Remote sensing method for detecting concentration of non-fluorescent chromophoric dissolved organic matter (cdom) by combining hyperspectral and lif techniques

Abstract

The application discloses a CDOM non-fluorescent substance concentration remote sensing detection method combining hyperspectral technology and LIF technology, and combines the hyperspectral technology to detect CDOM in seawater, utilizes measured hyperspectral data and measured sampling point CDOM concentration to establish a CDOM concentration inversion model, and obtains the CDOM concentration; the laser-induced fluorescence (LIF) remote sensing detection technology is combined to detect fluorescent dissolved organic matter (FDOM) in seawater and measure the sampling FDOM concentration, an FDOM concentration inversion model is established, and the FDOM concentration is obtained; concentration data obtained by the CDOM inversion model and concentration data obtained by the FDOM inversion model are analyzed to obtain the concentration of non-fluorescent substances in the CDOM.

Description

TECHNICAL FIELD

[0001] The present application relates to a CDOM non-fluorescent substance concentration remote sensing detection method combining hyperspectral and LIF technologies, and belongs to the fields of remote sensing and ocean science. BACKGROUND

[0002] Colored dissolved organic matter (CDOM) is also known as yellow substance, which is part of the colored group of dissolved organic matter (DOM). Some CDOM can emit fluorescence after absorbing ultraviolet light and blue light, and this kind of CDOM is called fluorescent dissolved organic matter (FDOM). FDOM is divided into two categories: protein-like and humic-like fluorescent substances. Current researches on the composition, structure and source of CDOM are relatively simple and systematic. In addition, researches on the fluorescence properties of FDOM are also becoming increasingly active.

[0003] However, in addition to FDOM, there is also a part of non-fluorescent substances in CDOM. The relationship between CDOM and non-fluorescent substances in CDOM is jointly controlled by environmental factors such as diagenesis (photodegradation and biodegradation), biological productivity (phytoplankton primary productivity) and hydrology (endmember mixing). At present, the research on non-fluorescent substances in CDOM is relatively less because the research in this aspect is relatively complex and is easily affected by many factors. However, the research on non-fluorescent substances in CDOM not only helps water quality managers to manage water quality, but also provides convenience for scholars to study marine ecosystems, and has very important research value.

[0004] Therefore, it is necessary to provide a method for detecting the concentration of non-fluorescent substances in CDOM in seawater. SUMMARY

[0005] To solve the above technical problems, the present application provides a CDOM non-fluorescent substance concentration remote sensing detection method combining hyperspectral and LIF technologies, which can detect the concentration of non-fluorescent substances in CDOM in seawater, help to determine the best position for monitoring the water quality of the target water area by three-dimensional fluorescence spectroscopy technology, and also help to monitor and evaluate the water quality and ecological environment of the water body.

[0006] To solve the above technical problems, the technical solution adopted by the present application is as follows:

[0007] The CDOM non-fluorescent substance concentration remote sensing detection method combining hyperspectral and LIF technologies detects the concentration of non-fluorescent substances in colored dissolved organic matter (CDOM) in seawater based on hyperspectral remote sensing detection technology and laser-induced fluorescence (LIF) remote sensing detection technology, and specifically includes the following steps:

[0008] Step one: control unmanned aerial vehicle hyperspectral detector to detect CDOM and fluorescent laser radar to detect FDOM, match the data obtained by hyperspectral detection of CDOM with the data obtained by fluorescent laser radar detection of FDOM; and simultaneously carry out water quality monitoring work at the sampling point, measure the CDOM and FDOM concentrations at the sampling point;

[0009] Step two: using the measured hyperspectral data and the measured CDOM concentration at the sampling point, a CDOM concentration inversion model is established to obtain the CDOM concentration;

[0010] Step three: using the fluorescent data detected by the fluorescent laser radar and the measured FDOM concentration at the sampling point, an FDOM concentration inversion model is established to obtain the FDOM concentration;

[0011] Step four: the data obtained by the CDOM inversion model and the data obtained by the FDOM inversion model are analyzed to obtain the concentration of non-fluorescent substances in CDOM.

