CN-121540641-B - Turbidity extraction method and device based on remote sensing data

CN121540641BCN 121540641 BCN121540641 BCN 121540641BCN-121540641-B

Abstract

The application provides a turbidity extraction method and device based on remote sensing data, and belongs to the field of satellite remote sensing water quality monitoring. The method comprises the steps of obtaining atmospheric corrected remote sensing data of a target sea area and turbidity values to be fitted of a plurality of sampling points, screening a plurality of target remote sensing bands according to fitting results of single-band remote sensing reflectivity data in the atmospheric corrected remote sensing data and the turbidity values to be fitted, determining a turbidity prediction model sample according to fitting results of combination of any two target remote sensing bands and the turbidity values to be fitted, modifying band composition in the turbidity prediction model sample, determining a target turbidity prediction model according to the modified instantiation turbidity prediction model and fitting results of the turbidity values to be fitted, and predicting real-time turbidity of the target sea area according to real-time remote sensing data of the target sea area and the target turbidity prediction model. The method and the device provided by the application can solve the problems that the accuracy of the offshore area single-band inversion is insufficient, the influence of interference factors such as aerosol is large, and the suitability of the existing model is limited.

Inventors

CAI LINA
ZHANG BEIBEI

Assignees

浙江海洋大学

Dates

Publication Date: 20260508
Application Date: 20260120

Claims (7)

