Construction method of space-time carbon flux digital map

Abstract

The invention discloses a space-time carbon flux digital map construction method, which comprises the steps of three-dimensional vector information acquisition, space-time carbon flux accounting database establishment, carbon flux space-time analysis, carbon flux influence factor identification and space-time carbon flux digital map display. Real-time and high-precision carbon flux data of a target area are measured through a carbon flux measurement system, a space-time carbon flux accounting database is constructed, carbon flux time analysis and space analysis are carried out to judge the carbon flux space-time distribution trend of the measured area, and a carbon flux space-time geographic weighting model is used for identifying carbon flux influence factors. A carbon flux digital map is displayed on the large digital screen in real time; according to the method, high-time and high-space-precision carbon flux accounting and influence factor identification of the measurement area can be realized, and detection, management and prediction of the space-time carbon flux of the target area are realized in a precise quantification, visualization and artificial intelligence manner.

Description

Technical Field

[0001] The invention belongs to the technical field of low-carbon city planning, and in particular relates to a method for constructing a spatiotemporal carbon flux digital map. Background Art

[0002] At present, spatial construction, economic activities and human actions are the main sources of carbon emissions, and carbon flux data is a key indicator connecting ecosystems, urban environments and human activities. It is one of the core indicators for measuring carbon peak and carbon neutrality. How to reasonably control carbon flux has become a difficult problem, including how to set up carbon flux measurement equipment, accurately calculate carbon flux, analyze spatiotemporal trends and analyze influencing factors. It has become a key issue in the current dual carbon goals.

[0003] At present, most carbon flux measurement methods are based on manual measurement of a single carbon flux measurement device, with low temporal resolution, making it difficult to capture carbon flux changes in a short period of time and respond to rapid responses and short-term fluctuations; insufficient spatial resolution makes it difficult to capture carbon source and carbon sink space in complex terrain and urban environments, and cannot guarantee full coverage of the measurement area; data assimilation and model simulation data are uncertain, and lack complex data processing and analysis data integration technology, and cannot identify the multiple influencing factors of carbon flux including economic factors and human factors from the perspective of urban planning; lack of data connection with the data middle platform and visualization screen, cannot intuitively display the carbon flux measurement results, and cannot accurately and automatically translate the carbon flux data measurement and spatiotemporal analysis into spatial development issues and spatial planning response plans.

[0004] Therefore, it is urgent to solve the above problems. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to provide a method for constructing a spatiotemporal carbon flux digital map. The present invention can measure the carbon flux with high precision, multi-scale and multi-dimensionality, identify influencing factors and display the spatiotemporal trend of the carbon flux.

[0006] Technical solution: To achieve the above objectives, the present invention discloses a method for constructing a spatiotemporal carbon flux digital map, comprising the following steps:

[0007] (1) Obtaining the building data, road data and boundary data of the target area to form a geographic space basic database, obtaining land use data, population data, economic data and business data to form an impact factor database, matching the geographic space basic database and the impact factor database to form a vector information data set, and storing it;

[0008] (2) Obtain carbon concentration and wind speed in the target area, calculate the carbon flux accounting data of the target area, link the carbon flux accounting data to the vector information data set, and form a spatiotemporal carbon flux accounting database;

[0009] (3) Based on the spatiotemporal carbon flux accounting database of step (2), the target area is subjected to carbon flux temporal analysis, spatial analysis and outlier identification, wherein the temporal analysis includes day-night comparison, quarterly comparison, monthly comparison, annual comparison, identification of extreme values ​​and the time when extreme values ​​appear, and the temporal distribution trend of carbon flux; the spatial analysis includes the overall distribution map of carbon flux, identification of extreme values ​​and the location where extreme values ​​appear, and the spatial distribution trend of carbon flux;

[0010] (4) According to the influencing factor database in step (1) and the spatiotemporal carbon flux accounting database in step (2), the carbon flux measurement results of the target area are extracted and the carbon flux influencing factors are numerically calculated. The extracted data are spatially matched in the geographic information system, and the land use indicators, business indicators, population indicators and economic indicators of each grid are calculated. The non-numerical indicators are numerically processed, and the numerical data are standardized to the maximum and minimum values. According to the carbon flux measurement results and the calculation results of the influencing factor indicators of the target area, the spatial influence fitting coefficient of each influencing factor in each grid is calculated through the carbon flux spatiotemporal geographic weighted model to form a fitting coefficient data set. The first quartile Q1, the second quartile Q2 and the third quartile Q3 in the fitting coefficient data set are set as critical values, and the fitting coefficient data set is divided into four major intervals: positive first-level influence area U1, positive second-level influence area U2, negative first-level influence area U3, and negative second-level influence area U4, so as to obtain the influence degree of the carbon flux influencing factor on each grid. Using Random The Forest algorithm calculates the contribution of each influencing factor to the overall carbon flux of the target area and uploads it to the data center;

[0011] (5) Based on the spatiotemporal carbon flux accounting database of step (2), the results of carbon flux time analysis, spatial analysis and outlier identification in step (3), and the spatial impact fitting coefficient and contribution value in step (4), digital map layers with different attributes are established, and the spatiotemporal carbon flux digital map is formed after data superposition.

[0012] Optionally, in step (1), building data, road data and boundary data of the target area are obtained from the open map platform to form a geospatial basic database; the building data includes the boundary of the building, the number of floors of the building, and the function of the building; the road data includes the shape of the road centerline, the road length and the road grade.

