Urban built-up area greenable land intervention zoning method

Abstract

The invention discloses an urban built-up area greenable land intervention zoning method, which comprises the steps of calculating a thermal risk index, calculating a thermal exposure index and a thermal vulnerability index, calculating based on the three indexes to obtain a comprehensive thermal risk index, extracting urban form data by using a high-resolution satellite remote sensing image, and marking greenable lands. The method comprises the following steps: dividing a research area into a plurality of decision units in a square grid, calculating a morphological index for each decision unit, counting the average surface temperature of each decision unit, training a machine learning regression model, and predicting the cooling potential of the units by using the regression model. The risk-potential combination type is constructed for the decision unit based on the comprehensive thermal risk index and the cooling potential, intervention partitions are divided, effective connection between risk identification and decision intervention is achieved, the thermal risk level and the cooling potential level are combined, and the risk identification accuracy is improved. And a clear quantitative basis is provided for newly-added green land configuration under the limited resource condition of the stock built-up area.

Description

Technical Field

[0001] This invention relates to the field of urban green space planning and construction, and in particular to a method for intervening in the zoning of greenable land in urban built-up areas. Background Technology

[0002] In recent years, the combined effects of extreme high-temperature events and the urban heat island effect have significantly increased heat exposure levels among people in built-up areas, thereby increasing the risk of heat-related health problems such as heatstroke and cardiovascular diseases. Numerous studies have shown that urban green spaces can reduce surface and near-surface air temperatures at different spatial scales through various means, including shading to reduce shortwave radiation, converting net radiation into latent heat and reducing sensible heat flux through evapotranspiration, improving local air exchange, and reducing surface heat storage. These green spaces have a clear cooling effect and value in mitigating heat risks. In urban renewal and existing built-up areas, newly added green spaces often take the form of pocket green spaces, corner greening, and marginal land greening. These spaces are spatially dispersed and subject to strong constraints from land use and implementation conditions. How to achieve more effective cooling interventions under limited resources has become a key issue in urban governance.

[0003] While existing technologies have developed a thermal risk assessment framework that combines thermal hazard, thermal exposure, and thermal vulnerability, and have obtained the spatial distribution of thermal risk through normalization and weighted aggregation, this framework still lacks effective integration with urban green space renewal decisions. This makes it difficult to directly translate risk identification results into actionable intervention zones. On the other hand, existing assessments of the cooling effect of green spaces often employ idealized greening scenarios or infer cooling benefits solely based on the current distribution of green spaces. They rarely conduct quantitative assessments of the cooling potential of physically modifiable fragmented spaces in existing urban areas after feasible greening renovations. Consequently, they struggle to support the refined zoning and strategy formulation of newly added green spaces in urban renewal. Summary of the Invention

[0004] Purpose of the invention: The purpose of this invention is to provide a method for intervening in the zoning of greenable land in urban built-up areas, in order to solve the problems existing in the prior art, conduct urban thermal risk assessment, identify the potential cooling potential of newly added green spaces, and realize green space intervention decisions oriented towards urban zoning.

[0005] Technical Solution: The urban built-up area greenable land intervention zoning method of the present invention includes obtaining the surface temperature of the urban built-up area, calculating the thermal hazard index representing the neighborhood-scale thermal environment, obtaining spatial distribution data of the residential and employed population of the urban built-up area, calculating the thermal exposure index, obtaining spatial distribution data of the elderly population and housing prices of the urban built-up area, constructing a thermal vulnerability index, multiplying the thermal hazard index, exposure index, and vulnerability index and taking the cube root to obtain a comprehensive thermal risk index, extracting urban morphological data using high-resolution satellite remote sensing imagery, marking areas with specific area and shape constraints as greenable land, and dividing the study area into several square grids. For each decision-making unit, morphological indicators are calculated for green spaces, impermeable surfaces, water bodies, open spaces, and buildings. Simultaneously, the average surface temperature of each decision-making unit is statistically analyzed. Using the average surface temperature of the decision-making unit as the dependent variable and urban morphological indicators as independent variables, a machine learning regression model is trained. The morphological indicators of the decision-making unit are calculated under two scenarios: the current situation and a greening scenario where all available green space is considered as newly added green space. These are then input into the machine learning model to calculate the predicted surface temperature under both scenarios. The difference between the two scenarios is taken as the cooling potential of the decision-making unit. Based on the comprehensive thermal risk index and cooling potential, a risk-potential combination type is constructed for the decision-making unit, and intervention zones are defined.