[0012] The further improvement of the technical scheme of the application is that in step one, the microcomputer on the unmanned aerial vehicle is connected to a wireless router through WiFi, the signal of the router is amplified through a signal amplifier, and a local area network is established through an AP receiver;

[0013] The ground station computer is connected to the AP receiver through a network cable, and the network of the ground station computer and the airborne computer is set in the same local area network, and the remote desktop is used to realize the control of the airborne microcomputer on the ground;

[0014] Through the ground station computer, the unmanned aerial vehicle is controlled to fly near the detection target, the ground station controls the microcomputer on the unmanned aerial vehicle through the local area network remote desktop to synchronously transmit the POS information into the hyperspectral detector and the fluorescent laser radar system, so that the hyperspectral detector and the fluorescent laser radar system have the same geographical position; the hyperspectral detector is controlled to collect hyperspectral images, and the fluorescent laser radar system laser is controlled to emit laser at the same time, so that the spectrum collection is realized; the collected data is stored in the microcomputer and is synchronously transmitted back to the ground station computer.

[0015] The further improvement of the technical scheme of the application is that in step one, when the hyperspectral detector detects CDOM, the CDOM hyperspectral data directly obtained by the unmanned aerial vehicle hyperspectral remote sensing platform is DN value data of remote sensing image pixels, and pretreatment operation is required, including radiation calibration and geometric correction steps;

[0016] Radiometric calibration: place a calibration plate under the lens of the hyperspectral detector, obtain the remote sensing image of the calibration plate, and the calibration plate is used for reflectivity calibration; the DN value of the water body is converted into hyperspectral reflectivity information of the water body surface by using the radiation calibration method;

[0017] The hyperspectral reflectance information is obtained by the reflected light intensity and the incident light intensity, the reflected light intensity is the DN value of the current environment, and the incident light intensity is the incident light intensity of the calibration board, because the reflectivity and the reflected light intensity of the calibration board are known, so the incident light intensity of the current environment is obtained; the hyperspectral reflectance calculation formula is as follows:

[0018]

[0019] In the formula, ρ t , DN t are the reflectivity and the DN value to be converted respectively, the calibration board reflectivity is ρ1, and DN1 is the DN value of the calibration board;

[0020] After the reflectivity is obtained, normalization processing is performed on the reflectivity, so as to reduce the influence of weather and measurement angle on the reflectivity, and the normalization formula is as follows:

[0021]

[0022] In the formula, L N (λ i ) represents the reflectivity after the band normalization; λ i represents the wavelength of the i band; L(λ i ) is the initial value of the i band reflectivity; and n is the number of contained bands;

[0023] In order to be able to show the spectral characteristics of the water quality parameter, effectively eliminate some interference, such as the sea surface roughness and environmental noise, the band combination factor in the water quality parameter is analyzed;

[0024] Using the iterative algorithm, the normalized reflectivity is compared one by one, Pearson correlation analysis is performed on the CDOM concentration, the correlation coefficient is obtained, and the band ratio with high correlation is obtained from the correlation coefficient; the expression of the correlation coefficient r is as follows:

[0025]

[0026] In the formula, y i is the concentration of the water quality parameter of the i sampling point; and x i is the reflectivity of the i sampling point;

[0027] Geometric correction: the differential GPS technology and the inertial measurement unit IMU are integrated into one, combined with the sensor, the position and attitude parameters of the sensor can be provided, the geographical positioning of the image can be quickly and accurately performed, and the image is corrected to a proper geographical position;

[0028] In the process of detecting FDOM by using the fluorescence laser radar system, the fluorescence laser radar system uses a 355nm laser as an excitation light source;

[0029] The laser of the fluorescence lidar system emits a laser beam, the laser beam is reflected by the emission mirror of the system to the sea surface at a certain angle, and the FDOM in the seawater emits fluorescence after being irradiated by the laser beam; the fluorescence signal of the FDOM is received by the telescope of the fluorescence lidar system, the parallel light after the collimating lens passes through the optical filter to reach the photosensitive sensor, and finally is processed by the system.

[0030] Further improvement of the technical scheme of the present application is that the matching of the CDOM data and the FDOM data in step one is to match the image resolution of the spectral image data and the spatial resolution of the point cloud data by using the resampling method.

[0031] Further improvement of the technical scheme of the present application is that in step one, the water quality monitoring work of the sampling points is that the CDOM concentration value of each sampling point is obtained by using the measurement result of the spectrophotometer; the FDOM concentration value of the corresponding point is obtained by measuring 25 times and taking the average value by using the EXO multi-parameter water quality analyzer.