[0013] Optionally, in step (1), the land use data is obtained from satellite remote sensing images, and the land use data includes land properties, land cover types and land use conditions; the population data is obtained from LBS data obtained through mobile phone operators, and the population data includes population size, population distribution and employment rate; the economic data is obtained from statistical yearbooks, and the economic data includes GDP, per capita income level and consumption level; the business data is obtained from POI data obtained from open map service providers, and the business data includes business name, business category, number of visits and user evaluation.

[0014] Optionally, in step (1), the target area boundary is divided into 100m×100m grids, each grid is labeled, the coordinates and elevations of the basic geographic spatial data and the influencing factor data are matched, the data format is unified and input into the geographic information system platform to form a vector information data set, and stored in the data center.

[0015] Optionally, in step (2), measuring points are set in the target area and a carbon flux measurement system is established. The carbon flux measurement system includes configuring carbon concentration measurement equipment, measuring point layout, measurement height setting, time resolution setting, spatial resolution setting, wind environment measurement setting and carbon flux calculation; the carbon concentration measurement equipment includes laser radar equipment and drones, and the laser radar equipment with a measurement accuracy of 100m×100m is used to measure the carbon concentration and wind field of the target area every 30 minutes; and 4 drones equipped with carbon concentration sensors and wind meters are used to measure the wind speed, wind direction and carbon concentration inside the grid with an aircraft altitude of 30m and 60m as the reference, correct and make up for the missing concentration values ​​outside the circular measurement range of the laser radar equipment, and control the measurement accuracy to be less than or equal to 1ppm; the measuring point layout is to set multiple carbon concentration measurement equipment in the target area according to their maximum range to meet the full coverage monitoring of the target area; the measurement height setting is to scan the highest height of the building at each measuring point by a three-dimensional laser scanning device. If the highest height is greater than 60m, the measurement height is set to the ground floor, 3 meters above the ground, and 100 meters below the ground. 0m, 60m from the ground. If the highest height is greater than 30m and less than 60m, the measurement height is set to the ground layer and 30m from the ground. If the highest height is less than 30m, the measurement height is set to the ground layer and the top floor of the highest building. The time resolution is set to 30min, with 24 hours as a measurement record. The spatial resolution is set to 100m, and the position is aligned with the 100m grid in step (1). The wind environment measurement is set to real-time wind field detection mode, including horizontal wind speed, vertical wind speed, average wind speed at the height, etc. speed and wind direction; carbon flux calculation refers to establishing a connection between the carbon concentration measurement equipment and the digital middle platform through the API interface, regularly calibrating the CO2 concentration measurement results of each measuring point, marking abnormally high values, abnormally low values ​​and blank values, calling the carbon flux calculation formula to calculate the carbon flux accounting data of each measuring point during the measurement period, forming a carbon flux accounting data set, calling the vector information data set in step (1), linking the carbon flux accounting data of each time slice to the corresponding grid, forming a spatiotemporal carbon flux accounting database and storing it in the data middle platform.

[0016] Optionally, the outlier identification in step (3) refers to calculating five statistics of the spatiotemporal carbon flux accounting data set based on the spatiotemporal carbon flux accounting data set, the five statistics including the minimum carbon flux x min , first quartile x 25％ , the second quartile x 50％ , the third quartile x 75％ and the maximum value x max , calculate the interquartile range ρ and determine the limits of outliers, the formula is:

[0017] ρ=x 75％ -x 25％

[0018] ε lower =x 25％ -1.5×ρ

[0019] ε upper =x 75％ +1.5×ρ

[0020] Among them, ε lower is the lowest outlier limit, ε upper It is the upper limit of the outlier. The carbon flux measurement data below the lower limit and above the upper limit are outliers. The corresponding time data clearly indicate the time when the outliers occur.

[0021] Optionally, the carbon flux time distribution trend in step (3) refers to the spatiotemporal carbon flux accounting data x ti , i represents the measurement point number, t represents the measurement time point, and the moving average SMA is calculated t The method eliminates the random fluctuation of carbon flux data, imports the carbon flux digital map to construct and display the accurate and smooth carbon flux time distribution curve, and clearly expresses the trend of carbon flux time distribution; moving average SMA t0 The formula is:

[0022]

[0023] Among them, SMA t is the moving average of carbon flux at the tth measurement time point, n is the moving average calculation window, x ti It is the original data point of carbon flux at the ith measurement point at the tth measurement time point; for the value of n calculation window, the value of n is determined to be 3-5 carbon flux data points based on the periodicity and time resolution of carbon flux data acquisition of 30 minutes, and multiple iterations are performed.

[0024] Optionally, the spatial distribution trend of carbon flux in step (3) refers to the spatial distribution trend of carbon flux accounting data x ti All observation points adjacent to observation point i, the number of adjacent observation points is N i , based on the actual distance, land properties, building height, green space ratio, and building density data, the spatial weight w of observation point i and its adjacent observation point j is calculated ij , calculate the aggregation value G of each observation point data i , the formula is:

[0025]

[0026] in, represents the actual distance between observation point i and adjacent observation point j, α and β are parameters that control the distance and the influence of land properties, building height, green space ratio, and building density, L ijis the difference data of land properties, building height, green space ratio and building density between observation point i and adjacent observation point j. is the average value of land properties, building height, green area ratio and building density, and γ is the parameter that controls the influence of differences in land properties, building height, green area ratio and building density;

[0027]

[0028] Among them, z j is the standardized value of the adjacent observation point j, j is 1 to N i , is the average of the standardized values ​​of all adjacent observation points;

[0029] Define the aggregation value threshold τ, G i >τ, then the observation point i is a high-value cluster; G i <τ, the observation point i is a low-value cluster, and the spatial hot and cold spot distribution maps of carbon flux in the measurement area at different time periods are superimposed to express the spatial distribution trend of carbon flux.

