Construction method of urban ventilation corridors based on local climate zone classification and remote sensing inversion

Abstract

The method for constructing urban ventilation corridors based on local climate zone classification and remote sensing inversion belongs to the technical field of urban climate regulation and space optimization. The present application constructs a multi-scale urban thermal environment analysis framework, from macro, meso to micro level, to identify the coupling mechanism of urban form and thermal effect, and to realize the fine modeling and regulation path identification of thermal environment. At the same time, combined with machine learning and explainable analysis method, the scientificity and transparency of thermal environment driving factor identification are enhanced. Through remote sensing inversion, form factor construction and three-dimensional simulation simulation, a complete technical path of "cold source identification-resistance modeling-path optimization-microclimate response evaluation" is put forward, which effectively improves the ventilation and cooling potential identification ability of blue-green infrastructure, and is suitable for ventilation corridor planning and climate adaptive updating design of various urban spaces, and has wide application prospect.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of urban climate regulation and space optimization, and particularly relates to a method for constructing urban ventilation corridors based on local climate zone classification and remote sensing inversion. BACKGROUND

[0002] With the continuous advancement of global urbanization, the problem of urban heat island effect (UHI) is becoming increasingly serious. UHI refers to the phenomenon that the surface and air temperature of urban areas are significantly higher than those of surrounding non-urban areas, mainly affected by factors such as impervious surface expansion, building density increase, green space reduction, and anthropogenic heat emission, which poses challenges to the ecological system, energy consumption, and residents' health. Existing research has confirmed that urban spatial structure, land use type, and building form are key factors affecting the urban thermal environment.

[0003] The local climate zone classification system is a standardized framework for classifying cities according to their surface and structural characteristics, providing a unified spatial scale basis for urban thermal environment research. Different LCZ types have significant differences in building height, density, vegetation coverage, and surface material, which directly lead to spatial heterogeneity of land surface temperature (LST) and urban heat island effect (UHI). The LCZ classification method has been widely applied in the fields of urban heat island mechanism identification, thermal environment assessment, and urban climate adaptability planning.

[0004] Blue-green infrastructure, as a natural infrastructure solution, plays an important role in regulating urban thermal environment. Green spaces can reduce local temperature through mechanisms such as evapotranspiration, shading, and increasing air humidity, while water bodies can also have a significant cooling effect due to their high specific heat and evaporation potential. A complex network of blue-green spaces can also improve urban ventilation, creating a connected structure of cold island cores and wind corridor channels. Related research shows that blue-green infrastructure can effectively alleviate the intensity of urban heat islands and has good ecological and social synergistic benefits. SUMMARY

[0005] To address the above problems in the prior art, the purpose of the present application is to provide a method for optimizing wind corridors and assessing thermal environment based on local climate zone (LCZ) classification, which realizes systematic identification of urban thermal environment response mechanisms and optimization of blue-green infrastructure layout by constructing a multi-source data system and a microclimate simulation model.

[0006] The present application provides the following technical solution: a method for constructing urban ventilation corridors based on local climate zone classification and remote sensing inversion, comprising the following steps:

[0007] S1, obtaining remote sensing images, building form data, and meteorological observation data of the study area, and preprocessing the obtained data;

[0008] S2. Extract urban morphology factors from the preprocessed data, and perform spatial aggregation and normalization on the local climate zone LCZ standard grid. Use the K-means++ algorithm to classify LCZ and obtain the LCZ type map.

[0009] S3. Based on remote sensing data, land surface temperature (LST) is retrieved, and the difference between single pixel LST and regional average LST (ΔLST) is calculated to achieve cold and heat island identification.

[0010] S4. Based on the morphological factors extracted in S2 and the LST obtained in S3, a random forest nonlinear regression model is established to obtain the ranking of variable importance.

[0011] S5. Construct the ventilation resistance surface and use the weighted cost path algorithm to identify the primary and secondary ventilation corridor network;

[0012] S6. Select the LCZ area to conduct ENVI-met three-dimensional microclimate simulation to evaluate the thermal environment response and thermal comfort changes under different corridor layouts.

