A chlorophyll concentration inversion method based on relative reflection depth

Through the inversion method based on relative reflection depth, the problems of low inversion accuracy and poor robustness of chlorophyll concentration in nearshore waters are solved. The model is constructed through the characteristic band and relative reflection depth, which significantly improves the inversion accuracy and robustness and reduces the error .

CN114088640BInactive Publication Date: 2024-11-26HEBEI NORMAL UNIVERSITY OF SCIENCE & TECHNOLOGY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111259231.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-10-27
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing chlorophyll concentration inversion model has low accuracy and poor robustness in nearshore waters, especially due to the large error caused by the difference in the remote sensing reflection spectrum and the characteristics of water bodies, making it difficult to effectively invert chlorophyll concentration.

Method used

Using the inversion method based on relative reflection depth, the hyperspectral data is standardized, the characteristic bands are selected and the inversion model of relative reflection depth and its ratio is constructed. The specific steps include laboratory measurement of the surface chlorophyll concentration and measuring the reflectance of seawater irradiance , determine the characteristic band and relative reflection depth, and build an inversion model.

Benefits of technology

The inversion accuracy and model robustness of chlorophyll concentration in nearshore waters have been significantly improved, the relative error is reduced by 27%-50%, and the correlation coefficient is improved, which meets higher accuracy and stability requirements.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114088640B_ABST
    Figure CN114088640B_ABST
Patent Text Reader

Abstract

The present invention discloses a chlorophyll concentration inversion method based on relative reflection depth, characterized in that the inversion method comprises the following steps: (1) measuring the chlorophyll concentration of the seawater surface; (2) using a ground object hyperspectral measurement instrument to measure the irradiance reflectance of the seawater surface and sky light; (3) obtaining the water-leaving irradiance reflectance; (4) normalizing the water-leaving irradiance reflectance; (5) determining four chlorophyll characteristic bands of the normalized water-leaving irradiance reflectance; (6) establishing the relative reflection depth of the characteristic band; (7) constructing an inversion model based on the relative reflection depth and its ratio. The method can make full use of the chlorophyll characteristic band information, reduce data sensitivity, improve the signal-to-noise ratio of water color elements in nearshore waters, and improve the inversion accuracy and robustness of the model.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of chlorophyll concentration inversion in nearshore waters, and in particular relates to a chlorophyll concentration inversion method based on relative reflection depth. Background Art

[0002] The empirical inversion model of chlorophyll concentration in water bodies is usually constructed using the intensity ratio or difference of remote sensing reflectance spectra. At present, the general empirical inversion model for generating chlorophyll products in satellite remote sensing at home and abroad is the OCx (x=2~6) algorithm. The OCx algorithm is constructed based on the intensity ratio of remote sensing reflectance spectra in different bands. Among them, the OC2 algorithm is a band ratio method, which directly uses the ratio of remote sensing reflectance between the blue band and the green band to invert chlorophyll concentration; the OC3-6 algorithm is a maximum band ratio method, that is, the inversion model is constructed by selecting the maximum value of the remote sensing reflectance ratio between the purple, blue, blue-green, green bands and the green and red bands. When applied to Class I water bodies, the average relative error of the OCx algorithm on a global scale is about 35%. However, in Class II nearshore water bodies, the content of colored dissolved organic matter and non-algae particles is high and no longer has covariation characteristics with chlorophyll concentration, and the remote sensing reflectance spectrum is quite different from that of Class I water bodies. Therefore, the OCx algorithm based on spectral intensity ratio produces very large estimation errors and even fails. To date, improving the inversion accuracy of chlorophyll concentration in nearshore waters remains an international challenge.

