Method for compensating for lost atmospheric data from UAV monitoring based on real-time and historical data

Abstract

This invention discloses a method for compensating for lost atmospheric data in UAV monitoring based on real-time and historical data. This method compensates for data loss in real-time and historical datasets through real-time interpolation and historical interpolation methods, taking into account both the monitoring process of the UAV and the interpolation compensation of the dataset after monitoring. It can improve the quality of real-time monitoring data and optimize the real-time visualization performance of the ground mobile monitoring platform. It can also perform more detailed analysis and compensation of historical data after monitoring, helping researchers to more accurately understand and analyze changes in monitoring data. It makes up for the drawbacks caused by the loss of atmospheric data in UAV monitoring and is suitable for further promotion and application.

Description

Technical Field

[0001] This invention relates to the field of environmental science monitoring technology, specifically to a method for compensating for lost atmospheric data from unmanned aerial vehicle (UAV) monitoring based on real-time and historical data. Background Technology

[0002] With the continuous progress and accelerated development of cities, air pollution has become a major environmental issue of global concern, severely impacting human health and ecosystems. Monitoring air pollution data not only provides governments, businesses, and the public with real-time air quality information but also helps research institutions analyze pollution sources and transmission pathways, providing a scientific basis for formulating effective environmental protection policies. Furthermore, long-term data monitoring allows us to understand the trends and changes in air pollution, thereby predicting future environmental risks and taking appropriate countermeasures. In short, monitoring air pollution data is a crucial step in ensuring human health, maintaining ecological balance, and achieving sustainable development.

[0003] Fixed monitoring stations are the traditional method for air pollution monitoring. They are typically located in specific areas and provide long-term, stable data. However, their distribution may not be uniform, making it impossible to know the pollution levels in some areas. With technological advancements, portable monitoring devices have emerged, allowing people to carry these devices while walking or cycling to monitor their environment in real time. This method offers higher spatial resolution, helping us better explore pollution hotspots and microenvironments in cities. However, mobile measurements with portable devices are more susceptible to human influences such as breathing and particulate resuspension during walking, thus reducing measurement accuracy.

[0004] The introduction of drone technology into air pollution monitoring leverages its high efficiency, maneuverability, and emission-free operation to collect air quality data in areas inaccessible by traditional methods by mounting portable equipment on the drones. However, in actual monitoring, data loss is a common problem due to various reasons such as equipment failure, maintenance, or other external interference. This data loss leads to a decline in the effectiveness of ground-based monitoring platforms, resulting in misinterpretations of pollution levels and thus impacting decision-making and public health.

[0005] To address this issue, interpolating missing atmospheric pollution data becomes crucial. By using appropriate statistical methods and algorithms, we can estimate the missing data, thereby ensuring the integrity and continuity of the dataset. However, how to interpolate datasets in real-time using drone-borne equipment to maintain high-quality monitoring data, improve the visualization performance of ground-based mobile monitoring platforms, or perform more complex and comprehensive interpolation on historical datasets to help researchers more accurately understand and analyze past atmospheric changes remains a problem to be solved. Summary of the Invention

[0006] To address the aforementioned issues, this invention provides a UAV-based atmospheric data interpolation method that spans the entire monitoring process. This method compensates for data loss in real-time and historical datasets through both real-time and historical interpolation approaches, thereby optimizing the visualization of real-time atmospheric monitoring and mitigating the drawbacks caused by data loss in UAV-based atmospheric monitoring.

[0007] To achieve the above-mentioned technical objectives, the technical solution adopted by this invention is as follows:

[0008] A method for compensating for lost atmospheric data from unmanned aerial vehicle (UAV) monitoring based on real-time and historical data includes:

[0009] 1) The drone carries sensors and flies over the area to be tested, while monitoring is carried out through a ground-based mobile monitoring platform; the ground-based mobile monitoring platform can be a pedestrian backpack mobile monitoring platform, a bicycle mobile monitoring platform, or a car-mounted monitoring platform.

