit significant differences in urban morphology and cannot accurately reflect the thermal environment and pollution diffusion characteristics under different building forms and land cover conditions. This method of spatial unit division makes it difficult to identify high-risk areas within cities where the risk of combined high temperature and PM2.5 pollution is highly concentrated, thus limiting the application value of the assessment results in refined regulation.

[0004] 3. Existing technologies are insufficient to quantitatively reveal the mechanisms by which urban morphology influences the risk of combined high-temperature PM2.5 pollution. Although existing research has shown that factors such as building height (BH), building cover (BSF), impervious surface ratio (ISA), green space coverage (GSR), and sky visibility factor (SVF) can affect the formation of high temperatures and the diffusion of pollutants, current technologies are mostly limited to correlation analysis or qualitative description. They lack technical means to quantify the contribution of different urban morphology indicators to the risk of combined high temperature-PM2.5 pollution, making it difficult to provide clear technical basis for urban spatial optimization.

[0005] 4. Existing technology assessment results are difficult to directly translate into implementable space control measures. Existing risk assessment technologies mostly output results in the form of risk levels or distribution maps, lacking a parametric expression method that is directly related to the regulation of urban physical space. This makes it difficult to answer questions such as how to adjust building layout, land cover, or green space configuration within specific spatial units, resulting in a disconnect between assessment results and actual planning implementation. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a method and system for assessing the risk of urban composite high-temperature PM2.5 pollution based on LCZ.

[0007] To achieve the above objectives, the present invention is implemented according to the following technical solution: The first technical solution provided by this invention is a method for assessing the risk of urban composite high-temperature PM2.5 pollution based on LCZ, comprising the following steps: S1. Obtain multi-source urban physical characteristic data of the study area to be evaluated, and after unified coordinate transformation, spatial registration and standardization preprocessing, construct multiple local climate zone (LCZ) spatial units; the multi-source urban physical characteristic data includes data such as building height, building coverage, land cover type and remote sensing imagery; specifically: (1) Unified coordinate system transformation In this invention, heterogeneous multi-source data such as building height, building coverage, land cover, and remote sensing imagery are acquired. It needs to be uniformly projected to using a mathematical mapping operator. The general projected coordinate system (such as Guangdong WGS84 / UTM Zone 49N) of the city to be evaluated is used, and the specific implementation is as follows: Perform high-precision Bursa-Wolf seven-parameter transformation on heterogeneous coordinate systems to eliminate the source data coordinate system With the target coordinate system To account for translations, rotations, and scale deviations, this invention employs the Bursa-Wolf seven-parameter transformation model. Let the coordinates of the source data spatial points be... The converted target point coordinates are The transformation equation is expressed as: ; in: There are three translation parameters; For three rotation angle parameters (in radians); m The above formula ensures geometric alignment between building vectors and environmental raster data at the cell level, providing a precise analytical space for the subsequent extraction of physical indicators from LCZ spatial units.

[0008] (2) Spatial registration To address the geometric distortion of remote sensing images during the imaging process, this invention introduces control points (GCPs) and employs a second-order polynomial fitting method for nonlinear registration, with its mapping function... f The definition is as follows: ; ; In the formula, The original image coordinates, For reference map coordinates, , These are the registration coefficients calculated using the least squares method. The root mean square error (RMSE) after registration is controlled within 0.5 pixels.

[0009] (3) Standardized pretreatment To eliminate the impact of dimensional differences in features of different dimensions (such as building height (meters) and building density (%)) on spatial clustering algorithms, this invention modifies each feature component... Perform Min-Max Scaling: ; The mapped feature variables are .

[0010] (4) Construct multiple local climate zones (LCZ) spatial units Local climate zoning was completed by combining rule-based reclassification and spatial clustering, generating multiple LCZ spatial units. Each unit was assigned a unique spatial identifier and a structured attribute table was constructed, providing a precise spatial technology carrier for subsequent risk assessment and control of complex high temperature-PM2.5 pollution.

[0011] S2. Acquire multi-source environmental and urban morphology data of the urban study area to be evaluated, and perform temporal scale unification, spatial interpolation, or resampling preprocessing to accurately map them to the corresponding LCZ spatial units, constructing a structured dataset with LCZ spatial units as the core index; the multi-source environmental and urban morphology data includes surface meteorological observations, PM2.5 monitoring, remote sensing inversion, and reanalysis meteorological data; specifically: (1) Uniform time scale: For environmental data with heterogeneous sources (point sites, area grids, linear movement trajectories) and different frequencies (second-level, hour-level, day-level), hierarchical homogenization processing technology is adopted to ensure that it is highly compatible with LCZ spatial units in the spatiotemporal dimension.

[0012] To meet the needs of hourly dynamic analysis of complex events, this invention establishes a unified time reference axis for the alignment and interpolation of multi-scale time series. Then proceed with the following steps: High-frequency data downsampling: for second-level motion measurement data The sliding window averaging method is used to calculate the arithmetic mean with a step size of 1 hour.

[0013] Low-frequency data reconstruction: For daily remote sensing inversion data (such as MODIS), introduce a time-varying correction function based on station observations. The cubic spline interpolation is used to interpolate it into an hourly continuous series, as expressed by the formula: ; in The dynamic weighting coefficients are based on the meteorological background field to ensure smoothness over time.

[0014] (2) Precise resampling and interpolation operators for heterogeneous spatial elements. Due to the large spatial span between the reanalysis data (e.g., ERA5, 0.25°) and the station data, this invention employs the following spatial downscaling and interpolation methods, with the following steps: Spatialization of Discrete Sites: For PM2.5 monitoring station data, ordinary Kriging interpolation is used. The variogram is then fitted. Calculate the center point of the LCZ spatial cell grid to be interpolated. Environmental values: ; Among them, weight Satisfying the unbiased estimation condition Furthermore, it minimizes the estimation variance, thereby capturing the spatial autocorrelation characteristics of air pollutants.