[0006] Preferably, the calculation of the thermal hazard index representing the neighborhood-scale thermal environment includes: acquiring summer clear-sky thermal infrared remote sensing images covering the study area; inverting the images to obtain a surface temperature grid; and using the surface temperature grid as input for each grid cell. In a radius of Kernel smoothing is performed within the neighborhood of the target to obtain thermal environment values ​​characterizing the neighborhood-scale thermal environment. The smoothing process uses a kernel-weighted average method, and the calculation formula is as follows:

[0007]

[0008] in, Number the target grid cells. Number the grid cells in the neighborhood of the target grid cell. For grid cells The original surface temperature value; For grid cells With grid cells The center distance; The kernel function bandwidth corresponds to approximately the daily neighborhood activity range of residents. The kernel function is of the fourth order, and its expression is:

[0009]

[0010] Finally, the obtained neighborhood-scale thermal environment values ​​were analyzed. Linear normalization was performed to obtain the thermal hazard index.

[0011] Preferably, the calculation of the heat exposure index includes: obtaining spatial distribution data of the resident and employed population in the study area based on regional mobile phone signaling data or population census data, and calculating the spatial distribution data for each population distribution location. Extract the number of residents respectively With the number of employed people The percentage of daily activity time of residents in the statistical area is used to calculate the weighted population value for each population location by weighting the residential exposure and employment exposure. The weighted summation formula is as follows:

[0012]

[0013] In the formula, This is a weighted population calculation value. The residential exposure weight is calculated as the percentage of time residents spend in their place of residence. The employment exposure weight is calculated as the percentage of time residents spend in their place of employment, using the same spatial scale as the land surface temperature grid, and applied to the weighted population value. Kernel density estimation is performed to transform discrete weighted population values ​​into a continuous spatial distribution of population exposure, resulting in a continuous surface of population exposure. The population exposure density at each analysis grid cell is calculated using the following formula:

[0014]

[0015] in, Population exposure density To analyze the grid cell numbering, For population location With analysis grid cells The center distance, For kernel function bandwidth, The kernel function is of the fourth order, and finally the population exposure density is analyzed for all grid cells. After normalization, the thermal exposure index is obtained.

[0016] Preferably, the construction of the thermal vulnerability index includes: acquiring population point data with an age field; using a time-weighted method; calculating the residential and work-weighted exposure of the elderly population at each population location based on the population point data; using the same unit spatial scale as the land surface temperature raster; and employing a kernel function. Kernel density smoothing is applied to the elderly population exposure grid to obtain an elderly population exposure continuity surface. The proportion of the elderly population in each analysis grid cell is calculated, which is the ratio of the elderly population exposure density to the total population exposure density in that grid cell. Housing price point data for the study area are obtained, and a housing price continuity grid is generated using ordinary Kriging interpolation and aligned with the surface temperature grid. The housing price grid at the grid scale is then reverse-processed. Cells with lower housing prices correspond to higher socioeconomic vulnerability. The proportion of the elderly population and the reverse-processed housing price data are normalized and then averaged with equal weights to obtain the thermal vulnerability index.

[0017] Preferably, the extraction of urban morphology data includes: classifying land features in the study area based on high-resolution satellite remote sensing image data, identifying vector or raster mask layers, overlaying the mask layers corresponding to all core land feature categories to construct a comprehensive exclusion mask for non-greenable areas, and, in conjunction with planning data and manual interpretation results of remote sensing images, including construction sites and areas with short-term land use uncertainty in the comprehensive exclusion mask. In the remaining areas outside the comprehensive exclusion mask, spatial filtering methods are used to remove narrow, fragmented areas and isolated small patches, and the retained continuous areas are defined as greenable land.

[0018] Preferably, the classification of land features is based on buildings, roads and their right-of-way, water bodies, existing green spaces, and sports fields.

[0019] Preferably, the morphological indicators include two-dimensional morphological feature indicators and three-dimensional morphological feature indicators, including at least one or more of area or area ratio, boundary features, shape complexity, connectivity, and height.

[0020] Preferably, the machine learning regression model is a support vector regression model, and hyperparameter optimization is performed through spatial grouped cross-validation. The model performance is evaluated by the coefficient of determination R2, root mean square error, and mean absolute error.

