Sky-ground remote sensing integrated lake and reservoir water quality parameter quantitative inversion method

Abstract

The invention belongs to the technical field of water quality remote sensing monitoring, and particularly relates to a sky-ground remote sensing integrated lake and reservoir water quality parameter quantitative inversion method. Qualitative inversion is carried out on water quality parameters through medium-high resolution satellite remote sensing images, and ground sampling points are reasonably arranged. And carrying out ground and air collaborative operation, including on-site water body sampling and unmanned aerial vehicle flight, so as to ensure the consistency of sampled water quality parameters and unmanned aerial vehicle image spectral characteristics. The sampling point water body sample is sent to a laboratory to be tested, and measured data are obtained. An unmanned aerial vehicle image is subjected to preprocessing including splicing and radiation correction, and a reflectivity image map is obtained. Water vector data are extracted by using an unmanned aerial vehicle image, a remote sensing quantitative inversion model of water quality parameters is established, and the model is analyzed and optimized through an optimal wave band combination. And in combination with the unmanned aerial vehicle image, the inversion model and the water body vector data, obtaining a water quality parameter quantitative inversion result and spatial distribution characteristics in the lake and reservoir range. The method can improve ground sampling rationality and time coordination of lake and reservoir water quality monitoring, and supports long-term high-precision dynamic monitoring.

Description

Technical Field

[0001] The invention belongs to the technical field of water quality remote sensing monitoring, and specifically relates to a quantitative inversion method for lake water quality parameters by integrating sky-ground remote sensing. Background Art

[0002] Water resources are important resources that are indispensable for human survival and social development.

[0003] Lake-reservoir water sources are a very important type of water source, and their water quality is directly related to the production and living water of the surrounding residents. Water quality monitoring is an important prerequisite for water resources planning, management and protection. Timely grasping the temporal and spatial distribution characteristics of lake and reservoir water quality, and then taking reasonable measures and means to treat its pollution sources and pollutants in a targeted manner, is of great significance to achieving good water quality in lakes and reservoirs, ensuring water safety and sustainable use.

[0004] Since the lake is small in area, has complex components, and is severely affected by human activities, it requires long-term, continuous, and high-precision dynamic monitoring. Although the traditional fixed-point water quality sampling and testing monitoring method can accurately measure the content of water quality parameters, it is affected by factors such as manpower, material resources, financial resources, and lake environment, resulting in many restrictions on the time, frequency, and quantity of water sample collection. At the same time, these sampling points can only represent the water characteristics of local waters and typical seasons, and cannot represent the actual situation of the entire lake, nor can it dynamically reflect the changes in the water quality of the lake in time and space.

[0005] Compared with traditional monitoring methods, remote sensing technology has the characteristics of wide monitoring range, low cost and strong timeliness. It can break through the constraints of time and space to achieve large-scale continuous dynamic monitoring. It can not only provide spatial distribution characteristics at different times and make up for the defects of traditional monitoring data in spatial discontinuity, but also can be used to monitor and analyze the dynamic changes of water quality, providing a new method for grasping the water quality status and dynamic changes of the entire lake and reservoir.

[0006] At present, high-precision quantitative remote sensing inversion of lake and reservoir water quality parameters is mainly based on the semi-empirical method of analyzing the spectral characteristics of water quality parameters. Through correlation analysis with the measured water quality parameter concentrations of lakes and reservoirs, statistical analysis methods or emerging mathematical methods such as artificial neural networks and support vector machines are used to establish a water quality parameter inversion model. The model is then applied to the remote sensing images of the entire water area to ultimately obtain the spatial distribution of water quality in the entire lake or reservoir.

[0007] Therefore, the inversion of lake and reservoir water quality mainly includes two types of data, one is the measured data of water quality parameters, and the other is remote sensing image data; and due to the requirements of periodic and high-precision dynamic monitoring of lake and reservoir water quality, higher requirements are put forward for both measured data and remote sensing image data. First, in terms of measured data, considering the sampling cost, efficiency and effectiveness, there are high requirements for the layout location of sample points, the number of sampling points and the representativeness of sample points. How to reasonably arrange sampling points, collect water samples efficiently and at low cost, and ensure that the sampling points are representative is an important link to improve the accuracy of lake and reservoir water quality monitoring; secondly, the essence of constructing a water quality parameter inversion model is to establish a relationship model between water quality parameters and remote sensing image spectral characteristics. It is necessary to ensure that the water quality parameters are consistent with the spectral characteristics. Therefore, it is necessary to ensure that the ground sampling time and the remote sensing image acquisition time are highly coordinated.