[0013] Furthermore, in S1, the remote sensing data selected is Landsat-8 imagery, the building morphology data includes building outline, roof area and height, and the meteorological data selected is temperature, wind speed, and humidity parameters from a typical high-temperature day; the data preprocessing process is as follows:

[0014] For Landsat 8 remote sensing imagery: radiometric calibration, atmospheric correction, and projection unification were performed sequentially; land surface temperature (LST) was retrieved, and normalized vegetation index (NDVI) was calculated. For building morphology data: after projection transformation, building density (BD), average height (AH), and sky openness (SVF) were calculated on the grid and normalized. For meteorological data: the format was standardized, and temperature, humidity, and wind parameters matching the remote sensing transit time were extracted as boundary conditions. All data were finally spatially registered and cropped to form a comprehensive dataset with standardized scale and spatial alignment.

[0015] Furthermore, in S2, building height (AH), building density (BD), normalized difference vegetation index (NDVI), and sky view factor (SVF) are calculated using remote sensing and building vector data, respectively. Spatial aggregation processing is performed within a set grid scale, and unsupervised classification is performed using the K-means++ clustering algorithm optimized by Mahalanobis distance. Building and non-building patches are distinguished by combining the building density threshold, and finally, an LCZ type layer is generated.

[0016] Furthermore, in S3, Landsat8 thermal infrared band is used to retrieve LST, and non-surface areas are excluded through atmospheric correction and NDVI masking. The difference ΔLST between each pixel's LST and the regional average LST is calculated, and the area is divided into cold island area and non-cold island area based on the positive and negative values, thereby realizing the spatial thermal environment zoning.

[0017] Furthermore, in step S4, relevant variables within the LCZ unit are selected as input factors, LST is the output variable, the random forest algorithm is used to train the model, the fitting performance is evaluated by out-of-bag error, and the contribution of each variable to the change of LST is extracted and ranked to identify the dominant factors affecting the thermal environment.

[0018] Furthermore, in S5, relevant factors are introduced when constructing the ventilation resistance surface, and resistance weights are assigned according to different surface types; combined with the prevailing wind direction data of the city, the minimum cumulative cost path (LCP) algorithm is used to identify and visualize the optimal paths of primary and secondary ventilation corridors, thereby realizing the identification of the thermal-wind spatial structure.

[0019] Furthermore, in S6, LCZ sample areas are selected based on cold island and heat island regions, three-dimensional models of buildings, green spaces, and water bodies are constructed, imported into ENVI-met for high-temperature day simulation, meteorological boundary conditions and thermal parameters are set, results are output, and the differences in microclimate response under different wind corridor structures and greening configurations are compared to evaluate their cooling effect and ventilation efficiency.

[0020] The method of this invention constructs a three-level data coupling architecture of "macro-meso-micro", extracts thermal environment influencing factors in layers at different spatial scales, and realizes data transmission from top to bottom and result verification from bottom to top in steps S1 to S6, forming a mutually verifying and closed-loop feedback urban thermal environment regulation system.

[0021] At the macro level, ecological thermal environment indicators such as land surface temperature (LST) and normalized difference vegetation index (NDVI) are obtained through Landsat 8 remote sensing images. Cold island and heat island areas are identified by combining ΔLST difference calculation, forming a regional scale cold source distribution map and ventilation corridor skeleton network, providing a basis for identifying thermal environment patterns from a global perspective, mainly corresponding to steps S1, S3 and S5.

[0022] At the meso-level, based on the macro-identification results, urban morphology factors (AH, BD, SVF, NDVI) are extracted using the LCZ standard grid (100m×100m). A nonlinear random forest model is used to evaluate the driving effect of each variable on LST. Furthermore, a ventilation resistance surface is constructed and the wind corridor path structure is optimized to achieve meso-scale modeling of the regional wind environment structure and thermal environment mechanism, which mainly corresponds to steps S2, S4 and S5.