[0032] Further improvement of the technical scheme of the present application is that in step two, the water quality remote sensing inversion model is constructed according to the relationship between the CDOM concentration and the wave band ratio, the wave band ratio is taken as the independent variable, the CDOM concentration is taken as the dependent variable, and the CDOM concentration inversion model is constructed.

[0033] In the research area, according to the distribution and number of sampling points, part of the sampling points are selected for training the model, and the remaining sampling points are used for verifying the model.

[0034] Further improvement of the technical scheme of the present application is that in step three, the obtained fluorescence spectrum information needs to be denoised; mainly adopting the S-G filtering algorithm to pretreat the fluorescence spectrum, while filtering the noise, the signal width and shape are ensured to be unchanged.

[0035] Further improvement of the technical scheme of the present application is that in step three, the inversion concentration model is constructed by the FDOM fluorescence intensity, and the relationship between the fluorescence intensity of the FDOM and the concentration of the FDOM can be regarded as a linear relationship, that is:

[0036] c FDOM = KF,

[0037] In the formula, K is a constant related to the structure of the FDOM itself and environmental factors, F is the fluorescence intensity value, and c FDOM is the concentration of the FDOM in seawater, if the value of the constant K is measured in advance, the concentration of the FDOM can be inverted by the formula;

[0038] In the research area, according to the distribution and number of sampling points, part of the sampling points are selected for training the model, and the remaining sampling points are used for verifying the model.

[0039] Further improvement of the technical scheme of the present application is that the concentration of non-fluorescent substances in CDOM can be obtained by analyzing the concentration obtained by the CDOM inversion model and the concentration obtained by the FDOM inversion model in step four.

[0040] c NFDOM = c CDOM -c FDOM ,

[0041] In the above formula, c NFDOM is the concentration of non-fluorescent substances in CDOM, c CDOM is the concentration of CDOM, and c FDOM is the concentration of FDOM.

[0042] Thanks to the above technical scheme, the present application has the following technical progress:

[0043] The present application can detect the concentration of non-fluorescent substances in CDOM in seawater, which helps to determine the best position for monitoring the water quality of the target water area by three-dimensional fluorescence spectroscopy, and also helps to monitor and evaluate the water quality and ecological environment of the water body. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 is the flow chart of the CDOM remote sensing detection method combining hyperspectral and LIF technologies according to the present application;

[0045] Figure 2 is the overall design diagram of the unmanned aerial vehicle airborne CDOM detection system;

[0046] Figure 3 is the matching process of hyperspectral data and fluorescence laser radar point cloud data;

[0047] Figure 4 is the step of unmanned aerial vehicle hyperspectral data acquisition and data preprocessing;

[0048] Figure 5 is the technical route diagram of the hyperspectral remote sensing detection CDOM technology according to the present application;

[0049] Figure 6 is the technical route diagram of the LIF remote sensing detection FDOM technology. DETAILED DESCRIPTION

[0050] In order to make the technical means, creative features, purposes and effects of the present application easy to understand, the following will describe the embodiments of the present application in detail with reference to the drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments.

[0051] The present application will be further described below with reference to the drawings and specific embodiments, but not as a limitation of the present application.

[0052] Referring to Figures 1-6 , the specific embodiment adopts the following technical solution: a CDOM non-fluorescent substance concentration remote sensing detection method combining hyperspectral and LIF technologies, comprising the following steps:

[0053] In combination Figure 1 As shown in the drawings, the present application designs a CDOM non-fluorescent substance concentration remote sensing detection method combining hyperspectral and LIF technologies, which is based on hyperspectral remote sensing detection technology and LIF remote sensing detection technology to detect and analyze the concentration of CDOM non-fluorescent substances in seawater.

[0054] Step one: simultaneously control the unmanned aerial vehicle hyperspectral detector to detect CDOM and the fluorescence laser radar to detect FDOM, and simultaneously carry out water quality monitoring work at the sampling point to measure the CDOM and FDOM concentrations at the sampling point;

[0055] Step two: use the measured hyperspectral data and the measured CDOM concentration at the sampling point to establish a CDOM concentration inversion model to obtain the CDOM concentration;

[0056] Step three: use the fluorescence data detected by the fluorescence laser radar and the measured FDOM concentration at the sampling point to establish an FDOM concentration inversion model to obtain the FDOM concentration;

[0057] Step four: analyze the concentration data obtained by the CDOM inversion model and the concentration data obtained by the FDOM inversion model to obtain the concentration of CDOM non-fluorescent substances.