[0030] Optionally, the first-level influencing factors in step (4) include land use indicators, business indicators, population indicators and economic indicators. The land use indicators include construction land, roads, forest land, grassland, water bodies, farmland and other non-construction land, with a total of 7 second-level influencing factors; the business indicators include business POI index, commercial POI index, residential POI index, public service POI index and industrial POI index, with a total of 5 second-level influencing factors; the population indicators include the number of residents and the number of employed people, with a total of 2 second-level influencing factors; the economic indicators include the GDP per unit land index and the consumption per unit population index, with a total of 2 second-level influencing factors;

[0031]

[0032] Among them, IF p is the first-level impact factor value, IF pk is the secondary impact factor value, p is the primary impact factor number, K is the number of secondary impact factors, and k is the secondary impact factor number;

[0033] The spatiotemporal geographic weighted carbon flux model uses the flux measurement data of each grid and the influencing factors of each grid, such as land use indicators, business indicators, population indicators, and economic indicators, and calculates the fitting coefficient of each grid influencing factor and carbon flux, capturing the spatial and temporal heterogeneity of carbon flux while clarifying the influence of the influencing factors.

[0034] Each measuring point i corresponds to a grid, so the grid number is also i, and the longitude value of the grid is determined to be u i , the latitude value is v i, the observation time is measured as t, and the spatiotemporal geographic weighted model of carbon flux is constructed. The formula is:

[0035]

[0036]

[0037] Among them, x i is the carbon flux value of the i-th group, j is the adjacent grid number of grid i, ε i is the random error, u i is the longitude of the i-th grid, t i is the carbon flux measurement time of the i-th grid, v i is the latitude of the i-th grid, u0 is the longitude of the initial grid, v0 is the latitude of the initial grid, t0 is the initial time of measurement, IF ipk is the value of the pkth influencing factor at the i-th grid, q0(u0,v0,t0) is the fitting coefficient under the initial measurement position longitude, latitude and time, x j (t ij ) is the adjacent grid j at time t ij The carbon flux value, q pk (u i ,v i ,t i ) is the fitting coefficient of the pkth influencing factor in the i-th carbon flux measurement grid, e is the constant value of the exponential function, w ij (t ij ) is the weight that takes into account the time and space distance, x j (t ij ) is the adjacent grid j at time t j The carbon flux value, d ij is the spatial distance between grid i and grid j, t ij is the time interval between two point measurements, σ ​​and τ are the spatial and temporal bandwidth parameters, respectively.

[0038] Optionally, in step (4), the Random Forest algorithm is used to calculate the contribution of each influencing factor to the overall carbon flux of the target area, that is, the land use index IF1, the business index IF2, the population index IF3 and the economic index IF4 are constructed into IF p The data set is used to construct four decision trees and calculate the feature importance value of the secondary indicator in each decision tree. pk , where land use indicators include construction land IF 11 , road IF 12 、Woodland IF 13 、Grassland IF 14 、Water IF 15 、Farmland IF 16 and other non-construction land IF17 , business indicators include business poi index IF 21 , Business Poi Index IF 22 , Residential Poi Index IF 23 , public service poi index IF 24 and industrial poi index IF 25 , population indicators include the number of residents IF 31 and the number of employed people IF 32 , economic indicators include the GDP per unit of land IF 41 and per capita consumption index IE 42 , and the importance values ​​VI of all secondary indicators pk Perform average calculation to obtain the overall contribution value VI of the secondary index p That is, the contribution value of the first-level indicator, the formula is:

[0039]

[0040] And the sum is normalized to 1, the formula is: p∈(1,2,3,4) can explain the contribution of each first-level indicator to the model prediction results. The higher the value, the greater the contribution.

[0041] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0042] (1) The vector information dataset of the present invention breaks through the data information that only focuses on a single spatial dimension or time dimension, and realizes the spatiotemporal characteristics of carbon flux data; at the same time, it matches spatiotemporal data with multiple elements, breaking through the traditional carbon flux data results that are difficult to reflect and establish connections with economic factors and human factors, and provides basic data connections for the identification and analysis of carbon flux influencing factors;

[0043] (2) The carbon flux measurement system constructed by the present invention realizes real-time measurement and full coverage of carbon flux measurement regardless of terrain and complex urban environment, greatly improving the spatiotemporal accuracy of carbon flux measurement, completing the real-time acquisition, automatic data processing and visual expression of carbon flux data, and establishing a spatiotemporal carbon flux accounting database, breaking through the limitations of the original human calculation, measurement and control, and using advanced artificial intelligence technology and geographic information systems, combined with machine learning algorithms, to achieve automatic measurement and quantitative analysis of carbon flux in a specific area, establish an intelligent and automated process for carbon flux data measurement and processing, and improve the efficiency, accuracy and timeliness of carbon flux data acquisition;

[0044] (3) The present invention retrieves information from the spatiotemporal carbon flux accounting database, performs temporal and spatial analysis of carbon flux, determines and corrects outliers through spatiotemporal geographic weighted analysis, and further clarifies the temporal and spatial distribution trends of carbon flux, connects to the digital middle platform to display the spatiotemporal distribution trend curve of carbon flux, and constructs different layers based on time attributes and spatial attributes, thereby realizing horizontal and vertical comparisons of different time domains and spatial domains, breaking through the limitations of traditional single-time or single-space carbon flux comparisons, and realizing the visualization of the spatiotemporal distribution trend of carbon flux; at the same time, the present invention retrieves the influencing factor library to identify carbon flux influencing factors, and uses the Random Forest algorithm to perform numerical calculations of the impact, clarifies the factors that affect the change of carbon flux and the contribution values ​​of each influencing factor, and delineates four levels of impact areas, breaking through the subjective judgment of traditional carbon flux influencing factors, and at the same time realizing the identification of the numerical judgment of the impact of multiple influencing factors including economic factors and human activity factors on carbon flux from the perspective of urban planning, thereby improving the comprehensiveness and objectivity of carbon flux influencing factors;