[0023] At the micro level, typical LCZ sample areas at the meso scale are selected to carry out ENVI-met three-dimensional microclimate simulation, construct a spatial physical model including buildings, green spaces and water bodies, simulate the temperature, wind speed and thermal comfort response under different wind corridor and greening configurations, and verify the regulatory effect of macro identification and meso modeling, which mainly corresponds to step S6.

[0024] The three types of data logically form a multi-level linkage mechanism of "macroscopic identification – mesoscopic analysis – microscopic verification": the macroscopic level provides problem location and overall pattern, the mesoscopic level reveals causal mechanism and proposes optimization path, and the microscopic level conducts physical verification and effect evaluation. Together, the three constitute a unified, efficient, and feedback-iterative urban thermal environment regulation technology system.

[0025] By employing the above-described technology, the beneficial effects of the present invention compared to the prior art are as follows:

[0026] 1) This invention constructs a multi-scale urban thermal environment analysis framework, systematically identifying the coupling mechanism between urban morphology and thermal effects from macroscopic, mesoscopic to microscopic levels, and realizing refined modeling and control path identification of the thermal environment;

[0027] 2) This invention combines machine learning and interpretability analysis methods to enhance the scientific rigor and transparency of identifying thermal environment driving factors. By combining remote sensing inversion, morphological factor construction, and 3D simulation, it proposes a complete technical path of "cold source identification - resistance modeling - path optimization - microclimate response assessment", which effectively improves the ability to identify the ventilation and cooling potential of blue-green infrastructure. It is applicable to wind corridor planning and climate-adaptive renewal design of various urban spaces and has broad prospects for promotion and application. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating the operation of the urban ventilation corridor construction method based on local climate zone (LCZ) classification and remote sensing inversion according to the present invention.

[0029] Figure 2 Visualization of building height (AH) in this invention;

[0030] Figure 3 This is a map showing the distribution of the LCZ in the study area of ​​this invention.

[0031] Figure 4 This invention defines the hot and cold islands in the research area.

[0032] Figure 5 Visualization of the wind corridor in the study area in this invention;

[0033] Figure 6 The temperature within each LCZ in this invention;

[0034] Figure 7 This is a scatter plot comparing the actual and predicted values ​​in the test set of the XGBoost model of this invention.

[0035] Figure 8 This is a scatter plot comparing the actual values ​​and predicted values ​​in the test set in the Random Forest (RF) model of this invention.

[0036] Figure 9This is a scatter plot comparing the actual and predicted values ​​in the test set in the Gradient Boosting Decision Tree (GBDT) of this invention.

[0037] Figure 10 This is a schematic diagram of SHAP values ​​in the XGBoost+SHAP nonlinear correlation analysis of this invention;

[0038] Figure 11 This is a schematic diagram of the absolute SHAP mean in the XGBoost+SHAP nonlinear correlation analysis of this invention;

[0039] Figure 12 This is a schematic diagram of the mean absolute SHAP value in the XGBoost+SHAP nonlinear correlation analysis of this invention;

[0040] Figure 13 This invention utilizes LCP to construct wind corridors;

[0041] Figure 14 This is the Euclidean distance drag model in this invention;

[0042] Figure 15 In this invention, BD is used as a function of the buffer distance;

[0043] Figure 16 The temperature simulation results for the LCZ2 non-cold island at 15:00 on August 10, 2023, in this invention;

[0044] Figure 17 This is the MLST dependency graph in this invention;

[0045] Figure 18 This is the NDVI dependency graph in this invention;

[0046] Figure 19 This is the NDWI variant in this invention, and the MNDWI dependency graph;

[0047] Figure 20 This is the temperature simulation result of the non-cold island cluster at 13:00 in this invention;

[0048] Figure 21 This is the simulation result of the wind speed of the non-cold island cluster at 13:00 in this invention;

[0049] Figure 22 The wind speed simulation results for the LCZ2 non-cold island at 15:00 on August 10, 2023 are shown in this invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0051] Conversely, this invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of the invention as defined in the claims. Furthermore, to provide a better understanding of the invention, certain specific details are described in detail below. However, those skilled in the art will fully understand the invention even without these detailed descriptions.