[0058] In the embodiment, the hyperspectral detector and the fluorescence laser radar system are fixedly connected as an integral frame to be fixed on the unmanned aerial vehicle as a detection system, as shown in Figure 2 It is an overall design drawing of the unmanned aerial vehicle on-board CDOM detection system, and the on-board part mainly comprises a detection system, a microcomputer, a differential GPS mobile station and a radio thereof, a router, a signal amplifier, and batteries required for power supply of various components. The ground station part mainly comprises a differential GPS reference station and a radio thereof, a flight control radio, an AP receiver, and the like.

[0059] Further, in the step one, the microcomputer on the unmanned aerial vehicle is connected to the wireless router through WiFi, the signal of the router is amplified through the signal amplifier, and a local area network is established through the AP receiver;

[0060] The ground station computer is connected to the AP receiver through a network cable, the network settings of the ground station computer and the on-board computer can be set in the same local area network, and the remote desktop is used to realize the control of the on-board microcomputer on the ground;

[0061] Through the ground station computer, the unmanned aerial vehicle is controlled to fly near the detection target, the ground station controls the microcomputer on the unmanned aerial vehicle through a local area network remote desktop to synchronously transmit the POS information into the hyperspectral detector and the fluorescent laser radar system, so that the hyperspectral detector and the fluorescent laser radar system have the same geographical position; the hyperspectral detector is controlled to collect hyperspectral images and the LIF radar system laser is controlled to emit laser, so that the spectrum is collected; the collected data is stored in the microcomputer and can be synchronously transmitted back to the ground station computer.

[0062] In further, in the step one, when the hyperspectral detector detects CDOM, the CDOM hyperspectral data directly acquired by the unmanned aerial vehicle hyperspectral remote sensing platform is DN value data of a remote sensing image pixel brightness, that is, the pixel brightness value of the hyperspectral remote sensing image, the unmanned aerial vehicle hyperspectral data needs to be preprocessed, including radiation calibration of the DN value data, geometric correction, etc.

[0063] Radiation calibration: before the unmanned aerial vehicle takes off, a calibration board (known reflectivity) is placed under the lens of the unmanned aerial vehicle hyperspectral detector, and a calibration board remote sensing image is acquired, the calibration board can be used for reflectivity calibration of the unmanned aerial vehicle hyperspectral;

[0064] Geometric correction: the positioning and orientation system (POS) of the unmanned aerial vehicle integrates differential GPS technology and inertial measurement unit (IMU) technology together, the sensor is integrated together, the position and attitude parameters of the sensor can be provided, the geographical positioning of the image can be directly and quickly performed, and the image can be corrected to the correct geographical position;

[0065] In the process of using the LIF to detect FDOM, the fluorescent laser radar system is adopted, the fluorescent laser radar system adopts a 355nm laser as an excitation light source;

[0066] The laser of the fluorescent laser radar system emits a laser, the laser is reflected by a system emission mirror and hits the sea surface at a certain angle, the FDOM in the seawater emits fluorescence after being irradiated by the laser; the fluorescence signal of the FDOM is received by the telescope of the fluorescent laser radar system, the parallel light after the collimating lens passes through the optical filter to the photosensitive sensor, and finally is processed by the system.

[0067] Further, in the step one, the hyperspectral data image data and the fluorescent laser radar point cloud data need to be matched, the principle of data matching is to make the hyperspectral image data and the fluorescent laser radar point cloud data spatial consistency, that is, the hyperspectral image data and the fluorescent laser radar point cloud data spatially correspond; the method of resampling is used to match the image resolution of the spectral image data and the spatial resolution of the point cloud data;

[0068] If the image resolution of the spectral image is lower than the spatial resolution of the fluorescence lidar point cloud data, the spectral image data is resampled by interpolation method, and if the image resolution of the spectral image is higher than the spatial resolution of the fluorescence lidar point cloud data, the fluorescence lidar point cloud data is resampled by density enhancement of up-sampling;

[0069] After resampling, the fluorescence lidar point cloud image is converted into a two-dimensional image to obtain a point cloud grid; the intensity value of each pixel in the point cloud grid corresponds to the average intensity of each foot point in the point cloud data, and the point cloud grid serves as a reference image for spectral image data registration; the spectral image data matching the waveband of the fluorescence lidar detector in the spectral image serves as a template image; image data matching is performed, so that the spectral image data and the fluorescence lidar point cloud data are spatially corresponding, and the data matching process is as shown in Figure 3 .