1.A turbidity extraction method based on remote sensing data, the method comprising: Acquiring remote sensing data after atmospheric correction of a target sea area and turbidity values to be fitted of a plurality of sampling points; screening a plurality of target remote sensing bands according to a first fitting result of the single-band remote sensing reflectivity data in the atmospheric corrected remote sensing data and the turbidity value to be fitted; Determining a turbidity prediction model sample according to the combination of any two target remote sensing wave bands and the second fitting result of the turbidity value to be fitted; modifying the wave band composition in the turbidity prediction model sample, and determining a target turbidity prediction model according to the modified instantiation turbidity prediction model and the second fitting result of the turbidity value to be fitted; Predicting the real-time turbidity of the target sea area according to the real-time remote sensing data of the target sea area and the target turbidity prediction model; And determining a turbidity prediction model sample according to the combination of any two target remote sensing bands and the second fitting result of the turbidity value to be fitted, wherein the method comprises the following steps: Pairing the screened target remote sensing wave bands in pairs; For each pair of wave bands, constructing a plurality of mathematical relations to form a plurality of candidate wave band combinations, wherein each mathematical form is used for carrying out numerical operation on two target remote sensing wave bands in each pair of wave bands; performing model fitting on each candidate wave band combination and the turbidity value to be fitted; Comparing the goodness of fit and error indexes of all candidate wave band combinations, and selecting the mathematical relationship corresponding to the candidate wave band combination with the highest goodness of fit and the smallest error index as the turbidity prediction model sample; The turbidity prediction model sample is a ratio, the modifying the wave band composition in the turbidity prediction model sample, and determining a target turbidity prediction model according to the modified instantiated turbidity prediction model and the second fitting result of the turbidity value to be fitted comprises the following steps: Determining a candidate wave band according to the first fitting result, and respectively adjusting the numerator, denominator and additional terms of the turbidity prediction model according to the candidate wave band to construct a plurality of instantiation turbidity prediction model subsets; Respectively carrying out model fitting on all models in the instantiation turbidity prediction model subset and the turbidity value to be fitted, selecting the instantiation turbidity prediction model with the highest fitting goodness and the smallest error from all second fitting results, and determining the corresponding remote sensing wave band combination as an optimized wave band combination; And taking the numerical value of the optimized wave band combination as an independent variable, respectively carrying out function fitting with a plurality of preset function forms, and selecting the function form with the highest fitting goodness and the smallest error and the corresponding coefficient from fitting results of all the function forms to form the target turbidity prediction model.
2. The method of claim 1, wherein predicting the real-time turbidity of the target sea area based on the real-time remote sensing data of the target sea area and the target turbidity prediction model comprises: Selecting a subarea in which the optical thickness distribution of the aerosol in the target sea area is consistent with the characteristic of the distribution of the aerosol during the sampling period of acquiring the turbidity value to be fitted; And extracting real-time remote sensing data of the subareas, and inputting the real-time remote sensing data into the target turbidity prediction model to obtain a real-time turbidity prediction result of the subareas.
3. The method of claim 1, wherein the screening the plurality of target remote sensing bands based on the first fitting result of the single band remote sensing reflectivity data in the atmospheric corrected remote sensing data and the turbidity value to be fitted comprises: respectively calculating a correlation coefficient between each single-band remote sensing reflectivity data and the turbidity value to be fitted; calculating the average correlation coefficient of each single-band and the turbidity value to be fitted for all sampling points, and sequencing bands from high to low according to the average correlation coefficient; and selecting a single band which is ranked forward and has an average correlation coefficient larger than a preset threshold value as the target remote sensing band.
4. The method of claim 1, wherein the step of determining the position of the substrate comprises, Based on the wave band composition of the target turbidity prediction model, extracting the corresponding wave band reflectivity from the remote sensing data after the atmospheric correction and calculating a combined value; calculating a quadratic polynomial of the combined value, the quadratic polynomial comprising a quadratic term, a first order term and a constant term; performing power operation by taking 10 as a base number and taking the calculation result of the quadratic polynomial as an index; the result of the power operation is taken as the final turbidity prediction value.
5. The method according to claim 1, further comprising applying the target turbidity prediction model to remote sensing data of other satellites, in particular comprising: acquiring sensor characteristics of a target satellite and remote sensing data of the target satellite after atmospheric correction, wherein the sensor characteristics at least comprise a spectral response function; Correcting the function coefficient of the target turbidity prediction model according to the difference between the spectral response function of the target satellite and the spectral response function of the HY-1C/D CZI sensor; And applying the corrected target turbidity prediction model to the atmospheric corrected remote sensing data of the target satellite, and calculating the real-time turbidity of the target sea area.
6. The method of claim 1, wherein the single band remote sensing reflectance data varies with aerosol concentration, and wherein atmospheric corrected remote sensing data of a specified type of weather and aerosol optical thickness below a preset threshold is selected as the real-time remote sensing data.
7. The turbidity extraction device based on the remote sensing data is characterized by comprising an acquisition module, a screening module, a determination module and a prediction module; The acquisition module is used for acquiring remote sensing data after atmospheric correction of the target sea area and turbidity values to be fitted of a plurality of sampling points; The screening module is used for screening a plurality of target remote sensing bands according to the single-band remote sensing reflectivity data in the atmospheric corrected remote sensing data and the first fitting result of the turbidity value to be fitted; The determining module is used for determining a turbidity prediction model sample according to the combination of any two target remote sensing wave bands and the second fitting result of the turbidity value to be fitted; The determining module is further used for modifying the wave band composition in the turbidity prediction model sample, and determining a target turbidity prediction model according to the modified instantiation turbidity prediction model and the second fitting result of the turbidity value to be fitted; The prediction module is used for predicting the real-time turbidity of the target sea area according to the real-time remote sensing data of the target sea area and the target turbidity prediction model; And determining a turbidity prediction model sample according to the combination of any two target remote sensing bands and the second fitting result of the turbidity value to be fitted, wherein the method comprises the following steps: Pairing the screened target remote sensing wave bands in pairs; For each pair of wave bands, constructing a plurality of mathematical relations to form a plurality of candidate wave band combinations, wherein each mathematical form is used for carrying out numerical operation on two target remote sensing wave bands in each pair of wave bands; performing model fitting on each candidate wave band combination and the turbidity value to be fitted; Comparing the goodness of fit and error indexes of all candidate wave band combinations, and selecting the mathematical relationship corresponding to the candidate wave band combination with the highest goodness of fit and the smallest error index as the turbidity prediction model sample; The turbidity prediction model sample is a ratio, the modifying the wave band composition in the turbidity prediction model sample, and determining a target turbidity prediction model according to the modified instantiated turbidity prediction model and the second fitting result of the turbidity value to be fitted comprises the following steps: Determining a candidate wave band according to the first fitting result, and respectively adjusting the numerator, denominator and additional terms of the turbidity prediction model according to the candidate wave band to construct a plurality of instantiation turbidity prediction model subsets; Respectively carrying out model fitting on all models in the instantiation turbidity prediction model subset and the turbidity value to be fitted, selecting the instantiation turbidity prediction model with the highest fitting goodness and the smallest error from all second fitting results, and determining the corresponding remote sensing wave band combination as an optimized wave band combination; And taking the numerical value of the optimized wave band combination as an independent variable, respectively carrying out function fitting with a plurality of preset function forms, and selecting the function form with the highest fitting goodness and the smallest error and the corresponding coefficient from fitting results of all the function forms to form the target turbidity prediction model.

