Method and system for retrieving chlorophyll-a in stratified light field of eutrophic lake

Abstract

The present application relates to the technical field of water environment remote sensing monitoring, solves the technical problem that the traditional method regards the water body as an optical uniform single-layer parameter, leading to systematic overestimation or underestimation under the condition of algal bloom outbreak or strong stratification, and particularly relates to a stratified light field correction chlorophyll-a remote sensing inversion method and system for eutrophic lakes, which uses multi-spectral / high-spectral remote sensing reflectivity and synchronous chlorophyll-a vertical profile and diffusion attenuation coefficient to perform vertical type identification and three-layer stratification, calculate the light field weight and light path weighted concentration of each layer, construct a stratified light field correction coefficient, perform stratified light field correction on the remote sensing reflectivity, and establish an empirical chlorophyll-a inversion relationship to generate a chlorophyll-a spatial distribution map and an algal bloom risk map. The present application corrects the water surface remote sensing reflectivity by constructing a stratified light field correction coefficient with clear physical meaning, thereby improving the inversion accuracy and robustness under the condition of strong stratification and complex optical conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of water environment remote sensing monitoring, and particularly relates to a layered light field correction method for quantitative remote sensing inversion of chlorophyll-a concentration of eutrophic lake water and a remote sensing inversion system thereof, which is suitable for application scenarios such as lake eutrophication monitoring, algal bloom early warning and hydrological-water ecological comprehensive evaluation. BACKGROUND

[0002] With the intensification of global climate change and human activities in the basin, the problem of inland lake eutrophication is becoming increasingly prominent. Algal blooms frequently occur in large shallow lakes, water transparency decreases, and the ecosystem service function significantly degrades. As the core indicator of phytoplankton biomass and nutritional status, chlorophyll-a is a key water quality parameter for evaluating the degree of lake eutrophication and the intensity of algal blooms. It is also one of the most commonly used inversion targets in water color remote sensing monitoring. Compared with traditional point monitoring, using satellite, aerial or unmanned aerial vehicle remote sensing to carry out chlorophyll-a inversion can obtain lake water quality information at a larger spatial scale and higher temporal resolution, and has become an important technical means for lake water environment management and hydrological-water ecological comprehensive assessment.

[0003] Existing chlorophyll-a remote sensing inversion methods can be broadly divided into three categories: empirical models, semi-analytical models and analytical models based on bio-optical theory. Empirical models are usually directly based on the statistical relationship between water body remote sensing reflectance and measured chlorophyll-a concentration, such as multi-band ratio, narrow-band index, principal component regression, etc. They have the advantages of simple form and easy implementation, but the physical mechanism is not clear enough, and the portability is limited. Semi-analytical models enhance the physical constraints by introducing simplified optical parameters (such as water absorption coefficient, backscattering coefficient), achieving a compromise between theory and experience. Analytical models attempt to start from the radiation transfer equation for inversion, which is theoretically strong, but requires higher parameters and prior information, and still has difficulties in engineering application in complex inland water bodies. Overall, a large number of researches and engineering practices have mainly focused on the two-dimensional relationship between water surface remote sensing reflectance and bulk average chlorophyll-a concentration.

[0004] At present, a variety of remote sensing inversion methods for lake chlorophyll-a have been disclosed. For example, the publication number CN102200576B discloses a chlorophyll-a concentration inversion method and system, which obtains environmental satellite multispectral remote sensing data, selects between seasonal type inversion model and band optimization based model according to observation date, realizes automatic calculation and business operation of chlorophyll-a concentration, and solves the problem of time and space limitation. However, this method is still essentially based on the empirical or semi-empirical relationship between water surface multispectral reflectance and chlorophyll-a bulk concentration, and the water body is regarded as a whole with relatively uniform optical properties in the vertical direction, without explicitly considering the vertical stratification structure commonly found in eutrophic lakes and its influence on the water light field. The publication number CN111781147A proposes a two-band inland lake water body chlorophyll-a concentration remote sensing inversion model and method, which uses two bands of 678 nm and 724 nm to construct an index Index, and obtains a linear relationship in the logarithmic space by least squares fitting, thereby forming a simple two-band inversion model with small operation amount and easy promotion. This method significantly improves the precision and usability of inland lake chlorophyll-a inversion, but it is still based on the empirical relationship between single water surface reflectance and chlorophyll-a concentration, and assumes that the water body is approximately uniform in optics, without special treatment for typical vertical stratification conditions such as surface algal bloom aggregation and subsurface chlorophyll maximum layer in eutrophic lakes. The publication number CN112881293A discloses an inland lake clean water body chlorophyll-a concentration inversion method based on Gaofen-1 satellite, which tests the performance of single band and band combination under different function forms through multi-point sampling and spectral measurement combined with Gaofen-1 satellite image, and selects the optimal function and optimal band combination for clean water body chlorophyll-a inversion. This method has good effect in clean water body with high transparency, weak suspended matter and CDOM (colored dissolved organic matter) interference, but it is also based on the assumption of uniform water column to construct the inversion relationship, and for eutrophic lakes with highly complex optical properties and obvious vertical structure, systematic deviation may occur when applied. For eutrophic lakes, the publication number CN104390917B proposes a MODIS satellite high-precision monitoring method for eutrophic lake water body chlorophyll-a, which selects an index (NDBI) sensitive to chlorophyll-a concentration change and less affected by high suspended solids, simulates and establishes the quantitative relationship between NDBI and chlorophyll-a by means of biological optical model, and realizes high-precision monitoring of eutrophic lake chlorophyll-a on MODIS image through conversion between ground spectrum and satellite Rayleigh scattering corrected reflectance.This method combines index construction and radiation transfer simulation, and has certain advantages in offsetting the interference of suspended solids, but still regards the chlorophyll-a concentration in the entire effective water depth as a single variable, and inversely estimates the overall concentration through the water surface spectral index, without establishing a light field layered correction mechanism for the vertical stratification structure of "surface algal bloom-subsurface maximum-deep background" commonly existing in eutrophic lakes.

