A remote sensing monitoring method and system for water quality parameters of brackish water lakes

Abstract

The invention provides a water quality parameter remote sensing monitoring method and system for a brackish water lake, belonging to the field of remote sensing monitoring. The water quality parameter remote sensing monitoring method comprises: collecting water body spectra and corresponding water samples in multiple areas of a lake to be monitored to obtain multiple water body spectra and multiple water samples; analyzing each water sample to determine the water quality parameter of the corresponding area; performing correlation analysis on each water body spectrum and the water quality parameter of each area, and performing one-dimensional linear fitting to determine multiple water quality parameter monitoring models; collecting a hyperspectral image of the lake to be monitored, and extracting spectral information of each water body pixel; for any water body pixel, according to the spectral information of the water body pixel and based on each water quality parameter monitoring model, determining multiple initial water quality parameters of the water body pixel; determining the water quality parameter of the water body pixel according to the average value of the multiple initial water quality parameters of the water body pixel, so that the water quality parameter of the lake can be quickly detected.

Description

Technical Field

[0001] The invention relates to the field of remote sensing monitoring, and in particular to a remote sensing monitoring method and system for water quality parameters of a brackish water lake. Background Art

[0002] Water quality parameters are the limiting parameters of the surface water quality of brackish lakes. For example, the chemical oxygen demand content in Lake Bosteng is the main limiting parameter. At present, chemical testing is generally used to measure the water quality parameters of lakes. The chemical testing measurement method is time-consuming and laborious, pollutes the environment and has poor timeliness, resulting in low monitoring efficiency and lag.

[0003] Based on the above problems, a new water quality parameter monitoring method is urgently needed to improve monitoring efficiency. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for remote sensing monitoring of water quality parameters of a brackish water lake, which can improve the monitoring efficiency of the water quality parameters of the lake.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A remote sensing monitoring method for water quality parameters of a brackish water lake, comprising:

[0007] Collect water spectra and corresponding water samples in multiple areas of the lake to be monitored to obtain multiple water spectra and multiple water samples;

[0008] Analyze each water sample to determine the water quality parameters of the corresponding area;

[0009] Conduct correlation analysis between the spectra of each water body and the water quality parameters of each region, and perform linear fitting to determine multiple water quality parameter monitoring models;

[0010] Collect hyperspectral images of the lake to be monitored and extract the spectral information of each water pixel;

[0011] For any water body pixel, multiple initial water quality parameters of the water body pixel are determined according to the spectral information of the water body pixel and based on various water quality parameter monitoring models; the water quality parameter of the water body pixel is determined according to the average value of the multiple initial water quality parameters of the water body pixel.

[0012] Optionally, a portable ground object spectrometer is used to collect water spectra of multiple areas in the lake to be monitored.

[0013] Optionally, the water quality parameter is the chemical oxygen demand content; analyzing each water sample to determine the water quality parameter of the corresponding area specifically includes: analyzing each water sample using a rapid digestion spectrophotometry method to determine the corresponding chemical oxygen demand content.

[0014] Optionally, the correlation analysis of each water body spectrum and the water quality parameters of each region is performed, and a one-dimensional linear fitting is performed to determine multiple water quality parameter monitoring models, specifically including:

[0015] Derivative the spectra of each water body to determine the first-order spectrum, second-order spectrum and third-order spectrum of each water body spectrum;

[0016] The single-band reflectance of each water body spectrum was analyzed for correlation with each water quality parameter, and a one-dimensional linear fitting was performed to determine the single-band monitoring model;

[0017] The correlation analysis was conducted between all band ratios of each first-order spectrum and each water quality parameter, and a one-dimensional linear fitting was performed to determine multiple first-order monitoring models;

[0018] The correlation analysis was conducted between all band ratios of each second-order spectrum and each water quality parameter, and a one-dimensional linear fitting was performed to determine multiple second-order monitoring models;

[0019] The correlation analysis was conducted between all band ratios of each third-order spectrum and each water quality parameter, and a one-dimensional linear fitting was performed to determine multiple third-order monitoring models;

[0020] From the single-band monitoring model, multiple first-order monitoring models, multiple second-order monitoring models and multiple third-order monitoring models, multiple monitoring models whose root mean square error is less than the first set threshold and whose average relative deviation is less than the second set threshold are selected as water quality parameter monitoring models.

[0021] Optionally, the correlation analysis of all band ratios of each first-order spectrum and each water quality parameter is performed, and a one-dimensional linear fitting is performed to determine multiple first-order monitoring models, specifically including:

[0022] For any first-order spectrum, all band ratios of the first-order spectrum are calculated, and correlation analysis is performed between all band ratios and various water quality parameters, and a one-dimensional linear fitting is performed to obtain multiple initial first-order monitoring models;

[0023] Calculate the R of each initial first-order monitoring model 2 index;

[0024] According to the R 2 Indicator, select R from multiple initial first-order monitoring models 2 First-order monitoring model when the indicator is greater than the third set threshold.