Description

Turbidity extraction method and device based on remote sensing data Technical Field The application relates to the technical field of satellite remote sensing water quality monitoring, in particular to a turbidity extraction method and device based on remote sensing data. Background The sea water turbidity is a key optical parameter for representing the light scattering and absorption effects of suspended particles in a water body, and the quantitative inversion of the sea water turbidity is of great significance to sea environment monitoring, estuary sediment transportation research, aquaculture area planning and water body eutrophication evaluation. Traditionally, turbidity monitoring mainly relies on ship navigation or site measurement of fixed-point stations, but the method is high in accuracy, time-consuming, labor-consuming, high in cost, and difficult to realize large-scale and synchronous sea area monitoring, especially in coastal areas with changeable environments, and the time-space representativeness is limited. The satellite remote sensing technology has the advantages of wide coverage, short revisit period, traceable historical data and the like, and provides an effective means for large-scale and dynamic monitoring of sea turbidity. At present, the turbidity inversion method based on remote sensing data is mostly a semi-analysis method based on a water body biological optical model or an empirical algorithm based on statistical regression, and the semi-analysis method based on the water body biological optical model realizes parameter inversion by carrying out physical modeling on absorption and backward scattering characteristics of each component of the water body, but has the characteristics of complex model, high requirement on input parameter precision, insufficient stability in II water bodies (especially offshore areas) with complex optical characteristics, and direct establishment of mathematical relations between remote sensing reflectivity or derivative quantity thereof and measured turbidity values based on the empirical algorithm of statistical regression, and has the characteristics of simple form and high calculation efficiency, so the method is widely applied in practice. However, the existing turbidity remote sensing inversion method based on statistical regression still faces a plurality of problems when applied to offshore areas. First, the accuracy of the single-band inversion model is generally inadequate. The optical signal of the near-shore water body is a mixed signal under the combined action of a plurality of components such as phytoplankton, non-algae particles, colored dissolved organic matters and the like, and the spectrum characteristics of turbidity change are difficult to effectively capture only by means of a single wave band, so that the model has poor robustness and limited applicability in different areas or different seasons. Secondly, the influence of atmospheric interference factors such as aerosol is great. The type of aerosol above the coastal sea is complex and spatially diverse, and its scattering contribution is an important component of the remote sensing signal. Although the L2A-level data has been Rayleigh-scattering corrected, the residual aerosol scattering effect, especially in the blue band, can still severely interfere with the extraction of the water-leaving radiation signal, thereby introducing significant turbidity inversion errors. Finally, the suitability of existing empirical models is limited. Most turbidity inversion models are constructed based on data of specific sensors in specific areas and time periods, and the band combination and model coefficients of the turbidity inversion models have strong scene dependence. When it is directly applied to different sensors or different empty scenes of the same sensor, the model performance tends to drop sharply due to the difference of the spectral response functions of the sensors and the change of the regional water color and the atmospheric conditions, and the necessary generalization capability is lacking. Therefore, a turbidity inversion method capable of realizing high precision and high stability under the offshore complex optics and the atmospheric environment needs to be studied, and the turbidity inversion method has good adaptability of crossing sensors and crossing time and space. Disclosure of Invention In view of the above, the application provides a turbidity extraction method and device based on remote sensing data, which can solve the problems of insufficient accuracy of single-band inversion in offshore areas, large influence of interference factors such as aerosol and the like, and limited suitability of the existing model. Specifically, the application is realized by the following technical scheme: the first aspect of the application provides a turbidity extraction method based on remote sensing data, which comprises the following steps: Acquiring remote sensing data afte