[0021] Preferably, the calculation of the predicted surface temperature under the two scenarios includes: under the current situation scenario, using the morphological indicators within each decision-making unit as input features, constructing the current situation morphological feature set corresponding to each decision-making unit. ,in Assign a number to each decision-making unit, and treat all greenable land identified in the aforementioned greenable land identification step as newly added green space. Based on this greening scenario, recalculate and update the landscape pattern and morphological indicators of each decision-making unit to obtain the morphological feature set of each decision-making unit under the greening scenario. The set of current morphological features and Greening Scene Morphological Feature Set Input the best-performing machine learning regression model trained and selected in the previous model training steps to obtain the predicted surface temperature for each decision unit under the current scenario. And predicted surface temperature under greening scenarios The difference between the predicted surface temperatures under the two scenarios is calculated as the cooling potential of the corresponding decision-making unit. The cooling potential is calculated using the following formula:

[0022]

[0023] in, For the first The cooling potential of a decision-making unit is a measure of its ability to reduce temperature. The higher the value, the greater the marginal cooling benefit that the unit can obtain after implementing greening transformation.

[0024] Preferably, the step of constructing risk-potential combination types for decision-making units includes: for each decision-making unit, obtaining the thermal risk index and cooling potential of the decision-making unit obtained in the steps described above; based on the distribution of thermal risk indices of all decision-making units in the study area, using the first and second ternary digits as grading thresholds, dividing the thermal risk index of each decision-making unit into three levels: low risk, medium risk, and high risk; based on the distribution of cooling potential of all decision-making units in the study area, using the first and second ternary digits as grading thresholds, dividing the cooling potential of each decision-making unit into three levels: low potential, medium potential, and high potential; forming nine standardized thermal risk-cooling potential combination types based on thermal risk level and cooling potential level; and mapping each decision-making unit to the thermal risk-cooling potential combination type to form differentiated greening intervention zones.

[0025] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: 1. It realizes the effective connection between risk identification and decision intervention, and can form zoning results that can be directly applied; 2. It predicts the maximum cooling potential of each unit through machine learning regression model; 3. It combines thermal risk level and cooling potential level to provide a clear quantitative basis for the allocation of new green space under the limited resource conditions of existing built-up areas. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating the data synchronization process of the present invention.

[0027] Figure 2 This is a schematic diagram of the spatial distribution of the urban thermal hazard index according to the present invention;

[0028] Figure 3 This is a schematic diagram of the spatial distribution of the urban heat exposure index according to the present invention;

[0029] Figure 4 This is a schematic diagram of the spatial distribution of the urban thermal vulnerability index according to the present invention;

[0030] Figure 5 This is a schematic diagram of the spatial distribution of the urban comprehensive thermal risk index according to the present invention;

[0031] Figure 6 This is a schematic diagram of the process for identifying greenable land use according to the present invention;

[0032] Figure 7 This is a schematic diagram of the spatial distribution of urban cooling potential according to the present invention;

[0033] Figure 8 This is a schematic diagram of the urban thermal risk-cooling potential combination zoning according to the present invention. Detailed Implementation

[0034] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0035] like Figure 1 As shown, this embodiment of the invention provides a method for intervening in the zoning of greenable land in urban built-up areas, combining thermal risk and cooling potential, including the following steps:

[0036] (1) Obtain the surface temperature of the study area and calculate the thermal hazard index representing the neighborhood-scale thermal environment. Specifically, obtain Landsat9 satellite thermal infrared remote sensing images of the summer clear sky conditions covering the study area, perform radiometric calibration and atmospheric correction on the images, and obtain the surface temperature grid by using the radiative transfer equation and Planck formula based on the surface emissivity estimation results.

[0037] Using the inverted surface temperature grid as input, for each grid cell In a radius of Kernel smoothing is performed within the neighborhood of the target to obtain thermal environment values ​​characterizing the neighborhood-scale thermal environment. The smoothing process uses a kernel-weighted average method, calculated using the following formula:

[0038]

[0039] In the formula, Number the target grid cells. Number the grid cells in the neighborhood of the target grid cell; For grid cells The original surface temperature value; For grid cells With grid cells The center distance; The kernel bandwidth is set to 400m (equivalent to approximately a 5-minute walk, representing the daily neighborhood activity range of residents). For the kernel function, a fourth-order kernel function is used, and its expression is:

[0040]

[0041] Neighborhood-scale thermal environment values Linear normalization is performed to obtain thermal hazard indicators, such as Figure 2 As shown.