[0008] At present, ground water quality sampling generally selects different water depths within the lake or reservoir or uses the grid method to evenly distribute sampling points. This method has a certain rationality, but there is no reliable theoretical basis for how many samples to distribute and whether the samples are representative. On the one hand, satellite remote sensing technology has a re-entry period of at least several days, and on the other hand, it is easily affected by cloudy and foggy weather, which greatly reduces the possibility of satellites taking effective images during transit, causing great difficulties for satellite-ground coordination and increasing uncertainty. In extreme cases, it may be found that the satellite has not taken any images after ground sampling and testing, resulting in a huge waste of manpower and material resources.

[0009] In recent years, with the development of UAV technology, the conditions for UAVs to serve as a new type of remote sensing platform have gradually matured. Equipped with various advanced remote sensing equipment, such as high-resolution cameras and spectrometers, UAVs are playing an increasingly important role in acquiring spatial information. UAVs have the characteristics of strong maneuverability, high timeliness, flexible and simple operation, and are not affected by clouds. They have unparalleled advantages over satellite remote sensing in terms of satellite-ground coordination, and can ensure the consistency of ground sampling time and image acquisition time to the greatest extent, providing more convenience for continuous, high-precision dynamic monitoring of lake and reservoir water quality. Summary of the invention

[0010] The purpose of the present invention is mainly to address the problems of rationality and representativeness of the layout of ground sampling points in the process of continuous and high-precision dynamic monitoring of lake and reservoir water quality, as well as the problem that it is difficult to ensure the coordination between satellite and ground in time between ground sampling and remote sensing image acquisition. A quantitative inversion method for lake and reservoir water quality parameters integrating sky-ground remote sensing is proposed. On the one hand, this method makes the layout of ground sampling points more reasonable and representative, and on the other hand, it ensures the coordination between satellite and ground in time between ground sampling and image acquisition to the greatest extent, providing strong support for the long-term, continuous and high-precision dynamic monitoring of lake and reservoir water quality.

[0011] This plan specifically includes the following steps:

[0012] S1: Layout of ground sampling points: Based on medium- and high-resolution satellite images such as Sentinel-2, and using the empirical formula for water quality parameter inversion, first conduct a qualitative inversion of the water quality of lakes and reservoirs, and then reasonably layout ground sampling points based on the qualitative inversion results to make the sample point selection more representative. This includes satellite remote sensing image acquisition, remote sensing image preprocessing, qualitative inversion of water quality using empirical formulas, and layout of ground sampling points.

[0013] S2: Collaborative ground-air operations include on-site water sampling and on-site drone flights. Water sampling and drone flights are carried out simultaneously in near real time to ensure the consistency of sampled water quality parameters with the spectral characteristics of drone images.

[0014] S3: Send water samples for testing. The water samples collected on site are sent to the laboratory for testing to obtain the actual measured data of water quality parameters.

[0015] S4: UAV image preprocessing, including image stitching, aerial triangulation and radiation correction, etc., to obtain the UAV reflectivity image of the lake area.

[0016] S5: Extraction of lake and reservoir water body vector data: Extract water body vectors based on drone reflectivity images using visual interpretation methods.

[0017] S6: Establish a lake and reservoir parameter inversion model, conduct an optimal band combination analysis of water quality parameters based on the measured water quality parameter data and the drone image reflectance values ​​corresponding to the sampling points, screen out the optimal band combination for water quality parameter inversion, establish a lake and reservoir water quality parameter inversion model, and verify the accuracy of the inversion model.

[0018] S7: Combining the UAV images of lakes and reservoirs, the inversion model of lake and reservoir water quality parameters and the lake and reservoir water body vector, the quantitative inversion results of water quality parameters in the entire lake and reservoir range and their spatial distribution characteristics are obtained.

[0019] As a further technical solution of the present invention, step one specifically comprises: collecting recent Sentinel-2 satellite remote sensing images of the lake and reservoir area, and performing preprocessing such as radiation calibration, atmospheric correction, and orthorectification to obtain a Sentinel-2 orthophoto base map of the lake and reservoir area;

[0020] Determine the empirical formula for water quality parameter inversion and obtain the empirical formula for water quality parameter inversion using Sentinel-2 images. The empirical formula is:

[0021] YTP=0.0234*exp(0.7508*[(float(b5)-float(b4)) / (float(b5)+float(b4))]-0.4707)

[0022] Among them, b5 represents the reflectance of the 5th band of Sentinel-2 image, b4 represents the reflectance of the 4th band of Sentinel-2 image, and YTP represents the total phosphorus parameter concentration in mg / L;

[0023] The above empirical formula is used to qualitatively invert water quality parameters. Qualitative inversion means that the water quality parameter status is inverted through empirical formulas and Sentinel-2 images, showing the relative high and low concentrations of water quality parameters in spatial distribution.

