Water bloom monitoring method of XGB inversion model based on unmanned aerial vehicle hyperspectral data and satellite remote sensing data

Abstract

The invention discloses a water bloom monitoring method of an XGB inversion model based on unmanned aerial vehicle hyperspectral data and satellite remote sensing data, and the method comprises the following operation steps: 1, extracting a water body region without flare interference as an analysis object in an unmanned aerial vehicle hyperspectral image; 2, calculating the median of the DN values of each column of pixels along the longitudinal direction of the strip, and carrying out normalization correction on the DN values of other columns by taking the median DN value of the middle column as a reference standard; step 3, on the basis of the step 2, eliminating the radiation attenuation effect of the edge of the strip by using a median alignment mode, and improving the overall brightness consistency and splicing smoothness of the image; compared with the prior art, the method has the advantages that the stability and the precision of inversion are effectively improved by optimizing a waveband selection and feature input scheme and through model hyper-parameter tuning and cross validation.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of water bloom monitoring, in particular to a water bloom monitoring method based on an XGB inversion model of unmanned aerial vehicle hyperspectral data and satellite remote sensing data. BACKGROUND

[0002] Water bloom is a common eutrophication phenomenon in lakes, reservoirs and nearshore sea areas, and its main feature is the abnormal proliferation and aggregation of phytoplankton (especially blue-green algae) in water. Water bloom outbreak not only leads to the decline of dissolved oxygen in water and the imbalance of the ecosystem, but also threatens the safety of drinking water and fishery resources. Chlorophyll a (Chl-a) as a key optical parameter representing algal biomass and the degree of water bloom occurrence is an important indicator for water bloom monitoring and early warning.

[0003] Traditional chlorophyll a monitoring mainly relies on manual sampling and laboratory analysis, which has high accuracy but is limited in time and space, and cannot meet the real-time dynamic monitoring needs of large-scale water bodies. With the development of remote sensing technology, the use of hyperspectral sensors carried by unmanned aerial vehicles to realize rapid and continuous observation of the optical characteristics of water bodies has become an important direction of chlorophyll a inversion research. Hyperspectral remote sensing data has high spectral resolution and rich spectral information, which can capture the absorption and reflection characteristics of chlorophyll a in the range of 400-900 nm. However, the spectral signal of water body is complex and is significantly disturbed by suspended solids, yellow substances (CDOM) and water depth, etc., making the inversion relationship of chlorophyll a concentration nonlinear and multidimensional coupling. Therefore, a single band or a simple empirical model cannot accurately describe the complex relationship between spectrum and chlorophyll a. SUMMARY

[0004] The technical problem to be solved by the present application is to overcome the above technical defects.

[0005] In order to solve the above problems, the technical scheme of the present application is as follows: a water bloom monitoring method based on an XGB inversion model of unmanned aerial vehicle hyperspectral data and satellite remote sensing data, comprising the following operation steps:

[0006] Step one: in the unmanned aerial vehicle hyperspectral image, extract the water area without flare interference as the analysis object;

[0007] Step two: calculate the median of the DN values of each column of pixels along the longitudinal direction of the strip, and take the median DN value of the middle column as the reference standard to normalize and correct the DN values of other columns;

[0008] Step three: based on the basis of step two, eliminate the radiation attenuation effect of the strip edge by using the median alignment method to improve the overall brightness consistency and splicing smoothness of the image;

[0009] Based on the steps one to three, the specific research area water body optical characteristics are used to construct a regional XGB chlorophyll a inversion model, the band selection and feature input scheme are optimized, the model hyperparameter tuning and cross validation are used to effectively improve the stability and precision of the inversion.

[0010] Further: the purpose of the step two is to make the statistical brightness level of each column consistent with the center column.

[0011] Further: in the step one, the main parameters of the analysis object are the absorption and reflection characteristics in the range of 400-900nm, including the influence of suspended solids, yellow substances (CDOM) and water depth.

[0012] Further: the method model in the steps one to two is a water bloom monitoring model for small inland water bodies.

[0013] Further: the main parameters of the analysis object in the step one are the absorption and reflection characteristics in the range of 400-900nm, including the influence of suspended solids, yellow substances (CDOM) and water depth.

[0014] Compared with the prior art, the present application has the following advantages:

[0015] The algorithm in the present application can significantly improve the splicing quality of the unmanned aerial vehicle hyperspectral image without losing the water body spectral feature information, and provides more stable and reliable image data basis for subsequent chlorophyll a concentration inversion. DETAILED DESCRIPTION

[0016] The specific embodiments of the present application will be further described below in combination with examples.