[0045] (4) The present invention realizes the construction of a digital map platform by establishing digital map layers with different attributes and superimposing data and image information according to geographic coordinates. It completes the automatic and intelligent processing and display of the entire process of carbon flux from data collection, data processing, data analysis, dynamic real-time detection, real-time feedback, strategy response and result output, providing urban managers with an intuitive, easy-to-understand, convenient and efficient spatial carbon flux display platform, realizing the interactive decision-making of spatial planning strategies and dynamic monitoring and analysis of carbon flux, and providing an interactive experience of two-dimensional layers and three-dimensional digital sandbox. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a flow chart of the present invention;

[0047] Figure 2 It is a framework diagram of the carbon flux measurement system in the present invention;

[0048] Figure 3 This is a graph showing the results of the carbon flux time analysis in the target area of ​​the present invention;

[0049] Figure 4 This is a graph showing the spatial analysis results of carbon flux in the target area of ​​the present invention;

[0050] Figure 5 This is a result diagram of identifying factors affecting carbon flux in the target area in the present invention;

[0051] Figure 6 It is a large screen for digitally displaying the spatiotemporal carbon flux in the present invention. DETAILED DESCRIPTION

[0052] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0053] like Figure 1 As shown, the present invention discloses a method for constructing a spatiotemporal carbon flux digital map, comprising the following steps:

[0054] S1. Acquisition of three-dimensional vector information: Acquire the building data, road data and boundary data of the target area to form a basic geographic space database; acquire land use data, population data, economic data and business data to form an influencing factor database; match the basic geographic space database and the influencing factor database to form a vector information data set and store it.

[0055] The building data, road data and boundary data of the target area are obtained from the open map platform to form a basic geographic space database; building data includes the boundary of the building, the number of floors of the building, and the function of the building; road data includes the shape of the road centerline, the length of the road and the road grade. Among them, land use data is obtained from satellite remote sensing images, and land use data includes land nature, land cover type and land use status; population data is obtained from LBS data through mobile phone operators, and the specific steps are data collection, base station positioning, AFLT positioning, hybrid positioning, conversion of longitude and latitude and data transmission. Population data includes population size, population distribution and employment rate; economic data is obtained through statistical yearbooks, and economic data includes GDP, per capita income level and consumption level; business data is obtained from POI data through open map service providers, and business data includes business name, business category, visit volume and user evaluation.

[0056] The target area boundary is divided into 100m×100m grids, each grid is labeled, the coordinates and elevations of the basic geographic spatial data and the influencing factor data are matched, the data format is unified and input into the geographic information system platform to form a vector information data set, and then stored in the data center.

[0057] S2. Establish a spatiotemporal carbon flux accounting database: obtain carbon concentration and wind speed in the target area, calculate the carbon flux accounting data of the target area, link the carbon flux accounting data to the vector information data set, and form a spatiotemporal carbon flux accounting database;

[0058] like Figure 2As shown, measuring points are set in the target area and a carbon flux measurement system is established. The carbon concentration measurement equipment includes laser radar equipment and drones. The laser radar equipment with a measurement accuracy of 100m×100m is used to measure the carbon concentration and wind field in the target area every 30 minutes. Four drones equipped with carbon concentration sensors and wind meters are used to measure the wind speed, wind direction and carbon concentration inside the grid with an aircraft altitude of 30m and 60m as the reference, and the missing concentration values ​​outside the circular measurement range of the laser radar equipment are corrected and supplemented to control the measurement accuracy to be less than or equal to 1ppm. The carbon concentration measurement equipment is connected to the digital middle station through the API interface, and the CO2 concentration measurement results of each measuring point are calibrated regularly, abnormally high values, abnormally low values ​​and blank values ​​are marked, and the carbon flux calculation formula is called to calculate the carbon flux accounting data of each measuring point during the measurement period to form a carbon flux accounting data set, and the vector information data set in S1 is retrieved to link the carbon flux accounting data of each time slice to the corresponding grid to form a spatiotemporal carbon flux accounting database and store it in the data middle station.

[0059] The carbon flux measurement system includes the configuration of carbon concentration measurement equipment, measurement point layout, measurement height setting, time resolution setting, spatial resolution setting, wind environment measurement setting and carbon flux calculation; the measurement point layout is to set up multiple carbon concentration measurement equipment according to their maximum range in the target area, and use laser detection radar and drones to meet the full coverage monitoring of the target area; the measurement height setting is to scan the highest height of the building at each measurement point through a three-dimensional laser scanning device, reasonably plan the scanning route, ensure that the scanning equipment points maintain a 30%-40% overlap area, improve the scanning accuracy, and establish data transmission and processing with the CNC platform. If the maximum height of the scanning result is greater than 60m, the measurement height is set to the ground The measurement height is set to the ground layer, 30m from the ground, and 60m from the ground. If the highest height is greater than 30m and less than 60m, the measurement height is set to the ground layer, 30m from the ground. If the highest height is less than 30m, the measurement height is set to the ground layer and the top floor of the highest building. The time resolution is set to 30min, with 24 hours as a measurement record. The spatial resolution is set to 100m, and the position is aligned with the 100m grid in step (1). The wind environment measurement is set to real-time wind field detection mode, and the drone is equipped with an intelligent wind measuring instrument to avoid measuring space blind spots and reduce errors. The wind environment measurement settings include horizontal wind speed, vertical wind speed, average wind speed and wind direction at the height. The carbon flux calculation formula is F co2 =ρ co2 ×v×A, where F co2 is the CO2 flux, ρ co2 is the CO2 concentration, v is the wind speed, and A is the cross-sectional area of ​​the measurement area.