[0025] Optionally, the correlation analysis of all band ratios of each second-order spectrum and each water quality parameter is performed, and a one-dimensional linear fitting is performed to determine multiple second-order monitoring models, specifically including:

[0026] For any second-order spectrum, all band ratios of the second-order spectrum are calculated, and correlation analysis is performed between all band ratios and various water quality parameters, and a one-dimensional linear fitting is performed to obtain multiple initial second-order monitoring models;

[0027] Calculate the R of each initial second-order monitoring model 2 index;

[0028] According to the R 2 Indicator, select R from multiple initial second-order monitoring models 2 Second-order monitoring model where the indicator is greater than the third set threshold.

[0029] Optionally, the correlation analysis of all band ratios of each third-order spectrum and each water quality parameter is performed, and a one-dimensional linear fitting is performed to determine multiple third-order monitoring models, specifically including:

[0030] For any third-order spectrum, all band ratios of the third-order spectrum are calculated, and correlation analysis is performed between all band ratios and various water quality parameters, and a one-dimensional linear fitting is performed to obtain multiple initial third-order monitoring models;

[0031] Calculate the R of each initial third-order monitoring model 2 index;

[0032] According to the R 2 Indicator, select R from multiple initial third-order monitoring models 2 A third-order monitoring model where the indicator is greater than the third set threshold.

[0033] To achieve the above object, the present invention also provides the following solution:

[0034] A remote sensing monitoring system for water quality parameters of a brackish water lake, comprising:

[0035] A data acquisition unit, used to collect water spectra and corresponding water samples in multiple areas of the lake to be monitored, to obtain multiple water spectra and multiple water samples;

[0036] A water sample analysis unit, connected to the data acquisition unit, for analyzing each water sample to determine the water quality parameters of the corresponding area;

[0037] A model building unit, connected to the data acquisition unit and the water sample analysis unit, respectively, for performing correlation analysis on the spectrum of each water body and the water quality parameters of each region, and performing a one-dimensional linear fitting to determine multiple water quality parameter monitoring models;

[0038] An image acquisition unit is used to acquire the hyperspectral image of the lake to be monitored and extract the spectral information of each water pixel;

[0039] The content determination unit is connected to the model building unit and the image acquisition unit respectively, and is used to determine, for any water body pixel, multiple initial water quality parameters of the water body pixel according to the spectral information of the water body pixel and based on various water quality parameter monitoring models; and determine the water quality parameter of the water body pixel according to the average value of the multiple initial water quality parameters of the water body pixel.

[0040] Optionally, the model building unit includes:

[0041] A derivation module, connected to the data acquisition unit, for derivation of each water body spectrum to determine the first-order spectrum, the second-order spectrum and the third-order spectrum of each water body spectrum;

[0042] A single-band model determination module is connected to the derivation module and the water sample analysis unit, respectively, and is used to perform correlation analysis on the reflectance of a single band of each water body spectrum and each water quality parameter, and perform a one-dimensional linear fitting to determine a single-band monitoring model;

[0043] A first-order model determination module, connected to the derivation module and the water sample analysis unit, respectively, for performing correlation analysis on all band ratios of each first-order spectrum and each water quality parameter, and performing a one-dimensional linear fitting to determine multiple first-order monitoring models;

[0044] A second-order model determination module, connected to the derivation module and the water sample analysis unit, respectively, for performing correlation analysis on all band ratios of each second-order spectrum and each water quality parameter, and performing a one-dimensional linear fitting to determine multiple second-order monitoring models;

[0045] A third-order model determination module is connected to the derivation module and the water sample analysis unit, respectively, and is used to perform correlation analysis on all band ratios of each third-order spectrum and each water quality parameter, and perform a one-dimensional linear fitting to determine multiple third-order monitoring models;

[0046] The monitoring model determination module is respectively connected to the single-band model determination module, the first-order model determination module, the second-order model determination module and the third-order model determination module, and is used to select multiple monitoring models whose root mean square error is less than a first set threshold and whose average relative deviation is less than a second set threshold from the single-band monitoring model, multiple first-order monitoring models, multiple second-order monitoring models and multiple third-order monitoring models as water quality parameter monitoring models.

[0047] Optionally, the first-order model determination module includes:

[0048] The initial first-order model determination submodule is connected to the derivation module and the water sample analysis unit respectively, and is used to calculate all the band ratios of the first-order spectrum for any first-order spectrum, and to perform correlation analysis on all the band ratios and various water quality parameters, and to perform one-dimensional linear fitting to obtain multiple initial first-order monitoring models;

[0049] The first indicator calculation submodule is connected to the initial first-order model determination submodule and is used to calculate the R of each initial first-order monitoring model. 2 index;

[0050] The first selection submodule is connected to the initial first-order model determination submodule and the first indicator calculation submodule respectively, and is used to select the first indicator according to the R of each initial first-order monitoring model. 2 Indicator, select R from multiple initial first-order monitoring models 2 First-order monitoring model when the indicator is greater than the third set threshold.