[0005] In recent years, there have also been attempts to use machine learning methods to improve the robustness of lake chlorophyll-a inversion. For example, CN110598251B proposes a lake chlorophyll-a concentration inversion method based on Landsat-8 data and machine learning. Based on the Landsat-8 inland lake reflectance dataset and the measured chlorophyll-a concentration, an XGBoost model is constructed to estimate the chlorophyll-a concentration of a large range of lake groups. This method has made progress in improving the universality of the model and is suitable for large-scale spatio-temporal evaluation, but its inversion process still stays at the fitting level between two-dimensional remote sensing reflectance and chlorophyll-a scalar, and removes algal bloom coverage pixels through threshold, rather than depicting and correcting the vertical light field structure of the water body.

[0006] Overall, existing chlorophyll-a remote sensing inversion techniques mainly focus on:

[0007] 1) Selecting appropriate band combinations and empirical / semi-empirical function forms to improve the fitting accuracy under specific satellite or specific lake conditions;

[0008] 2) Using time series models, seasonal typing or machine learning algorithms to improve the spatio-temporal applicability of the model;

[0009] 3) In the context of eutrophic lakes, improving through index form and atmospheric correction to adapt to complex optical water bodies.

[0010] However, these techniques mostly implicitly or explicitly assume that the water column is approximately uniform in the vertical direction, or rely heavily on the spectral response of the surface water body, regarding chlorophyll-a concentration as a "single layer" average parameter. In the face of typical stratification phenomena such as surface algal bloom floating layer, subsurface chlorophyll maximum layer, and deep low chlorophyll background layer commonly existing in eutrophic lakes, the diffusion attenuation coefficient and upward radiation field of the water body show obvious non-uniformity in the vertical direction, and the water surface remote sensing reflectance is contributed by different water layers. The traditional "single layer" model cannot accurately reflect the real water column chlorophyll-a distribution, often leading to systematic overestimation or underestimation under algal bloom outbreak or strong stratification conditions.

[0011] Therefore, in the existing technology, there is a lack of a method and supporting system that can explicitly introduce the vertical stratification information of eutrophic lake water bodies, construct stratified light field correction coefficients based on stratified light field weights and optical path weighted chlorophyll-a, and perform physically meaningful stratified correction of water surface remote sensing reflectance, thereby improving the accuracy and robustness of chlorophyll-a remote sensing inversion of eutrophic lakes. Summary of the Invention

[0012] To address the shortcomings of existing technologies, this invention provides a stratified optical field-corrected chlorophyll-a remote sensing inversion method and system for eutrophic lakes. This method solves the technical problem that traditional methods treat water bodies as optically homogeneous single-layer parameters, leading to systematic overestimation or underestimation under algal blooms or strong stratification conditions.

[0013] Specifically, this invention addresses the common problem in remote sensing monitoring of chlorophyll-a in eutrophic lakes: significant vertical stratification and highly non-uniform optical properties. Existing inversion methods still approximate the water column as a "single-layer homogeneous body," resulting in a systematic deviation between the remotely sensed reflectance of the water surface and the actual chlorophyll-a distribution of the water column under typical stratification conditions such as surface algal blooms and subsurface chlorophyll maxima layers. This leads to inversion results that are prone to overestimation or underestimation, making it difficult to meet the needs of refined water quality assessment and algal bloom early warning for eutrophic lakes.

[0014] The core technical problem to be solved by this invention is: how to explicitly introduce chlorophyll-a vertical stratification information without significantly increasing the complexity of the inversion process, construct a physically meaningful stratified light field weight and optical path weighted chlorophyll-a characterization, form a stratified light field correction coefficient that can be used to correct the water surface remote sensing reflectance, and establish a chlorophyll-a remote sensing inversion method and supporting system suitable for eutrophic lakes, so as to improve the inversion accuracy and robustness under complex optical conditions such as strong stratification, high turbidity and high CDOM, and realize reliable quantitative inversion and spatial characterization of chlorophyll-a in eutrophic lakes.

[0015] To address the aforementioned technical problems, this invention proposes a layered light field correction chlorophyll-a remote sensing inversion method and supporting system for eutrophic lakes. Based on the traditional two-dimensional relationship of "water surface reflectance - volume average chlorophyll-a", the invention introduces the concepts of vertical layered light field weight and optical path weighted average chlorophyll-a concentration. By constructing layered light field correction coefficients with clear physical meaning, the remote sensing reflectance of the water surface is corrected, thereby improving the inversion accuracy and robustness under strong layering and complex optical conditions.

[0016] The preferred method includes the following technical solutions: First, remote sensing images of eutrophic lake water bodies are acquired using a satellite platform, airborne platform, or UAV platform equipped with multispectral or hyperspectral sensors. The remote sensing images are then subjected to radiometric calibration, atmospheric correction, solar flare suppression, and water body mask extraction to obtain the remote sensing reflectance of each water body pixel in multiple bands from visible light to near-infrared. Simultaneously, multi-depth chlorophyll fluorescence probes, optical profilers, and / or multi-parameter water quality profilers were used to conduct synchronous or near-synchronous in-situ observations before and after remote sensing transit to obtain vertical chlorophyll-a concentration profiles of the water body. Transparency and water diffusion attenuation coefficient The isooptic parameter profile provides a data foundation for subsequent layered modeling and correction coefficient calibration.

[0017] Regarding vertical stratification of water bodies, the present invention preferably uses a vertical chlorophyll-a concentration profile of the water body. Based on the morphological and gradient variation characteristics, the effective observation water depth is divided into L water layers, and any water layer The layer depth (upper and lower boundary depth) is And calculate the water layer thickness and average chlorophyll-a concentration ,get Discrete representation. Based on this, preferably, the geometric mean chlorophyll-a concentration over the entire observation depth is calculated. Used to compare subsequent optical path weighting results, i.e.:

[0018]

[0019] To balance automation and physical rationality, this invention further analyzes the vertical chlorophyll-a concentration profile of the water body. After normalization, based on indicators such as the maximum depth of the profile, the ratio of the peak values ​​of the surface layer to the subsurface layer, and the transparency depth, the vertical structure is divided into several typical types, such as the surface algal bloom-dominated type, the subsurface maximum-dominated type, and the near-uniform mixed type. Different vertical types correspond to different water layers L and different rules for selecting the stratification depth, so that the stratification scheme matches the actual water stratification characteristics.