[0060] S3: Spatiotemporal analysis of carbon flux: Based on the spatiotemporal carbon flux accounting database of step S2, the target area is subjected to temporal analysis, spatial analysis and outlier identification. The temporal analysis results include day-night comparison, quarterly comparison, monthly comparison, annual comparison, identification of extreme values ​​and the time when extreme values ​​appear, and temporal distribution trend of carbon flux; the spatial analysis results include the overall distribution map of carbon flux, identification of extreme values ​​and the location where extreme values ​​appear, and spatial distribution trend of carbon flux;

[0061] Outlier identification refers to calculating and sorting five statistics of the spatiotemporal carbon flux accounting data set based on the spatiotemporal carbon flux accounting data set. The five statistics include the minimum carbon flux x min , first quartile x 25％ , the second quartile x 50％ , the third quartile x 75％ and the maximum value x max , calculate the interquartile range ρ to measure the dispersion of carbon flux data and determine the limit of outliers, the formula is:

[0062] ρ=x 75％ -x 25％

[0063] ε lower =x 25％ -1.5×ρ

[0064] ε upper =x 75％ +1.5×ρ

[0065] Among them, ε lower is the lowest outlier limit, ε upper It is the highest limit of outliers. Carbon flux measurement data below the lowest limit and above the highest limit are outliers. Corresponding to time data, the time of occurrence of outliers is clarified, and the extreme events and reasons for the errors are further analyzed to determine whether to retain the data and build an extreme event set storage data center;

[0066] like Figure 3 As shown in the figure, the temporal distribution trend of carbon flux refers to the spatial and temporal carbon flux accounting data x ti , i represents the measurement point number, t represents the measurement time point, and the moving average SMA is calculated t The method eliminates the random fluctuation of carbon flux data, imports the carbon flux digital map to construct and display the accurate and smooth carbon flux time distribution curve, and clearly expresses the trend of carbon flux time distribution; moving average SMA t0 The formula is:

[0067]

[0068] Among them, SMA tis the moving average of carbon flux at the tth measurement time point, n is the moving average calculation window, x ti is the carbon flux raw data point of the ith measurement point at the tth measurement time point; for the value of the n calculation window, the value of n is determined to be 3-5 carbon flux data points based on the periodicity and time resolution of carbon flux data acquisition of 30 minutes, and multiple iterations are performed;

[0069] like Figure 4 As shown in the figure, the spatial distribution trend of carbon flux refers to the spatial and temporal carbon flux accounting data x ti All observation points adjacent to observation point i, the number of adjacent observation points is N i , based on the actual distance, land properties, building height, green space ratio, and building density data, the spatial weight w of observation point i and its adjacent observation point j is calculated ij , calculate the aggregation value G of each observation point data i , the formula is:

[0070]

[0071] in, represents the actual distance between observation point i and adjacent observation point j, α and β are parameters that control the distance and the influence of land properties, building height, green space ratio, and building density, L ij is the difference data of land properties, building height, green space ratio and building density between observation point i and adjacent observation point j. is the average value of land properties, building height, green area ratio and building density, and γ is the parameter that controls the influence of differences in land properties, building height, green area ratio and building density;

[0072]

[0073] Among them, z j is the standardized value of the adjacent observation point j, j is 1 to N i , is the average of the standardized values ​​of all adjacent observation points;

[0074] Define the aggregation value threshold τ, G i >τ, then the observation point i is a high-value cluster; G i <τ, the observation point i is a low-value cluster, and the carbon flux spatial hot and cold spot distribution maps of the measurement area in different time periods are superimposed to accurately express the spatial distribution trend of carbon flux;

[0075] S4. Identification of factors affecting carbon flux: Figure 5As shown, according to the influencing factor database in step S1 and the spatiotemporal carbon flux accounting database in step S2, the carbon flux measurement results of the target area are extracted and the carbon flux influencing factors are numerically calculated, the extracted data are spatially matched in the geographic information system, the land use indicators, business indicators, population indicators and economic indicators of each grid are calculated, the non-numerical indicators are numerically processed, and the numerical data are standardized by the maximum and minimum values;

[0076] Firstly, according to the carbon flux measurement results and the calculation results of the influencing factor index in the target area, the spatial impact fitting coefficient of each influencing factor in each grid was calculated through the carbon flux spatiotemporal geographic weighted model. The first quartile Q1, the second quartile Q2 and the third quartile Q3 in the fitting coefficient data set were set as critical values, and the data set was divided into four major intervals: positive first-level impact area U1, positive second-level impact area U2, negative first-level impact area U3, and negative second-level impact area U4, to obtain the influence degree of carbon flux influencing factors on each grid;

[0077] Then, the Random Forest algorithm is used to calculate the contribution of each influencing factor to the overall carbon flux of the target area and upload it to the data center;