[0070] Further, the resampling principle of the spectral image data by interpolation method is as follows:

[0071] Let (x, y) be the pixel coordinates of the interpolation point in the resampled spectral image, and the neighborhood point coordinates be (i, j); let the gray value of (i, j) be f(x, y) on the (i, j) coordinate plane;

[0072] The cubic convolution interpolation function is defined as:

[0073]

[0074] In the formula, s is the distance between the interpolation point and the neighborhood point;

[0075] f(x, y) = f(i+x d ,j+y d ),

[0076] In the formula, x d , y d is the difference between the horizontal and vertical coordinates of the interpolation point and the neighborhood point;

[0077] Further, the resampling principle of the fluorescence lidar point cloud data by density enhancement is as follows:

[0078] For the foot point a of the fluorescence lidar point cloud data, the neighborhood search sphere radius R is set to have N neighborhood points; a neighborhood fitting plane is established by using the moving least square method;

[0079]

[0080] ρ a is the neighborhood point density of the foot point a;

[0081] The point cloud density threshold is set to ρ0, and when ρa <ρ0, randomly and uniformly project the foot point a of the point cloud data onto the fitting plane of a, and determine the intensity value of the projected point as the intensity value of its neighboring points; until the density of the neighboring points of foot point a satisfies ρ a ≥ρ0; Obtain the resampled point cloud image; Repeat the search for point cloud foot points and perform the above operations to complete the resampling of the fluorescent lidar point cloud;

[0082] Furthermore, the process of image data registration includes:

[0083] The SIFT algorithm is used to find the spectral image T that is closest to the wavelength of the fluorescent lidar detector, which is then used as the template image; R is used as the reference image, and an objective function for affine transformation spatial registration is established.

[0084]

[0085] In the above equation, b represents the corresponding matching feature point pair in T and R, T(b) is the pixel corresponding to b in T, and R(b) is the pixel corresponding to b in R. For an affine transformation acting on T, Similarity measure between R and R;

[0086] Affine transformation:

[0087]

[0088] The above formula, For rotation matrix, It is a translation matrix;

[0089] Use optimization algorithms to optimize the objective function. Optimization is performed to obtain the optimal rotation matrix and optimal translation matrix, which are then applied to all acquired spectral image data to complete image data registration.

[0090] Furthermore, in step one, water quality monitoring was also carried out simultaneously at the sampling points. The CDOM concentration value of each sampling point was obtained by measuring the results with a spectrophotometer; the FDOM concentration value of the corresponding point was obtained by measuring 25 times with an EXO multi-parameter water quality analyzer and taking the average value.

[0091] Furthermore, in step two, the CDOM hyperspectral data directly acquired by the UAV hyperspectral remote sensing platform is the pixel brightness (DN) value data of the remote sensing image, that is, the pixel brightness value of the hyperspectral remote sensing image. After preprocessing the UAV hyperspectral data (spectral calibration and geometric correction of the DN value data), reflectance calculation is required to finally generate reflectance data; the data acquisition and data preprocessing steps are as follows: Figure 4 As shown;

[0092] Since the water body detection is on-site detection, it is necessary to convert the DN value of the water body into the hyperspectral reflectance information of the water body surface by using radiation calibration;

[0093] The reflectance of the hyperspectral image of the unmanned aerial vehicle is the ratio of the reflected light intensity to the incident light intensity, the reflected light intensity can be considered as the DN value data, and the incident light intensity needs to be calibrated by the calibration plate to obtain, because the reflectivity of the calibration plate and the reflected light intensity are known, so the incident light intensity of the current environment can be obtained, then the reflectivity of the hyperspectral image of the unmanned aerial vehicle can be obtained through the formula:

[0094]

[0095] In the formula, ρ t , DN t are the reflectivity and DN value to be converted respectively, the reflectivity of the calibration plate is ρ1, and DN1 is the DN value of the calibration plate;

[0096] After the reflectivity is obtained, the normalization processing is carried out, which can reduce the influence of weather and measurement angle and the like on the reflectivity, the spectral range of the reflectivity curve on the hyperspectral image of the unmanned aerial vehicle is extracted and normalized (the wavelength range of the hyperspectral image shot by the unmanned aerial vehicle is located in the visible light and near-infrared region (400-900nm) of the electromagnetic spectrum), and the normalization formula is as follows:

[0097]

[0098] In the formula, L N (λ i ) represents the reflectivity after the wavelength normalization; λ i represents the wavelength of the i wavelength; L(λ i ) is the initial value of the i wavelength reflectivity; and n is the number of contained wavelengths.