[0020] In terms of optical field weighting and optical path weighting calculation, this invention preferably uses the water diffusion attenuation coefficient. Constructing the vertical light field weighting function This is used to characterize the relative contribution of different depths to the upward radiation at the water surface. The vertical optical field weighting function... It can be expressed in the form of the Beer-Lambert index, i.e.:

[0021]

[0022] Alternatively, a segmented approach based on the layer-average diffusion attenuation coefficient can be adopted within each water layer, i.e.:

[0023]

[0024] The or The results were obtained through in-situ optical observations or calibration via a water tank experiment. Furthermore, the optical field weighting coefficients for each water layer were integrated at each water level. ,Right now:

[0025]

[0026] And using the light field weighting coefficient The chlorophyll-a concentration of each water layer was weighted to obtain the optical path weighted average chlorophyll-a concentration. ,Right now:

[0027]

[0028] in, Water layer The average chlorophyll-a concentration.

[0029] By comparing the light path weighted average chlorophyll-a concentration With geometric mean chlorophyll-a concentration The present invention preferably constructs a wavelength-varying layered optical field correction coefficient. For example, using the ratio form:

[0030]

[0031] Or, in difference form:

[0032]

[0033] The layered light field correction coefficient This intuitively reflects the degree to which the optical weight of the water column for chlorophyll-a deviates from the uniformity assumption under vertical stratification conditions.

[0034] Preferably, for different vertical types and different transparency and turbidity ranges, the correction coefficient for the layered light field is adjusted through on-site samples and water tank experiments. Statistical summarization and smoothing processes are performed to create a layered optical field correction parameter library, providing a basis for rapid selection in engineering applications.

[0035] Regarding remote sensing reflectance correction, the layered light field correction coefficient Used for the original remote sensing reflectance Perform band-level correction. Preferably, a band-multiplication correction method can be used, i.e.:

[0036]

[0037] in, The pre-calibrated monotonic transform function is used to ensure that the corrected reflectivity is physically reasonable; it can also be used in the red, red-edge, and near-infrared sensitive bands with layered optical field correction coefficients. We use weights to combine multiple bands to construct a spectral index corrected by layered optical field.

[0038] Furthermore, it is preferable to select a narrowband or wideband combination in the 660–720 nm wavelength range that is sensitive to chlorophyll-a and has a certain robustness to interference from suspended matter and highly colored dissolved organic matter (CDOM) in order to reduce the influence of complex optical background.

[0039] In establishing the chlorophyll-a inversion relationship, this invention does not rely on complex general algorithms, but rather preferably employs an empirical regression model with a clear physical background. Specifically, the remote sensing reflectance corrected by the layered light field can be used... Alternatively, the corresponding spectral index can be used as the independent variable, and the synchronously measured chlorophyll-a concentration (either surface concentration or optical path weighted concentration can be selected) can be used as the dependent variable to establish empirical relationships in linear, logarithmic or power function form for different vertical types.

[0040] Furthermore, multiple regression can be used to combine multiple sensitive bands or indices as independent variables to improve the model's adaptability to complex optical conditions. By using genotyping or introducing type indicator parameters into a unified model, the inversion relationship can explicitly reflect the corrective effect of layered light field correction on the relationship between water surface spectrum and chlorophyll-a in water column.

[0041] At the system level, the present invention further provides a chlorophyll-a layered light field correction remote sensing inversion system for eutrophic lakes. The system preferably includes a remote sensing data acquisition and preprocessing unit, an in-situ vertical observation unit, a vertical layering and parameter calculation unit, a layered light field weight and correction coefficient calculation unit, a remote sensing reflectance correction unit, and a chlorophyll-a inversion and mapping unit.

[0042] The remote sensing data acquisition and preprocessing unit is used to acquire remote sensing images of eutrophic lake areas and perform atmospheric correction and water masking processing, outputting the remote sensing reflectance of each pixel. The in-situ vertical observation unit is used to obtain the vertical chlorophyll-a concentration profile of the water body. Transparency and water diffusion attenuation coefficient The vertical stratification and parameter calculation unit is used to divide the water body into multiple water layers and calculate the thickness of each water layer. Average chlorophyll-a concentration and geometric mean chlorophyll-a concentration The layered light field weighting and correction coefficient calculation unit is based on the water diffusion attenuation coefficient. Establish vertical light field weighting function The light field weighting coefficients of each water layer were obtained. Calculate the optical path weighted average chlorophyll-a concentration. and layered light field correction coefficient It can preferably be connected to a storage unit to retrieve or update reference parameters from the correction parameter library under different hydrological seasons and different lake conditions.

[0043] The remote sensing reflectivity correction unit is based on the layered light field correction coefficient. For remote sensing reflectance After correction, the remote sensing reflectance is obtained after layered light field correction. The spectral index of chlorophyll-a and its combination; the chlorophyll-a inversion and mapping unit calls the pre-established chlorophyll-a inversion relationship, calculates the chlorophyll-a concentration of each pixel and generates a spatial distribution map, and can further superimpose information such as transparency and turbidity to provide support for eutrophic lake water quality assessment and algal bloom early warning.

[0044] Through the above technical solution, this invention, while maintaining a clear inversion process structure and controllable computational complexity, organically integrates the vertical stratification information of eutrophic lake water with the optical properties of the water body. It constructs a stratified light field correction coefficient with stratified light field weights and optical path weighted chlorophyll-a as the core, and makes a physically meaningful correction to the water surface remote sensing reflectance. This significantly reduces the system bias under strong stratification, high turbidity and high CDOM conditions, and improves the accuracy and robustness of chlorophyll-a remote sensing inversion results.