[0015] Coarse-resolution raster downscaling: For reanalysis meteorological data, bilinear interpolation is used, and weighted mapping is performed using the four original grid center points around the target LCZ spatial unit to eliminate the jagged edge effect caused by gridding.

[0016] (3) Accurately map it to the corresponding LCZ spatial unit, and construct a structured dataset with LCZ spatial unit as the core index to provide highly adaptable data support for subsequent risk assessment and mechanism analysis of composite high temperature-PM2.5 pollution. The steps are as follows: Spatial grid partitioning statistical mapping The "spatial unit-attribute field" mapping operator is employed, specifically executed through the following algorithm: Geometric intersection operator: Using the LCZ spatial cell network vector boundary as a mask, interpolate the temperature... T and PM2.5 concentration P Zonal statistics are performed on the raster.

[0017] Mean extraction model: Let the first... Each LCZ space unit contains The environmental attributes of a cell after resampling at that time are... for: ; in: This refers to the environmental pixel values ​​falling within the LCZ spatial unit. For morphological indicators such as building density, a geometric area weighted method is used for extraction.

[0018] (4) Construction of Structured Dataset. After mapping, the system automatically constructs a structured dataset with LCZ_UID (unique identifier) ​​and Timestamp as composite primary keys. Its storage format is a multidimensional relational table or tensor matrix, and its logical structure is defined as follows: ; The dataset uses a row-oriented storage format to ensure the continuity of the time series and a column-oriented storage format to optimize statistical efficiency across cells. Each row represents a spatial-temporal observation point, deeply coupling static urban morphology parameters with dynamic environmental evolution indicators, and providing a standardized data input interface for subsequent calls to Copula functions for nonlinear correlation modeling.

[0019] S3. Based on the structured dataset, calculate the high-temperature hazard index and the PM2.5 pollution hazard index respectively. The results of both calculations are accurately stored numerically in the attribute table of the corresponding LCZ spatial unit; specifically: (1) Based on the high spatiotemporal resolution dataset (generated by the XGBoost-GWR model), the steps for quantifying the deviation of the elements are as follows.

[0020] High-resolution, refined reconstruction of environmental elements based on XGBoost-GWR. In this embodiment of the invention, to address the problem that discrete monitoring station data cannot cover the microscale spatial variability of the LCZ, a hybrid XGBoost-GWR modeling technique integrating global nonlinear mining and local spatial effect correction is adopted to achieve quantitative inversion of air temperature and PM2.5 concentration across a 300m grid.

[0021] Global Feature Extraction: Nonlinear Mapping Based on XGBoost. First, a high-dimensional feature space X={LST,AOD,NDVI,BH,ISA,BSF,….} is constructed. Extreme Gradient Boosting Tree (XGBoost) is then used to capture the complex nonlinear relationship between environmental variables and target variables (temperature / PM2.5).

[0022] Objective function construction: To prevent overfitting and improve generalization ability, an objective function including a regularization term is defined. : ; in, For loss functions (such as mean squared error). For regularization terms, The number of leaf nodes. The leaf weight.

[0023] Iterative optimization: By continuously fitting the residuals of the previous prediction using an additive model, a global baseline prediction value is generated. This model can effectively extract the global contribution characteristics of LCZ physical morphology indicators to environmental factors.

[0024] Local Effects Correction: Spatial Non-stationarity Modeling Based on Geographically Weighted Regression (GWR). Due to the significant spatial non-stationarity of the urban environment, the global XGBoost model may exhibit systematic biases in different LCZ spatial units. This invention introduces Geographically Weighted Regression (GWR) to perform secondary spatial correction on the XGBoost residuals.

[0025] Residual definition: Calculation of observed values Compared with XGBoost predictions residual .

[0026] Local regression equation: coordinates of the center of each LCZ spatial unit Establish a local correction model: ; in, It is the regression coefficient that varies with spatial location.

[0027] Spatial weighting function: The Gaussian kernel function is used to determine the influence weight of neighboring sites on the current cell. : ; in, Euclidean distance. b The optimal bandwidth is determined through optimization using the AICc information criteria.

[0028] The results are fused with the high-resolution dataset to generate the final high-resolution environmental element estimates. It is composed of the superposition of global prediction terms and local correction terms: ; Using this model, the present invention drills down the original kilometer-level (such as ERA5) or discrete point-like environmental data to a 300m resolution LCZ spatial cell network.

[0029] (2) Deviation Measurement (Function Deviation Calculation). Based on the above generated... With high-resolution datasets, this invention further calculates element deviation to characterize the hazard intensity of a single disaster.

[0030] High temperature deviation : ; in, For a specific unit temperature, For the regional background reference temperature, The standard deviation is the time-space standard deviation.

[0031] PM2.5 deviation : ; in, Unit concentration, To meet the threshold (e.g., 35) ).

[0032] Storage and Indexing: Calculated and The numerical values, along with parameters such as occurrence frequency and duration, are precisely written into the structured attribute table of the corresponding LCZ spatial unit in floating-point form (Float64), providing standardized input for subsequent edge distribution fitting of the Copula function.

[0033] The formula for calculating the high temperature hazard index is: ; in: The dynamic threshold for high temperature is determined by a combination of intensity and duration. Formula for calculating PM2.5 pollution risk index: ; in: This is the national secondary standard limit. This is the frequency factor exceeding the standard.