[0024] According to the qualitative inversion results of lake and reservoir water quality, ground sampling points are set up in different areas, water depths, and locations with different concentrations.

[0025] As a further technical solution of the present invention, step 2 specifically includes: ground-air collaborative operation planning, firstly, according to the layout of ground sampling points, design the route for water sampling, and prepare the tools and equipment related to water sampling and water sample preservation; then, according to the surrounding environment and terrain conditions of the lake, plan the flight route of the UAV, lay out a certain number of ground image control points, and prepare the calibration plate, the ground image control points are used for subsequent aerial triangulation encryption processing, and the calibration plate is used for subsequent image reflectivity correction;

[0026] Ground and air operations are carried out simultaneously. For water sampling, we take a boat to the sampling point according to the preset route of the ground sampling point, take samples on the water surface according to the technical specifications for water quality sampling, and use GPS to record the coordinates of the sampling point. While the water sampling work is being carried out, drone flights are carried out according to the pre-planned routes to obtain drone data within the lake and reservoir range, and RTK equipment is used to measure the coordinates of the preset ground image control point positions, including longitude, latitude and elevation values.

[0027] As a further technical solution of the present invention, step three specifically includes: measured parameter data of lake and reservoir water quality. After the water sampling is completed, it is immediately sent to the laboratory for water quality testing to obtain various water quality parameter concentration data of all ground sampling points, and the data are sorted and merged with the sampling point coordinates recorded by GPS to obtain a measured data table of lake and reservoir water quality parameters, including the sampling point number, longitude and latitude and concentration values ​​of various water quality parameters.

[0028] As a further technical solution of the present invention, step four is specifically as follows: using PIX4D software to pre-process the UAV flight data, including data preparation, data import, image stitching, aerial triangulation processing, radiation calibration and calibration plate reflectivity correction, to obtain the UAV reflectivity image of the lake area.

[0029] As a further technical solution of the present invention, step five is specifically as follows: based on the drone reflectivity image map of the lake and reservoir area, using ArcGIS software and adopting manual visual interpretation, the lake and reservoir water body is vectorized to obtain the lake and reservoir water body vector data.

[0030] As a further technical solution of the present invention, step six is ​​specifically as follows: based on the latitude and longitude data of the ground sampling points and the drone reflectivity image, the reflectivity values ​​of each band of the drone image corresponding to each ground sampling point are extracted, and combined with the measured data table of lake water quality parameters obtained in step 301, the measured water quality parameter data of all ground sampling points and the corresponding drone reflectivity value data table of each band are sorted out.

[0031] The optimal band combination analysis of water quality parameters, firstly, based on the obtained data table, the bands of the drone image are combined in a certain way, and the image reflectance value under each band combination is calculated. Specifically, any 4 bands of the remote sensing image are randomly combined according to the four operations of "addition, subtraction, multiplication and division" to obtain the image reflectance value under any 4 band combination; secondly, according to the measured water quality parameter content of each sampling point and the reflectance value under any 4 band combination of the corresponding sampling point, Pearson correlation analysis is performed to find the band combination with the highest correlation with each water quality index, that is, the optimal band combination of the water quality index;

[0032] The Pearson correlation coefficient is an important indicator to measure the correlation between two random variables x and y. Its calculation formula is:

[0033] r=∑[(xx mean )(yy mean )] / [sqrt(∑(xx mean ) 2 *sqrt(∑(yy mean ) 2 )]

[0034] Among them, x and y represent the values ​​of two variables respectively. mean and mean They represent the average values ​​of x and y respectively, and r represents the correlation coefficient, which is between (-1, 1). The larger the value, the higher the correlation.

[0035] To establish a water quality parameter inversion model and precision verification, the water quality sampling data was first divided into two parts at a ratio of 7:3, 70% for modeling the inversion model, and 30% for precision verification; using 70% of the water quality sampling data and the corresponding optimal band combination reflectivity calculation results, a linear regression modeling method was used to establish a lake water quality parameter inversion model; and the remaining 30% of the water quality sampling measured data was used to verify the precision of the inversion results;

[0036] As a further technical solution of the present invention, firstly, in ArcGIS software, the drone reflectivity image of the lake area is cropped using the lake water body vector to obtain the drone reflectivity image map of the lake water area; then, the water quality parameter inversion model established in step 603 is applied to the cropped drone reflectivity image map of the lake water area, thereby obtaining the remote sensing quantitative inversion results of the water quality parameters of the entire lake range, and the spatial distribution characteristics of the water quality parameters can be intuitively seen from the inversion results.