[0003] The accuracy of chlorophyll concentration inversion is mainly restricted by satellite sensor radiation calibration, atmospheric correction, white hat and flare removal, digital rounding error and inversion model construction method. -3 When the water color index method is used, it has high accuracy and strong algorithm robustness, and has become an alternative algorithm to OCx. Unlike the OCx algorithm, the CI method is constructed using the relative height of the reflectance spectrum, which can effectively filter out various DC and linear noises generated by instruments, atmospheric corrections, and remote sensing reflectance calculations. In addition, compared with the absolute height of the spectrum, the relative height is conducive to information extraction and is less sensitive to random noise. This has important guiding significance for improving the inversion accuracy of chlorophyll concentration in nearshore waters with complex optical properties and strong regionality. Summary of the invention

[0004] In order to solve the technical problems of low chlorophyll concentration inversion accuracy and poor model robustness in my country's nearshore waters, the present invention provides a high-precision and robust inversion method. The method normalizes the hyperspectral data, selects the characteristic bands, and constructs a chlorophyll concentration inversion model using the relative reflection depth and ratio of the characteristic bands.

[0005] In order to achieve the above-mentioned invention object, the present invention adopts the following technical solutions:

[0006] A chlorophyll concentration inversion method based on relative reflection depth, the inversion method adopts the following steps:

[0007] (1) Measure surface chlorophyll concentration in the laboratory according to NASA ocean survey specifications;

[0008] (2) Measure seawater irradiance reflectivity in accordance with GB12763.1~7-1991 "Ocean Survey Specifications" and "Ocean Optical Survey Technical Regulations" and the irradiance reflectance of sky light ;

[0009] (3) Obtain the reflectivity of the irradiance off water ;

[0010] (4) Normalize the reflectance of water-leaving irradiance;

[0011] (5) Determine four chlorophyll characteristic bands of normalized off-water irradiance reflectance; the four chlorophyll characteristic reflectance bands are characteristic reflectance valley band B 435 , characteristic reflection valley band B 664 , characteristic reflection peak band B 573 and characteristic reflection peak band B 706 ;

[0012] (6) Establishing the relative reflection depth of the characteristic band; the relative reflection depth is the relative reflection valley depth RRD 435 , relative reflection valley depth RRD 664 , relative reflection peak depth RRD 573 and relative reflection peak depth RRD 706 ;

[0013] (7) Construct an inversion model based on the relative reflection depth and its ratio.

[0014] Furthermore, the specific method of the normalization processing described in step (4) is to subtract the water-off-irradiance reflectance at 300 nm from the water-off-irradiance reflectance obtained in step (3), and then normalize it.

[0015] Furthermore, the reflection depths in step (6) are respectively the distance from the center point of the characteristic reflection band in step (5) to its baseline b. 421-458 、b 646-679 、b 526-609 、b 695-713 The vertical height of .

[0016] Furthermore, the characteristic reflection valley band B in step (5) 435The upper end point is 421 nm, the lower end point is 458 nm, and the center point is 435 nm; the characteristic reflection valley band B 664 The upper end point is 646 nm, the lower end point is 679 nm, and the center point is 664 nm; the characteristic reflection peak band B 573 The upper endpoint is 526 nm, the lower endpoint is 609 nm, and the center endpoint is 573 nm; the characteristic reflection peak band B 706 The upper endpoint is 695 nm, the lower endpoint is 713 nm, and the center endpoint is 706 nm. The line connecting the upper and lower endpoints is b 421-458 、b 646-679 、b 526-609 、b 695-713 are the baselines of the characteristic reflection bands respectively.

[0017] Furthermore, regression analysis was performed to construct a model based on relative reflection depth (RRD). 435 、RRD 706 and RRD 664 / RRD 573 The chlorophyll concentration inversion model of the ratio is C CHL =10 x , , where C CHL is the chlorophyll concentration; a0, a1, a2, a3, a4 are model parameters.