[0034] S4. Based on the calculation results of high temperature hazard indicators and PM2.5 pollution hazard indicators, a nonlinear coupled model of the combined probability distribution of high temperature and PM2.5 pollution hazard is constructed. The calculation results of high temperature hazard and PM2.5 pollution hazard are then input into the combined probability distribution model of high temperature and PM2.5 pollution hazard to calculate the combined high temperature and PM2.5 pollution hazard indicators for each LCZ spatial unit during the study period, and the hazard classification of the study area is completed according to preset rules. Specifically: This invention proposes a composite high-temperature-PM2.5 pollution risk control technology that integrates Copula theory and LCZ physical mechanism: The Copula function independently characterizes the marginal distribution and tail correlation of high temperature and PM2.5, constructing a composite hazard index (CHI) to quantify the synergistic effect; LCZ morphological parameters such as SVF (sky visibility factor) and ISA (impermeable surface ratio) are introduced to construct a modified model representing the contribution of the physical environment; innovatively, multi-objective optimization algorithms such as NSGA-II are used for back-calculation, aiming to minimize the composite hazard index (CHI), and combined with constraints such as building capacity to output the optimal morphological control threshold, achieving precise transformation from risk assessment to control indicators, breaking through the limitations of traditional forward modeling.

[0035] To address the shortcomings of traditional weighted methods in characterizing the risk of "extreme coexistence," this invention introduces the Copula function to establish a nonlinear coupling model. Through its sensitivity to "upper tail correlation," it can effectively capture the "extreme coexistence" characteristics of high temperature and high pollution occurring simultaneously.

[0036] (1) Marginal distribution fitting: The distribution tests of the high temperature hazard index HI and the PM2.5 pollution hazard index PI are performed respectively. Generally, the temperature follows the generalized extreme value distribution (GEV), and PM2.5 follows the log-normal distribution.

[0037] (2) Copula coupling modeling: The Gumbel Copula function, which is sensitive to the "upper tail correlation", is selected to calculate the joint probability: ; in: This is a correlation parameter that reflects the sensitivity of the synergistic amplification between the two.

[0038] (3) The formula for the nonlinear coupling model of the combined probability distribution of high temperature and PM2.5 pollution risk is as follows: ; in: This is a composite high-temperature-PM2.5 pollution hazard index containing a synergistic amplification index; if A score greater than 1 indicates that the physical environment significantly promotes the combined effect of high temperature and pollution, thus correcting the risk score of the combined high temperature-PM2.5 pollution.

[0039] Furthermore, the LCZ-based urban composite high-temperature-PM2.5 pollution hazard assessment method also includes: S5. Using urban morphological indicators such as building height (BH), building coverage (BSF), impervious surface ratio (ISA), green space coverage (GSR), and sky visibility factor (SVF) as inputs, and the composite high-temperature-PM2.5 pollution hazard index of each LCZ spatial unit as output, a SHAP attribution model is constructed. The SHAP value (contribution) of each urban morphological indicator is calculated to quantify the degree and direction of influence of each morphological indicator on the composite high-temperature-PM2.5 pollution hazard, and to accurately identify the core driving factors; specifically: (1) SHAP model diagnosis: based on the combined high temperature-PM2.5 pollution risk of each LCZ spatial unit. H Let (BH, BSF, ISA, GSR, SVF) be the objective function, and (BH, BSF, ISA, GSR, SVF) be the independent variables.

[0040] The objective function defined in this invention H It is not a simple linear superposition of the factors, but an exponential function based on energy balance and statistical synergistic effects. Its specific functional expression is as follows: ; , These are the standardized deviations from the high-temperature hazard and the PM2.5 pollution hazard. The joint probability distribution calculated using the aforementioned Gumbel-Copula function reflects the probability measure of their simultaneous occurrence. The composite high-temperature-PM2.5 pollution hazard index, which includes a synergistic amplification index, is used to characterize the nonlinear enhancement effect produced when high temperature and pollution are superimposed.

[0041] (2) Contribution quantification: Calculate the SHAP value (contribution) of each morphological factor to accurately identify whether "obstructed ventilation (low SVF)" or "surface thermal properties (high ISA)" dominates the surge in the risk of composite high temperature-PM2.5 pollution in a specific LCZ spatial unit (such as LCZ1).

[0042] To perform SHAP diagnostics, this invention first constructs a black-box mapping model. f Establish LCZ morphological factor vector X With objective function H Nonlinear functional relationship: ; in: Sky visibility factor; Building coverage; : Proportion of impermeable surfaces; Building height; Green coverage rate; : Random perturbation term.

[0043] SHAP value mechanism decomposition algorithm To decompose the H assessment results into understandable planning directives, this invention introduces the SHAP attribution model. For each specific LCZ spatial unit, its composite high-temperature-PM2.5 pollution hazard prediction value... H It can be decomposed into the sum of the marginal contributions of each morphological factor: ; (3) According to SHAP The formula for calculating the SHAP value (contribution) of each urban morphology indicator using the attribution model is as follows: ; in: This represents the total number of urban morphology indicators. =5; For factors not included All possible subsets of features; To add factors The change in the model's predicted value is the marginal contribution of that factor.

[0044] Results output and mechanism identification: The system automatically calculates and outputs the ranking of SHAP contribution values ​​for each LCZ spatial unit.

[0045] Positive driving factor identification: If Furthermore, the value is relatively large, indicating that in this unit, the impermeable surface area (ISA) is the dominant factor leading to the increased risk of combined high temperature-PM2.5 pollution. Negative regulatory factor identification: If (A negative SHAP value indicates that increasing the sky visibility factor can significantly suppress the growth of the risk of combined high temperature-PM2.5 pollution.)