[0042] (2) Obtain spatial distribution data of the resident population and the employed population, and calculate the heat exposure index. The specific process is as follows:

[0043] Acquire spatial distribution data of the residential and employed population in the study area (e.g., extracted from mobile phone signaling data or census data), and project it onto a spatial reference frame consistent with the analysis grid; for each population distribution location... Extract the number of residents respectively With the number of employed people (in (for population location numbering).

[0044] Based on the proportion of time residents spend in daily activities, a weighted sum of residential and employment exposures is calculated to obtain the weighted population value for each population location. The preferred method is to calculate using the following formula:

[0045]

[0046] In the formula, For residential exposure weighting, The employment exposure weight is calculated as the percentage of time residents spend in their place of residence and place of employment.

[0047] For the weighted population value Kernel density estimation is performed to transform discrete weighted population values ​​into a continuous spatial distribution of population exposure, resulting in a population exposure continuum; where each analysis grid cell... Population exposure density at the location ( The analysis grid cell number is calculated using the following formula:

[0048]

[0049] In the formula, For population location With analysis grid cells The center distance; The kernel bandwidth is preferably 400m (corresponding to approximately a 5-minute walking distance, matching the daily activity range of residents); It is the same kernel function as in step S1.

[0050] Population exposure density for all analyzed grid cells After normalization, heat exposure indices are obtained, such as... Figure 3 As shown.

[0051] (3) Obtain spatial distribution data of the elderly population and housing prices, and construct a thermal vulnerability index, specifically:

[0052] Retrieve population point set data with age field;

[0053] Using the same time-weighted method as in step S2, the residential and work-weighted exposures of the elderly population at each population location were calculated.

[0054] Using the same spatial scale as the raster analyzed in step S1, and using the same kernel function as in step S1, the elderly population exposure raster is subjected to kernel density smoothing to obtain the elderly population exposure continuity surface;

[0055] Calculate the proportion of elderly population in each analysis grid cell, which is the ratio of the elderly population exposure density in that grid cell to the total population exposure density calculated by S2.

[0056] Acquire housing price point data for the study area, generate a continuous housing price raster using ordinary kriging interpolation, and align it with the analyzed raster in step S1.

[0057] The housing price grid at the aforementioned grid scale is reversed so that lower housing prices correspond to higher socioeconomic vulnerability.

[0058] The proportion of the elderly population and the housing price data after reverse processing were normalized and then averaged with equal weights to obtain the thermal vulnerability index, such as... Figure 4 As shown;

[0059] (4) Multiply the thermal hazard index, exposure index, and vulnerability index together and take the cube root to obtain the comprehensive thermal risk index, such as... Figure 5 As shown; the comprehensive thermal risk index is composed of the thermal hazard index, thermal exposure index, and thermal vulnerability index through multiplication. This method, compared to the additive summation method, better captures the interdependencies between them. The specific calculation formula is as follows:

[0060]

[0061] (5) Using high-resolution satellite remote sensing images to extract urban morphology data, masking areas such as buildings, roads, water bodies, existing green spaces, and sports fields, and identifying the remaining areas that meet the area and shape constraints as land suitable for greening, the specific process is as follows:

[0062] Based on high-resolution remote sensing imagery, a deep learning semantic segmentation model is used to classify ground features and identify vector or raster mask layers for core ground feature categories such as buildings, roads and their right-of-way, water bodies, existing green spaces, and sports fields.

[0063] The mask layers corresponding to all the core land cover categories are overlaid to construct a comprehensive exclusion mask for non-green areas;

[0064] Based on planning data and manual interpretation of remote sensing images, construction sites and other areas with short-term land use uncertainty will be included in the comprehensive exclusion mask.

[0065] In the remaining area outside the comprehensive exclusion mask, a spatial filtering method is used to remove narrow, fragmented areas and isolated small patches;

[0066] The preserved continuous areas are designated as land suitable for greening, and the complete identification process is as follows: Figure 6 As shown.