[0052] Taking City A as the research object, a method for constructing urban ventilation corridors based on local climate zone (LCZ) classification and remote sensing inversion is provided for urban thermal environment regulation, specifically including the following:

[0053] Reference Figure 1 The method flow shown in this embodiment includes the following specific steps:

[0054] Taking the nine urban districts of City A as the study area, the typical landscape pattern and complex urban form provide a spatial basis for thermal environment analysis.

[0055] S1. Data Collection and Preprocessing: A multi-source data system was constructed, including Landsat 8 remote sensing imagery (30m), roof area data, 1m land use map, and 12.5m elevation data. Landsat data was used to retrieve land surface temperature (LST) and calculate NDVI, roof area was used to estimate building density (BD), and land use and elevation data were used for resistance surface and slope analysis, respectively. Remote sensing data underwent uniform radiometric and atmospheric correction, and the NDVI threshold method was used to distinguish between built and non-built areas. Average building height (AH), BD, NDVI, and SVF were spatially aggregated and normalized on a 100m × 100m grid. All data were cropped to the built-up area boundary to ensure consistency.

[0056] AH and BD values ​​are concentrated in the main urban area, while NDVI and SVF values ​​are higher in the peripheral areas, reflecting the distribution characteristics of green spaces and open spaces, providing input support for subsequent LCZ classification and thermal environment modeling. Figure 2 As shown, AH visualization is used as an example.

[0057] S2, LCZ Classification and Thermal Environment Analysis: In step S2, this embodiment uses morphological parameters such as building density (BD), building height (AH), normalized difference vegetation index (NDVI), and sky view factor (SVF) to perform LCZ classification using the K-means++ clustering method combined with Mahalanobis distance to generate an urban morphological type layer.

[0058] Classification results as follows Figure 3 As shown, the city is divided into ten main LCZ types, including compact high-rise, open mid-rise, and sparse low-rise, reflecting the spatial structural differences in urban form.

[0059] Building upon this, the present invention further proposes a method for classifying hot and cold islands based on relative surface temperature. For example... Figure 4 As shown, the average temperature of the entire area is calculated using the land surface temperature (LST) retrieved by remote sensing, and the difference between each pixel and the average value (ΔLST) is used as the basis for judging the cold and heat island effect. The calculation of ΔLST is as follows:

[0060] .

[0061] Where ΔLST<0 is defined as a cold island region, and ΔLST≥0 is defined as a non-cold island region.

[0062] By overlaying the LCZ type and the distribution results of hot and cold islands, typical sample groups are selected in each major LCZ to construct representative building blocks for microclimate simulation.

[0063] To identify the role and mechanism of blue-green infrastructure in urban thermal regulation, a cooling island network dominated by blue-green spaces was further constructed, such as... Figure 5 As shown, a complete cold island network map is formed by combining remote sensing inversion LST, morphological spatial pattern analysis (MSPA), and minimum cumulative drag model (MCR).

[0064] Based on the above LCZ classification results and the spatial pattern of hot and cold islands, the LST distribution characteristics within different LCZ types were statistically analyzed, and the results are as follows: Figure 6 As shown in the figure. Analysis shows that LCZ4 (open mid-rise building zone), LCZ6 (low-rise building zone), and LCZ9 (sparse building zone) have the highest overall temperature levels, exhibiting typical characteristics of the heat island effect.

[0065] S3. Mechanism modeling of morphological factors and LST: Pearson correlation analysis was conducted on factors such as building density (BD), building height (AH), and normalized difference vegetation index (NDVI) to preliminarily identify significant correlations among variables. Among them, building density (BD) and average building height (AH) are significantly positively correlated with LST, while normalized difference vegetation index (NDVI) is negatively correlated with LST. That is, high-density built environment significantly warms up, while vegetation cover has a significant cooling effect.