[0046] S6. Based on the LCZ spatial unit and the composite high temperature-PM2.5 pollution hazard assessment results, a composite high temperature-PM2.5 pollution hazard control unit is generated. This control unit corresponds to one or more LCZ spatial units in space and is associated with their hazard level, key urban morphology index values ​​and control targets in terms of attributes. Specifically, adjacent LCZ spatial units with a composite high temperature-PM2.5 pollution hazard index CHI containing the synergistic amplification index greater than the preset high-risk threshold and whose dominant factor SHAP value ranks in the top two are aggregated into key environmental risk control units.

[0047] S7. Under the constraint of the established target for reducing the risk of combined high temperature-PM2.5 pollution, the urban morphology control threshold corresponding to each risk control unit is calculated through model back-engineering and output as a numerical range. This threshold is used to guide the optimization and adjustment of the influencing factors (such as building density, building height, impermeable surface ratio, green space ratio, ventilation corridor width, etc.) that contribute the most to the risk of combined high temperature-PM2.5 pollution within the risk control unit, so as to realize the direct transformation of assessment results into practical control measures; specifically: (1) Multi-objective optimization algorithm NSGA-II: Target Algorithm: ; Constraints: ; (2) Parameter output: Through iterative calculation, the Pareto optimal solution set is obtained, and the suggested range of specific urban morphology indicators is output. For example: In order to reduce the risk of composite high temperature-PM2.5 pollution in this unit by 15%, it is recommended that the sky visibility factor (SVF) should be increased to above 0.65, and the impermeable surface ratio (ISA) should be controlled between 40% and 55%.

[0048] The second technical solution provided by this invention is an urban composite high-temperature-PM2.5 pollution hazard assessment system based on LCZ, used to perform the above method. This system includes: The LCZ spatial unit construction module is used to acquire multi-source urban physical characteristic data of the urban study area to be evaluated, and after unified coordinate transformation, spatial registration and standardization preprocessing, construct multiple local climate zone LCZ spatial units; the multi-source urban physical characteristic data includes data such as building height, building coverage, land cover type and remote sensing imagery. The multi-source data acquisition and fusion module is used to acquire multi-source environmental and urban morphology data of the urban study area to be evaluated, and to perform temporal scale unification, spatial interpolation or resampling preprocessing to accurately map them to the corresponding LCZ spatial units, and to construct a structured dataset with LCZ spatial units as the core index; the multi-source environmental and urban morphology data include ground meteorological observation, PM2.5 monitoring, remote sensing inversion and reanalysis meteorological data; The module for independent risk assessment of high temperature and PM2.5 pollution is used to calculate the risk index of high temperature and the risk index of PM2.5 pollution based on structured datasets. The results of both types of calculations are stored in numerical form in the attribute table of the corresponding LCZ spatial unit. The composite high-temperature-PM2.5 pollution hazard assessment module is used to construct a nonlinear coupled model of the joint probability distribution of composite high-temperature-PM2.5 pollution hazard based on the calculation results of high-temperature hazard and PM2.5 pollution hazard indicators. The calculation results of high-temperature hazard and PM2.5 pollution hazard indicators are input into the composite high-temperature-PM2.5 pollution hazard joint probability distribution model to calculate the composite high-temperature-PM2.5 pollution hazard indicators in each LCZ spatial unit during the study period, and complete the hazard classification of the study area according to preset rules. The urban morphology impact mechanism analysis module uses urban morphology indicators such as building height (BH), building coverage (BSF), impervious surface ratio (ISA), green space coverage (GSR), and sky visibility factor (SVF) as inputs, and the composite high-temperature-PM2.5 pollution hazard index of each LCZ spatial unit as output to construct... SHAP Attribution model, and calculate the SHAP value (contribution) of each urban morphology indicator, quantify the degree and direction of influence of each morphology indicator on the risk of compound high temperature-PM2.5 pollution, and accurately identify the core driving factors. The module for generating composite high-temperature-PM2.5 pollution hazard control units is used to generate composite high-temperature-PM2.5 pollution hazard control units based on LCZ spatial units and composite high-temperature-PM2.5 pollution hazard assessment results. Spatially, each control unit corresponds to one or more LCZ spatial units, and in terms of attributes, it is associated with its hazard level, key urban morphology index values, and control targets. The regulation threshold back-dive and parameter output module is used to calculate the urban morphology regulation threshold corresponding to each hazard control unit through model back-dive under the constraint of the established target of reducing the risk of composite high temperature-PM2.5 pollution, and output the numerical range. The threshold is used to guide the optimization and adjustment of the factors that contribute the most to the risk of composite high temperature-PM2.5 pollution within the hazard control unit (such as building density, building height, impermeable surface ratio, green space ratio, ventilation corridor width, etc.), so as to realize the direct transformation of assessment results into practical regulation measures.

[0049] Compared with the prior art, the present invention has the following beneficial effects: 1. Achieve quantitative and refined assessment of the pollution risks of combined high temperature PM2.5. This invention constructs LCZ spatial units by unifying the time scale and spatial benchmark, and uses a joint probability distribution model to quantitatively assess the combined high-temperature-PM2.5 pollution risk. This avoids the risk assessment bias caused by traditional simple superposition of multiple factors or linear weighting, and accurately reflects the risk distribution characteristics of combined high-temperature-PM2.5 pollution in cities. Compared with existing single-factor or coarse-scale evaluation methods, its combined high-temperature-PM2.5 pollution hazard results show significant improvements in spatial resolution and risk expression accuracy, demonstrating outstanding technical advantages.