[0045] By employing the above technical solution, the present invention provides a method and system for stratified light field correction of chlorophyll-a remote sensing inversion in eutrophic lakes, which has at least the following beneficial effects:

[0046] This invention, based on the traditional "water surface remote sensing reflectance - volume average chlorophyll-a" inversion framework, introduces the characterization of vertically layered light field weights and optical path-weighted chlorophyll-a for water bodies, constructing layered light field correction coefficients with clear physical meaning to perform layered correction of remote sensing reflectance:

[0047] First, it can explicitly characterize the different contributions of typical stratified structures such as surface algal blooms, subsurface chlorophyll maximum layers and deep background layers to upward radiation, effectively correct the systematic bias caused by the existing "single-layer homogeneity" assumption, and significantly improve the accuracy and robustness of chlorophyll-a remote sensing inversion in eutrophic lakes under complex optical conditions such as strong stratification, high turbidity and high CDOM.

[0048] Secondly, this invention adopts an optical field weighting function based on the diffusion attenuation coefficient and a simple, calibrable empirical inversion relationship, avoiding dependence on complex general algorithms and a large number of black box parameters, maintaining the physical interpretability and engineering feasibility of the method, and facilitating its transfer and promotion among different sensor platforms and different lake water bodies.

[0049] Third, this invention also provides a supporting layered light field correction remote sensing inversion system. Through the integration of functional modules such as vertical type identification, layered parameter calculation, correction coefficient library calling and updating, it realizes batch and near real-time remote sensing inversion of chlorophyll-a in eutrophic lakes. It provides a new technical approach with a clear structure, stable operation and easy operational deployment for lake eutrophication monitoring, algal bloom early warning and hydrological-aquatic ecological comprehensive evaluation. Attached Figure Description

[0050] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0051] Figure 1 This is a schematic diagram of the overall structure of the layered light field corrected chlorophyll-a remote sensing inversion system in this invention;

[0052] Figure 2 This is a flowchart of the layered light field correction chlorophyll-a remote sensing inversion method in this invention;

[0053] Figure 3 This is a diagram showing the relationship between vertical water stratification, light field weight, and optical path weighted concentration in this invention.

[0054] Figure (a) shows the three layers of the water column and the chlorophyll-a concentration levels in each layer, while Figure (b) shows the variation of the light field weighting function W(z) with depth and the weight contribution of each layer. , , and , Comparison chart;

[0055] Figure 4 This is a schematic diagram of the structure of the layered optical field correction coefficient library and its calling and updating relationships in this invention;

[0056] Figure 5 This is a schematic diagram of the vertical distribution and stratification of chlorophyll-a at a typical P3 station in this invention;

[0057] Figure 6 This is a comparison of scatter plots and fitted lines between the uncorrected model and the layered light field corrected model for the calibration set ln(Index) – chlorophyll-a relationship in this invention.

[0058] Figure 7 This is a graph showing the comparison between the predicted values ​​of the calibration set and the measured chlorophyll-a values ​​in this invention.

[0059] Figure 8 This is a schematic diagram of the vertical distribution and stratification of chlorophyll-a at a typical B1 station in this invention.

[0060] Figure 9 This is a graph showing the comparison between the predicted values ​​and the measured chlorophyll-a values ​​at 30 stations in this invention.

[0061] Figure 10 This is a comparison of the average relative errors of the three types of water bodies (algal bloom center / edge / background) in this invention. Detailed Implementation

[0062] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This will allow for a full understanding of how the present application uses technical means to solve technical problems and achieve technical effects, and to facilitate its implementation.

[0063] Example 1: Calibration of layered light field correction coefficients and construction of inversion model for a typical shallow eutrophic lake.

[0064] This embodiment uses a typical shallow eutrophic lake as an example to demonstrate how to construct a layered light field correction coefficient using the method of the present invention. In this way, a chlorophyll-a remote sensing inversion model suitable for the lake was established, and the difference in inversion accuracy before and after layered light field correction was quantitatively evaluated.

[0065] The lake has an area of ​​approximately 200 km². 2 The lake has an average depth of approximately 2.5 meters, with a maximum depth not exceeding 6 meters. It is significantly affected by nutrient input from rivers flowing into the lake, leading to frequent algal blooms during the summer and autumn. Eighteen observation stations were selected in the open water area in the central part of the lake and the eutrophic nearshore area. Two simultaneous in-situ observation and satellite remote sensing experiments were conducted in mid-July and early August, respectively. Each observation yielded 18 vertical profile samples, totaling 36 profile samples. Based on the satellite transit time and cloud cover, 27 sample points with good matching to satellite pixels and no significant cloud shadow interference were selected from the two experiments for use in the calibration of correction coefficients and the construction of the inversion model in this embodiment.

[0066] In terms of in-situ observation, multi-parameter water quality profilers and chlorophyll fluorescence probes were used at each station to obtain vertical chlorophyll-a concentration profiles of the water body within a water depth range of 0–5 m. By measuring water temperature, turbidity, and the light radiation attenuation at different depths, the water diffusion attenuation coefficient can be calculated. Simultaneously, surface water samples were collected at a depth of 0.2 m, and chlorophyll-a concentration was determined by laboratory 90% ethanol extraction colorimetric method as a reference value. Regarding remote sensing data, multispectral satellite data corresponding to the in-situ observation date was selected. Radiometric calibration and atmospheric correction were performed on the original Level-1C imagery, and the remote sensing reflectance of the water surface was obtained using standard inland water body processing procedures. The satellite sensor extracts satellite pixels corresponding to each station location. It features narrowband channels in the red-red-edge-near-infrared bands (665nm, 705nm, and 740nm), suitable for chlorophyll-a retrieval.