[0067] (6) The study area was divided into several decision units by a square grid of 150m x 150m. Morphological indicators of green space, impermeable surface, water body, open space and buildings within each decision unit were calculated. At the same time, the average surface temperature of each decision unit was calculated (the average value of the original surface temperature grid within the unit was taken without kernel smoothing). The morphological indicators used specifically included: green space area ratio, maximum green space patch index, number of green space patches, total length of green space boundary, total green space core area, effective grid area of ​​green space, green space shape index, building density, average building height, effective grid area of ​​impermeable surface, maximum impermeable surface patch index, impermeable surface shape index, water body shape index, effective grid area of ​​open space, water body area ratio, and average distance to the shoreline of the water body.

[0068] (7) Using the average surface temperature of the decision-making unit as the dependent variable and urban morphology indicators as the independent variables, a machine learning regression model was trained. The regression model was specifically a support vector regression model, which is suitable for relatively smooth nonlinear relationship models. When constructing the model, hyperparameter optimization was performed through spatial grouping cross-validation. The cross-validation fold was selected as 5 folds. The grouping basis was to divide the study area into spatial partition numbers with a coarser scale than the decision-making unit to reduce information leakage caused by spatial autocorrelation. The model performance was measured by the coefficient of determination R. 2 The root mean square error and mean absolute error are evaluated.

[0069] (8) Under the current situation and the greening scenario where all the greenable land is regarded as newly added green space, the morphological index of the decision unit is calculated respectively, and the result is input into the machine learning model to obtain the predicted surface temperature under the two scenarios. The difference between the two is taken as the cooling potential of the decision unit. The specific process is as follows:

[0070] In the current situation, using the morphological indicators within each decision-making unit as input features, a set of current situation morphological features corresponding to each decision-making unit is constructed. (in (Numbering the decision-making unit).

[0071] All greenable land identified in the aforementioned greenable land identification step is considered as newly added green space. Based on this greening scenario, the landscape pattern and morphological indicators of each decision-making unit are recalculated and updated to obtain the morphological feature set of each decision-making unit under the greening scenario. ;

[0072] The current state feature set and Greening Scene Morphological Feature Set Input the best-performing machine learning regression model trained and selected in the previous model training steps to obtain the predicted surface temperature for each decision unit under the current scenario. And predicted surface temperature under greening scenarios ;

[0073] The difference between the predicted surface temperatures under the two scenarios is calculated as the cooling potential of the corresponding decision-making unit. The cooling potential is calculated using the following formula:

[0074]

[0075] in, For the first The cooling potential of each decision-making unit is represented by a higher value, indicating a greater marginal cooling benefit that can be obtained after implementing greening renovations. The spatial distribution of the cooling potential is shown in the figure below. Figure 7 As shown.

[0076] (9) Based on the comprehensive thermal risk index and cooling potential, risk-potential combination types are constructed for decision-making units, and intervention zones are divided. The specific process is as follows:

[0077] For each decision-making unit, obtain the comprehensive thermal risk index and cooling potential of that decision-making unit obtained from the steps described above;

[0078] Based on the distribution of the comprehensive thermal risk index of all decision-making units in the study area, and using the first and second tertiles as the grading thresholds, the comprehensive thermal risk index of each decision-making unit is divided into three levels: low risk, medium risk, and high risk.

[0079] Based on the distribution of cooling potential of all decision-making units in the study area, and using the first and second tertiles as the grading thresholds, the cooling potential of each decision-making unit is divided into three levels: low potential, medium potential, and high potential.

[0080] Nine standardized risk-potential combination types are formed based on thermal health risk level and cooling potential level, specifically including: low risk-low potential, low risk-medium potential, low risk-high potential, medium risk-low potential, medium risk-medium potential, medium risk-high potential, high risk-low potential, high risk-medium potential, and high risk-high potential. Each decision-making unit is mapped to one of these thermal risk-cooling potential combination types to form differentiated greening intervention zones, such as... Figure 8 As shown;

[0081] Green space intervention zones can be used to develop differentiated intervention strategies. For high-risk units with high cooling potential, green space intervention is highly necessary and expected to have a significant cooling effect, making them key intervention targets. For high-risk units with low cooling potential, due to limitations in form and space, the marginal benefits of green space are low, and non-structural measures such as shading facilities, misting cooling, temporary cooling shelters, or emergency cooling services can be combined. For medium-risk units, phased or combined intervention strategies can be selected based on their potential level. For low-risk units with high cooling potential, they can be used as forward-looking adaptation reserve areas for preventative enhancement. For low-risk units with low cooling potential, long-term monitoring and routine maintenance can be emphasized.