[0050] 2. Provide spatial analysis and risk identification capabilities that are highly compatible with the urban physical form. This invention uses Local Climate Zones (LCZs) as the basic spatial unit, enabling environmental risk assessments to directly correspond to the specific physical characteristics of cities. This effectively overcomes the problem of spatial heterogeneity masking when administrative divisions or simple urban-rural classifications are used as analysis units. Through this technology, high-risk areas with significant concentrations of combined high-temperature and PM2.5 pollution within cities can be accurately identified, providing a reliable spatial basis for the refined management of urban combined high-temperature and PM2.5 pollution risks.

[0051] 3. Identify the key factors and directions of influence of urban morphology on the risk of combined high-temperature PM2.5 pollution. This invention constructs a quantitative model of urban morphological indicators and the risk of combined high temperature-PM2.5 pollution. It can accurately identify the relative importance and direction of different morphological factors on the formation of combined high temperature-PM2.5 pollution risk, breaking through the limitations of existing technologies that can only conduct correlation analysis or qualitative description, and providing clear and quantifiable scientific and technological basis for the optimization of urban spatial structure.

[0052] 4. Achieve direct conversion of environmental risk assessment results into spatial control parameters. This invention utilizes model-based back-analysis technology to accurately calculate the urban morphology control threshold required to achieve the target of reducing the risk of combined high-temperature PM2.5 pollution. This allows the assessment results to transcend the limitations of risk levels and distribution maps, directly transforming them into implementable spatial control parameters. This design effectively addresses the pain point of existing technologies where assessment results are difficult to guide specific planning, design, and spatial governance practices, significantly enhancing the application value of the technology.

[0053] 5. Construct a complete technological closed loop from risk identification to risk reduction. This invention organically integrates the risk assessment of combined high temperature and PM2.5 pollution, the analysis of the impact mechanism of urban morphology, the generation of risk control units, and the output of control parameters to construct a complete technical closed loop. This closed loop enables systematic identification, in-depth analysis, and precise intervention of urban combined high temperature and PM2.5 pollution risks, breaking through the fragmented limitations of existing technologies and significantly improving the overall efficiency and implementation effect of urban environmental risk prevention and control. The technology is highly practical and innovative.

[0054] 6. The technical solution has good feasibility and applicability for promotion. The technical solution of this invention has clear implementation steps and system structure, is compatible with existing urban environmental monitoring and spatial information technology conditions, and has good repeatability and operability. Furthermore, the solution is not limited by city size, data source, or specific application scenario, demonstrating outstanding versatility and the ability to be flexibly implemented in different urban environmental governance scenarios, thus possessing broad application value. Attached Figure Description

[0055] Figure 1 This is a structural block diagram of the LCZ-based urban composite high-temperature PM2.5 pollution hazard assessment system of the present invention.

[0056] Figure 2 This is a flowchart of the LCZ-based urban composite high-temperature PM2.5 pollution hazard assessment method of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0058] like Figure 1 As shown in the illustration, this embodiment exemplarily demonstrates an urban composite high-temperature-PM2.5 pollution hazard assessment system based on LCZ, the system comprising: The LCZ spatial unit construction module is used to acquire multi-source urban physical characteristic data of the urban study area to be evaluated, and after unified coordinate transformation, spatial registration and standardization preprocessing, construct multiple local climate zone LCZ spatial units; the multi-source urban physical characteristic data includes data such as building height, building coverage, land cover type and remote sensing imagery. The multi-source data acquisition and fusion module is used to acquire multi-source environmental and urban morphology data of the urban study area to be evaluated, and to perform temporal scale unification, spatial interpolation or resampling preprocessing to accurately map them to the corresponding LCZ spatial units, and to construct a structured dataset with LCZ spatial units as the core index; the multi-source environmental and urban morphology data include ground meteorological observation, PM2.5 monitoring, remote sensing inversion and reanalysis meteorological data; The module for independent risk assessment of high temperature and PM2.5 pollution is used to calculate the risk index of high temperature and the risk index of PM2.5 pollution based on structured datasets. The results of both types of calculations are stored in numerical form in the attribute table of the corresponding LCZ spatial unit. The composite high-temperature-PM2.5 pollution hazard assessment module is used to construct a nonlinear coupled model of the joint probability distribution of composite high-temperature-PM2.5 pollution hazard based on the calculation results of high-temperature hazard and PM2.5 pollution hazard indicators. The calculation results of high-temperature hazard and PM2.5 pollution hazard are input into the composite high-temperature-PM2.5 pollution hazard joint probability distribution model to calculate the composite high-temperature-PM2.5 pollution hazard indicators in each LCZ spatial unit during the study period, and complete the hazard classification of the study area according to preset rules. The Urban Morphology Impact Mechanism Analysis Module uses urban morphology indicators such as building height (BH), building coverage (BSF), impervious surface ratio (ISA), green space coverage (GSR), and sky visibility factor (SVF) as inputs, and composite high temperature-PM2.5 pollution hazard indicators of each LCZ spatial unit as outputs, to construct a SHAP attribution model, calculate the SHAP value of each urban morphology indicator (i.e., its contribution), quantify the degree and direction of influence of each morphology indicator on composite high temperature-PM2.5 pollution hazard, and accurately identify the core driving factors. The module for generating composite high-temperature-PM2.5 pollution hazard control units is used to generate composite high-temperature-PM2.5 pollution hazard control units based on LCZ spatial units and composite high-temperature-PM2.5 pollution hazard assessment results. Spatially, each control unit corresponds to one or more LCZ spatial units, and in terms of attributes, it is associated with its hazard level, key urban morphology index values, and control targets. The regulation threshold back-dive and parameter output module is used to calculate the urban morphology regulation threshold corresponding to each hazard control unit through model back-dive under the constraint of the established target of reducing the risk of composite high temperature-PM2.5 pollution, and output the numerical range. The threshold is used to optimize and adjust the influencing factors (such as building density, building height, impermeable surface ratio, green space ratio, ventilation corridor width, etc.) that contribute the most to the risk of composite high temperature-PM2.5 pollution within the hazard control unit, so as to realize the direct transformation of the assessment results into practical regulation measures.