[0066] Building upon linear relationship screening, a nonlinear ensemble learning model is further constructed to uncover complex response mechanisms among variables. Three nonlinear models—XGBoost, Random Forest (RF), and Gradient Boosting Decision Tree (GBDT)—are constructed to predict LST using BD, AH, FAR, NDVI, and SVF as input variables, as shown in Table 1. Figures 7-9 As shown, the model performance is good, with the RF model having the highest R². Using the SHAP value method to interpret variable contributions, NDVI, BD, and MLST were identified as the main driving factors, such as... Figures 10-12As shown, MLST (History Heat Accumulation) is the primary factor influencing the current LST, and the SHAP value remains positive, reflecting the continuous warming trend in high-temperature areas; NDVI contributes negatively in most samples, showing a significant cooling effect; building density (BD) is a typical positive variable, with densely built-up areas having a significant warming effect on local temperatures.

[0067] Table 1 Comparison of the fitting ability of different models on training and test data

[0068] .

[0069] S4. Based on the surface temperature data retrieved from remote sensing, first extract the blue-green spatial patches and their cores and bridging regions, such as... Figure 13 As shown, combining LST hotspot analysis results with LST thermal analysis and topographic NDVI data, a ventilation resistance surface based on the minimum cumulative resistance (MCR) model is constructed. The Linkage Mapper tool is used to simulate ventilation paths, generating multiple wind corridors connecting ecological cold sources.

[0070] Secondly, urban ventilation resistance surfaces are constructed based on DEM, land use, and NDVI, and the optimal wind corridor path is calculated using the minimum cumulative resistance (LCP) model. Wind corridors typically extend along green spaces and water bodies, conforming to the city's natural wind flow paths. For example... Figure 14 As shown, the Euclidean distance model is introduced to verify the continuity of the ventilation structure, and it is divided into three types: continuous, deflected and fractured, to guide the priority of wind corridor network repair.

[0071] Finally, combining microclimate simulation and Figure 15 The building form variables shown are analyzed to assess the cooling and ventilation response effects of each LCZ type under the wind corridor.

[0072] To further evaluate the role mechanism of wind corridors in different LCZ types, a characteristic dependency plot (PDP) and an accumulated local effect plot (ALE) were constructed to analyze the response relationships between multiple morphological and ecological variables. Figure 17 As shown, NDVI, building density, and thermal accumulation value (MLST) exhibit a nonlinear response trend to LST in different value ranges.

[0073] Based on the above results, further response structure analysis was conducted on the main variables. For example... Figure 18 As shown, NDVI significantly reduces temperature in low green coverage areas and tends to stabilize in high green coverage areas; as Figure 19 As shown, the NDWI variant series indices exhibit differences in temperature regulation capabilities among water body types, reflecting their heterogeneous response characteristics.

[0074] S5. Select 10 typical cold island and 10 non-cold island communities, model them using SketchUp, and import them into ENVI-met 5.7 for 3D simulation. Set the simulation horizontal resolution to 5m, and the heat wave period to 13:00–15:00 on August 10, 2023. As shown in Table 2, the input parameters include building structure, surface material, green space configuration, and meteorological conditions, etc.

[0075] Table 2 Overview of ENVI-met simulation parameters and experimental configuration

[0076] ;

[0077] The simulation results of temperature and wind speed at a height of 1.5m were extracted, and the ventilation regulation capabilities of different LCZ types were compared and analyzed. Figure 20 and Figure 16 As shown, the average temperature drop in the cold island area is 2.75–2.86°C, significantly better than that in the non-cold island area. Figure 21 and Figure 22 As shown, the wind speed increased by about 0.11–0.33 m / s, indicating improved ventilation. The results show that the open lower layer (LCZ6) has the best ventilation performance, while the compact middle layer (LCZ2) has a significant heat accumulation effect.

