Remote sensing mapping method for concentration of suspended particulate matters in water body of Songhua river

Through the SR data and random forest algorithm of Landsat series satellite sensors, a model of suspended particulate matter concentration inversion of the Songhua River water body was constructed, which solved the problem of monitoring the concentration of suspended matter in the river, and achieved high-precision space-time mapping to support water quality and ecological assessment.

CN119023623APending Publication Date: 2024-11-26NORTHEAST INST OF GEOGRAPHY & AGRIECOLOGY C A S
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411125308.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-16
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The prior art is difficult to effectively monitor and evaluate the dynamic changes in the concentration of suspended particulate matter in rivers, affecting water quality and water ecological environment, and lacking high-precision spatial mapping methods.

Method used

The SR data set of Landsat series satellite sensors is used to construct a suspended particle concentration inversion model of the Songhua River water body through a random forest algorithm, and the surface reflectivity data of different bands are used to map the suspended material concentration at different time and space scales.

Benefits of technology

The accurate estimation of the concentration of suspended particulate matter in the Songhua River river water body and the mapping of the spatial and temporal changes are achieved, and the river water quality and soil erosion are effectively evaluated, and ecological restoration performance evaluation is supported.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119023623A_ABST
    Figure CN119023623A_ABST
Patent Text Reader

Abstract

The invention relates to a remote-sensing mapping method for the concentration of suspended particulate matters in a water body of a Songhua river, in particular to a remote-sensing mapping method for the concentration of suspended particulate matters in a water body. The method comprises the following steps: 1, acquiring a target satellite scene; 2, using Landsat series satellite sensor SR surface reflectance data of different wavebands as random forest input variables; and 3, outputting the river water body suspended matter concentration by adopting a random forest algorithm, and realizing Songhua river water body suspended matter concentration mapping with different spatial and temporal scales. By adopting the method disclosed by the invention, remote sensing mapping of the concentration of the suspended particulate matters in the water body of the Songhua river can be realized, the model can accurately estimate the concentration of the suspended particulate matters in the water body of the Songhua river through verification of an actual measurement sample point model, and the river water quality condition, the drainage basin water and soil loss condition and the ecological restoration performance can be effectively evaluated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a remote sensing mapping method for the concentration of suspended particles in a river water body. Background Art

[0002] Suspended matter concentration (TSM) refers to the content of solid matter suspended in water bodies. It is one of the important parameters for describing the optical properties of water bodies, the degree of water pollution, and other water ecological environment evaluations. In water bodies with high suspended matter concentrations, solar radiation passing through the water body is refracted and absorbed by suspended particles, affecting the underwater transmittance of solar radiation, which in turn affects the productivity of phytoplankton, causing turbid water quality, and causing adverse effects on aquatic organisms and human drinking water safety. The concentration of suspended matter is affected by many factors, such as the size, shape, and material of the particles. Mastering the dynamic changes in the concentration of suspended matter in rivers is very necessary for understanding the changes in the degree of soil erosion in the basin and evaluating the impact of water conservancy and hydropower projects on river water quality. Summary of the invention

[0003] The invention provides a remote sensing mapping method for the concentration of suspended particles in the water body of the Songhua River.

[0004] The method for remote sensing mapping of suspended particle concentration in Songhua River water body of the present invention is carried out according to the following steps:

[0005] 1. Obtain the target satellite scene, i.e. the Landsat series satellite sensor SR data set, which is the surface reflectance data after atmospheric correction;

[0006] Second, the SR surface reflectance data of the Landsat series satellite sensors in different bands are used as random forest input variables;

[0007] The SR surface reflectance of the Landsat series satellite sensors in different bands include:

[0008] Band2, Band3, Band4, Band5, Band6, Band7, Band2+Band4, Band4+Band7 and Band5-Band7;

[0009] 3. The random forest algorithm is used to output the suspended matter concentration in the river water, so as to realize the mapping of the suspended matter concentration in the Songhua River water at different time and space scales.