[0067] Regarding vertical stratification and parameter calculation, this embodiment selects station P3, which exhibits typical strong stratification characteristics, for illustration. Figure 5 As shown. At station P3, the water depth is approximately 5m. During observations in mid-July, the measured chlorophyll-a concentrations (unit: µg / L) at each depth were as follows: 0.2m: 65, 0.5m: 62, 0.8m: 55, 1.2m: 40, 2.0m: 36, 2.8m: 30, 3.5m: 12, 4.5m: 8. It is evident that a significant algal bloom high value exists in the surface layer (0–1m), a subsurface chlorophyll maxima zone forms in the 1–3m range, and below 3m gradually transitions into a low-concentration background layer. Based on the stratification principle proposed in this invention, the 0–5m water body is divided into three layers: the surface algal bloom layer (0–1m, layer 1), the subsurface chlorophyll maxima layer (1–3m, layer 2), and the deep background layer (3–5m, layer 3), i.e. =0m, =1m, =3m, =5m.

[0068] The average chlorophyll-a concentration was calculated based on measured values ​​within each layer. and thickness That is, for the first layer (0-1m), the average chlorophyll-a concentration is obtained by averaging the values ​​at three points: 0.2m, 0.5m, and 0.8m. ≈60µg / L, water layer thickness =1.0m; for the second layer (1-3m), the average chlorophyll-a concentration was obtained by averaging the values ​​at three points: 1.2m, 2.0m, and 2.8m. ≈35µg / L, water layer thickness =2.0m; for the third layer (3-5m), the average chlorophyll-a concentration was obtained by averaging the values ​​at 3.5m and 4.5m. ≈10µg / L, water layer thickness =2.0m. Therefore, the geometrically average chlorophyll-a concentration of the entire water column is:

[0069]

[0070] Regarding the calculation of optical field weighting and optical path weighting, the calculation process of layered optical field weighting and optical path weighting of chlorophyll-a is illustrated using the 665nm band as an example. Based on the profile optical measurement results, the layered average diffusion attenuation coefficient of station P3 in the 665nm band is... (665) is approximately: the first layer , second floor , third floor The vertical light field weighting function in exponential form is:

[0071]

[0072] The light field weighting coefficients for each water layer can be calculated using the following formula:

[0073]

[0074] Substituting the values, we get:

[0075] The first layer (0-1m) is:

[0076]

[0077] The second floor (1-3m) is:

[0078]

[0079] The third layer (3-5 m) is:

[0080]

[0081] The sum of the weights of the three light fields is approximately:

[0082]

[0083] Based on this, the optical path-weighted average chlorophyll-a concentration at the 665nm wavelength can be calculated as follows:

[0084]

[0085] optical path weighted average chlorophyll-a concentration With geometric mean chlorophyll-a concentration In comparison, the layered optical field correction coefficient for the 665nm band can be defined as follows:

[0086]

[0087] Similarly, in the 705nm band, based on the measured average diffusion attenuation coefficient of the layer... , , The vertical light field weighting function can be obtained by calculating using the method described above. , , and optical path weighted average chlorophyll-a concentration The results showed that the optical path weighted average chlorophyll-a concentration Approximately 36.9 µg / L, corresponding to the layered optical field correction coefficient. It is evident that under strong separation conditions, the surface light field exhibits "preferential sampling" of the upper high-concentration water body, resulting in an optical path-weighted concentration that is significantly higher than the geometric mean concentration. This embodiment addresses this by using a layered light field correction coefficient. Quantify this deviation explicitly.

[0088] Regarding remote sensing reflectance correction, the remote sensing reflectance of the satellite pixel corresponding to station P3 in the 665nm and 705nm bands are respectively... , According to the layered light field correction concept of the present invention, a ratio-based correction function is preferably used. The optical path weighted average chlorophyll-a concentration Converted to the corresponding geometric mean chlorophyll-a concentration The spectral response was used to obtain the corrected remote sensing reflectance. :

[0089]

[0090]

[0091] For the 27 valid sample points in this embodiment, the average chlorophyll-a concentration of three layers or two layers was calculated respectively according to the above method. Water layer thickness Geometric mean chlorophyll-a concentration Light field weighting coefficient Path-weighted average chlorophyll-a concentration and layered light field correction coefficient After performing layered optical field correction on the 665nm and 705nm bands, the spectral indices before and after correction are constructed as follows:

[0092] The uncorrected spectral index is denoted as:

[0093]

[0094] The corrected spectral index is denoted as:

[0095]

[0096] like Figure 6 As shown, this embodiment uses the measured chlorophyll-a concentration at 27 sampling points as the dependent variable to establish two types of empirical inversion relationships. That is:

[0097] Uncorrected model:

[0098]

[0099] Revised model:

[0100]

[0101] in , , , The coefficients of determination R0 of the uncorrected model were obtained through least squares fitting. The fitting results show that the coefficients of determination R0 of the uncorrected model at 27 sample points are... 2 The coefficient of determination R0.80 is approximately 0.80, the root mean square error (RMSE) is approximately 9.2 µg / L, and the average relative error is approximately 28%. After correction for the layered light field, the coefficient of determination R0.80 of the model is approximately 9.2 µg / L. 2 The mean square error (RMSE) was reduced to approximately 6.3 µg / L, and the average relative error was reduced to approximately 18%. In particular, at sites with strong surface algal blooms and significant subsurface algal blooms, the systematic overestimation or underestimation phenomenon was significantly alleviated.

[0102] like Figure 7 As shown, in terms of model extrapolation validation, this embodiment uses the data from the remaining 9 stations in the second observation as an independent validation set to calculate the prediction error of the model before and after correction. The results show that the R-value of the model after layered light field correction on the validation set is higher. 2 The mean square error (RMSE) is approximately 0.88, which is a significant improvement compared to approximately 0.77 for the uncorrected model. The root mean square error (RMSE) is reduced from approximately 10.1 µg / L to approximately 7.0 µg / L, indicating that the hierarchical optical field correction method of the present invention has good applicability on independent samples.

[0103] As can be seen from this embodiment, the method for constructing layered optical field correction coefficients and its application process in chlorophyll-a remote sensing inversion provided by the present invention can realize a complete technical closed loop in typical shallow eutrophic lakes, from in-situ profile observation to layered modeling, optical path weighting calculation, correction coefficient calibration, and inversion model construction. While maintaining the simplicity of the method structure and the controllability of the computational load, it significantly reduces the systematic bias under strong layering conditions and improves the accuracy and robustness of chlorophyll-a remote sensing inversion results.