[0010] 2): The monitoring data of the UAV is received through the ground mobile monitoring platform. The interpolation method of the monitoring data is divided into real-time data interpolation and historical data interpolation. Real-time data interpolation is carried out throughout the entire monitoring process, while historical data interpolation is performed after the monitoring is completed.

[0011] 3): In the real-time data interpolation during the monitoring process, the ground mobile monitoring platform determines whether data loss has occurred each time it receives data, and classifies the data reception status at that moment into lost data and non-lost data.

[0012] If the data is not lost, it is saved directly; if the data is lost, it is marked as data to be interpolated and saved.

[0013] 4) After each data saving, determine whether there is any missing data with interpolation markers in the saved data. If so, determine whether the total number of non-lost data and interpolated missing data within the specified search radius centered on it reaches the minimum number. The search radius can be 5m, 10m, 20m, 30m or larger or smaller, and the minimum number can be 5, 10, 15 or 20 or larger or smaller. The search radius and the minimum number are not fixed and depend on the specific situation.

[0014] If so, IDW (Inverse Distance Weighted Method) is used to efficiently interpolate and compensate for the lost data, and the interpolated data is marked as such. This data then becomes the interpolated lost data, compensating for the lack of data on the ground mobile monitoring platform and providing real-time visualization. If not, the ground mobile monitoring platform will directly visualize the data.

[0015] 5): In the historical data interpolation after the monitoring is completed, the ground mobile monitoring platform saves the received historical dataset in the database, which becomes the historical dataset;

[0016] 6): Preprocess the historical dataset, and then perform normality tests and normal distribution transformations on the historical data;

[0017] 7): Calculate the distance and semivariance between each data point;

[0018] 8): Cross-validation is used to select the best-fitting model and calculate the relevant parameters;

[0019] 9): Use Kriging interpolation to perform interpolation and obtain the Kriging interpolation dataset;

[0020] 10): The ground mobile monitoring platform stores the datasets obtained by both real-time data interpolation and historical data interpolation methods in its database.

[0021] As a possible implementation method, the monitoring platform of the UAV can also be a hot air balloon high-altitude mobile monitoring platform; wherein, high altitude refers to the near-ground space above 3m and below 500m above the ground.

[0022] Furthermore, the ground-based mobile monitoring platform can receive data monitored by sensors mounted on the UAV and can visualize the monitoring data. This visualization utilizes computer graphics and image processing technology to convert the monitored data into graphics or images displayed on a screen, allowing for interactive processing.

[0023] Furthermore, the monitoring data mentioned in step 2) includes, but is not limited to, pollutant data, meteorological data, corresponding latitude and longitude coordinates, sampling time, and UAV altitude values. The data provided above are relatively general concepts; in practice, the data can be more specific. For example, pollutant data could include PM2.5. 2.5 The data, and other data are similar.

[0024] Furthermore, the real-time data interpolation method described in step 2) can perform IDW real-time interpolation on lost data based on the unlost data and the interpolated lost data during the monitoring process. IDW interpolation is rapid and suitable for real-time monitoring. Moreover, the real-time data interpolation method can be actively turned on or off on the ground mobile monitoring platform.

[0025] The historical data interpolation method described above can perform Kriging interpolation based on historical data after monitoring is completed. Kriging interpolation is accurate but more complex and time-consuming. It is suitable for use after monitoring is completed. It needs to be manually enabled and automatically disabled after interpolation is completed.