[0059] like Figure 2 As shown, the urban composite high-temperature-PM2.5 pollution hazard assessment method based on LCZ includes the following steps: S1. Obtain multi-source urban physical characteristic data of the urban study area to be evaluated, and after unified coordinate transformation, spatial registration and standardization preprocessing, construct multiple local climate zone LCZ spatial units; among which, multi-source urban physical characteristic data include data such as building height, density, land cover type and remote sensing imagery; S2. Acquire multi-source environmental and urban morphology data of the urban study area to be evaluated, and perform temporal scale unification, spatial interpolation or resampling preprocessing to accurately map them to the corresponding LCZ spatial units, and construct a structured dataset with LCZ spatial units as the core index; the multi-source environmental and urban morphology data include ground meteorological observation, PM2.5 monitoring, remote sensing inversion and reanalysis meteorological data; S3. Calculate the high temperature hazard index and PM2.5 pollution hazard index based on the structured dataset. The results of both types of calculations are accurately stored in the attribute table of the corresponding LCZ spatial unit in numerical form. S4. Based on the calculation results of high temperature hazard index and PM2.5 pollution hazard index, construct a nonlinear coupled model of the combined probability distribution of high temperature-PM2.5 pollution hazard, and input the calculation results of high temperature hazard and PM2.5 pollution hazard into the combined probability distribution model of high temperature-PM2.5 pollution hazard to calculate the combined high temperature-PM2.5 pollution hazard index in each LCZ spatial unit during the study period, and complete the hazard classification of the study area according to the preset rules; S5. Using urban morphological indicators such as building height (BH), building coverage (BSF), impervious surface ratio (ISA), green space coverage (GSR), and sky visibility factor (SVF) as inputs, and the composite high-temperature-PM2.5 pollution hazard index of each LCZ spatial unit as output, a SHAP attribution model is constructed, and the values ​​of each urban morphological indicator are calculated. SHAP The value represents the contribution level, quantifying the degree and direction of influence of various indicators on the risk of complex high temperature-PM2.5 pollution, and accurately identifying the core driving factors. S6. Based on the LCZ spatial unit and the composite high temperature-PM2.5 pollution risk assessment results, a risk control unit is generated. This control unit corresponds to one or more LCZ spatial units in space and is associated with their risk level, key urban morphology index values ​​and control targets in terms of attributes. S7. Under the constraint of the established target of reducing the risk of composite high temperature-PM2.5 pollution, the urban form control threshold corresponding to each risk control unit is calculated by back-engineering the model and output as a numerical range. The threshold is used to guide the optimization and adjustment of physical space parameters such as building layout, land cover and green space configuration, so as to realize the direct transformation of assessment results into practical control measures.

[0060] To verify the effectiveness of the method of the present invention, for example, Guangzhou, a subtropical megacity, was selected as the empirical research object, and the verification was carried out on the typical combined high temperature-PM2.5 event in its summer (June-August).

[0061] The following is a detailed explanation using monitoring data from a high-density area (typical LCZ2: compact mid-rise building zone) in Tianhe District, Guangzhou. Data acquisition was conducted in real time via API interface, collecting data from meteorological monitoring stations, the China National Environmental Monitoring Centre (CNEMC) air quality monitoring network, and remote sensing satellites such as Himawari-8. Local climate zone (LCZ) gridding and resampling were performed using the Python Geopandas toolkit.

[0062] (1) Construction of urban spatial LCZ spatial units and alignment with spatial benchmarks This invention first uses the Bursa-Wolf seven-parameter model to perform coordinate transformation between vector building data and raster environmental data in Guangzhou.

[0063] Empirical data from Guangzhou: Selected grid points in the Tianhe Road business district Convert it from latitude and longitude coordinates to target projected coordinates Error control within .

[0064] LCZ Identification: The building coverage ratio (BSF) of this unit is calculated to be 0.55, the impermeable surface area ratio (ISA) to be 0.85, and the building height to be 25m. The system automatically identifies its spatial identifier as GZ_TH_LCZ02_001.

[0065] (2) High-resolution environmental data generation based on XGBoost-GWR To address the uneven distribution of monitoring points in Guangzhou during the summer, the XGBoost-GWR algorithm was used to drill down meteorological reanalysis data at a resolution of 9km to 300m.

[0066] Input: ERA5 original temperature 34.2°C, AOD remote sensing inversion value.

[0067] Algorithm logic: XGBoost extracts global features, and GWR performs local corrections based on the residuals of the Tianhe District meteorological station.

[0068] Verification values: The refined temperature at 14:00 for this LCZ spatial cell was 37.2°C, and the PM2.5 concentration was 68 μg / m³.

[0069] (3) Quantification of the independent hazard of high temperature and PM2.5 and its deviation Based on Guangzhou summer threshold = 33°C, Calculate the deviation.

[0070] High temperature deviation : ; Pollution deviation : ; The calculation results show that the unit is in a "high-risk" state.

[0071] (4) Risk of combined high temperature-PM2.5 pollution based on Gumbel-Copula Build This is the core innovation of the invention, used to capture the "tail synergy probability" of high temperature and PM2.5 under extreme conditions.

[0072] Joint distribution modeling: Let the correlation parameters between the two in Guangzhou summer be set. (Represents a strong positive correlation).