[0104] Example 2: Application and effect evaluation of layered light field correction under strong layering conditions during algal blooms.

[0105] This embodiment, based on the layered light field correction coefficient library and chlorophyll-a inversion relationship constructed in Example 1, selects multi-temporal remote sensing images and in-situ observation data of the same shallow eutrophic lake during summer algal blooms to verify the applicability and effectiveness of the method of the present invention under conditions of strong stratification and highly complex optical properties. Steps and apparatus that are the same as those in Example 1 unless otherwise specified will not be repeated.

[0106] The lake has an area of ​​approximately 200 km². 2 The average water depth is approximately 2.5 m, and a large algal bloom floating zone forms on the leeward side of the prevailing wind in mid-to-late August. Preferably, 30 in-situ observation stations are deployed when the satellite passes overhead on August 18th, with 10 located in the central area of ​​the algal bloom (denoted as B1–B10), 12 in the edge area (denoted as E1–E12), and 8 in the relatively clear background area (denoted as C1–C8). Within two hours before and after the satellite pass, a multi-parameter water quality profiler is used to obtain vertical chlorophyll-a concentration profiles of 0.2–0.5 m resolution within a water depth range of 0–4 m. Water temperature, turbidity, and the water diffusion attenuation coefficient in the visible to red edge bands were simultaneously measured using underwater optical sensors. Simultaneously, surface water samples were collected at a depth of 0.2 m to determine chlorophyll-a concentration, serving as a reference value for evaluating algal bloom intensity and verifying inversion results. Satellite imagery was generated using the same multispectral sensor as in Example 1, and narrowband water surface remote sensing reflectance at 665 nm and 705 nm was obtained through radiometric calibration and atmospheric correction. It is spatially paired with 30 stations.

[0107] To highlight the role of layered light field correction in cases of strong layering, this embodiment selects one representative station each from the central region of the algal bloom, the edge region of the algal bloom, and the background region for vertical structure analysis. For example... Figure 8 and Figure 9 As shown, at the central algal bloom station B1, the water depth is approximately 4m. The measured chlorophyll-a concentrations (µg / L) at depths of 0.1m, 0.3m, 0.5m, 0.8m, 1.2m, 2.0m, 3.0m, and 4.0m are approximately 150, 130, 110, 80, 50, 40, 20, and 10, respectively. This indicates a thin layer of high concentration at depths of 0–0.5m, a mid-to-upper layer of slightly higher concentrations at depths of 0.5–2m, and a low-concentration background layer below depth 2m. Based on the vertical stratification principle of this invention, the 0–4m water column is divided into three layers: the surface algal bloom layer (0–0.5m, layer 1), the subsurface enhancement layer (0.5–2m, layer 2), and the deep background layer (2–4m, layer 3). The average chlorophyll-a concentration of each layer can then be obtained. ≈130µg / L ≈56.7µg / L ≈15µg / L, water layer thickness is respectively =0.5m =1.5m =2.0m, then the geometric mean chlorophyll-a concentration of the entire water column is:

[0108]

[0109] Regarding the calculation of optical field weights, this embodiment preferably uses the 665nm band as an example. Based on the diffusion attenuation coefficient profile, the layer-average diffusion attenuation coefficient of station B1 in layers 1-3 of the 665nm band is... (665) are approximately respectively , and The vertical light field weighting function in exponential form is:

[0110]

[0111] The light field weighting coefficients for each water layer can be calculated using the following formula:

[0112]

[0113] Substitution =0、 =0.5、 =2.0、 =4.0 and =1.8、 =1.0、 =0.5, therefore:

[0114] Level 1:

[0115] Level 2:

[0116] Level 3:

[0117] The sum of the weights of the three light fields is approximately Based on this, the optical path-weighted average chlorophyll-a concentration at the 665nm wavelength can be calculated as follows:

[0118]

[0119] In comparison, the optical path-weighted average chlorophyll-a concentration Significantly higher than the geometric mean chlorophyll-a concentration of the entire water column The concentration is approximately 45.0 µg / L, indicating that under strong stratification conditions, the weight of the upward-flowing light field on the high-concentration surface water is significantly amplified. Therefore, the stratification correction coefficient for the 665 nm band can be defined as:

[0120]

[0121] Similarly, for the 705nm band, , , The same calculation yields the optical path weighted average chlorophyll-a concentration. ≈59.1µg / L, corresponding to the layered optical field correction coefficient In this embodiment, for other algal bloom center sites B2 to B10, their respective layered light field correction coefficients can be calculated using the method described above. Furthermore, by combining the correction coefficient library constructed in Example 1, outliers are smoothed and grouped.

[0122] Regarding remote sensing reflectance correction, the original remote sensing reflectance of the water surface for the satellite pixel corresponding to station B1 in the 665nm and 705nm bands is approximately [missing information]. , Spectral index Based on the aforementioned layered optical field correction coefficients, this embodiment preferably uses a method of dividing by the wavelength band. Methods for remote sensing reflectance Make corrections, namely:

[0123]

[0124] Substitution , have to:

[0125]

[0126]

[0127] Corresponding corrected spectral index ≈1.45. For all pixels at the 30 stations, their respective layered optical field correction coefficients are used. and For remote sensing reflectance Layered light field correction is performed to obtain the corrected remote sensing reflectance. and spectral index .