[0018] Furthermore, in step (2), the irradiance reflectance of the sea surface and sky light is measured using a ground object hyperspectral measurement instrument. Beneficial Effects

[0019] In the inversion method disclosed by the present invention, in the Qinhuangdao coastal waters in the embodiment, the correlation coefficient between the model inversion value and the measured value is 0.88358, and the average relative error is 28.33%; the average relative error of the model is reduced by 27%-50% when compared with algorithms such as OCx; the average relative error of the model verification estimate is 31.17%. Therefore, constructing an inversion model based on relative reflection depth can make full use of chlorophyll characteristic band information, reduce data sensitivity, and improve the signal-to-noise ratio of water color elements in coastal waters, thereby greatly improving the inversion accuracy and robustness of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flow chart of the chlorophyll concentration inversion method based on relative reflection depth described in the present invention.

[0021] Figure 2 is the measured reflectance of the irradiance off water.

[0022] Figure 3It is the normalized water-off-irradiance reflectance and characteristic band.

[0023] Figure 4 It is a schematic diagram of the relative reflection depth construction of normalized off-water irradiance reflectance.

[0024] Figure 5 It is the inversion result of hyperspectral data.

[0025] Figure 6 It is the comparison result between the inversion model and the OCx and FLH algorithms.

[0026] Figure 7 It is the data verification result of the inversion model. DETAILED DESCRIPTION

[0027] The chlorophyll concentration inversion method based on relative reflection depth of the present invention is described in detail below in conjunction with the embodiments and drawings.

[0028] The process of the chlorophyll concentration inversion method based on relative reflection depth described in the present invention is as follows: Figure 1 shown.

[0029] In this embodiment, there are 40 survey stations, distributed in the Qinhuangdao waters within the range of 119.35°-119.88°E and 39.67°-39.98°N.

[0030] According to NASA ocean survey specifications, the surface chlorophyll concentration was measured in the laboratory; according to GB12763.1~7-1991 "Ocean Survey Specifications" and "Ocean Optical Survey Technical Regulations" the irradiance reflectance of seawater was measured. and the irradiance reflectance of sky light , the reflectance of the irradiance away from water is A total of 57 sets of corresponding chlorophyll concentration and water-off-irradiance reflectance data were obtained.

[0031] The measured reflectance of water-leaving irradiance is as follows: Figure 2 shown.

[0032] The off-water irradiance reflectance is normalized, that is, the off-water irradiance reflectance is subtracted from the off-water irradiance reflectance at 300nm and then normalized. The normalized off-water irradiance reflectance and the four chlorophyll characteristic reflection bands are: characteristic reflection valley band B 435 (421 nm -458 nm), B 664 (646 nm -679 nm) and characteristic reflection peak band B 573 (526 nm -609 nm) and B 706 (695 nm -713 nm), such as Figure 3Among them, 435nm, 664nm, 573nm, and 706nm are the center points of the above characteristic bands; 421nm, 646nm, 526nm, and 695nm are the upper endpoints of the above characteristic bands; and 458nm, 679nm, 609nm, and 713nm are the lower endpoints of the above characteristic bands.

[0033] Take the line connecting the upper and lower endpoints b 421-458 、b 646-679 、b 526-609 、b 695-713 They are the baselines of the characteristic reflection bands, and the center points of the characteristic reflection bands 435nm, 664nm, 573nm, 706nm are constructed to their baselines b 421-458 、b 646-679 、b 526-609 、b 695-713 The vertical height of the reflection valley is obtained by 435 、RRD 664 and relative reflection peak depth RRD 573 、RRD 706 like Figure 4 shown.

[0034] 38 sets of data were selected to construct the relative reflection depth (RRD) 435 、RRD 706 , and reflection valley depth RRD 435 Reflection peak depth RRD 573 The chlorophyll concentration inversion model is C CHL =10 x , .

[0035] Figure 5 The inversion results show that the correlation coefficient between the inversion value of the inversion model based on relative reflection depth and the measured value is 0.88358, and the root mean square error is 0.47924 mg•m -3 , the average relative error is 28.33%.