[0010] Further, in step 2, Band2 represents the surface reflectance of Band 2 of the Landsat satellite TM / ETM+ / OLI sensor;

[0011] Band3 represents the surface reflectance of Band 3 of the Landsat satellite TM / ETM+ / OLI sensor;

[0012] Band4 represents the surface reflectance of Band 4 of the Landsat satellite TM / ETM+ / OLI sensor;

[0013] Band5 represents the surface reflectance of Band 5 of the Landsat satellite TM / ETM+ / OLI sensor;

[0014] Band6 represents the surface reflectance of Band 6 of the Landsat satellite TM / ETM+ / OLI sensor;

[0015] Band7 represents the surface reflectance of Band 7 of the Landsat satellite TM / ETM+ / OLI sensor;

[0016] Band2+Band4 represents the sum of the surface reflectance of Band 2 and Band 4 of the Landsat satellite TM / ETM+ / OLI sensor;

[0017] Band4+Band7 represents the difference between the surface reflectance of Band 4 and Band 7 of the Landsat satellite TM / ETM+ / OLI sensor.

[0018] Furthermore, the input variables of the random forest algorithm described in step 2 are based on the inversion model of suspended particle concentration in the Songhua River water body. The construction process of the inversion model of suspended particle concentration in the Songhua River water body is as follows:

[0019] a. Collect the measured data of suspended matter concentration of water samples with a surface depth of 0-0.5m in Songhua River. The suspended matter concentration is determined by weight method. Each sample point is repeated three times, and the average value is recorded and calculated as the measured value of the point, in mg / L;

[0020] b. Obtain the SR surface reflectance data of the Landsat series satellite sensors corresponding to the geographic coordinates of the sampling points of the measured suspended matter concentration data. The remote sensing data acquisition method uses ±7 days as the time window and 3 pixels as the space window to screen the effective matching data; quickly screen out the effective matching data based on the binary quality assessment data, and screen out the remote sensing reflectance data without cloud cover and mountain shadows that matches the measured suspended matter concentration data, and obtain the remote sensing reflectance of Landsat series satellite sensors in different bands;

[0021] c. Based on the measured suspended matter concentration data and the band reflectance and band reflectance combination of Landsat series satellite sensors that effectively match the measured suspended matter concentration data, an inversion model for suspended particulate matter concentration in the Songhua River water body is established:

[0022] Randomly select the measured suspended matter concentration data and the matching Landsat series satellite sensor band reflectivity as the model construction correction data set (N=2 / 3 data), analyze the linear correlation between the remote sensing reflectivity of different bands and the measured suspended matter concentration data, select the bands and combinations with the Pearson correlation coefficient range of [-0.4, 0.4], and the confidence level is 0.01 on both sides; use the remaining data set as the validation data set (N=1 / 3 data), test the linear correlation between different combinations and the measured suspended matter concentration, and use the relevant bands and band combinations as random forest input variables to establish the inversion model of suspended particle concentration in the Songhua River water body;

[0023] The linear correlation analysis was implemented in SPSS 16.0 software platform, and the Pearson correlation coefficient r was expressed as follows:

[0024]

[0025] Among them, x i is the measured value of suspended solids concentration, y i is the reflectance of the Landsat series satellite sensor band that effectively matches the measured suspended matter concentration, and N is the number of samples.

[0026] Further, the random forest input variables are verified for accuracy: the measured suspended matter concentration data remaining after the above selection and the Landsat series satellite sensor band reflectance data are used to verify the accuracy of the constructed Songhua River water body suspended particle concentration inversion model;

[0027] The expression for the accuracy verification is as follows:

[0028]

[0029]

[0030] Where RMSE is the root mean square error, MAE is the mean absolute percentage error, N is the number of samples, and y i is the measured value of suspended solids concentration, y i ' is the estimated value of suspended matter concentration.

[0031] The present invention is based on the remote sensing mapping method of suspended particle concentration in the Songhua River water body, obtains Landsat series satellite remote sensing SR reflectance products on geographic computing cloud platforms such as GEE, builds a random forest model through independent programming, inputs the input variables determined by the present invention, and can estimate the suspended particle concentration in the Songhua River water body, and perform spatial mapping at different time and space scales.