[0026] Furthermore, the expression for the IDW (Inverse Distance Weighted Interpolation) method is as follows:

[0027] Distance formula:

[0028]

[0029] Inverse distance weighted interpolation formula:

[0030]

[0031]

[0032] Among them, D 0i For the predicted point X0 and the known observation point X i Euclidean distance between them; X i X0 is the longitude coordinate of the i-th known data point; Y0 is the longitude coordinate of the predicted point; i Let X be the latitude coordinate of the i-th known data point; Y0 is the latitude coordinate of the predicted point; Z(X0) is the estimated value of the predicted point X0; Z(X0) is the estimated value of the predicted point X0. i (x) represents the known observation point X. i The value of μ; i It is to give the observed value Z(X) i The weights assigned; N is the total number of known observation points; p is a positive real number that determines the rate at which the weights change with distance, with a commonly used value of 2, which can be adjusted as needed.

[0033] Furthermore, the data that was not lost is defined as the data received by the ground mobile monitoring platform in a single data reception, in which the pollutant data, meteorological data, corresponding latitude and longitude coordinates, sampling time, and UAV altitude value were not lost.

[0034] The lost data is defined as the data received by the ground mobile monitoring platform once, in which the corresponding latitude and longitude coordinates, UAV altitude value and sampling time are not lost, but pollutant data or meteorological data are lost.

[0035] Missing data with interpolation tags is defined as missing data with interpolation tags.

[0036] Interpolated missing data is defined as data that has been interpolated using the IDW method to fill in the missing data to be interpolated;

[0037] The historical dataset is defined as either a dataset that has undergone real-time data interpolation or a dataset that has not undergone real-time data interpolation.

[0038] Furthermore, the preprocessing method in step 6) includes:

[0039] Remove all pollutant data, meteorological data, corresponding latitude and longitude coordinates, sampling time, and outlier data with drone altitude values ​​of 0 or less from the historical dataset, making it empty. Then, transform all pollutant data and meteorological data in the historical dataset according to the following formula:

[0040] γ(x i )=ln(Z(x i (4)

[0041] Where Z(x) i ) represents the historical dataset; γ(x) i ) represents the transformed historical dataset; ln is the natural logarithm. The purpose of preprocessing is to make the historical dataset have a near-normal distribution, thereby improving the accuracy of subsequent interpolation.

[0042] Furthermore, the formula for calculating the distance between data points in step 7) is as follows:

[0043]

[0044] Where, x i x is the longitude coordinate of the i-th point; j Let y be the longitude coordinate of the j-th point; i Let y be the latitude coordinate of the i-th point; j Let j be the latitude coordinates of the j-th point;

[0045] The formula for calculating the semivariance between data points is as follows:

[0046]

[0047] Wherein, N(d) ij ) is a distance of d ij The number of data pairs, z(x) i ) and z(x j ) represents the observation value at the corresponding location.

[0048] Furthermore, in step 8), the cross-validation process splits the transformed historical dataset into a training set and a validation set. Then, using the training set data, a kriging model is fitted to each semivariogram function model. Next, each fitted kriging model is used to predict the validation set data. Finally, the prediction results of each kriging model are evaluated (the mean squared error (MSE) can be used as the evaluation criterion). Based on the evaluation results, the best-performing model is selected as the best model.

[0049] The semivariance function models include Gaussian models, linear models, spherical models, exponential models, and power function models, and the expressions for each model are as follows:

[0050] Gaussian model:

[0051]

[0052] Linear model:

[0053]

[0054] Spherical model:

[0055]

[0056] Exponential model:

[0057]

[0058] Power function model:

[0059] γ(d ij )=c0+c·d ij ∝ 0<∝<2 (11)

[0060] Where c0 is a nugget, representing the variance under small variations in non-zero distance; c depends on the variance or amplitude of the data, and a is a range parameter.

[0061] Furthermore, the Kriging interpolation formula mentioned in step 9) is as follows:

[0062] Prediction formula:

[0063]

[0064] Kriging weights:

[0065]

[0066] Where z(x0) is the predicted value corresponding to the predicted point, z(x i ) is the observation value of the i-th known point, λ i It is the weight of the i-th known point.