[0037] The present invention has the following beneficial effects:

[0038] This method is based on satellite remote sensing images and uses empirical formulas for water quality parameter inversion to qualitatively invert lake and reservoir water quality, providing a theoretical basis for the selection of ground sampling points, making the layout of ground sampling points more reasonable and representative, and making water body sampling work more scientific and efficient.

[0039] This method uses drones to replace satellites, and collects ground measured data and drone images of lakes and reservoirs through ground-air collaborative operations. On the one hand, it can avoid the invalid sampling work caused by the uncertainty of satellite remote sensing shooting, which causes waste of manpower and material resources; on the other hand, it can ensure the consistency of water body sampling time and image acquisition time to the greatest extent, ensure the correspondence between water quality parameters and image spectral characteristics, and can greatly improve the accuracy of quantitative inversion of lake water quality parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a technical flow chart of the quantitative inversion method of lake water quality parameters based on integrated sky-ground remote sensing;

[0041] Figure 2 UAV images and water body vector distribution map of Jinyinqiao Reservoir;

[0042] Figure 3 This is the quantitative inversion result diagram of total phosphorus index in Jinyinqiao Reservoir;

[0043] Figure 4 It is a quantitative inversion method for lake water quality parameters based on integrated sky-ground remote sensing. DETAILED DESCRIPTION

[0044] The present invention is described below in conjunction with the accompanying drawings. A quantitative inversion method for lake water quality parameters integrating sky-ground remote sensing includes the following steps:

[0045] Step 1: Layout of ground sampling points: Based on Sentinel-2 satellite remote sensing images and using the empirical formula for water quality parameter inversion, first conduct a qualitative inversion of lake water quality conditions, and then reasonably layout ground sampling points based on the qualitative inversion results to make the sample point selection more representative. This includes satellite remote sensing image acquisition, remote sensing image preprocessing, qualitative inversion of water quality using empirical formulas, and layout of ground sampling points.

[0046] Step 101, collect recent Sentinel-2 satellite remote sensing images of the lake area, and use ENVI software to perform radiometric calibration, atmospheric correction, orthorectification and other preprocessing to obtain the Sentinel-2 orthophoto base map of the lake area. Among them, the Sentinel-2 images are downloaded from the official website of the European Space Agency (https: / / dataspace.copernicus.eu / );

[0047] Step 102: Determine the empirical formula for water quality parameter inversion. Consult relevant literature to obtain the empirical formula for water quality parameter inversion using Sentinel-2 images, such as total phosphorus, which is:

[0048] YTP=0.0234*exp(0.7508*[(float(b5)-float(b4)) / (float(b5)+float(b4))]-0.4707)

[0049] Among them, b5 represents the reflectance of the 5th band of Sentinel-2 image, b4 represents the reflectance of the 4th band of Sentinel-2 image, and YTP represents the total phosphorus parameter concentration in mg / L.

[0050] Step 103: Use the above empirical formula to perform qualitative inversion of water quality parameters. Qualitative inversion refers to the inversion of water quality parameter conditions through empirical formulas and Sentinel-2 images. However, since there is no measured water quality data involved, the results only represent the high and low concentration relationship of water quality parameters in spatial distribution, and the concentration value has no practical meaning. Therefore, only qualitative evaluation of water quality conditions can be performed.

[0051] Step 104, based on the qualitative inversion results of lake water quality, ground sampling points are arranged in different areas, different water depths, and different concentrations. For example, in areas with higher concentrations of water quality parameters, more sampling points can be arranged relatively densely, while in areas with lower concentrations of water quality parameters, some sampling points can be arranged appropriately dispersedly.

[0052] Step 2: Ground-air collaborative operations include on-site water sampling and on-site drone flights. Water sampling and drone flights are carried out simultaneously in near real time to ensure the consistency of water quality parameters with the spectral characteristics of drone images.

[0053] Step 201, ground-air collaborative operation planning, first design the water sampling route according to the layout of ground sampling points, prepare relevant tools and equipment for water sampling and water sample preservation, such as sampling bottles, water sample protectants and refrigeration measures, etc.; then plan the UAV flight route according to the surrounding environment and terrain conditions of the lake reservoir, deploy a certain number of ground image control points, and prepare calibration plates. The ground image control points are used for subsequent aerial triangulation processing, and the calibration plates are used for subsequent image reflectivity correction.