[0051] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects: it is only necessary to collect water spectra and water samples from several areas in the lake, conduct correlation analysis, and establish multiple water quality parameter monitoring models. Based on the multiple water quality parameter monitoring models, multiple initial water quality parameters of each water pixel in the hyperspectral image of the lake can be determined, and the water quality parameters of the water pixel are determined according to the average value of the multiple initial water quality parameters, and then the water quality parameters of the entire lake are determined. Through remote sensing monitoring, the water quality parameters of the lake can be quickly detected, and the monitoring area is large, which will not pollute the environment of the lake. Moreover, by combining the preferred multiple water quality parameter monitoring models for monitoring, the monitoring accuracy of water quality parameters is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0053] Figure 1 A flow chart of a remote sensing monitoring method for water quality parameters of a brackish water lake according to the present invention;

[0054] Figure 2 It is a schematic diagram of the module structure of the remote sensing monitoring system for water quality parameters of brackish water lakes of the present invention.

[0055] Explanation of symbols:

[0056] Data acquisition unit-1, water sample analysis unit-2, model building unit-3, image acquisition unit-4, content determination unit-5. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] The purpose of the present invention is to provide a remote sensing monitoring method and system for water quality parameters of brackish water lakes. By means of remote sensing monitoring, the monitoring efficiency of water quality parameters is improved, the monitoring area is large, and the environment of the lake will not be polluted. Combined monitoring by multiple models is used to improve monitoring accuracy.

[0059] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0060] like Figure 1 As shown, the remote sensing monitoring method for water quality parameters of brackish water lakes of the present invention comprises:

[0061] S1: Collecting water spectra and corresponding water samples in multiple areas of the lake to be monitored to obtain multiple water spectra and multiple water samples. In this embodiment, a portable ground object spectrometer is used to collect water spectra in multiple areas of the lake to be monitored.

[0062] S2: Analyze each water sample to determine the water quality parameter of the corresponding area. In this embodiment, the water quality parameter is the chemical oxygen demand content, and each water sample is analyzed by rapid digestion spectrophotometry to determine the corresponding chemical oxygen demand content.

[0063] S3: Conduct correlation analysis on the spectra of each water body and the water quality parameters of each region, and perform one-dimensional linear fitting to determine multiple water quality parameter monitoring models.

[0064] S4: Collect hyperspectral images of the lake to be monitored and extract the spectral information of each water pixel. Specifically, collect multiple hyperspectral images of the lake to be detected, perform radiation calibration, atmospheric correction, orthorectification and image stitching on each hyperspectral image to obtain the final hyperspectral image.

[0065] S5: For any water body pixel, determine multiple initial water quality parameters of the water body pixel according to the spectral information of the water body pixel and based on various water quality parameter monitoring models; determine the water quality parameter of the water body pixel according to the average value of the multiple initial water quality parameters of the water body pixel.

[0066] Furthermore, the water quality of the lake to be monitored is determined according to the water quality parameters of each water pixel, and then classified according to the international surface water quality classification standard GB3838-2002 to determine the most polluted area in the lake.

[0067] Furthermore, step S3 specifically includes:

[0068] S31: Deriving each water body spectrum to determine the first-order spectrum, second-order spectrum and third-order spectrum of each water body spectrum. Specifically, in MATLAB, the water body spectrum is preprocessed using the first-order derivative, the second-order derivative and the third-order derivative to obtain the corresponding first-order spectrum, the second-order spectrum and the third-order spectrum.

[0069] S32: Correlation analysis is performed on the reflectance of a single band of each water body spectrum and each water quality parameter, and a one-dimensional linear fitting is performed to determine a single-band monitoring model. Specifically, SPSS (Statistical Product Service Solutions) software is used to perform correlation analysis, determine the best single band with the strongest correlation, and use the reflectance of the best single band as the independent variable, and a one-dimensional linear fitting is performed to obtain a single-band monitoring model. The maximum correlation coefficient is 1, and the minimum is -1. The larger the absolute value of the correlation coefficient, the stronger the correlation.

[0070] S33: Perform correlation analysis on all band ratios of each first-order spectrum and each water quality parameter, and perform one-dimensional linear fitting to determine multiple first-order monitoring models.

[0071] Specifically, step S33 includes:

[0072] S331: For any first-order spectrum, all band ratios of the first-order spectrum are calculated, and correlation analysis is performed between all band ratios and various water quality parameters, and a one-dimensional linear fitting is performed to obtain multiple initial first-order monitoring models.

[0073] S332: Calculate the R of each initial first-order monitoring model 2 index.

[0074] S333: According to the R of each initial first-order monitoring model 2 Indicator, select R from multiple initial first-order monitoring models 2 The first-order monitoring model has an indicator greater than a third set threshold. In this embodiment, the third set threshold is 0.65.

[0075] In this embodiment, all band ratios (|b1 / b2|) and (|(b1-b2) / (b1+b2)|) of the first-order spectrum are subjected to correlation analysis with the corresponding 50 water quality parameters in MATLAB, and the band ratio with the highest correlation is determined according to the size of the correlation coefficient. The absolute value of the band ratio is used as the independent variable, and a one-dimensional linear fitting is performed to obtain a monitoring model for all bands. Among them, b1 and b2 are both reflectances of the band. The error value R of the monitoring model for all bands is calculated 2 , Among them, y i is the measured value of the chemical oxygen demand content of the ith element in the modeling set; y' i is the inverted value of the i-th chemical oxygen demand content in the modeling set; is the measured average value of the chemical oxygen demand content in the modeling set; n is the number of samples in the modeling set. 2 The monitoring model with a value greater than 0.65 is used as the first-order monitoring model. The modeling set includes the water quality parameters determined in step S2.