[0067] By adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art:

[0068] The UAV monitoring atmospheric data loss compensation method based on real-time and historical data provided by this invention takes into account both the interpolation compensation of the dataset during and after the UAV monitoring process. It can improve the quality of real-time monitoring data and optimize the real-time visualization performance of the ground mobile monitoring platform. It can also perform more detailed analysis and compensation of historical data after the monitoring is completed, helping researchers to more accurately understand and analyze the changes in past monitoring data. Attached Figure Description

[0069] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0070] Figure 1 This is a simplified flowchart of the present invention.

[0071] Figure 2 This is a schematic diagram of the IDW search radius based on interpolation using a real-time dataset.

[0072] Figure 3 This is a schematic diagram illustrating the results of interpolation based on a real-time dataset, as described in an embodiment of the present invention.

[0073] Figure 4 This is a schematic diagram illustrating the results of interpolation based on historical datasets in an embodiment of the present invention. Detailed Implementation

[0074] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be particularly noted that the following embodiments are for illustrative purposes only and do not limit the scope of the invention. Similarly, the following embodiments are only some, not all, embodiments of the present invention, and all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0075] See attached document Figure 1 As shown, this embodiment provides a method for compensating for lost atmospheric data from unmanned aerial vehicle (UAV) monitoring based on real-time and historical data, including:

[0076] (1) Use a drone equipped with sensors to conduct monitoring flights in the area to be tested, while the ground mobile monitoring platform starts monitoring.

[0077] (2) The ground mobile monitoring platform begins to receive monitoring data from the UAV.

[0078] (3) The interpolation methods for monitoring data are divided into real-time data interpolation and historical data interpolation. Real-time data interpolation is performed throughout the entire monitoring process, while historical data interpolation is performed after the monitoring is completed.

[0079] (4) In the real-time data interpolation during the monitoring process, the ground mobile monitoring platform will determine whether the data is lost each time it receives the data, and will classify the data reception status at that moment into lost data and non-lost data.

[0080] (5) If the data is not lost, save the data directly; if the data is lost, mark the lost data as to be interpolated, and the data becomes the lost data to be interpolated and save it.

[0081] (6) After each data saving, determine whether there is any missing data with interpolation mark in the saved data. If so, determine whether the total number of non-lost data and interpolated lost data within the specified search radius centered on it reaches the minimum number. If so, use IDW (Inverse Distance Weighted Method) to quickly interpolate and compensate for the lost data, and mark it as interpolated lost data to make up for the real-time visualization when the ground mobile monitoring platform is missing data. If not, the ground mobile monitoring platform will directly visualize it.

[0082] (7) In the historical data interpolation after the monitoring is completed, the ground mobile monitoring platform saves the received historical dataset in the database, which becomes the historical dataset.

[0083] (8) Preprocess the historical dataset, and then perform normality test and normal distribution transformation on the historical data.

[0084] (9) Calculate the distance and semivariance between each data point.

[0085] (10) Cross-validation selects the best-fitting model (Gaussian model, linear model, spherical model) and calculates relevant parameters.

[0086] (11) Use the Kriging interpolation method to perform interpolation and obtain the Kriging interpolation dataset.

[0087] (12) The ground mobile monitoring platform will save the datasets obtained by real-time data interpolation method and historical data interpolation method in the database.

[0088] Based on the area to be monitored by the ground-based mobile monitoring platform, this case study uses PM2.5 as an example. The corresponding sensors are mounted on a drone, which flies to the area to be monitored, while the ground-based mobile monitoring platform simultaneously begins monitoring. This case study involves two independent monitoring experiments conducted in two separate study areas. In both experiments, the drones flew in a spiral trajectory. The first experiment monitored pollution levels near a school, and the second experiment monitored pollution levels in an industrial park.

[0089] Once the ground-based mobile monitoring platform starts monitoring, it will receive monitoring data from the drone.