[0073] objective function H calculate: ; Empirical values: Calculated (Range 0-1), which belongs to the extremely high risk level, proves the existence of the synergistic amplification effect.

[0074] (5) Urban morphology mechanism diagnosis based on SHAP Using the SHAP attribution model, the objective function H The high values ​​are decomposed into specific morphological factors.

[0075] Diagnostic results: For the GZ_TH_LCZ02_001 unit, the morphological contribution ranking is as follows: ISA (Impervious Surface Area Ratio): (Main cause of disaster); SVF (Sky Visibility Factor): (Obstructed ventilation exacerbates the risk); GSR (Green Space Coverage Ratio): (Insufficient relief).

[0076] (6) Backward deduction of control parameters (based on NSGA-II algorithm) After determining ISA (impermeable surface area ratio) as the dominant factor, the system is set... The objective is to reduce the value by 20% (to 0.65), and then use the NSGA-II multi-objective optimization algorithm to back-calculate the result.

[0077] Constraints: The building height remains unchanged at 25m, and the floor area ratio fluctuates by ≤5%.

[0078] Reverse deduction output (adjustment threshold): ISA (Percentage of impervious surfaces): Reduce from 0.85 to [0.65, 0.72] (It is recommended to increase permeable paving and small green spaces).

[0079] SVF (Sky Visibility Factor): By adjusting the building layout, the SVF can be increased from 0.35 to over 0.50.

[0080] Finally, through the above operations, the aforementioned control parameters are automatically mapped to the Guangzhou City detailed planning land parcel attribute table.

[0081] Practicality is reflected in: The assessment results are directly used as the basis for the "Climate Adaptive City Special Plan", while the control ranges of ISA (Impervious Surface Ratio) and SVF (Sky Visibility Factor) are transformed into rigid control indicators in the "Detailed Control Plan".

[0082] By applying the reverse inference algorithm of this invention to high-risk control units such as high-density commercial districts like Tianhe, the assessment results are transformed into rigid or flexible indicators at the land parcel level. The detailed planning guidelines specify the upper limit of impervious surface ratio (ISA), the lower limit of sky visibility factor (SVF), and the greening rate requirements for newly built land parcels within the control unit. Through refined morphological optimization, the risk of local compound high temperature-PM2.5 pollution is directly reduced.

[0083] The technical solutions of the present invention are not limited to the specific embodiments described above. Any technical modifications made in accordance with the technical solutions of the present invention fall within the protection scope of the present invention.

Claims

1. A method for assessing the risk of urban complex high temperature-PM2.5 pollution based on LCZ, characterized in that, Includes the following steps: S1. Obtain multi-source urban physical characteristic data of the study area of ​​the city to be evaluated, and after unified coordinate transformation, spatial registration and standardization preprocessing, construct multiple local climate zone LCZ spatial units. The multi-source urban physical feature data includes building height, building coverage, land cover type and remote sensing image data; S2. Acquire multi-source environmental and urban morphology data of the urban study area to be evaluated, and perform temporal scale unification, spatial interpolation or resampling preprocessing to accurately map them to the corresponding LCZ spatial units, and construct a structured dataset with LCZ spatial units as the core index; the multi-source environmental and urban morphology data include ground meteorological observation, PM2.5 monitoring, remote sensing inversion and reanalysis meteorological data; S3. Calculate the high-temperature hazard index and the PM2.5 pollution hazard index based on the structured dataset. The results of both calculations are stored precisely in numerical form in the attribute table of the corresponding LCZ spatial unit. The formula for calculating the high-temperature hazard index is: ； in: The dynamic threshold for high temperature is determined by a combination of intensity and duration. Formula for calculating PM2.5 pollution risk index: ； in: This is the national secondary standard limit. This is the frequency factor exceeding the standard; S4. Based on the calculation results of high temperature hazard index and PM2.5 pollution hazard index, a nonlinear coupled model of the joint probability distribution of composite high temperature-PM2.5 pollution hazard is constructed. The calculation results of high temperature hazard and PM2.5 pollution hazard are input into the joint probability model to obtain the composite high temperature-PM2.5 pollution hazard index for each LCZ spatial unit during the study period and complete the hazard classification. S41. High-temperature hazard indicators HI and PM2.5 pollution risk indicators PI Distribution tests were performed separately. Generally, temperature follows a generalized extreme value distribution (GEV), while PM2.5 follows a log-normal distribution. S42. Copula coupling modeling selects the Gumbel Copula function, which is sensitive to upper tail correlation, to calculate the joint probability: ； in: This is a correlation parameter that reflects the sensitivity of the synergistic amplification between the two. The formula for the nonlinear coupling model of the combined probability distribution of high temperature and PM2.5 pollution hazards (S43) is as follows: ； in: This is a composite high-temperature-PM2.5 pollution hazard index containing a synergistic amplification index; if >1 indicates that the physical environment significantly promotes the superimposed effect of high temperature and pollution, thus correcting the risk score of combined high temperature-PM2.5 pollution; S5. Using urban morphology indicators such as building height (BH), building coverage ratio (BSF), impervious surface ratio (ISA), green space coverage ratio (GSR), and sky visibility factor (SVF) as inputs, and the composite high-temperature-PM2.5 pollution hazard index of each LCZ spatial unit as output, a SHAP attribution model is constructed. The SHAP value (contribution) of each urban morphology indicator is calculated to quantify the degree and direction of influence of each morphology indicator on the composite high-temperature-PM2.5 pollution hazard, accurately identifying the core driving factors. The formula for calculating the SHAP value (contribution) of each urban morphology indicator according to the SHAP attribution model is as follows: ； in: This represents the total number of urban morphology indicators. =5; For factors not included All possible subsets of features; To add factors The change in the model's predicted value is the marginal contribution of that factor; S6. Based on the LCZ spatial unit and the composite high temperature-PM2.5 pollution hazard assessment results, generate composite high temperature-PM2.5 pollution hazard control units. Spatially, each control unit corresponds to one or more LCZ spatial units, and in terms of attributes, it is associated with its hazard level, key urban morphology index values, and control targets. Specifically, this includes aggregating adjacent LCZ spatial units with a composite high temperature-PM2.5 pollution hazard index (CHI) value greater than the preset high-risk threshold and whose dominant factor (SHAP) value ranks in the top two as key environmental risk control units. S7. Under the constraint of the established target of reducing the risk of combined high temperature-PM2.5 pollution, the urban morphology control threshold corresponding to each risk control unit is calculated by back-engineering the model and output as a numerical range. This threshold is used to guide the optimization and adjustment of the influencing factors that contribute the most to the risk of combined high temperature-PM2.5 pollution within the combined high temperature-PM2.5 pollution risk control unit, so as to realize the direct transformation of the assessment results into practical control measures. S71, Multi-objective optimization algorithm NSGA-II: Objective function: ; Constraints: ; S72. Parameter Output: Through iterative calculation, the Pareto optimal solution set is obtained, and the suggested range of specific urban morphology indicators is output.