[0076] S34: Conduct correlation analysis on all band ratios of the second-order spectrum and various water quality parameters, and perform one-dimensional linear fitting to determine multiple second-order monitoring models.

[0077] Specifically, step S34 includes:

[0078] S341: For any second-order spectrum, all band ratios of the second-order spectrum are calculated, and correlation analysis is performed between all band ratios and various water quality parameters, and a one-dimensional linear fitting is performed to obtain multiple initial second-order monitoring models.

[0079] S342: Calculate the R of each initial second-order monitoring model 2 index;

[0080] S343: According to the R of each initial second-order monitoring model 2 Indicator, select R from multiple initial second-order monitoring models 2 Second-order monitoring model where the indicator is greater than the third set threshold.

[0081] S35: Conduct correlation analysis on all band ratios of each third-order spectrum and each water quality parameter, and perform one-dimensional linear fitting to determine multiple third-order monitoring models.

[0082] Specifically, step S35 includes:

[0083] S351: for any third-order spectrum, calculating all band ratios of the third-order spectrum, and performing correlation analysis on all band ratios and various water quality parameters, performing one-dimensional linear fitting, and obtaining multiple initial third-order monitoring models;

[0084] S352: Calculate the R of each initial third-order monitoring model 2 index.

[0085] S353: According to the R of each initial third-order monitoring model 2 Indicator, select R from multiple initial third-order monitoring models 2 A third-order monitoring model where the indicator is greater than the third set threshold.

[0086] The specific method for establishing the second-order monitoring model and the third-order monitoring model is the same as the method for establishing the first-order monitoring model, and will not be repeated here.

[0087] S36: Select multiple monitoring models whose root mean square error is less than the first set threshold and whose average relative deviation is less than the second set threshold from the single-band monitoring model, multiple first-order monitoring models, multiple second-order monitoring models and multiple third-order monitoring models as water quality parameters.

[0088] Specifically, the RMSE (Root Mean Squared Error) and the average relative deviation of the single-band monitoring model, each first-order monitoring model, each second-order monitoring model and each third-order monitoring model are calculated respectively. In this embodiment, the first threshold value is set to 2.5 mg / L, and the second threshold value is set to 10%.

[0089] Furthermore, the chemical oxygen demand monitoring model may include multiple monitoring models or only one monitoring model, depending on the R 2 , root mean square error and mean relative deviation.

[0090] In order to better understand the solution of the present invention, it is further described below in conjunction with specific embodiments.

[0091] This embodiment is described by taking monitoring the chemical oxygen demand content of Bosten Lake as an example.

[0092] Step 1: Design the sampling points and use the ASD spectrometer to collect 50 water spectra and 50 water samples in the corresponding area in Bosten Lake. After the spectral curve is resampled according to the spectral resolution of the hyperspectral image to obtain a new hyperspectral curve, the first-order derivative, second-order derivative and third-order derivative processing are performed to obtain the first-order spectrum, second-order spectrum and third-order spectrum. The laboratory obtains the corresponding 50 chemical oxygen demand contents through the "Rapid Hydrolysis Spectrophotometer for Determination of Chemical Oxygen Demand in Water Quality".

[0093] In SPSS, the correlation analysis of 50 water spectra and the corresponding 50 chemical oxygen demands was performed. The maximum correlation coefficient was 1 and the minimum was -1. The larger the absolute value of the correlation coefficient, the stronger the correlation. SPSS software obtained the best single band with the strongest correlation as 824nm. With the reflectivity of the best single band as the independent variable, the single-band monitoring model obtained by one-dimensional linear fitting was y1=358.22*x1+20.302, R 2 =0.03. According to R 2 It can be seen that the accuracy of this model is not high, and it is necessary to continue to build a model with high accuracy.

[0094] In MATALB, the code was used to perform correlation analysis on all the band ratios (|b1 / b2|) and (|(b1-b2) / (b1+b2)|) of the first-order spectrum, the second-order spectrum, and the third-order spectrum with the corresponding 50 chemical oxygen demands. The band ratio with the highest correlation was obtained by the size of the correlation coefficient, and R was selected. 2 The monitoring models with a value greater than 0.65 are: the ratio of the reflectivity of band 791 to the reflectivity of band 814 in the first-order spectrum: (791-814) / (791+814), the ratio of the reflectivity of band 583 to the reflectivity of band 730 in the first-order spectrum: (583 / 730), the ratio of the reflectivity of band 618 to the reflectivity of band 649 in the second-order spectrum: (618-649) / (618+649), the ratio of the reflectivity of band 706 to the reflectivity of band 730 in the second-order spectrum: (706 / 730), the ratio of the reflectivity of band 707 to the reflectivity of band 825 in the third-order spectrum: (707-825) / (707+825), and the ratio of the reflectivity of band 424 to the reflectivity of band 773 in the third-order spectrum: (424 / 773).