[0090] Data interpolation methods for monitoring data are divided into real-time data interpolation and historical data interpolation. Real-time data interpolation is performed throughout the entire monitoring process, while historical data interpolation is performed after monitoring is completed. Real-time data interpolation can perform IDW (In-Time Data Wrapping) interpolation on lost data during monitoring based on both unlost data and interpolated lost data. IDW interpolation is rapid and suitable for real-time monitoring. Furthermore, real-time data interpolation can be actively enabled or disabled on the ground-based mobile monitoring platform. This example uses the first UAV monitoring instance to illustrate real-time data interpolation. In this embodiment, the IDW search radius is calculated based on the real-time dataset, such as... Figure 2 As shown.

[0091] The first monitoring session yielded 100 data points, of which 8 were lost during the monitoring process. The lost data was successfully recovered through IDW real-time data interpolation. Figure 3 As shown, the red circle represents the data after interpolation compensation, and the rightmost legend represents the pollution level index.

[0092] The formula for the IDW (Inverse Distance Weighted Interpolation) method is as follows:

[0093] Distance formula:

[0094]

[0095] Inverse distance weighted interpolation formula:

[0096]

[0097]

[0098] Among them, D 0i For the predicted point X0 and the known observation point X i Euclidean distance between them; X i X0 is the longitude coordinate of the i-th known data point; Y0 is the longitude coordinate of the predicted point; i Let X be the latitude coordinate of the i-th known data point; Y0 is the latitude coordinate of the predicted point; Z(X0) is the estimated value of the predicted point X0; Z(X0) is the estimated value of the predicted point X0. i (x) represents the known observation point X. i The value of μ; i It is to give the observed value Z(X) i The weights assigned; N is the total number of known observation points; p is a positive real number that determines the rate at which the weights change with distance, and its commonly used value is 2, which can be adjusted as needed.

[0099] Among them, lost data is defined as data received by the ground mobile monitoring platform once, in which the corresponding latitude and longitude coordinates, drone altitude value and sampling time are not lost, but pollutant data or meteorological data are lost.

[0100] Unlost data is defined as data received by the ground mobile monitoring platform in a single instance where pollutant data, meteorological data, corresponding latitude and longitude coordinates, sampling time, and drone altitude value are all intact.

[0101] Missing data with interpolation tags is defined as missing data with interpolation tags.

[0102] Interpolated missing data is defined as data that has been interpolated using the IDW method to fill in the missing data to be interpolated.

[0103] Historical data interpolation can perform Kriging interpolation based on historical data after monitoring is completed. Kriging interpolation is accurate but more complex and time-consuming, making it suitable for use after monitoring is finished. The historical dataset is defined as either a dataset that has undergone real-time data interpolation or one that has not. In this case, the historical dataset consists of data obtained from the second drone monitoring operation and has not undergone real-time data interpolation.

[0104] This historical dataset consists of 180 data points, of which 27 are missing. Through historical data interpolation, the missing data was successfully compensated for. Figure 4 As shown, the red circle represents the data after interpolation compensation, and the rightmost legend represents the pollution level index.

[0105] First, the historical dataset is preprocessed. The preprocessing methods include: removing all pollutant data, meteorological data, corresponding latitude and longitude coordinates, sampling time, and outlier data with drone altitude values ​​of 0 or less from the historical dataset, making it empty. Then, all pollutant data and meteorological data in the historical dataset are transformed according to the following formula:

[0106] γ(x i )=ln(Z(x i (4)

[0107] Where Z(x) i ) represents the historical dataset; γ(x) i ) represents the transformed historical dataset; ln is the natural logarithm. The purpose of preprocessing is to make the historical dataset have a near-normal distribution, thereby improving the accuracy of subsequent interpolation.

[0108] Next, as described in step (9), calculate the distance between each data point using the following equation:

[0109]

[0110] Where, x i x is the longitude coordinate of the i-th point; j Let y be the longitude coordinate of the j-th point; i Let y be the latitude coordinate of the i-th point;j Let be the latitude coordinates of the j-th point.