[0078] The first-level influencing factors include land use indicators, business indicators, population indicators and economic indicators. The land use indicators include construction land, roads, forest land, grassland, water bodies, farmland and other non-construction land, with a total of 7 second-level influencing factors; the business indicators include business POI index, commercial POI index, residential POI index, public service POI index and industrial POI index, with a total of 5 second-level influencing factors; the population indicators include the number of residents and the number of employed people, with a total of 2 second-level influencing factors; the economic indicators include the GDP per unit of land and the consumption per unit of population, with a total of 2 second-level influencing factors;

[0079]

[0080] Among them, IF p is the first-level impact factor value, IF pk is the secondary impact factor value, p is the primary impact factor number, K is the number of secondary impact factors, and k is the secondary impact factor number;

[0081] The spatiotemporal geographic weighted carbon flux model uses the flux measurement data of each grid and the influencing factors of each grid, such as land use indicators, business indicators, population indicators, and economic indicators, to calculate the spatiotemporal comprehensive weights, and calculate the fitting coefficient of each grid influencing factor and carbon flux, so as to capture the spatial and temporal heterogeneity of carbon flux and clarify the influence of the influencing factors.

[0082] Each measuring point i corresponds to a grid, so the grid number is also i, and the longitude value of the grid is determined to be ui , the latitude value is v i , the observation time is measured as t, and the spatiotemporal geographic weighted model of carbon flux is constructed. The formula is:

[0083]

[0084]

[0085] Among them, x i is the carbon flux value of the i-th group, j is the adjacent grid number of grid i, ε i is the random error, u i is the longitude of the i-th grid, t i is the carbon flux measurement time of the i-th grid, v i is the latitude of the i-th grid, u0 is the longitude of the initial grid, v0 is the latitude of the initial grid, t0 is the initial time of measurement, IF ipk is the value of the pkth influencing factor at the i-th grid, q0(u0,v0,t0) is the fitting coefficient under the initial measurement position longitude, latitude and time, x j (t ij ) is the adjacent grid j at time t ij The carbon flux value, q pk (u i ,v i ,t i ) is the fitting coefficient of the pkth influencing factor in the i-th carbon flux measurement grid, e is the constant value of the exponential function, w ij (t ij ) is the weight that takes into account the time and space distance, x j (t ij ) is the adjacent grid j at time t j The carbon flux value, d ij is the spatial distance between grid i and grid j, t ij is the time interval between two point measurements, σ ​​and τ are the spatial and temporal bandwidth parameters respectively;

[0086] According to the calculated fitting coefficient, the first quartile Q1, the second quartile Q2 and the third quartile Q3 in the fitting coefficient data set are selected to divide the data set into four major intervals: positive first-level influence area U1, positive second-level influence area U2, negative first-level influence area U3, and negative second-level influence area U4;

[0087] The Random Forest algorithm is used to calculate the contribution of each influencing factor to the overall carbon flux of the target area, that is, the land use index IF1, the business index IF2, the population index IF3 and the economic index IF4 are constructed into IF pThe data set is used to construct four decision trees and calculate the feature importance value of the secondary indicator in each decision tree. pk , where land use indicators include construction land IF 11 , road IF 12 、Woodland IF 13 、Grassland IF 14 、Water IF 15 、Farmland IF 16 and other non-construction land IF 17 , business indicators include business poi index IF 21 , Business Poi Index IF 22 , Residential Poi Index IF 23 , public service poi index IF 24 and industrial poi index IF 25 , population indicators include the number of residents IF 31 and the number of employed people IF 32 , economic indicators include the GDP per unit of land IF 41 and per capita consumption index IF 42 , and the importance values ​​VI of all secondary indicators pk Perform average calculation to obtain the overall contribution value VI of the secondary index p That is, the contribution value of the first-level indicator, the formula is:

[0088]

[0089] And the sum is normalized to 1, the formula is: p∈(1,2,3,4), which can explain the contribution of each first-level indicator to the model prediction results. The higher the value, the greater the contribution;

[0090] S5. Digital map display of spatiotemporal carbon flux: Figure 6 As shown, digital map layers with different attributes are established for the carbon flux spatiotemporal data of step S2, the carbon flux spatiotemporal analysis results of step S3, and the carbon flux influencing factor identification results of step S4. After the data and image information are superimposed according to the geographic coordinates, a 24-hour dynamic spatiotemporal carbon flux digital map is displayed on a digital large screen, and the results of each module of the digital map are output in csv format according to user needs.

Claims

1. A method for constructing a spatiotemporal carbon flux digital map, characterized in that: The steps include: (1) Obtain the building data, road data and boundary data of the target area to form a geospatial basic database, obtain land use data, population data, economic data and business data to form an impact factor database, and match the geospatial basic database and the impact factor database to form a vector information data set; (2) Obtain carbon concentration and wind speed in the target area, calculate the carbon flux accounting data of the target area, link the carbon flux accounting data to the vector information data set, and form a spatiotemporal carbon flux accounting database; (3) Based on the spatiotemporal carbon flux accounting database of step (2), the target area is subjected to carbon flux temporal analysis, spatial analysis and outlier identification, wherein the temporal analysis includes day-night comparison, quarterly comparison, monthly comparison, annual comparison, identification of extreme values ​​and the time when extreme values ​​appear, and the temporal distribution trend of carbon flux; the spatial analysis includes the overall distribution map of carbon flux, identification of extreme values ​​and the location where extreme values ​​appear, and the spatial distribution trend of carbon flux; (4) Extract the carbon flux measurement results of the target area and calculate the carbon flux influencing factors according to the influencing factor database in step (1) and the spatiotemporal carbon flux accounting database in step (2), perform spatial matching on the extracted data in the geographic information system, calculate the land use index, business index, population index and economic index of each grid, digitize the non-numerical index, and perform maximum and minimum value standardization on the numerical data; According to the carbon flux measurement results and the calculation results of the influencing factor index in the target area, the spatial impact fitting coefficient of each influencing factor in each grid is calculated through the carbon flux spatiotemporal geographic weighted model to form a fitting coefficient data set. The first quartile Q1, the second quartile Q2 and the third quartile Q3 in the fitting coefficient data set are set as critical values, and the fitting coefficient data set is divided into four major intervals: positive first-level impact area U1, positive second-level impact area U2, negative first-level impact area U3, and negative second-level impact area U4, and the influence degree of the carbon flux influencing factor on each grid is obtained; the Random Forest algorithm is used to calculate the contribution value of each influencing factor to the overall carbon flux of the target area; (5) Based on the spatiotemporal carbon flux accounting database of step (2), the results of carbon flux time analysis, spatial analysis and outlier identification in step (3), and the spatial impact fitting coefficient and contribution value in step (4), digital map layers with different attributes are established, and the spatiotemporal carbon flux digital map is formed after data superposition.