[0032] The method of the present invention can realize remote sensing mapping of suspended particle concentration in the water body of the Songhua River. Through verification of the measured sample point model, it is found that the model can accurately estimate the concentration of suspended particle in the water body of the Songhua River, and can effectively evaluate the water quality of the river, the soil erosion situation in the basin and the performance of ecological restoration. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The distribution map of the sampling points of the Songhua River water body in Example 1;

[0034] Figure 2 It is the calibration map of the remote sensing inversion model of Huajiang River in Example 1;

[0035] Figure 3 It is a verification diagram of the remote sensing inversion model of Huajiang River in Example 1;

[0036] Figure 4 is a flow chart of Example 1; DETAILED DESCRIPTION

[0037] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0039] Specific implementation method 1: The remote sensing mapping method of the Songhua River water body suspended particle concentration in this implementation method is carried out according to the following steps:

[0040] 1. Obtain the target satellite scene, i.e. the Landsat series satellite sensor SR data set, which is the surface reflectance data after atmospheric correction;

[0041] Second, the SR surface reflectance data of the Landsat series satellite sensors in different bands are used as random forest input variables;

[0042] The SR surface reflectance of the Landsat series satellite sensors in different bands include:

[0043] Band2, Band3, Band4, Band5, Band6, Band7, Band2+Band4, Band4+Band7 and Band5-Band7;

[0044] 3. The random forest algorithm is used to output the suspended matter concentration in the river water, so as to realize the mapping of the suspended matter concentration in the Songhua River water at different time and space scales.

[0045] The input variables of the random forest algorithm in step 2 of this embodiment are based on the inversion model of the suspended particle concentration in the Songhua River water body. The construction process of the inversion model of the suspended particle concentration in the Songhua River water body is as follows:

[0046] a. Collect the measured data of suspended matter concentration in water samples with a surface depth of 0-0.5m in Songhua River, Figure 1 This is the distribution map of sampling points in the Songhua River. The suspended matter concentration was determined by the weight method. Each sampling point was repeated three times, and the average value was recorded and calculated as the measured value of the point. The unit is mg / L, a total of 378;

[0047] b. Obtain the SR surface reflectance data of the Landsat series satellite sensors corresponding to the geographic coordinates of the sampling points of the measured suspended matter concentration data. The remote sensing data acquisition method uses ±7 days as the time window and 3 pixels as the space window to screen the effective matching data; quickly screen out the effective matching data based on the binary quality assessment data, and screen out the remote sensing reflectance data without cloud cover and mountain shadows that matches the measured suspended matter concentration data, and obtain the remote sensing reflectance of Landsat series satellite sensors in different bands;

[0048] c. Based on the measured suspended matter concentration data and the band reflectance and band reflectance combination of Landsat series satellite sensors that effectively match the measured suspended matter concentration data, an inversion model for suspended particulate matter concentration in the Songhua River water body is established:

[0049] Randomly select the measured suspended matter concentration data and the matching Landsat series satellite sensor band reflectivity as the model construction correction data set (N=2 / 3 data), analyze the linear correlation between the remote sensing reflectivity of different bands and the measured suspended matter concentration data, select the bands and combinations with the Pearson correlation coefficient range of [-0.4, 0.4], and the confidence level is 0.01 on both sides; use the remaining data set as the validation data set (N=1 / 3 data), test the linear correlation between different combinations and the measured suspended matter concentration, and use the relevant bands and band combinations as random forest input variables to establish the inversion model of suspended particle concentration in the Songhua River water body. Figure 2 This is the correction map of the remote sensing inversion model of the Songhua River;

[0050] The linear correlation analysis was implemented in SPSS 16.0 software platform, and the Pearson correlation coefficient r was expressed as follows:

[0051]

[0052] Among them, x iis the measured value of suspended solids concentration, y i is the reflectance of the Landsat series satellite sensor band that effectively matches the measured suspended matter concentration, and N is the number of samples.

[0053] The random forest input variables described in this embodiment are verified for accuracy: the measured suspended matter concentration data remaining after the above selection and the Landsat series satellite sensor band reflectance data are used to verify the accuracy of the constructed Songhua River water body suspended particle concentration inversion model;

[0054] The expression for the accuracy verification is as follows:

[0055]

[0056]

[0057] Where RMSE is the root mean square error, MAE is the mean absolute percentage error, N is the number of samples, and y i is the measured value of suspended solids concentration, y i ' is the estimated value of suspended matter concentration.