[0054] Step 202, ground and air operations are carried out simultaneously. For water sampling, the sampling point is reached by boat according to the preset route of the ground sampling point. According to the technical specifications for water quality sampling, sampling is carried out 30-50 cm below the water surface, and the coordinates of the sampling point are recorded using GPS. While the water sampling work is being carried out, the drone is flown according to the pre-planned route to obtain drone data within the lake and reservoir range, and the coordinates of the preset ground image control point positions are measured using RTK equipment, including longitude, latitude and elevation values.

[0055] Step 3: Send water samples for testing. Send the water samples collected on site to the laboratory for testing to obtain actual measured data of water quality parameters.

[0056] Step 301, measured parameter data of lake and reservoir water quality. After the water sampling is completed, it is immediately sent to the laboratory for water quality testing to obtain the concentration data of various water quality parameters of all ground sampling points, and the data are sorted and merged with the sampling point coordinates recorded by GPS to obtain the measured data table of lake and reservoir water quality parameters, including the sampling point number, longitude and latitude and concentration values ​​of various water quality parameters.

[0057] Step 4: UAV image preprocessing, including image stitching, aerial triangulation processing and radiation correction, to obtain the UAV reflectivity image of the lake area.

[0058] Step 401: Use PIX4D software to pre-process the UAV flight data, including data preparation, data import, image stitching, aerial triangulation, radiation calibration, and calibration plate reflectivity correction, to obtain a UAV reflectivity image of the lake area.

[0059] Step 5: Extract lake and reservoir water body vector data. Extract water body vectors based on drone reflectivity images using visual interpretation methods.

[0060] Step 501: Based on the drone reflectivity image of the lake area, ArcGIS software is used to vectorize the lake water body by manual visual interpretation to obtain the lake water body vector data.

[0061] Step 6. Establish a lake and reservoir parameter inversion model. According to the measured water quality parameter data and the drone image reflectance values ​​corresponding to the sampling points, conduct an analysis of the best band combination for water quality parameter inversion, screen out the best band combination for water quality parameter inversion, establish a lake and reservoir water quality parameter inversion model, and verify the accuracy of the inversion model.

[0062] Step 601: Based on the latitude and longitude data of the ground sampling points and the drone reflectivity image, the reflectivity values ​​of each band of the drone image corresponding to each ground sampling point are extracted, and combined with the measured data table of lake water quality parameters obtained in step 301, the measured water quality parameter data of all ground sampling points and the corresponding drone reflectivity value data table of each band are sorted out.

[0063] Step 602, analysis of the best band combination of water quality parameters. First, based on the data table obtained in step 601, the bands of the drone image are combined in a certain way, and the image reflectance value under each band combination is calculated. Specifically, any 4 bands of the remote sensing image are randomly combined according to the four arithmetic operations of "addition, subtraction, multiplication and division" to obtain the image reflectance value under any 4 band combination. Secondly, according to the measured water quality parameter content of each sampling point and the reflectance value under any 4 band combination of the corresponding sampling point, Pearson correlation analysis is performed to find the band combination with the highest correlation with each water quality index, that is, the best band combination of the water quality index.

[0064] The Pearson correlation coefficient is an important indicator to measure the correlation between two random variables x and y. Its calculation formula is:

[0065] r=∑[(xx mean )(yy mean )] / [sqrt(∑(xx mean ) 2 *sqrt(∑(yy mean ) 2 )]

[0066] Among them, x and y represent the values ​​of two variables respectively. mean and mean They represent the average values ​​of x and y respectively, and r represents the correlation coefficient, which is between (-1, 1). The larger the value, the higher the correlation.

[0067] Step 603: Establish a water quality parameter inversion model and verify its accuracy. First, divide the water quality sampling data into two parts according to a ratio of 7:3, 70% for modeling the inversion model, and 30% for accuracy verification. Use 70% of the water quality sampling data and the corresponding optimal band combination reflectivity calculation results to establish a lake water quality parameter inversion model using a linear regression modeling method. Use the remaining 30% of the water quality sampling measured data to verify the accuracy of the inversion result.

[0068] Step 7. Combine the lake and reservoir UAV images, lake and reservoir water quality parameter inversion model and lake and reservoir water body vector to obtain the quantitative inversion results of water quality parameters and their spatial distribution characteristics for the entire lake and reservoir.

[0069] Step 701. First, in ArcGIS software, the UAV reflectivity image of the lake area is cropped using the lake water body vector to obtain the UAV reflectivity image of the lake water area; then, the water quality parameter inversion model established in step 603 is applied to the cropped UAV reflectivity image of the lake water area, thereby obtaining the quantitative inversion results of the water quality parameters in the entire lake range. The spatial distribution characteristics of the water quality parameters can be intuitively seen from the inversion results.