Claims

1. A method for intervening in the zoning of greenable land in urban built-up areas, characterized in that, The method specifically includes the following steps: (1) Obtain the surface temperature of the urban built-up area and calculate the thermal hazard index representing the neighborhood-scale thermal environment; (2) Obtain spatial distribution data of the urban built-up area's residential and employed population, and calculate heat exposure indicators; (3) Obtain spatial distribution data of the elderly population and housing prices in urban built-up areas, and construct a thermal vulnerability index; (4) Multiply the thermal hazard index, exposure index and vulnerability index together and take the cube root to obtain the comprehensive thermal risk index; (5) Use high-resolution satellite remote sensing images to extract urban morphology data and mark areas with specific area and shape constraints as greenable land; (6) Divide the study area into several decision units using a square grid, calculate the morphological indicators of green space, impermeable surface, water body, open space and buildings within each decision unit, and simultaneously calculate the average surface temperature of each decision unit. (7) Using the average surface temperature of the decision-making unit as the dependent variable and the urban morphology index as the independent variable, train a machine learning regression model; (8) Calculate the decision unit morphology index under the current situation and the greening scenario where all the greenable land is regarded as new green space, respectively, and input it into the machine learning model to calculate the predicted surface temperature under the two scenarios. The difference between the two is used as the cooling potential of the decision unit. (9) Based on the comprehensive thermal risk index and cooling potential, risk-potential combination types are constructed for decision-making units, and intervention zones are divided.

2. The method for intervening in the zoning of greenable land in urban built-up areas according to claim 1, characterized in that, The thermal hazard indicators representing the neighborhood-scale thermal environment include: (1) Acquire thermal infrared remote sensing images of summer clear sky conditions covering the study area, and invert the influence to obtain the surface temperature grid; (2) Using the surface temperature grid as input, for each grid cell In a radius of Kernel smoothing is performed within the neighborhood of the target to obtain thermal environment values ​​characterizing the neighborhood-scale thermal environment. The smoothing process uses a kernel-weighted average method, and the calculation formula is as follows: in, Number the target grid cells. Number the grid cells in the neighborhood of the target grid cell. For grid cells The original surface temperature value; For grid cells With grid cells The center distance; The kernel function bandwidth corresponds to approximately the daily neighborhood activity range of residents. The kernel function is of the fourth order, and its expression is: (3) The obtained neighborhood-scale thermal environment values Linear normalization was performed to obtain the thermal hazard index.

3. The method for intervening in the zoning of greenable land in urban built-up areas according to claim 1, characterized in that, The calculated heat exposure index includes: (1) Based on regional mobile phone signaling data or population census data, obtain spatial distribution data of the resident population and employed population in the study area, and for each population distribution location Extract the number of residents respectively With the number of employed people ; (2) The proportion of daily activity time of residents in the statistical area is calculated, and the residential exposure and employment exposure are weighted and summed to obtain the weighted population value for each population location. The weighted summation formula is as follows: in, This is a weighted population calculation value. The residential exposure weight is calculated as the percentage of time residents spend in their place of residence. The employment exposure weight is calculated as the percentage of time residents spend in their place of employment. (3) Using the same unit spatial scale as the surface temperature grid, the weighted population value is... Kernel density estimation is performed to transform discrete weighted population values ​​into a continuous spatial distribution of population exposure, resulting in a continuous surface of population exposure. The population exposure density at each analysis grid cell is calculated using the following formula: in, Population exposure density To analyze the grid cell numbering, For population location With analysis grid cells The center distance, For kernel function bandwidth, It is a fourth-order kernel function; (4) Population exposure density for all analyzed grid cells After normalization, the thermal exposure index is obtained.