[0095] Taking the absolute value of the band ratio as the independent variable, a one-dimensional linear fitting is performed, and the monitoring models are obtained as follows:

[0096] First-order monitoring model: y2 = -0.3012*x2 + 21.302, R 2 =0.82.

[0097] The second first-order monitoring model: y3 = -0.0267*x3 + 21.052, R 2 =0.79.

[0098] The first and second order monitoring model: y4 = -0.0963*x4 + 21.175, R 2 =0.82.

[0099] Second-order monitoring model: y5 = -0.158*x5 + 21.217, R 2 =0.72.

[0100] The first and third order monitoring model: y6 = -0.4179*x6 + 21.356, R 2 =0.83.

[0101] Second and third order monitoring model: y7 = -0.0499*x7 + 21.208, R 2 =0.72.

[0102] Among them, y1 is the predicted value of chemical oxygen demand content by the single-band monitoring model, and x1 is the optimal single band of the single-band monitoring model; y2 is the predicted value of chemical oxygen demand content by the first-order monitoring model, and x2 is the absolute value of the reflectivity ratio of band 791 to band 814 of the first-order spectrum; y3 is the predicted value of chemical oxygen demand content by the second-order monitoring model, and x3 is the absolute value of the reflectivity ratio of band 583 to band 730 of the first-order spectrum; y4 is the predicted value of chemical oxygen demand content by the first-order monitoring model, and x4 is the absolute value of the reflectivity ratio of band 583 to band 730 of the second-order spectrum. The absolute value of the reflectivity ratio of band 618 to band 649; y5 is the predicted value of the chemical oxygen demand content by the second-order monitoring model, and x5 is the absolute value of the reflectivity ratio of band 706 to band 730 of the second-order spectrum; y6 is the predicted value of the chemical oxygen demand content by the first third-order monitoring model, and x6 is the absolute value of the reflectivity ratio of band 707 to band 825 of the third-order spectrum; y7 is the predicted value of the chemical oxygen demand content by the second third-order monitoring model, and x7 is the absolute value of the reflectivity ratio of band 424 to band 773 of the third-order spectrum.

[0103] Step 2: Perform model adaptation based on the selected hyperspectral satellite: The band with the smallest difference between the hyperspectral satellite band and the ASD selected band is the band adapted by the hyperspectral satellite. The first-order monitoring model, the first-order monitoring model, the second-order monitoring model, the second-order monitoring model, and the second-order monitoring model are selected from the above 7 models based on RMSE (less than 2.5 mg / L) and average relative deviation (less than 10%). Specifically, the chemical oxygen demand content of the validation set is substituted into the 7 models respectively, the chemical oxygen demand content is calculated, and compared with the validation data to evaluate the accuracy. The top four models with the highest accuracy are selected, namely the first-order monitoring model, the first-order monitoring model, the second-order monitoring model, the second-order monitoring model, and the second-order monitoring model.

[0104] The ASD spectrometer model band, the Ziyuan-1 02D model band, and the Gaofen-5 satellite GF5 model band are shown in Table 1.

[0105] Table 1 Hyperspectral satellite adaptation bands based on ASD

[0106]

[0107] Step 3: The COD content value is the average value of the four model estimates in step 2 after removing the abnormal values ​​less than 0 and greater than 40. That is, y = (y2 + y4 + y5 + y7) / 4. This increases the estimation accuracy and avoids accidental errors. y is the final COD content.

[0108] Step 4: After preprocessing the hyperspectral image, the selected quantitative inversion model is used to calculate the chemical oxygen demand of each water pixel on the lake image. The four values ​​of the same water pixel are eliminated and averaged to obtain the chemical oxygen demand content of the water pixel. The range of outliers is less than 0 and greater than 40, and all four estimated values ​​of pixels without outliers are involved in the calculation. Then, the chemical oxygen demand content of each water pixel is analyzed to obtain the chemical oxygen demand of the entire lake, and it is classified according to the national standard surface water quality classification standard GB3838-2002 to find the most polluted area of ​​the lake.

[0109] The establishment of a chemical oxygen demand monitoring model based on the spectrum of the ground feature spectrometer can ensure the accuracy of the chemical oxygen demand content, and the monitoring area is large without polluting the environment. In addition, the chemical oxygen demand content can be traced back through historical images.

[0110] like Figure 2 As shown, the water quality parameter remote sensing monitoring system of the brackish water lake of the present invention comprises: a data acquisition unit 1, a water sample analysis unit 2, a model building unit 3, an image acquisition unit 4 and a content determination unit 5.

[0111] The data acquisition unit 1 is used to collect water spectra and corresponding water samples in multiple areas of the lake to be monitored, and obtain multiple water spectra and multiple water samples.

[0112] The water sample analysis unit 2 is connected to the data acquisition unit 1, and the water sample analysis unit 2 is used to analyze each water sample to determine the water quality parameters of the corresponding area.