[0017] In order to make the content of the present application more easily understood, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the embodiments of the present application.

[0018] The water bloom monitoring method based on the XGB inversion model of the unmanned aerial vehicle hyperspectral data and satellite remote sensing data comprises the following operation steps:

[0019] Step one: in the unmanned aerial vehicle hyperspectral image, the water body region without flare interference is extracted as the analysis object; in the step one, the main parameters of the analysis object are the absorption and reflection characteristics in the range of 400-900nm, including the influence of suspended solids, yellow substances (CDOM) and water depth.

[0020] Step two: calculate the median of DN values of each column of pixels in the longitudinal direction of the strip, and take the median DN value of the middle column as the reference standard to normalize and correct the DN values of other columns; the purpose of using normalization correction in step two is to make the statistical brightness level of each column consistent with the center column, and the method model in steps one to two is a water bloom monitoring model for small inland water bodies.

[0021] Step three: based on step two, use the median alignment method to eliminate the radiation attenuation effect of the strip edge and improve the overall brightness consistency and splicing smoothness of the image.

[0022] Based on the above steps one to three, a regional XGB chlorophyll-a inversion model is constructed using the optical properties of the water body in the specific study area, the band selection and feature input scheme are optimized, and the stability and accuracy of the inversion are effectively improved through model hyperparameter tuning and cross-validation.

[0023] In specific use, a regional XGB chlorophyll-a inversion model is constructed for the optical properties of the water body in the specific study area. In the model establishment process, based on the measured sample data and the hyperspectral reflectance obtained by the unmanned aerial vehicle, combined with the typical spectral characteristics of the water body in the region, the band selection and feature input scheme are optimized. Through model hyperparameter tuning and cross-validation, the stability and accuracy of the inversion are effectively improved. The results show that the XGB model can fully capture the spectral-biochemical characteristic relationship of the water body in the study area, and is significantly better than the traditional linear model and the general machine learning algorithm, with high accuracy and good regional adaptability, providing reliable technical support for water bloom monitoring of small inland water bodies.

[0024] The above describes the present application and its embodiments, which are not limiting, and the actual structure is not limited thereto. In summary, if a person skilled in the art is inspired by it, without departing from the purpose of the present application, without creative design, similar structure and embodiments of the technical solution can be obtained, which should belong to the protection scope of the present application.

Claims

1. A method for monitoring algal blooms based on an XGB inversion model using UAV hyperspectral data and satellite remote sensing data, characterized in that: The following steps are included: Step 1: Extract water bodies free from flare interference from UAV hyperspectral imagery as the analysis target; Step 2: Calculate the median DN value of each column of pixels along the longitudinal direction of the strip, and use the median DN value of the middle column as a reference standard to normalize and correct the DN values ​​of other columns. Step 3: Based on Step 2, median alignment is used to eliminate the radiation attenuation effect at the edges of the stripes, thereby improving the overall brightness consistency and stitching smoothness of the image. Based on steps one through three above, a regionalized XGB chlorophyll a retrieval model is constructed using the specific optical characteristics of the water body in the study area. By optimizing the band selection and feature input scheme, and through model hyperparameter tuning and cross-validation, the stability and accuracy of the retrieval are effectively improved.

2. The method for monitoring algal blooms based on the XGB inversion model using UAV hyperspectral data and satellite remote sensing data as described in claim 1, characterized in that: The purpose of normalization correction in step two is to ensure that the statistical brightness level of each column is consistent with that of the center column.

3. The algal bloom monitoring method based on the XGB inversion model using UAV hyperspectral data and satellite remote sensing data as described in claim 1, characterized in that: The construction of the XGB chlorophyll a inversion model requires combining measured sample data with hyperspectral reflectance obtained by UAVs, and incorporating typical spectral characteristics of water bodies in the region.

4. The algal bloom monitoring method based on the XGB inversion model using UAV hyperspectral data and satellite remote sensing data as described in claim 1, characterized in that: The method model in steps one and two is a monitoring model for algal blooms in small inland water bodies.

5. The method for monitoring algal blooms based on the XGB inversion model using UAV hyperspectral data and satellite remote sensing data according to claim 1, characterized in that: The main parameters of the analysis object in step one are the absorption and reflection characteristics in the range of 400–900 nm, including suspended matter, yellow substance (CDOM) and water depth.