[0058] This implementation is based on the remote sensing mapping method of suspended particle concentration in the Songhua River water body. The Landsat series satellite remote sensing SR reflectance products are obtained on geographic computing cloud platforms such as GEE. A random forest model is built through autonomous programming, and the input variables determined by the present invention are input. The suspended particle concentration in the Songhua River water body can be estimated, and spatial mapping at different time and space scales can be performed.

[0059] Specific implementation method 2: The difference between this implementation method and specific implementation method 1 is that: in step 2, Band2 represents the surface reflectivity of band 2 of the Landsat satellite TM / ETM+ / OLI sensor;

[0060] Band3 represents the surface reflectance of Band 3 of the Landsat satellite TM / ETM+ / OLI sensor;

[0061] Band4 represents the surface reflectance of Band 4 of the Landsat satellite TM / ETM+ / OLI sensor;

[0062] Band5 represents the surface reflectance of Band 5 of the Landsat satellite TM / ETM+ / OLI sensor;

[0063] Band6 represents the surface reflectance of Band 6 of the Landsat satellite TM / ETM+ / OLI sensor;

[0064] Band7 represents the surface reflectance of Band 7 of the Landsat satellite TM / ETM+ / OLI sensor;

[0065] Band2+Band4 represents the sum of the surface reflectance of Band 2 and Band 4 of the Landsat satellite TM / ETM+ / OLI sensor;

[0066] Band4+Band7 represents the difference between the surface reflectance of Band 4 and Band 7 of the Landsat satellite TM / ETM+ / OLI sensor. Other steps and parameters are the same as those in the first embodiment.

[0067] Example 1: Mapping the suspended matter concentration in the Songhua River using the method of the present invention

[0068] Here’s how:

[0069] 1. Obtain the target satellite scene, such as the Landsat series satellite sensor SR data set of the Songhua River water body, which is the surface reflectance data after atmospheric correction;

[0070] Second, the SR surface reflectance data of the Landsat series satellite sensors in different bands are used as random forest input variables;

[0071] The SR surface reflectance of the Landsat series satellite sensors in different bands include:

[0072] Band2, Band3, Band4, Band5, Band6, Band7, Band2+Band4, Band4+Band7 and Band5-Band7;

[0073] Band2 represents the surface reflectance of Band 2 of the Landsat satellite TM / ETM+ / OLI sensor;

[0074] Band3 represents the surface reflectance of Band 3 of the Landsat satellite TM / ETM+ / OLI sensor;

[0075] Band4 represents the surface reflectance of Band 4 of the Landsat satellite TM / ETM+ / OLI sensor;

[0076] Band5 represents the surface reflectance of Band 5 of the Landsat satellite TM / ETM+ / OLI sensor;

[0077] Band6 represents the surface reflectance of Band 6 of the Landsat satellite TM / ETM+ / OLI sensor;

[0078] Band7 represents the surface reflectance of Band 7 of the Landsat satellite TM / ETM+ / OLI sensor;

[0079] Band2+Band4 represents the sum of the surface reflectance of Band 2 and Band 4 of the Landsat satellite TM / ETM+ / OLI sensor;

[0080] Band4+Band7 represents the difference between the surface reflectance of Band 4 and Band 7 of the Landsat satellite TM / ETM+ / OLI sensor;

[0081] 3. The random forest algorithm is used to output the suspended matter concentration in the river water, so as to realize the mapping of the suspended matter concentration in the Songhua River water at different time and space scales.