[0070] Example

[0071] Taking the Jinyinqiao Reservoir in Zigong City as an example, a quantitative inversion method for lake and reservoir water quality parameters integrating sky-ground remote sensing is explained below.

[0072] The Jinyinqiao Reservoir is located in Yantan District, Zigong City. It is a comprehensive small reservoir integrating flood control, water storage and irrigation, and urban water supply. In 2018, it was adjusted from a backup water source to a township drinking water source, and officially provided drinking water for rural residents in Yong'an Town. The daily water supply is 4,000 tons, and the water supply population is 12,600.

[0073] This case takes the total phosphorus index in the water quality parameters as an example, and adopts the method of the present invention to quantitatively invert the total phosphorus index of the Jinyinqiao Reservoir.

[0074] Firstly, the Sentinel-2 satellite remote sensing images of the Jinyinqiao Reservoir area were downloaded from the official website of the European Space Agency (https: / / dataspace.copernicus.eu / ), and preprocessed using the ENVI software. The total phosphorus index content of the Jinyinqiao Reservoir was qualitatively inverted using the empirical formula for total phosphorus index parameter inversion.

[0075] The empirical formula for inverting total phosphorus from Sentinel-2 images is:

[0076] YTP=0.0234*exp(0.7508*[(float(b5)-float(b4)) / (float(b5)+float(b4))]-0.4707)

[0077] Among them, b5 represents the reflectance of the 5th band of Sentinel-2 image, b4 represents the reflectance of the 4th band of Sentinel-2 image, and YTP represents the total phosphorus parameter concentration in mg / L.

[0078] According to the qualitative inversion results, 40 ground sampling points were arranged as evenly as possible in different areas of the Jinyinqiao Reservoir and at different locations with different total phosphorus contents. The qualitative inversion results of the total phosphorus index of the Jinyinqiao Reservoir and the layout of the ground sampling points are shown in the figure. Figure 1 shown.

[0079] Then, ground and air coordinated operations were carried out to conduct on-site water sampling and on-site drone flights. The water sampling work and drone flights were carried out simultaneously in near real time, obtaining water samples from 40 sampling points and drone flight data in the Jinyinqiao Reservoir area.

[0080] Then, on the day of sampling, the water samples were preserved and sent to the laboratory for testing to obtain quantitative measured data of various water quality parameters, including chlorophyll a, total phosphorus, total nitrogen, ammonia nitrogen, suspended solids concentration, turbidity, etc. The results are shown in Table 1:

[0081] Table 1 Laboratory test results of water quality parameters of Jinyinqiao Reservoir

[0082]

[0083]

[0084] Secondly, the UAV flight data was prepared, imported, stitched, processed by air triangulation, calibrated by radiation, and corrected by the calibration plate reflectivity using PIX4D software to obtain the UAV reflectivity image of the lake area. Based on the UAV reflectivity image, ArcGIS software was used to vectorize the lake water body by manual visual interpretation to obtain the lake water body vector data. The results are as follows: Figure 2 shown.

[0085] Next, based on the latitude and longitude data of the ground sampling points and the UAV reflectivity images, the reflectivity values ​​of each band of the UAV images corresponding to each ground sampling point were extracted. Combined with the water quality parameter data tested in the laboratory, the measured data of the total phosphorus index of 40 ground sampling points and the corresponding UAV reflectivity value data table of each band were obtained. The data of the 40 sampling points were divided into two parts in a ratio of 7:3. 28 points were used to build the inversion model and 12 points were used for accuracy verification, as shown in Table 2.

[0086] Among them, B1, B2, B3, B4, and B5 are the reflectivity values ​​of the drone’s blue band, green band, red band, red edge band, and infrared band, respectively.

[0087] Table 2 Measured data of total phosphorus index at 40 sampling points in Jinyinqiao Reservoir and corresponding reflectivity values ​​of each band of drone

[0088]

[0089]

[0090] Furthermore, the total phosphorus index parameter inversion modeling was carried out, specifically: 4 bands were randomly selected from the 5 bands, and they were randomly combined according to the four operations of "addition, subtraction, multiplication and division" to obtain the image reflectance value under any combination of 4 bands, and the reflectance result of the combination was compared with the total phosphorus measured data, and the Pearson correlation analysis was performed to find the band combination with the highest correlation with the total phosphorus index. When the total phosphorus index inversion model was constructed, the 10 band combinations with the highest correlation were shown in Table 3:

[0091] Table 3 The top ten band combinations and corresponding Pearson correlation coefficients of the total phosphorus index inversion model of Jinyinqiao Reservoir

[0092]

[0093]

[0094] From the above table, we can see that the band combination with the highest correlation with the total phosphorus index is: b4 / b2+b4 / b5. The reflectivity results calculated by this band combination and the corresponding measured data of the total phosphorus index are subjected to linear regression analysis, and the inversion model of the total phosphorus index is obtained as follows:

[0095] P TP =(-0.7854)×R TP +1.7591

[0096] Among them, the model determination coefficient R 2 It is 0.8352.