[0099] In order to show the spectral characteristics of the water quality parameters and effectively eliminate some interference such as the roughness of the seawater surface and environmental noise and the like, the wavelength combination factor in the water quality parameters needs to be analyzed; the wavelength combination factor can average and randomize the error caused by the cross influence of the non-characteristic wavelength and the characteristic wavelength and other water quality parameters;

[0100] The present application uses an iterative algorithm to perform the ratio of the normalized reflectivity one by one, and performs Pearson correlation analysis on the CDOM concentration, to obtain a correlation coefficient, from which the wavelength with high correlation can be obtained to participate in modeling; the expression of the correlation coefficient r is as follows:

[0101]

[0102] In the formula, y iis the concentration of water quality parameter of the i th sampling point; x i is the reflectivity of the i th sampling point;

[0103] According to the relationship between the CDOM concentration and the wave band ratio, a water quality remote sensing inversion model is constructed, the wave band ratio is used as the independent variable, and the CDOM concentration is used as the dependent variable, and the model is used for water quality remote sensing inversion.

[0104] Commonly used models are established based on statistical regression analysis methods, including polynomial, linear, exponential, logarithmic and multiple linear regression, etc.

[0105] In the research area, according to the distribution and number of sampling points and the like, a part of the measured points is selected for model training, and the remaining measured points are used for verifying the model; the detection CDOM concentration technical route is as shown in Figure 5 .

[0106] Further, in order to obtain the required detection target information more accurately, it is very important to effectively extract the fluorescence signal, therefore, the obtained fluorescence spectrum information needs to be denoised;

[0107] The present application mainly adopts the S-G filtering algorithm to pretreat the fluorescence spectrum, and this denoising method is based on local polynomial least square fitting in time domain, and the maximum feature is that the signal width and shape can be ensured unchanged while filtering noise.

[0108] Further, the S-G filtering algorithm includes:

[0109] Suppose that a column of data x[n] is a circle with n=0 as the center, contains 2M+1 data, and if it can be fitted by an N-order polynomial:

[0110]

[0111] The residual error of least square fitting is:

[0112]

[0113] The constant term of the fitting polynomial is obtained by using convolution operation:

[0114]

[0115] According to calculus, if ε is minimum, the partial derivative of ε to each parameter is 0, that is:

[0116]

[0117] In order to facilitate calculation, an auxiliary matrix A of (2M+1) rows and (N+1) columns is introduced:

[0118] A={an,i},

[0119] a n,i =n i ,-M≤n≤M,0≤i≤N,

[0120] In order to set an auxiliary matrix B, make:

[0121] B=A T A,

[0122] Definition

[0123] Then we can get:

[0124] Ba=A T Aa=A T x,

[0125] a=(A T A) -1 A T x=Hx,

[0126] The first row of H is the required convolution coefficient.

[0127] Further, by constructing the inversion concentration model of FDOM fluorescence intensity, the relationship between fluorescence intensity and fluorescence substance concentration can be regarded as the linear relationship between the fluorescence intensity of FDOM and the concentration of FDOM, that is:

[0128] c FDOM =KF,

[0129] In the formula, K is a constant related to the structure of FDOM itself and environmental factors, F is the fluorescence intensity value, and c FDOM is the concentration of FDOM in seawater, if the value of constant K is measured in advance, the concentration of FDOM can be inverted by the formula;

[0130] With the increase of FDOM concentration, the absorption capacity of FDOM to laser becomes stronger, which leads to the increase of the probability of FDOM fluorophore being excited, so that the fluorescence intensity of FDOM becomes stronger. The value of K can be obtained by fitting a part of measured sampling point FDOM concentration and fluorescence intensity;

[0131] In the research area, according to the distribution and number of sampling points and other conditions, a part of the measured points are selected for model training, and the remaining measured points are used for model verification. The detection FDOM concentration technical route is as Figure 6 shown.