[0111] Next, calculate the semivariance between the data points using the following equation:

[0112]

[0113] Wherein, N(d) ij ) is a distance of d ij The number of data pairs, z(x) i ) and z(x j ) represents the observation value at the corresponding location.

[0114] Then, cross-validation is performed to obtain the best model. Cross-validation involves splitting the transformed historical dataset into training and validation sets. Then, using the training set data, a kriging model is fitted to each semivariogram function model. Next, each fitted kriging model is used to predict the validation set data. Finally, the prediction results of each kriging model are evaluated (using mean squared error (MSE) as the evaluation metric). Based on the evaluation results, the best-performing model is selected as the optimal model.

[0115] The semivariance function models mentioned above include Gaussian models, linear models, spherical models, exponential models, and power function models. The equations for each model are as follows:

[0116] Gaussian model:

[0117]

[0118] Linear model:

[0119]

[0120] Spherical model:

[0121]

[0122] Exponential model:

[0123]

[0124] Power function model:

[0125] γ(d ij )=c0+c·d ij ∝ 0<∝<2 (11)

[0126] Where c0 is a nugget, representing the variance under small changes in non-zero distance; c depends on the variance or amplitude of the data, and a is a range parameter.

[0127] Then, the Kringing interpolation formula is used for calculation, as follows:

[0128] Prediction formula:

[0129]

[0130] Kriging weights:

[0131]

[0132] Where z(x0) is the predicted value corresponding to the predicted point, z(x i ) is the observation value of the i-th known point, λ i It is the weight of the i-th known point.

[0133] The parameters involved in the above steps need to be set according to different scenarios, and there are no absolutely fixed values.

[0134] The above descriptions are only some embodiments of the present invention and do not limit the scope of protection of the present invention. Any equivalent device or equivalent process transformation made by using the contents of the description and drawings of the present invention, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for compensating for lost atmospheric data from unmanned aerial vehicle (UAV) monitoring based on real-time and historical data, characterized in that, include: 1): The drone, equipped with sensors, flies over the area to be tested, while monitoring is conducted through a ground-based mobile monitoring platform; 2): The monitoring data of the UAV is received through the ground mobile monitoring platform. The interpolation method of the monitoring data is divided into real-time data interpolation and historical data interpolation. Real-time data interpolation is carried out throughout the entire monitoring process, while historical data interpolation is performed after the monitoring is completed. 3): In the real-time data interpolation during the monitoring process, the ground mobile monitoring platform determines whether data loss has occurred each time it receives data, and classifies the data reception status at any given time into lost data and non-lost data. If no data is lost, save the data directly; If the data is missing, mark it as missing data to be interpolated, and save it. 4) After each data saving, determine whether there is any missing data with interpolation markers in the saved data. If so, determine whether the total number of non-lost data and interpolated missing data within the specified search radius centered on it has reached the minimum number. If so, the IDW is used to interpolate and compensate for the lost data, and it is marked as interpolated lost data; if not, the ground motion monitoring platform directly visualizes it. 5): In the historical data interpolation after monitoring is completed, the ground mobile monitoring platform saves the received historical dataset in the database, which becomes the historical dataset; 6) Preprocess the historical dataset, then perform normality tests and normal distribution transformations on the historical data, including: Remove all pollutant data, meteorological data, corresponding latitude and longitude coordinates, sampling time, and outlier data with drone altitude values ​​of 0 or less from the historical dataset, making it empty. Then, transform all pollutant data and meteorological data in the historical dataset according to the following formula: (4) in, For historical datasets; The transformed historical dataset; It is the natural logarithm; 7): Calculate the distance and semivariance between each data point; 8): Cross-validation selects the best-fitting model and calculates relevant parameters. The cross-validation process splits the transformed historical dataset into a training set and a validation set. Then, the training set data is used to fit a kriging model to each semivariogram function model. Next, each fitted kriging model is used to predict the validation set data. Finally, the prediction results of each kriging model are evaluated, and the best-performing model is selected as the best model based on the evaluation results. The semivariance function models include Gaussian models, linear models, spherical models, exponential models, and power function models, and the expressions for each model are as follows: Gaussian model: (7) Linear model: (8) Spherical model: (9) Exponential model: (10) Power function model: (11) in It's a nugget of gold, representing the variance under small variations in non-zero distance; c depends on the variance or amplitude of the data. It is a range parameter; 9): Use Kriging interpolation to perform interpolation and obtain the Kriging interpolation dataset; 10): The ground mobile monitoring platform stores the datasets obtained by both real-time data interpolation and historical data interpolation methods in its database. The data that was not lost is defined as the data received by the ground mobile monitoring platform once in which the pollutant data, meteorological data, corresponding latitude and longitude coordinates, sampling time, and UAV altitude value were not lost. The lost data is defined as the data received by the ground mobile monitoring platform once, in which the corresponding latitude and longitude coordinates, UAV altitude value and sampling time are not lost, but pollutant data or meteorological data are lost. Missing data with interpolation tags is defined as missing data with interpolation tags. Interpolated missing data is defined as data that has been interpolated using the IDW method to fill in the missing data to be interpolated; The historical dataset is defined as either a dataset that has undergone real-time data interpolation or a dataset that has not undergone real-time data interpolation.