[0097] The accuracy of the model was verified using 12 validation data, and it was found that the average accuracy of the model was 96.21%, as shown in Table 4 below.

[0098] Table 4 Total phosphorus index inversion model accuracy verification table

[0099]

[0100] Finally, the quantitative inversion results and spatial distribution of the total phosphorus index of the Jinyinqiao Reservoir were obtained by combining the drone images, the total phosphorus index inversion model and the reservoir water body vector. Figure 3 shown.

Claims

1. A quantitative inversion method for lake water quality parameters based on integrated sky-ground remote sensing, characterized in that: The following steps are involved: S1: Ground sampling point layout: Based on the medium- and high-resolution satellite remote sensing images of Sentinel-2, using the empirical formula for water quality parameter inversion, the water quality of lakes and reservoirs is first qualitatively inverted, and the ground sampling points are reasonably arranged according to the qualitative inversion results; this includes satellite remote sensing image acquisition, remote sensing image preprocessing, qualitative inversion of water quality using empirical formulas, and ground sampling point layout; S2: Ground-air collaborative operations, including on-site water sampling and on-site drone flights, with water sampling and drone flights being performed simultaneously in near real time to ensure consistency between water quality parameters and spectral characteristics of drone images; S3: Send water samples for testing. Send the water samples collected on site to the laboratory for testing to obtain measured data of water quality parameters; S4: UAV image preprocessing, including image stitching, aerial triangulation and radiation correction, to obtain the UAV reflectivity image of the lake area; S5: Extraction of lake and reservoir water body vector data: Extract water body vectors based on drone reflectivity images using visual interpretation methods; S6: Establish a lake parameter inversion model, perform optimal band combination analysis of water quality parameters based on measured water quality parameter data and the reflectivity value of the drone image corresponding to the sampling point, screen out the optimal band combination for water quality parameter inversion, establish a lake water quality parameter inversion model, and verify the accuracy of the inversion model; S7: Combining the UAV images of lakes and reservoirs, the inversion model of lake and reservoir water quality parameters and the lake and reservoir water body vector, the quantitative inversion results of water quality parameters in the entire lake and reservoir range and their spatial distribution characteristics are obtained.

2. The quantitative inversion method for lake water quality parameters based on integrated sky-ground remote sensing according to claim 1 is characterized in that: S1 specifically includes the following steps: S 101 : Collect recent Sentinel-2 satellite remote sensing images of the lake and reservoir area, and perform pre-processing such as radiation calibration, atmospheric correction, and orthorectification to obtain the Sentinel-2 orthophoto base map of the lake and reservoir area; S 102 :Determine the empirical formula for water quality parameter inversion and obtain the empirical formula for water quality parameter inversion using Sentinel-2 images. The empirical formula is: YTP=0.0234*exp(0.7508*[(float(b5)-float(b4)) / (float(b5)+float(b4))]-0.4707) Among them, b5 represents the reflectance of the 5th band of Sentinel-2 image, b4 represents the reflectance of the 4th band of Sentinel-2 image, and YTP represents the total phosphorus parameter concentration in mg / L; S 103 :Use the above empirical formula to perform qualitative inversion of water quality parameters; qualitative means inverting the water quality parameter status through empirical formula and Sentinel-2 images, and presenting the relative high and low concentrations of water quality parameters in spatial distribution; S 104 :Based on the qualitative inversion results of lake and reservoir water quality, ground sampling points are arranged in different areas, water depths, and concentrations.

3. The quantitative inversion method for lake water quality parameters based on integrated sky-ground remote sensing according to claim 1 is characterized in that: S2 specifically includes the following steps: S 201 : Ground-air collaborative operation planning: First, according to the layout of ground sampling points, design the water sampling route and prepare the relevant tools and equipment for water sampling and water sample preservation; then, according to the surrounding environment and terrain conditions of the lake, plan the UAV flight route, deploy a certain number of ground image control points, and prepare calibration plates. The ground image control points are used for subsequent aerial triangulation encryption processing, and the calibration plates are used for subsequent image reflectivity correction; S 202 : Ground and air operations are carried out simultaneously. For water sampling, we follow the preset route of the ground sampling point and take a boat to the sampling point. According to the technical specifications for water quality sampling, we take samples on the water surface and use GPS to record the coordinates of the sampling points. While the water sampling work is being carried out, drone flights are carried out according to the pre-planned routes to obtain drone data within the lake and reservoir range, and RTK equipment is used to measure the coordinates of the preset ground image control point positions, including longitude, latitude and elevation values.