[0113] The model building unit 3 is connected to the data acquisition unit 1 and the water sample analysis unit 2 respectively. The model building unit 3 is used to perform correlation analysis on each water body spectrum and each water quality parameter, and perform a one-dimensional linear fitting to determine multiple water quality parameter monitoring models.

[0114] The image acquisition unit 4 is used to acquire the hyperspectral image of the lake to be monitored and extract the spectral information of each water body pixel.

[0115] The content determination unit 5 is connected to the model building unit 3 and the image acquisition unit 4 respectively. The content determination unit 5 is used to determine, for any water body pixel, multiple initial water quality parameters of the water body pixel according to the spectral information of the water body pixel and based on various water quality parameter monitoring models; and determine the water quality parameters of the water body pixel according to the average value of the multiple initial water quality parameters of the water body pixel.

[0116] Furthermore, the model building unit 3 includes: a derivation module, a single-band model determination module, a first-order model determination module, a second-order model determination module, a third-order model determination module and a monitoring model determination module.

[0117] The derivation module is connected to the data acquisition unit 1, and is used to derive each water body spectrum to determine the first-order spectrum, the second-order spectrum and the third-order spectrum of each water body spectrum.

[0118] The single-band model determination module is connected to the derivation module and the water sample analysis unit 2 respectively. The single-band model determination module is used to perform correlation analysis on the single band of each water body spectrum and each water quality parameter, and perform a one-dimensional linear fitting to determine the single-band monitoring model.

[0119] The first-order model determination module is connected to the derivation module and the water sample analysis unit 2 respectively. The first-order model determination module is used to perform correlation analysis on all band ratios of each first-order spectrum and each water quality parameter, and perform a one-dimensional linear fitting to determine multiple first-order monitoring models.

[0120] Specifically, the first-order model determination module includes: an initial first-order model determination submodule, a first indicator calculation submodule and a first selection submodule.

[0121] Among them, the initial first-order model determination submodule is respectively connected to the derivation module and the water sample analysis unit. The initial first-order model determination submodule is used to calculate all band ratios of the first-order spectrum for any first-order spectrum, and perform correlation analysis on all band ratios and various water quality parameters, and perform univariate linear fitting to obtain multiple initial first-order monitoring models.

[0122] The first indicator calculation submodule is connected to the initial first-order model determination submodule, and the first indicator calculation submodule is used to calculate the R of each initial first-order monitoring model. 2 index.

[0123] The first selection submodule is connected to the initial first-order model determination submodule and the first indicator calculation submodule respectively, and the first selection submodule is used to determine the initial first-order monitoring model according to the R 2 Indicator, select R from multiple initial first-order monitoring models2 First-order monitoring model when the indicator is greater than the third set threshold.

[0124] The second-order model determination module is connected to the derivation module and the water sample analysis unit 2 respectively. The second-order model determination module is used to perform correlation analysis on all band ratios of each second-order spectrum and each water quality parameter, and perform a one-dimensional linear fitting to determine multiple second-order monitoring models.

[0125] Specifically, the second-order model determination module includes: an initial second-order model determination submodule, a second indicator calculation submodule and a second selection submodule.

[0126] Among them, the initial second-order model determination submodule is respectively connected to the derivation module and the water sample analysis unit. The initial second-order model determination submodule is used to calculate all band ratios of the second-order spectrum for any second-order spectrum, and perform correlation analysis on all band ratios and various water quality parameters, and perform univariate linear fitting to obtain multiple initial second-order monitoring models.

[0127] The second indicator calculation submodule is connected to the initial second-order model determination submodule, and the second indicator calculation submodule is used to calculate the R of each initial second-order monitoring model. 2 index.

[0128] The second selection submodule is connected to the initial second-order model determination submodule and the second indicator calculation submodule respectively, and the second selection submodule is used to determine the initial second-order model according to the R 2 Indicator, select R from multiple initial second-order monitoring models 2 Second-order monitoring model where the indicator is greater than the third set threshold.

[0129] The third-order model determination module is connected to the derivation module and the water sample analysis unit 2 respectively. The third-order model determination module is used to perform correlation analysis on all band ratios of each third-order spectrum and each water quality parameter, and perform a one-dimensional linear fitting to determine multiple third-order monitoring models.

[0130] Specifically, the third-order model determination module includes: an initial third-order model determination submodule, a third indicator calculation submodule and a third selection submodule.

[0131] Among them, the initial third-order model determination submodule is respectively connected to the derivation module and the water sample analysis unit. The initial third-order model determination submodule is used to calculate all band ratios of the third-order spectrum for any third-order spectrum, and perform correlation analysis on all band ratios and various water quality parameters, and perform univariate linear fitting to obtain multiple initial third-order monitoring models.

[0132] The third indicator calculation submodule is connected to the initial third-order model determination submodule, and the third indicator calculation submodule is used to calculate the R of each initial third-order monitoring model. 2 index.

[0133] The third selection submodule is connected to the initial third-order model determination submodule and the third indicator calculation submodule respectively, and the third selection submodule is used to select the initial third-order monitoring model according to the R 2 Indicator, select R from multiple initial third-order monitoring models 2 A third-order monitoring model where the indicator is greater than the third set threshold.