4. The method for intervening in the zoning of greenable land in urban built-up areas according to claim 1, characterized in that, The thermal vulnerability indicators include: (1) Obtain population point set data with age field, and use time weighting method to calculate the residential and working weighted exposure of the elderly population at each population location based on the population point set data; (2) The same spatial scale as the surface temperature grid is used, and the kernel function is employed. Kernel density smoothing is applied to the elderly population exposure grid to obtain a continuous surface of elderly population exposure. (3) Calculate the proportion of elderly population in each analysis grid cell, that is, the ratio of the elderly population exposure density in that grid cell to the total population exposure density; (4) Obtain housing price point data for the study area, generate a continuous housing price grid using ordinary kriging interpolation, and align it with the surface temperature grid. (5) The housing price grid at the grid scale is processed in reverse. The lower the housing price, the higher the socioeconomic vulnerability of the corresponding unit. (6) Normalize the proportion of elderly population and the housing price data after reverse processing, and then perform equal weight averaging to obtain the thermal vulnerability index.

5. The method for intervening in the zoning of greenable land in urban built-up areas according to claim 1, characterized in that, The extracted urban morphology data includes: (1) Based on high-resolution satellite remote sensing image data, land cover classification was carried out in the study area to identify vector or raster mask layers; (2) Overlay the mask layers corresponding to all the core land cover categories to construct a comprehensive exclusion mask for non-green areas; (3) Combining planning data with the results of manual interpretation of remote sensing images, construction sites and areas with short-term land use uncertainty are included in the comprehensive exclusion mask; (4) In the remaining area outside the comprehensive exclusion mask, a spatial filtering method is used to remove narrow, fragmented areas and isolated small patches; (5) Define the retained continuous area as land suitable for greening.

6. The method for intervening in the zoning of greenable land in urban built-up areas according to claim 5, characterized in that, The classification of land features is based on buildings, roads and their right-of-way, water bodies, existing green spaces, and sports fields.

7. The method for intervening in the zoning of greenable land in urban built-up areas according to claim 1, characterized in that, The morphological indicators include two-dimensional morphological feature indicators and three-dimensional morphological feature indicators, including at least one or more of the following: area or area ratio, boundary features, shape complexity, connectivity, and height.

8. The method for intervening in the zoning of greenable land in urban built-up areas according to claim 1, characterized in that, The machine learning regression model is a support vector regression model, and hyperparameter optimization is performed through spatial grouped cross-validation. The model performance is measured by the coefficient of determination R. 2 The root mean square error and mean absolute error are evaluated.

9. The method for intervening in the zoning of greenable land in urban built-up areas according to claim 1, characterized in that, The calculated predicted surface temperatures under the two scenarios include: (1) Under the current situation, the morphological indicators within each decision-making unit are used as input features to construct the current situation morphological feature set corresponding to each decision-making unit. ,in Number the decision-making units; (2) All the greenable land identified in the aforementioned greenable land identification step is regarded as newly added green space. Based on this greening scenario, the landscape pattern and morphological indicators of each decision-making unit are recalculated and updated to obtain the morphological feature set of each decision-making unit under the greening scenario. ; (3) The set of current morphological features and Greening Scene Morphological Feature Set Input the best-performing machine learning regression model trained and selected in the previous model training steps to obtain the predicted surface temperature for each decision unit under the current scenario. And predicted surface temperature under greening scenarios ; (4) Calculate the difference in predicted surface temperature under the two scenarios, which is taken as the cooling potential of the corresponding decision-making unit. The cooling potential is calculated by the following formula: in, For the first The cooling potential of a decision-making unit is a measure of its ability to reduce temperature. The higher the value, the greater the marginal cooling benefit that the unit can obtain after implementing greening renovations.

10. The method for intervening in the zoning of greenable land in urban built-up areas according to claim 1, characterized in that, The types of risk-potential combinations constructed for decision-making units include: (1) For each decision-making unit, obtain the comprehensive thermal risk index and cooling potential of the decision-making unit obtained in the steps described above; (2) Based on the distribution of the comprehensive thermal risk index of all decision-making units in the study area, the comprehensive thermal risk index of each decision-making unit is divided into three levels: low risk, medium risk and high risk, according to the first tertile and the second tertile as the grading threshold. (3) Based on the distribution of cooling potential of all decision-making units in the study area, the cooling potential of each decision-making unit is divided into three levels: low potential, medium potential and high potential, according to the first tertile and the second tertile as the grading threshold. (4) Nine standardized thermal risk-cooling potential combination types are formed by thermal risk level and cooling potential level; (5) Map each decision unit to the heat risk-cooling potential combination type to form differentiated greening intervention zones.

Images