[0082] The random forest input variable described in step 2 of this embodiment, i.e., the SR surface reflectance of the Landsat series satellite sensors in different bands, is constructed based on the inversion model of the suspended particle concentration in the Songhua River water body. The construction process of the suspended particle concentration model in the Songhua River water body is as follows:

[0083] a. Collect the measured data of suspended matter concentration of 378 water samples in the surface layer (0-0.5m) of Songhua River. The suspended matter concentration determination method is repeated three times for each sample point, and the average value is recorded and calculated as the measured value of the point, in mg / L;

[0084] b. Obtain the SR surface reflectance data of the Landsat series satellite sensors corresponding to the measured suspended matter concentration data. The remote sensing data acquisition method uses ±7 days as the time window and 3 pixels as the space window to obtain the remote sensing reflectance of the Landsat series satellite sensors in different bands;

[0085] c. Based on the measured suspended matter concentration data and the reflectivity of the Landsat series satellite sensor bands that match the measured suspended matter concentration data,

[0086] e. The remaining (N = 1 / 3 data) measured suspended matter concentration data and the matching Landsat series satellite sensor band reflectivity are used as the model verification data set, and input into the random forest algorithm to verify the accuracy of the remote sensing inversion model of suspended particulate matter concentration in the Songhua River water body. Figure 3 This is a validation diagram of the remote sensing inversion model for the Songhua River.

[0087] Model testing data set linear fit R 2 =0.91, RMSE is 12.77 mg / L, MAE is 7.88 mg / L, which meets the internationally accepted remote sensing inversion accuracy requirements (<45%).

[0088] The expression for the accuracy verification is as follows:

[0089]

[0090]

[0091] Where RMSE is the root mean square error, MAE is the mean absolute percentage error, N is the number of samples, and y i is the measured value of suspended solids concentration, y i ' is the estimated value of suspended matter concentration.

[0092] The Landsat satellite SR remote sensing reflectance of different bands and their reflectance band combinations proposed in this embodiment are input as random forest algorithm input variables to estimate the concentration of suspended particles in the water body of the Songhua River, and the results are quickly mapped in a standardized manner to accurately obtain the temporal and spatial variation law of the concentration of suspended particles in the water body of the Songhua River. The flow chart of this embodiment is as follows Figure 4 shown.

[0093] In this embodiment, a total of 378 water samples covering the Songhua River water body and the suspended matter concentration measured in the synchronous laboratory were used. The random forest inversion model of suspended particle concentration in the Songhua River water body was trained in combination with the Landsat series satellite remote sensing SR reflectivity, which can better represent the complex and changing optical properties of the river water body and ensure that the developed model has good applicability.

[0094] The purpose of the present invention is to solve the technical problem of spatial mapping of suspended matter concentration in the water body of Songhua River, and to provide a method for mapping the suspended matter concentration in the water body of Songhua River.

Claims

1. A remote sensing mapping method for the concentration of suspended particles in the Songhua River water body, characterized in that The remote sensing mapping method for the concentration of suspended particulate matter in the Songhua River water body is carried out in the following steps:

1. Obtain the target satellite scene, i.e. the Landsat series satellite sensor SR data set, which is the surface reflectance data after atmospheric correction; Second, the SR surface reflectance data of the Landsat series satellite sensors in different bands are used as random forest input variables; The SR surface reflectance of the Landsat series satellite sensors in different bands include: Band2, Band3, Band4, Band5, Band6, Band7, Band2+Band4, Band4+Band7 and Band5-Band7; 3. The random forest algorithm is used to output the suspended matter concentration in the river water, so as to realize the mapping of the suspended matter concentration in the Songhua River water at different time and space scales.

2. The method for remote sensing mapping of suspended particle concentration in the Songhua River water body according to claim 1 is characterized by: Band2 in step 2 represents the surface reflectance of band 2 of the Landsat satellite TM / ETM+ / OLI sensor; Band3 represents the surface reflectance of Band 3 of the Landsat satellite TM / ETM+ / OLI sensor; Band4 represents the surface reflectance of Band 4 of the Landsat satellite TM / ETM+ / OLI sensor; Band5 represents the surface reflectance of Band 5 of the Landsat satellite TM / ETM+ / OLI sensor; Band6 represents the surface reflectance of Band 6 of the Landsat satellite TM / ETM+ / OLI sensor; Band7 represents the surface reflectance of Band 7 of the Landsat satellite TM / ETM+ / OLI sensor; Band2+Band4 represents the sum of the surface reflectance of Band 2 and Band 4 of the Landsat satellite TM / ETM+ / OLI sensor; Band4+Band7 represents the difference between the surface reflectance of Band 4 and Band 7 of the Landsat satellite TM / ETM+ / OLI sensor.