2. The method for assessing the risk of urban composite high-temperature PM2.5 pollution based on LCZ according to claim 1, characterized in that, In step S1: The unified coordinate system transformation is performed using a high-precision Bursa-Wolf seven-parameter model, and then uniformly projected onto the target coordinate system. That is, the projected coordinate system in which the city to be evaluated is located; Spatial registration is performed using control points (GCPs) and a second-order polynomial fitting method for nonlinear registration. The root mean square error (RMSE) after registration is controlled within 0.5 pixels. Standardization preprocessing is performed on each feature component. Performing Min-Max Scaling, the mapped feature variables are: ; The construction of multiple local climate zone (LCZ) spatial units is achieved by combining rule-based reclassification and spatial clustering to generate multiple LCZ spatial units. Each unit is assigned a unique spatial identifier and a structured attribute table is constructed.

3. The method for assessing the risk of urban composite high-temperature PM2.5 pollution based on LCZ according to claim 2, characterized in that, In step S2, the structured dataset, which uses a unique identifier LCZ_UID and a timestamp as a composite primary key, is stored in the form of a multidimensional relational table (Table) or a tensor matrix (Tensor). Its logical structure is defined as follows: 。 4. A city-wide composite high-temperature PM2.5 pollution hazard assessment system based on LCZ, characterized in that, The system for performing the method according to any one of claims 1-3 includes: The LCZ spatial unit construction module is used to acquire multi-source urban physical characteristic data of the urban study area to be evaluated, and after unified coordinate transformation, spatial registration and standardization preprocessing, construct multiple local climate zone LCZ spatial units; the multi-source urban physical characteristic data includes building height, building coverage, land cover type and remote sensing image data. The multi-source data acquisition and fusion module is used to acquire multi-source environmental and urban morphology data of the urban study area to be evaluated, and to perform temporal scale unification, spatial interpolation or resampling preprocessing to accurately map them to the corresponding LCZ spatial units, and to construct a structured dataset with LCZ spatial units as the core index; the multi-source environmental and urban morphology data include ground meteorological observation, PM2.5 monitoring, remote sensing inversion and reanalysis meteorological data; The module for independent risk assessment of high temperature and PM2.5 pollution is used to calculate the risk index of high temperature and the risk index of PM2.5 pollution based on structured datasets. The results of both types of calculations are stored in numerical form in the attribute table of the corresponding LCZ spatial unit. The composite high-temperature-PM2.5 pollution hazard assessment module is used to construct a nonlinear coupled model of the joint probability distribution of composite high-temperature-PM2.5 pollution hazard based on the calculation results of high-temperature hazard and PM2.5 pollution hazard indicators. It inputs the calculation results of high-temperature hazard and PM2.5 pollution hazard into the composite high-temperature-PM2.5 pollution hazard joint probability distribution model to calculate the composite high-temperature-PM2.5 pollution hazard indicators in each LCZ spatial unit during the study period, and completes the hazard classification of the study area according to preset rules. The Urban Morphology Impact Mechanism Analysis Module uses urban morphology indicators such as building height (BH), building coverage (BSF), impervious surface ratio (ISA), green space coverage (GSR), and sky visibility factor (SVF) as inputs, and composite high temperature-PM2.5 pollution hazard indicators of each LCZ spatial unit as outputs. It constructs a SHAP attribution model, calculates the SHAP value (contribution) of each urban morphology indicator, quantifies the degree and direction of influence of each morphology indicator on composite high temperature-PM2.5 pollution hazard, and accurately identifies the core driving factors. The module for generating composite high-temperature-PM2.5 pollution hazard control units is used to generate composite high-temperature-PM2.5 pollution hazard control units based on LCZ spatial units and composite high-temperature-PM2.5 pollution hazard assessment results. Spatially, each control unit corresponds to one or more LCZ spatial units, and in terms of attributes, it is associated with its hazard level, key urban morphology index values, and control targets. The regulation threshold back-dive and parameter output module is used to calculate the urban morphology regulation threshold corresponding to each risk control unit through model back-dive under the constraint of the established target of reducing the risk of compound high temperature-PM2.5 pollution, and output the numerical range. The threshold is used to guide the optimization and adjustment of the influencing factors that contribute the most to the risk of compound high temperature-PM2.5 pollution within the risk control unit, so as to realize the direct transformation of the assessment results into practical regulation measures.

Images