2. The method for constructing a spatiotemporal carbon flux digital map according to claim 1, characterized in that: In the step (1), the building data, road data and boundary data of the target area are obtained from the open map platform to form a geospatial basic database; the building data includes the boundary of the building, the number of floors of the building, and the function of the building; the road data includes the shape of the road centerline, the road length and the road grade.

3. The method for constructing a spatiotemporal carbon flux digital map according to claim 1, characterized in that: In the step (1), the land use data is obtained from satellite remote sensing images, and the land use data includes land properties, land cover types and land use conditions; the population data is obtained from LBS data obtained from mobile phone operators, and the population data includes population size, population distribution and employment rate; the economic data is obtained from statistical yearbooks, and the economic data includes GDP, per capita income level and consumption level; the business data is obtained from POI data obtained from open map service providers, and the business data includes business name, business category, number of visits and user evaluation.

4. The method for constructing a spatiotemporal carbon flux digital map according to claim 1, characterized in that: In the step (1), the target area is divided into 100m×100m grids according to its boundary, each grid is numbered, the coordinates and elevations of the basic geographic space data and the influencing factor data are matched, the data format is unified and input into the geographic information system platform to form a vector information data set, and then stored in the data center.

5. The method for constructing a spatiotemporal carbon flux digital map according to claim 1, characterized in that: In the step (2), measuring points are set in the target area and a carbon flux measurement system is established. The carbon flux measurement system includes configuring carbon concentration measurement equipment, measuring point layout, measurement height setting, time resolution setting, spatial resolution setting, wind environment measurement setting and carbon flux calculation; the carbon concentration measurement equipment includes laser radar equipment and drones, and the laser radar equipment with a measurement accuracy of 100m×100m is used to measure the carbon concentration and wind field of the target area every 30 minutes; and 4 drones equipped with carbon concentration sensors and wind meters are used to measure the wind speed, wind direction and carbon concentration inside the grid with an aircraft altitude of 30m and 60m as the reference, correct and make up for the missing concentration values ​​outside the circular measurement range of the laser radar equipment, and control the measurement accuracy to be less than or equal to 1ppm; the measuring point layout is to set multiple carbon concentration measurement equipment in the target area according to their maximum range to meet the full coverage monitoring of the target area; the measurement height setting is to scan the highest height of the building at each measuring point by a three-dimensional laser scanning device. If the highest height is greater than 60m, the measurement height is set to the ground floor, 30m above the ground, and 100m below the ground. m, 60m from the ground. If the highest height is greater than 30m and less than 60m, the measurement height is set to the ground layer, 30m from the ground. If the highest height is less than 30m, the measurement height is set to the ground layer, the top floor of the highest building; the time resolution is set to 30min, with 24 hours as a measurement record; the spatial resolution is set to 100m, and the position is aligned with the 100m grid in step (1); the wind environment measurement is set to real-time wind field detection mode, including the horizontal wind speed, vertical wind speed, average wind speed and wind direction at the height; carbon flux calculation refers to establishing a connection between the carbon concentration measurement equipment and the digital middle station through the API interface, regularly calibrating the CO2 concentration measurement results of each measuring point, marking abnormally high values, abnormally low values ​​and blank values, calling the carbon flux calculation formula to calculate the carbon flux accounting data of each measuring point during the measurement period, forming a carbon flux accounting data set, calling the vector information data set in step (1), linking the carbon flux accounting data of each time slice to the corresponding grid, forming a spatiotemporal carbon flux accounting database and storing it in the data middle station.

6. The method for constructing a spatiotemporal carbon flux digital map according to claim 1, characterized in that: In the step (3), outlier identification refers to calculating five statistics of the spatiotemporal carbon flux accounting data set based on the spatiotemporal carbon flux accounting data set, the five statistics including the minimum carbon flux x min , first quartile x 25％ , the second quartile x 50％ , the third quartile x 75％ and the maximum value x max , calculate the interquartile range ρ and determine the limits of outliers, the formula is: p=x 75％ -x 25％ e lower =x 25％ -1.5×r e upper =x 75％ +1.5×r Among them, ε lower is the lowest outlier limit, ε upper It is the upper limit of the outlier. The carbon flux measurement data below the lower limit and above the upper limit are outliers. The corresponding time data clearly indicate the time when the outliers occur.