[0134] The monitoring model determination module is respectively connected to the single-band model determination module, the first-order model determination module, the second-order model determination module and the third-order model determination module. The monitoring model determination module is used to select multiple monitoring models whose root mean square error is less than a first set threshold and whose average relative deviation is less than a second set threshold from the single-band monitoring model, multiple first-order monitoring models, multiple second-order monitoring models and multiple third-order monitoring models as water quality parameter monitoring models.

[0135] Compared with the prior art, the remote sensing monitoring system for the chemical oxygen demand content of brackish water lakes of the present invention has the same beneficial effects as the above-mentioned remote sensing monitoring method for the chemical oxygen demand content of brackish water lakes, which will not be described in detail here.

[0136] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0137] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A remote sensing monitoring method for water quality parameters of brackish water lakes, characterized in that: The remote sensing monitoring method for water quality parameters of the brackish water lake comprises: Collect water spectra and corresponding water samples in multiple areas of the lake to be monitored to obtain multiple water spectra and multiple water samples; Analyze each water sample to determine the water quality parameters of the corresponding area, specifically including: analyze each water sample using a rapid digestion spectrophotometer to determine the corresponding chemical oxygen demand content; the water quality parameter is the chemical oxygen demand content; The correlation analysis between the spectra of each water body and the water quality parameters of each region was carried out, and a linear fitting was performed to determine multiple water quality parameter monitoring models, including: Derivative the spectra of each water body to determine the first-order spectrum, second-order spectrum and third-order spectrum of each water body spectrum; The single-band reflectance of each water body spectrum was analyzed for correlation with each water quality parameter, and a one-dimensional linear fitting was performed to determine the single-band monitoring model; The correlation analysis was conducted between all band ratios of each first-order spectrum and each water quality parameter, and a one-dimensional linear fitting was performed to determine multiple first-order monitoring models; The correlation analysis was conducted between all band ratios of each second-order spectrum and each water quality parameter, and a one-dimensional linear fitting was performed to determine multiple second-order monitoring models; The correlation analysis was conducted between all band ratios of each third-order spectrum and each water quality parameter, and a one-dimensional linear fitting was performed to determine multiple third-order monitoring models; Among them, the ratio of all bands is specifically (|b1 / b2|), (|(b1-b2) / (b1+b2)|); b1 and b2 are the reflectivity of the band; Select multiple monitoring models whose root mean square error is less than a first set threshold and whose average relative deviation is less than a second set threshold from the single-band monitoring model, multiple first-order monitoring models, multiple second-order monitoring models and multiple third-order monitoring models as water quality parameter monitoring models; Collect hyperspectral images of the lake to be monitored and extract the spectral information of each water pixel; For any water body pixel, multiple initial water quality parameters of the water body pixel are determined according to the spectral information of the water body pixel and based on various water quality parameter monitoring models; the water quality parameter of the water body pixel is determined according to the average value of the multiple initial water quality parameters of the water body pixel.

2. The remote sensing monitoring method for water quality parameters of brackish water lakes according to claim 1, characterized in that: A portable ground spectrometer is used to collect water spectra in multiple areas of the lake to be monitored.

3. The remote sensing monitoring method for water quality parameters of brackish water lakes according to claim 1, characterized in that: The correlation analysis of all band ratios of each first-order spectrum and each water quality parameter is performed, and a one-dimensional linear fitting is performed to determine multiple first-order monitoring models, specifically including: For any first-order spectrum, all band ratios of the first-order spectrum are calculated, and correlation analysis is performed between all band ratios and various water quality parameters, and a one-dimensional linear fitting is performed to obtain multiple initial first-order monitoring models; Calculate the R of each initial first-order monitoring model 2 index: Among them, y i is the measured value of the chemical oxygen demand content of the ith element in the modeling set; y' i is the inverted value of the i-th chemical oxygen demand content in the modeling set; is the measured average value of the chemical oxygen demand content in the modeling set; n is the number of samples in the modeling set; According to the R 2 Indicator, select R from multiple initial first-order monitoring models 2 First-order monitoring model when the indicator is greater than the third set threshold.

4. The remote sensing monitoring method for water quality parameters of brackish water lakes according to claim 1, characterized in that: The correlation analysis of all band ratios of each second-order spectrum and each water quality parameter is performed, and a one-dimensional linear fitting is performed to determine multiple second-order monitoring models, specifically including: For any second-order spectrum, all band ratios of the second-order spectrum are calculated, and correlation analysis is performed between all band ratios and various water quality parameters, and a one-dimensional linear fitting is performed to obtain multiple initial second-order monitoring models; Calculate the R of each initial second-order monitoring model 2 index: Among them, y i is the measured value of the chemical oxygen demand content of the ith element in the modeling set; y' i is the inverted value of the i-th chemical oxygen demand content in the modeling set; is the measured average value of the chemical oxygen demand content in the modeling set; n is the number of samples in the modeling set; According to the R 2 Indicator, select R from multiple initial second-order monitoring models 2 Second-order monitoring model where the indicator is greater than the third set threshold.