[0132] Further, the concentration of non-fluorescent substances in CDOM can be obtained by analyzing the concentration obtained by the CDOM inversion model and the concentration obtained by the FDOM inversion model in step four.

[0133] c NFDOM = c CDOM - c FDOM ,

[0134] where c NFDOM is the concentration of non-fluorescing material in the CDOM, c CDOM is the concentration of CDOM, and c FDOM is the concentration of FDOM.

[0135] While the present application has been described with reference to specific implementations thereof, it should be understood that these implementations are illustrative only and in no way limit the scope of the present application as defined in the appended claims. It should also be understood that the various claims hereinafter presented are not mutually exclusive and can be combined in other ways.

Claims

1. A method for detecting concentration of non-fluorescent substances of CDOM by remote sensing combining hyperspectral and LIF technologies, characterized in that: Based on hyperspectral remote sensing detection technology and laser-induced fluorescence LIF remote sensing detection technology, the concentration of non-fluorescent substances in colored dissolved organic matter CDOM in seawater is detected, which comprises the following steps: Step one: control the unmanned aerial vehicle hyperspectral detector to detect CDOM and the fluorescence laser radar to detect FDOM, match the data obtained by the hyperspectral detection of CDOM with the data obtained by the fluorescence laser radar detection of FDOM, and simultaneously carry out water quality monitoring work at the sampling point to measure the CDOM and FDOM concentrations at the sampling point; in the step one, when the hyperspectral detector detects CDOM, the CDOM hyperspectral data directly obtained by the unmanned aerial vehicle hyperspectral remote sensing platform is DN value data of remote sensing image pixel brightness, which needs to be preprocessed, including radiation calibration and geometric correction steps; Radiation calibration: place the calibration plate under the hyperspectral detector lens to obtain the remote sensing image of the calibration plate, and the calibration plate is used for reflectivity calibration; the DN value of the water body is converted into hyperspectral reflectivity information of the water surface by using the radiation calibration method; The hyperspectral reflectivity information is obtained through the reflected light intensity and the incident light intensity, the reflected light intensity is the current environmental DN value, and the incident light intensity is the incident light intensity of the calibration plate, because the reflectivity and the reflected light intensity of the calibration plate are known, so the incident light intensity of the current environment is obtained; the hyperspectral reflectivity calculation formula is as follows: where ρ t , DN t are the reflectance and DN values, respectively, that need to be converted, and ρ1, DN1are the reflectance and DN values, respectively, of the calibration panel. After obtaining the reflectivity, it is normalized to reduce the influence of weather and measurement angle on the reflectivity, and the normalization formula is as follows: In the formula, L N (λ i ) represents the reflectivity after the wavelength band is normalized; λ i represents the wavelength of the i wavelength band; L(λ i ) is the initial value of the reflectivity of the i wavelength band; n is the number of contained wavelength bands; In order to show the spectral characteristics of the water quality parameters and effectively eliminate the interference of seawater surface roughness and environment noise, the band combination factor in the water quality parameters is analyzed; Using the iterative algorithm, the normalized reflectivity is compared one by one, the Pearson correlation analysis is carried out between the CDOM concentration and the normalized reflectivity, the correlation coefficient is obtained, and the band ratio with high correlation is obtained from the correlation coefficient; the expression of the correlation coefficient r is as follows: In the formula, y i is the concentration of the water quality parameter of the ith sampling point; x i is the reflectivity of the ith sampling point; Geometric correction: the differential GPS technology and the inertial measurement unit IMU are integrated into one and combined with the sensor to provide the position and attitude parameters of the sensor, so that the geographical positioning of the image can be quickly and accurately carried out; the image is corrected to the appropriate geographical position; In the process of detecting FDOM by using the fluorescence laser radar system, the fluorescence laser radar system uses a 355nm laser as an excitation light source; The laser of the fluorescence laser radar system emits a laser beam, the laser is reflected by the emission mirror of the system to hit the sea surface at a certain angle, and the FDOM in the seawater emits fluorescence after being irradiated by the laser; The fluorescence signal of the FDOM is received by the telescope of the fluorescence laser radar system, the parallel light after the collimating lens passes through the optical filter to the photosensitive sensor, and finally is processed by the system; Step two: using the measured hyperspectral data and the measured CDOM concentration at the sampling point, a CDOM concentration inversion model is established to obtain the CDOM concentration; Step three: using the fluorescence data detected by the fluorescence laser radar and the measured FDOM concentration at the sampling point, an FDOM concentration inversion model is established to obtain the FDOM concentration; Step four: analyzing the data obtained by the CDOM inversion model and the data obtained by the FDOM inversion model to obtain the concentration of non-fluorescent substances in CDOM.