7. The method for constructing a spatiotemporal carbon flux digital map according to claim 1, characterized in that: The carbon flux time distribution trend in step (3) refers to the calculation data x based on the spatiotemporal carbon flux. ti , i represents the measurement point number, t represents the measurement time point, and the moving average SMA is calculated t The method eliminates the random fluctuation of carbon flux data, imports the carbon flux digital map to construct and display the accurate and smooth carbon flux time distribution curve, and clearly expresses the trend of carbon flux time distribution; moving average SMA t0 The formula is: Among them, SMA t is the moving average of carbon flux at the tth measurement time point, n is the moving average calculation window, x ti It is the original data point of carbon flux at the ith measurement point at the tth measurement time point; for the value of n calculation window, the value of n is determined to be 3-5 carbon flux data points based on the periodicity and time resolution of carbon flux data acquisition of 30 minutes, and multiple iterations are performed.

8. The method for constructing a spatiotemporal carbon flux digital map according to claim 7, characterized in that: The spatial distribution trend of carbon flux in step (3) refers to the spatial distribution trend of carbon flux according to the spatiotemporal carbon flux accounting data x ti All observation points adjacent to observation point i, the number of adjacent observation points is N i , based on the actual distance, land properties, building height, green space ratio, and building density data, the spatial weight w of observation point i and its adjacent observation point j is calculated ij , calculate the aggregation value G of each observation point data i , the formula is: in, represents the actual distance between observation point i and adjacent observation point j, α and β are parameters that control the distance and the influence of land properties, building height, green space ratio, and building density, L ij is the difference data of land properties, building height, green space ratio and building density between observation point i and adjacent observation point j. is the average value of land properties, building height, green area ratio and building density, and γ is the parameter that controls the influence of differences in land properties, building height, green area ratio and building density; Among them, z j is the standardized value of the adjacent observation point j, j is 1 to N i , is the average of the standardized values ​​of all adjacent observation points; Define the aggregation value threshold τ, G i >τ, then the observation point i is a high-value cluster; G i <τ, the observation point i is a low-value cluster, and the spatial hot and cold spot distribution maps of carbon flux in the measurement area at different time periods are superimposed to express the spatial distribution trend of carbon flux.

9. The method for constructing a spatiotemporal carbon flux digital map according to claim 8, characterized in that: In step (4), the primary influencing factors include land use indicators, business indicators, population indicators and economic indicators. The land use indicators include construction land, roads, forest land, grassland, water bodies, farmland and other non-construction land, with a total of 7 secondary influencing factors; the business indicators include business POI index, commercial POI index, residential POI index, public service POI index and industrial POI index, with a total of 5 secondary influencing factors; the population indicators include the number of residents and the number of employed people, with a total of 2 secondary influencing factors; the economic indicators include the GDP per unit land index and the consumption per unit population index, with a total of 2 secondary influencing factors; Among them, IF p is the first-level impact factor value, IF pk is the secondary impact factor value, p is the primary impact factor number, K is the number of secondary impact factors, and k is the secondary impact factor number; The spatiotemporal geographic weighted carbon flux model uses the flux measurement data of each grid and the influencing factors of each grid, such as land use indicators, business indicators, population indicators, and economic indicators, and calculates the fitting coefficient of each grid influencing factor and carbon flux, capturing the spatial and temporal heterogeneity of carbon flux while clarifying the influence of the influencing factors. Each measuring point i corresponds to a grid, so the grid number is also i, and the longitude value of the grid is determined to be u i , the latitude value is v i , the observation time is measured as t, and the spatiotemporal geographic weighted model of carbon flux is constructed. The formula is: Among them, x i is the carbon flux value of the i-th group, j is the adjacent grid number of grid i, ε i is the random error, u i is the longitude of the i-th grid, t i is the carbon flux measurement time of the i-th grid, v i is the latitude of the i-th grid, u0 is the longitude of the initial grid, v0 is the latitude of the initial grid, t0 is the initial time of measurement, IF ipk is the value of the pkth influencing factor at the i-th grid, q0(u0,v0,t0) is the fitting coefficient under the initial measurement position longitude, latitude and time, x j (t ij ) is the adjacent grid j at time t ij The carbon flux value, q pk (u i ,v i ,t i ) is the fitting coefficient of the pkth influencing factor in the i-th carbon flux measurement grid, e is the constant value of the exponential function, w ij (t ij ) is the weight that takes into account the time and space distance, x j (t ij ) is the adjacent grid j at time t j The carbon flux value, d ij is the spatial distance between grid i and grid j, t ij is the time interval between two point measurements, σ ​​and τ are the spatial and temporal bandwidth parameters, respectively.

10. The method for constructing a spatiotemporal carbon flux digital map according to claim 9, characterized in that: In step (4), the Random Forest algorithm is used to calculate the contribution of each influencing factor to the overall carbon flux of the target area, that is, the land use index IF1, the business index IF2, the population index IF3 and the economic index IF4 are constructed into IF p The data set is used to construct four decision trees and calculate the feature importance value of the secondary indicator in each decision tree. pk , where land use indicators include construction land IF 11 , road IF 12 、Woodland IF 13 、Grassland IF 14 、Water IF 15 、Farmland IF 16 and other non-construction land IF 17 , business indicators include business poi index IF 21 , Business Poi Index IF 22 , Residential Poi Index IF 23 , public service poi index IF 24 and industrial poi index IF 25 , population indicators include the number of residents IF 31 and the number of employed people IF 32 , economic indicators include the GDP per unit of land IF 41 and per capita consumption index IF 42 , and the importance values ​​VI of all secondary indicators pk Perform average calculation to obtain the overall contribution value VI of the secondary index p That is, the contribution value of the first-level indicator, the formula is: And the sum is normalized to 1, the formula is: p∈(1,2,3,4) can explain the contribution of each first-level indicator to the model prediction results. The higher the value, the greater the contribution.

Images