[0128] Regarding chlorophyll-a inversion, this embodiment directly calls the empirical inversion relationship established in Embodiment 1 based on the spectral index after layered light field correction, and uses the spectral index... Using the measured chlorophyll-a concentration in the surface layer as the independent variable and the measured chlorophyll-a concentration in the surface layer as the dependent variable, the chlorophyll-a concentration was retrieved from 30 stations and compared with that of stations that used spectral indices directly without layered light field correction. A comparison was made with the established traditional empirical model. The results show that:

[0129] At the central sites of the algal bloom, the measured chlorophyll-a concentration in the surface layer generally ranged from 120 to 180 µg / L. For example, the measured value at site B1 was approximately 150 µg / L, and the inversion result given by the uncorrected model was approximately 100 µg / L, with a deviation of approximately -33%. After applying the layered light field correction model of this invention, the inversion result was approximately 135 µg / L, with a deviation of approximately -10%. At site B2 at the edge of the algal bloom (measured at approximately 80 µg / L in the surface layer), the uncorrected model, due to the mixing effect of the thin surface layer and the subsurface maximum layer, gave an inversion value of approximately 120 µg / L, with a relative error of approximately +50%. After layered light field correction, the inversion result was approximately 86 µg / L, and the relative error was reduced to approximately +7.5%. At background station B3 (surface measured at approximately 30 µg / L), the uncorrected model underestimated the light field due to its primary control by the relatively clear water in the upper layer, resulting in an inversion value of approximately 20 µg / L and a relative error of approximately -33%. After stratified correction, the inversion value of the model was approximately 28 µg / L, with a relative error of approximately -6.7%. The above comparison of typical stations demonstrates that the method of this invention can effectively reduce systematic bias in three typical water bodies: algal bloom centers, high gradient edges, and relatively clear water backgrounds.

[0130] Furthermore, the overall statistical results for the 30 stations are as follows: Figure 10 As shown. The coefficient of determination R between the uncorrected model predictions and the measured chlorophyll-a concentrations. 2 The coefficient of determination R is approximately 0.81, the root mean square error (RMSE) is approximately 15.2 µg / L, and the average relative error is approximately 38%. After applying the layered optical field correction of this invention, the coefficient of determination R... 2 The accuracy was improved to approximately 0.92, the root mean square error (RMSE) was reduced to approximately 9.7 µg / L, and the average relative error was reduced to approximately 19.6%. In particular, in high-value pixels of algal blooms with chlorophyll-a concentrations greater than 80 µg / L, the uncorrected model generally showed an underestimation or overestimation of more than 30%, while the method of this invention can control the absolute error to around 10%, significantly improving the reliability of chlorophyll-a remote sensing inversion during algal blooms.

[0131] In terms of spatial applications, this embodiment uses the entire lake remote sensing image from August 18th as the object. The layered light field correction coefficient library and chlorophyll-a inversion relationship constructed in Embodiment 1 are embedded into the batch processing workflow. Layered light field correction and chlorophyll-a inversion are performed on all lake water pixels, and area statistics are calculated using 50µg / L and 100µg / L as thresholds for medium-to-high algal bloom levels. The results show that the area of ​​severe algal bloom with chlorophyll-a > 100µg / L obtained without correction is approximately 63 km². 2 The area of ​​severe algal bloom obtained by the method of this invention is approximately 48 km². 2 The latter, along with the results of the on-site visual investigation (approximately 45km), 2The results are closer to the traditional "uniform single-layer" assumption, which significantly exaggerates the area of ​​algal blooms under strong stratification conditions. This invention effectively alleviates this problem through stratified light field correction. Meanwhile, in moderate algal bloom regions with chlorophyll-a levels between 50 and 100 µg / L, the results from both methods are similar in overall area. However, the spatial distribution obtained by the method of this invention is more consistent with the orientation and outer edge of the algal bloom zone revealed by profile observations, and the boundaries are smoother.

[0132] In summary, this embodiment demonstrates that under conditions of algal blooms and strong water stratification, the "stratified light field correction" chlorophyll-a remote sensing inversion method for eutrophic lakes proposed in this invention can construct physically meaningful stratified light field correction coefficients by explicitly introducing vertical stratified light field weights and optical path weighted chlorophyll-a concentrations. This allows for reasonable correction of the water surface remote sensing reflectance, thereby significantly improving the accuracy and spatial characterization of chlorophyll-a inversion and providing more reliable technical support for the monitoring and early warning of algal blooms in eutrophic lakes.

[0133] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0134] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. Since the above embodiments are substantially similar to the method embodiments, their descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0135] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A layered light field-corrected chlorophyll-a remote sensing inversion method for eutrophic lakes, characterized in that, The method includes the following steps: S1. Obtain the remote sensing reflectance of the target eutrophic lake in multiple bands based on remote sensing images of the lake. Furthermore, vertical chlorophyll-a concentration profiles of water bodies were collected synchronously or nearly synchronously before and after the remote sensing transit. Transparency and water diffusion attenuation coefficient ; S2. Based on the vertical chlorophyll-a concentration profile of the water body Based on the variation characteristics, the effective observation depth of the water body is divided into L water layers vertically, and the geometric mean chlorophyll-a concentration within the entire observation depth is calculated. ; S3, Based on the water diffusion attenuation coefficient of each water layer or layer average diffusion attenuation coefficient Calculate the optical path weighted average chlorophyll-a concentration and the geometric mean chlorophyll-a concentration By performing ratio or difference operations, the layered optical field correction coefficients that vary with wavelength are obtained. ; S4. Based on the layered light field correction coefficient For remote sensing reflectance Perform band-by-band product correction and / or band reweighting correction to obtain the layered optical field corrected remote sensing reflectance. ; S5, Remote sensing reflectance corrected by layered light field Alternatively, the spectral index constructed from it can be used as the independent variable, and the chlorophyll-a concentration at the corresponding sampling point can be used as the dependent variable. An empirical regression method can be used to establish the chlorophyll-a inversion relationship. S6. Apply the chlorophyll-a inversion relationship to the remote sensing image of the eutrophic lake to be inverted, and use the remote sensing reflectance corrected by the layered light field. The spatial distribution of chlorophyll-a concentration in the lake water was calculated.