[0078] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing urban ventilation corridors based on local climate zone classification and remote sensing inversion, characterized in that, Includes the following steps: S1. Acquire remote sensing images, building morphology data, and meteorological observation data of the study area, and preprocess the acquired data; S2. Extract urban morphology factors from the preprocessed data, and perform spatial aggregation and normalization on the local climate zone LCZ standard grid. Use the K-means++ algorithm to classify LCZ and obtain the LCZ type map. S3. Based on remote sensing data, land surface temperature (LST) is retrieved, and the difference between single pixel LST and regional average LST (ΔLST) is calculated to achieve cold and heat island identification. S4. Based on the morphological factors extracted in S2 and the LST obtained in S3, a random forest nonlinear regression model is established to obtain the ranking of variable importance. S5. Construct the ventilation resistance surface and use the weighted cost path algorithm to identify the primary and secondary ventilation corridor network; S6. Select the LCZ area to conduct ENVI-met three-dimensional microclimate simulation to evaluate the thermal environment response and thermal comfort changes under different corridor layouts.

2. The method for constructing urban ventilation corridors based on local climate zone classification and remote sensing inversion according to claim 1, characterized in that, In step S1, the remote sensing data selected is Landsat-8 imagery, the building morphology data includes building outline, roof area and height, and the meteorological data selected is temperature, wind speed and humidity parameters from a typical high-temperature day; the data preprocessing process is as follows: For Landsat 8 remote sensing imagery: radiometric calibration, atmospheric correction, and projection unification are performed sequentially; land surface temperature (LST) is retrieved, and normalized vegetation index (NDVI) is calculated. For building morphology data: after projection transformation, the building density (BD), average height (AH), and sky openness (SVF) are calculated on the grid and normalized. For meteorological data: the format is standardized and temperature, humidity, and wind parameters that match the remote sensing transit time are extracted as boundary conditions. All data are finally spatially registered and cropped to form a comprehensive dataset with standardized scale and spatial alignment.

3. The method for constructing urban ventilation corridors based on local climate zone classification and remote sensing inversion according to claim 1, characterized in that, In step S2, building height (AH), building density (BD), normalized difference vegetation index (NDVI), and sky view factor (SVF) are calculated using remote sensing and building vector data, respectively. Spatial aggregation is performed within a set grid scale, and unsupervised classification is performed using the K-means++ clustering algorithm optimized by Mahalanobis distance. Building and non-building patches are distinguished by combining the building density threshold, and finally, an LCZ type layer is generated.

4. The method for constructing urban ventilation corridors based on local climate zone classification and remote sensing inversion according to claim 2, characterized in that, In S3, Landsat 8 thermal infrared band is used to retrieve LST. Non-surface areas are excluded by atmospheric correction and NDVI masking. The difference ΔLST between each pixel's LST and the regional average LST is calculated. Based on the positive and negative values, the area is divided into cold island area and non-cold island area to achieve spatial thermal environment zoning.

5. The method for constructing urban ventilation corridors based on local climate zone classification and remote sensing inversion according to claim 1, characterized in that, In step S4, relevant variables within the LCZ unit are selected as input factors, LST is the output variable, the random forest algorithm is used to train the model, the fitting performance is evaluated by out-of-bag error, and the contribution of each variable to the change of LST is extracted and ranked to identify the dominant factors affecting the thermal environment.

6. The method for constructing urban ventilation corridors based on local climate zone classification and remote sensing inversion according to claim 1, characterized in that, In S5, relevant factors are introduced when constructing the ventilation resistance surface, and resistance weights are assigned according to different surface types. Combined with the prevailing wind direction data of the city, the minimum cumulative cost path (LCP) algorithm is used to identify and visualize the optimal paths of primary and secondary ventilation corridors, thereby realizing the identification of the thermal-wind spatial structure.

7. The method for constructing urban ventilation corridors based on local climate zone classification and remote sensing inversion according to claim 1, characterized in that, In S6, LCZ sample areas are selected based on cold island and heat island regions, three-dimensional models of buildings, green spaces and water bodies are constructed, imported into ENVI-met for high-temperature day simulation, meteorological boundary conditions and thermal parameters are set, results are output, and the differences in microclimate response under different wind corridor structures and greening configurations are compared to evaluate their cooling effect and ventilation efficiency.

Images