5. The remote sensing monitoring method for water quality parameters of brackish water lakes according to claim 1, characterized in that: The correlation analysis of all band ratios of each third-order spectrum and each water quality parameter is performed, and a one-dimensional linear fitting is performed to determine multiple third-order monitoring models, specifically including: For any third-order spectrum, all band ratios of the third-order spectrum are calculated, and correlation analysis is performed between all band ratios and various water quality parameters, and a one-dimensional linear fitting is performed to obtain multiple initial third-order monitoring models; Calculate the R of each initial third-order monitoring model 2 index: Among them, y i is the measured value of the chemical oxygen demand content of the ith element in the modeling set; y' i is the inverted value of the i-th chemical oxygen demand content in the modeling set; is the measured average value of the chemical oxygen demand content in the modeling set; n is the number of samples in the modeling set; According to the R 2 Indicator, select R from multiple initial third-order monitoring models 2 A third-order monitoring model where the indicator is greater than the third set threshold.

6. A remote sensing monitoring system for water quality parameters of brackish water lakes, characterized in that: The water quality parameter remote sensing monitoring system of the brackish water lake comprises: A data acquisition unit, used to collect water spectra and corresponding water samples in multiple areas of the lake to be monitored, to obtain multiple water spectra and multiple water samples; A water sample analysis unit is connected to the data acquisition unit and is used to analyze each water sample to determine the water quality parameters of the corresponding area, specifically including: using a rapid digestion spectrophotometry method to analyze each water sample to determine the corresponding chemical oxygen demand content; the water quality parameter is the chemical oxygen demand content; A model building unit, connected to the data acquisition unit and the water sample analysis unit, respectively, for performing correlation analysis on the spectrum of each water body and the water quality parameters of each region, and performing a one-dimensional linear fitting to determine multiple water quality parameter monitoring models; The model building unit comprises: A derivation module, connected to the data acquisition unit, for derivation of each water body spectrum to determine the first-order spectrum, the second-order spectrum and the third-order spectrum of each water body spectrum; A single-band model determination module is connected to the derivation module and the water sample analysis unit, respectively, and is used to perform correlation analysis on the reflectance of a single band of each water body spectrum and each water quality parameter, and perform a one-dimensional linear fitting to determine a single-band monitoring model; The first-order model determination module is connected to the derivation module and the water sample analysis unit respectively, and is used to perform correlation analysis on all band ratios of each first-order spectrum and each water quality parameter, and perform one-dimensional linear fitting to determine multiple first-order monitoring models; wherein all band ratios are specifically (|b1 / b2|), (|(b1-b2) / (b1+b2)|); b1 and b2 are both reflectances of the bands; A second-order model determination module, connected to the derivation module and the water sample analysis unit, respectively, for performing correlation analysis on all band ratios of each second-order spectrum and each water quality parameter, and performing a one-dimensional linear fitting to determine multiple second-order monitoring models; A third-order model determination module is connected to the derivation module and the water sample analysis unit, respectively, and is used to perform correlation analysis on all band ratios of each third-order spectrum and each water quality parameter, and perform a one-dimensional linear fitting to determine multiple third-order monitoring models; A monitoring model determination module, connected to the single-band model determination module, the first-order model determination module, the second-order model determination module and the third-order model determination module, respectively, for selecting a plurality of monitoring models whose root mean square error is less than a first set threshold and whose average relative deviation is less than a second set threshold from the single-band monitoring model, multiple first-order monitoring models, multiple second-order monitoring models and multiple third-order monitoring models as water quality parameter monitoring models; An image acquisition unit is used to acquire the hyperspectral image of the lake to be monitored and extract the spectral information of each water pixel; The content determination unit is connected to the model building unit and the image acquisition unit respectively, and is used to determine, for any water body pixel, multiple initial water quality parameters of the water body pixel according to the spectral information of the water body pixel and based on various water quality parameter monitoring models; and determine the water quality parameter of the water body pixel according to the average value of the multiple initial water quality parameters of the water body pixel.

7. The remote sensing monitoring system for water quality parameters of brackish water lakes according to claim 6, characterized in that: The first-order model determination module includes: The initial first-order model determination submodule is connected to the derivation module and the water sample analysis unit respectively, and is used to calculate all the band ratios of the first-order spectrum for any first-order spectrum, and to perform correlation analysis on all the band ratios and various water quality parameters, and to perform one-dimensional linear fitting to obtain multiple initial first-order monitoring models; The first indicator calculation submodule is connected to the initial first-order model determination submodule and is used to calculate the R of each initial first-order monitoring model. 2 index: Among them, y i is the measured value of the chemical oxygen demand content of the ith element in the modeling set; y' i is the inverted value of the i-th chemical oxygen demand content in the modeling set; is the measured average value of the chemical oxygen demand content in the modeling set; n is the number of samples in the modeling set; The first selection submodule is connected to the initial first-order model determination submodule and the first indicator calculation submodule respectively, and is used to select the first indicator according to the R of each initial first-order monitoring model. 2 Indicator, select R from multiple initial first-order monitoring models 2 First-order monitoring model when the indicator is greater than the third set threshold.

Images