2. The method for compensating for lost atmospheric data from unmanned aerial vehicle (UAV) monitoring based on real-time and historical data according to claim 1, characterized in that, The ground-based mobile monitoring platform can receive data monitored by sensors mounted on the UAV and visualize the monitoring data.

3. The method for compensating for lost atmospheric data from unmanned aerial vehicle (UAV) monitoring based on real-time and historical data according to claim 1, characterized in that, The monitoring data mentioned in step 2) includes pollutant data, meteorological data, corresponding latitude and longitude coordinates, sampling time, and drone altitude value.

4. The method for compensating for lost atmospheric data from unmanned aerial vehicle (UAV) monitoring based on real-time and historical data according to claim 1, characterized in that, The real-time data interpolation method described in step 2) enables real-time IDW interpolation of lost data based on both unlost data and interpolated lost data during the monitoring process; The historical data interpolation method described above can perform Kriging interpolation based on historical data after monitoring is completed.

5. The method for compensating for lost atmospheric data from unmanned aerial vehicle (UAV) monitoring based on real-time and historical data according to claim 1, characterized in that, The expression for the IDW method is as follows: Distance formula: (1) Inverse distance weighted interpolation formula: (2) (3) in, For prediction points With known observation points The Euclidean distance between them; For the first The longitude coordinates of one known data point; The longitude coordinates of the predicted point; For the first Latitude coordinates of one known data point; These are the latitude coordinates of the predicted point. For prediction points The estimated value; For known observation points The value; It is to give the observation value The assigned weights; It is the total number of known observation points; It is a positive real number that determines how quickly the weight changes with distance.

6. The method for compensating for lost atmospheric data from unmanned aerial vehicle (UAV) monitoring based on real-time and historical data according to claim 1, characterized in that, The formula for calculating the distance between data points in step 7) is as follows: (5) in, For the first The longitude coordinates of each point; For the first The longitude coordinates of each point; For the first The latitude coordinates of each point; For the first The latitude coordinates of each point; The formula for calculating the semivariance between data points is as follows: (6) in, The distance is The number of data pairs and These are the observations at the corresponding locations.

7. The method for compensating for lost atmospheric data from unmanned aerial vehicle (UAV) monitoring based on real-time and historical data according to claim 1, characterized in that, The Kriging interpolation formula mentioned in step 9) is as follows: Prediction formula: (12) Kriging weights: (13) in, The predicted value corresponding to the predicted point. It is Observations at known points It is The weights of the known points.

Images