2. The layered light field corrected chlorophyll-a remote sensing inversion method according to claim 1, characterized in that, The vertical stratification of the water body includes: Vertical chlorophyll-a concentration profile of water body After normalization, based on the maximum depth of the profile, the ratio of the peak values ​​of the surface layer to the subsurface layer, and the transparency characteristics, the vertical structure of the water body is classified into at least one of the following: surface algal bloom-dominated type, subsurface algal bloom-dominated type, and near-uniform mixed type. Different vertical types correspond to different water layers L and different stratification depths. Selection rules; The geometric mean chlorophyll-a concentration The calculation process includes: Based on any water layer Layer depth Calculate water layer thickness and average chlorophyll-a concentration ; According to thickness and average chlorophyll-a concentration Calculate the geometric mean chlorophyll-a concentration used to compare with subsequent optical path weighting results. The calculation formula is: ； Among them, thickness .

3. The layered light field corrected chlorophyll-a remote sensing inversion method according to claim 1, characterized in that, The layered light field correction coefficient The calculation process includes: The vertical optical field weighting function is established using the Beer–Lambert approximation or the two-stream approximation. The light field weighting coefficients for each water layer are obtained by integrating the light fields across the water layers. ,Right now: ； Using the light field weighting coefficient The chlorophyll-a concentration of each water layer was weighted to obtain the optical path weighted average chlorophyll-a concentration. ,Right now: ； in, Water layer The average chlorophyll-a concentration; The optical path weighted average chlorophyll-a concentration With geometric mean chlorophyll-a concentration By performing ratio or difference operations, the wavelength-dependent layered optical field correction coefficients are obtained. ,Right now: The ratio form is as follows: ； Using the difference form: ； The layered light field correction coefficient This intuitively reflects the degree to which the optical weight of the water column for chlorophyll-a deviates from the uniformity assumption under vertical stratification conditions.

4. The layered light field corrected chlorophyll-a remote sensing inversion method according to claim 3, characterized in that, The vertical light field weighting function It may take one of the following forms, including: Based on water diffusion attenuation coefficient The exponential form, namely: ； Based on layer-average diffusion attenuation coefficient The piecewise constant form is: ； Based on the simplified function form of the two-stream approximation, the function form is determined by outdoor or experimental water tank calibration to ensure that the light field weight decreases with depth.

5. The layered light field corrected chlorophyll-a remote sensing inversion method according to claim 1, characterized in that, The layered optical field correction applies at least to the red, red-edge, and near-infrared bands, by selecting several sensitive bands within the 660–720 nm wavelength range. The corresponding remote sensing reflectance is corrected, and a chlorophyll-sensitive spectral index containing two-band ratio, three-band combination or narrowband index is constructed based on the corrected remote sensing reflectance. Remote sensing reflectance is obtained by using the band-by-band product correction. ,Right now: ； in, It is a monotonic transformation function that has been determined in advance through calibration.

6. The layered light field corrected chlorophyll-a remote sensing inversion method according to claim 1, characterized in that, The chlorophyll-a inversion relationship is at least one of a linear function, a logarithmic function, a power function, or a multiple regression empirical formula, and different inversion relationships can be established according to the vertical type, or a classification parameter representing the vertical type can be introduced into the unified inversion relationship.

7. The layered light field corrected chlorophyll-a remote sensing inversion method according to claim 3, characterized in that, Also includes: When eutrophic lakes contain high turbidity or high levels of colored dissolved organic matter, the vertical light field weighting function... and layered light field correction coefficient The determination of the chlorophyll-a remote sensing inversion accuracy under complex optical conditions takes into account both the concentration of suspended particulate matter and the absorption coefficient of high colored dissolved organic matter. The correction coefficient is graded and calibrated by field paired observations and / or flume experiments under different turbidity and high colored dissolved organic matter levels.

8. A layered light-field corrected chlorophyll-a remote sensing inversion system for eutrophic lakes, used to implement the layered light-field corrected chlorophyll-a remote sensing inversion method according to any one of claims 1-7, characterized in that, include: The remote sensing data acquisition and preprocessing unit is used to acquire remote sensing images of eutrophic lake areas, perform atmospheric correction and water masking, and output the remote sensing reflectance of each pixel. ; In-situ vertical observation unit is used to obtain vertical chlorophyll-a concentration profiles in water bodies. Transparency and water diffusion attenuation coefficient ; The vertical stratification and parameter calculation unit is used to divide the water body into multiple water layers and calculate the thickness of each water layer. Average chlorophyll-a concentration and geometric mean chlorophyll-a concentration ; The layered light field weighting and correction coefficient calculation unit is used to calculate the water diffusion attenuation coefficient. Establish vertical light field weighting function The light field weighting coefficients of each water layer were obtained. Calculate the optical path weighted average chlorophyll-a concentration. and layered light field correction coefficient ; Remote sensing reflectivity correction unit, used to adjust the reflectivity based on the layered light field correction coefficient. For remote sensing reflectance After correction, the remote sensing reflectance is obtained after layered light field correction. Spectral indices of themselves and their combinations; The chlorophyll-a inversion and mapping unit is used to call the pre-established chlorophyll-a inversion relationship, calculate the chlorophyll-a concentration of each pixel, and generate a spatial distribution map.

9. The layered light field corrected chlorophyll-a remote sensing inversion system according to claim 8, characterized in that, The in-situ vertical observation unit includes a multi-depth chlorophyll fluorescence probe, an optical profiler, and / or a multi-parameter water quality profiler, used to simultaneously record chlorophyll-a signals and water diffusion attenuation coefficients at multiple depths within the same water column, providing data support for the calculation of optical path-weighted chlorophyll-a concentration and stratified light field correction coefficients.

10. The layered light field corrected chlorophyll-a remote sensing inversion system according to claim 8, characterized in that, It includes a storage unit connected to the layered light field weight and correction coefficient calculation unit, wherein the storage unit pre-stores reference light field weight parameters or correction coefficient ranges for different vertical types, different transparency and turbidity conditions; When new in-situ vertical observation data is collected, the layered light field weight and correction coefficient calculation unit corrects or refines the reference parameters based on the new observation results and updates the parameters in the storage unit to improve the applicability under different seasons and different eutrophic lake water conditions.

Images