[0036] Irradiance reflectivity According to the style Convert to remote sensing reflectance , select 412nm, 443nm, 490nm, 510nm, 555nm and 670nm bands respectively and take them as the center value for calculation The average value in the nm band is used for OC4, OC5, and OC6 algorithm inversion; at the same time, The results of comparing the model of the present invention with the OCx algorithm are shown in the figure. Figure 6At the confidence level of 0.05, the correlation coefficient R, root mean square error RMSE and mean relative error MRE of the inversion values ​​of each algorithm and the measured value are shown in Table 1.

[0037] Table 1 Comparison of inversion algorithms

[0038]

[0039] Depend on Figure 6 As shown in Table 1, in the Qinhuangdao waters, the correlation coefficient between the inversion value of the OCx algorithm and the measured value is small, the root mean square error is greater than 1, the average relative error is about 78%-81%, and the inversion accuracy is low; although the error of the FLH algorithm is lower than that of the OCx algorithm, it is 27% higher than the average relative error of the model of the present invention.

[0040] Figure 7 The remaining 19 sets of data are used to verify the model of the present invention. When the significance level is 0.01, the correlation coefficient between the estimated value and the measured value is 0.58297, and the root mean square error is 0.93039 mg•m -3 The average relative error is 31.17%, which is lower than the international model construction standard of chlorophyll concentration inversion error <35%.

Claims

1. A chlorophyll concentration inversion method based on relative reflection depth, characterized in that: The inversion method uses the following steps: (1) Measure the chlorophyll concentration in the surface layer of seawater; (2) Measure the irradiance reflectance of the sea surface and sky light; (3) Obtaining the reflectance of water-leaving irradiance; (4) Normalize the reflectance of water-leaving irradiance; (5) Determine the chlorophyll characteristic reflectance band of the normalized water-leaving irradiance reflectance; The chlorophyll characteristic reflection bands are characteristic reflection valley band B 435 , characteristic reflection valley band B 664 , characteristic reflection peak band B 573 and characteristic reflection peak band B 706 ; The characteristic reflection valley band B 435 The upper end point is 421 nm, the lower end point is 458 nm, and the center point is 435 nm; the characteristic reflection valley band B 664 The upper end point is 646 nm, the lower end point is 679 nm, and the center point is 664 nm; the characteristic reflection peak band B 573 The upper endpoint is 526 nm, the lower endpoint is 609 nm, and the center endpoint is 573 nm; the characteristic reflection peak band B 706 The upper endpoint is 695 nm, the lower endpoint is 713 nm, and the center endpoint is 706 nm. The line connecting the upper and lower endpoints is b 421-458 , b 646-679 , b 526-609 , b 695-713 They are the baselines of the characteristic reflection bands respectively; (6) Establishing the relative reflection depth of the characteristic band; the relative reflection depth is the relative reflection valley depth RRD 435 , relative reflection valley depth RRD 664 , relative reflection peak depth RRD 573 and relative reflection peak depth RRD 706 The reflection depths are respectively from the center point of the characteristic reflection band to its baseline b 421-458 , b 646-679 , b 526-609 , b 695-713 The vertical height of (7) Perform regression analysis and construct a model based on relative reflection depth (RRD). 435 、RRD 706 and RRD 664 / RRD 573 The chlorophyll concentration inversion model of the ratio is C CHL =10 x , , where a0, a1, a2, a3 and a4 are model parameters.

2. The chlorophyll concentration inversion method based on relative reflection depth according to claim 1, characterized in that: The specific method of the normalization processing described in step (4) is to subtract the water-off-irradiance reflectance at 300 nm from the water-off-irradiance reflectance obtained in step (3), and then normalize it.

3. The chlorophyll concentration inversion method based on relative reflection depth according to claim 1, characterized in that: The irradiance reflectance of the sea surface and sky light in step (2) is measured using a ground object hyperspectral measurement instrument.