2. The method for detecting concentration of non-fluorescent CDOM by combining hyperspectral and LIF techniques according to claim 1, wherein: In step one, the microcomputer on the unmanned aerial vehicle connects the wireless router through WiFi, the signal of the router is amplified through the signal amplifier, and the local area network is established through the AP receiver; The ground station computer connects the AP receiver through the network cable, sets the network of the ground station computer and the airborne computer in the same local area network, and realizes the control of the airborne microcomputer on the ground using remote desktop; Through the ground station computer, the unmanned aerial vehicle is controlled to fly near the detection target, the ground station controls the microcomputer on the unmanned aerial vehicle to synchronize the POS information into the hyperspectral detector and the fluorescent laser radar system through the local area network remote desktop, so that the hyperspectral detector and the fluorescent laser radar system have the same geographical position; the hyperspectral detector is controlled to collect hyperspectral images, and the fluorescent laser radar system laser is controlled to emit laser to realize the collection of spectrum; the collected data is stored in the microcomputer and transmitted back to the ground station computer.

3. The method for detecting concentration of non-fluorescent CDOM by combining hyperspectral and LIF techniques according to claim 1, characterized in that: In step one, the matching of CDOM data and FDOM data is to match the image resolution of spectral image data and the spatial resolution of point cloud data using the resampling method.

4. The method for detecting concentration of non-fluorescent CDOM by combining hyperspectral and LIF techniques according to claim 1, wherein: In step one, the water quality monitoring work of the sampling point is: the CDOM concentration value of the corresponding point is obtained by measuring the results of each sampling point using a spectrophotometer; the FDOM concentration value of the corresponding point is obtained by measuring 25 times and taking the average value using an EXO multi-parameter water quality analyzer.

5. The method for detecting concentration of non-fluorescent CDOM by combining hyperspectral and LIF techniques according to claim 1, wherein: In step two, a water quality remote sensing inversion model is constructed according to the relationship between CDOM concentration and wave band ratio, the wave band ratio is taken as the independent variable, the CDOM concentration is taken as the dependent variable, and the CDOM concentration inversion model is constructed; In the study area, according to the distribution and number of sampling points, part of the sampling points are selected for training the model, and the remaining sampling points are used for verifying the model.

6. The method for detecting concentration of non-fluorescent CDOM by combining hyperspectral and LIF techniques according to claim 5, characterized in that: In step three, the obtained fluorescence spectrum information needs to be denoised; mainly using the S-G filtering algorithm to preprocess the fluorescence spectrum, while filtering out the noise, the signal width and shape are ensured to be unchanged.

7. The method for detecting concentration of non-fluorescent CDOM by combining hyperspectral and LIF techniques according to claim 6, characterized in that: In step three, the concentration model is constructed by the FDOM fluorescence intensity, and the relationship between the fluorescence intensity of FDOM and the concentration of FDOM is considered as a linear relationship, that is: c FDOM = KF, In the above equation, K is a constant related to the structure of FDOM itself and environmental factors, F is the fluorescence intensity value, c FDOM is the concentration of FDOM in seawater. If the value of constant K is measured in advance, the concentration of FDOM can be inverted by the formula; In the study area, according to the distribution and number of sampling points, part of the sampling points are selected for training the model, and the remaining sampling points are used for verifying the model.

8. The method for detecting concentration of non-fluorescent CDOM by combining hyperspectral and LIF techniques according to claim 7, characterized in that: In step four, the concentration obtained by the CDOM inversion model and the concentration obtained by the FDOM inversion model are analyzed, and the concentration of the non-fluorescent substance in CDOM can be obtained: c NFDOM = c CDOM - c FDOM , c is the concentration of the non-fluorescent fraction of CDOM in the above formula NFDOM c is the concentration of the non-fluorescent fraction of CDOM in the above formula CDOM c is the concentration of the non-fluorescent fraction of CDOM in the above formula FDOM c is the concentration of the non-fluorescent fraction of CDOM in the above formula

Images