4. The method for quantitative inversion of lake water quality parameters based on integrated sky-ground remote sensing according to claim 1 is characterized in that: S3 specifically includes the following steps: S 301 : Measured parameter data of lake and reservoir water quality. After the water sampling is completed, it is immediately sent to the laboratory for water quality testing to obtain the concentration data of various water quality parameters of all ground sampling points, and then sorted and merged with the sampling point coordinates recorded by GPS to obtain the measured data table of lake and reservoir water quality parameters, including the sampling point number, longitude and latitude and concentration values ​​of various water quality parameters.

5. The method for quantitative inversion of lake water quality parameters based on integrated sky-ground remote sensing according to claim 1 is characterized in that: S4 specifically includes the following steps: The UAV flight data was preprocessed using PIX4D software, including data preparation, data import, image stitching, aerial triangulation, radiation calibration, and calibration plate reflectivity correction to obtain the UAV reflectivity image of the lake and reservoir area.

6. The method for quantitative inversion of lake water quality parameters based on integrated sky-ground remote sensing according to claim 1 is characterized in that: S5 specifically includes the following steps: S 501 Based on the UAV reflectivity image of the lake and reservoir area, ArcGIS software was used to vectorize the lake and reservoir water body by manual visual interpretation to obtain the lake and reservoir water body vector data.

7. The method for quantitative inversion of lake water quality parameters based on integrated sky-ground remote sensing according to claim 1 is characterized in that: S6 specifically includes the following steps: S 601 : Based on the latitude and longitude data of the ground sampling points and the UAV reflectivity images, the reflectivity values ​​of each band of the UAV images corresponding to each ground sampling point are extracted and combined with S 301 The measured data table of lake and reservoir water quality parameters obtained in the experiment was used to sort out the measured water quality parameter data of all ground sampling points and the corresponding data table of the reflectivity values ​​of each band of the drone; S 602 :Based on S 601 Get the data table, combine the bands of the drone image in a certain way, and calculate the image reflectance value under each band combination; S 603 : Establish a water quality parameter inversion model and accuracy verification. First, divide the water quality sampling data into two parts in a 7:3 ratio, 70% for modeling the inversion model, and 30% for accuracy verification; use 70% of the water quality sampling data and the corresponding optimal band combination reflectivity calculation results, and use linear regression modeling to establish a lake water quality parameter inversion model; and use the remaining 30% of the water quality sampling measured data to verify the accuracy of the inversion results.

8. The method for quantitative inversion of lake water quality parameters based on integrated sky-ground remote sensing according to claim 7 is characterized in that: Calculating the image reflectance value under each band combination specifically includes: Using any four bands of remote sensing images, we randomly combined them according to the four operations of "addition, subtraction, multiplication and division" to obtain the image reflectance value under any combination of four bands; secondly, we conducted Pearson correlation analysis based on the measured water quality parameter content of each sampling point and the reflectance value under any combination of four bands of the corresponding sampling point to find the band combination with the highest correlation with each water quality index, that is, the optimal band combination of the water quality index; The Pearson correlation coefficient is an important indicator to measure the correlation between two random variables x and y. Its calculation formula is: r=∑[(x-x mean )(y-y mean )] / [sqrt(∑(x-x mean ) 2 *sqrt(∑(y-y mean ) 2 )] Among them, x and y represent the values ​​of two variables respectively. mean and mean They represent the average values ​​of x and y respectively, r represents the correlation coefficient, and r is between (-1, 1).

9. The method for quantitative inversion of lake water quality parameters based on integrated sky-ground remote sensing according to claim 1 is characterized in that: S7 specifically includes the following steps: In ArcGIS software, the UAV reflectivity image of the lake and reservoir area is clipped using the lake and reservoir water body vector to obtain the UAV reflectivity image map of the lake and reservoir water body area; S 603 The water quality parameter inversion model established in the paper is applied to the cropped drone reflectivity image of the lake and reservoir water area, so as to obtain the remote sensing quantitative inversion results of the water quality parameters in the entire lake and reservoir range. The spatial distribution characteristics of the water quality parameters can be intuitively seen from the inversion results.

Images