CN105300864A - Quantitative remote sensing method of suspended sediment - Google Patents

Abstract

The invention discloses a quantitative remote sensing method of suspended sediment. The quantitative remote sensing method comprises the following steps: step S1, carrying out field water body spectral measurement and suspended sediment content measurement; step S2, carrying out equivalent remote sensing reflectivity calculation; step S3, randomly dividing collected water samples into a modeling sample and a verification sample; step S4, establishing a sediment remote sensing quantitative model by the modeling sample; step S5, acquiring a satellite image and pre-processing the satellite image; and step S6, carrying out inversion to obtain the content of the suspended sediment in a research region. By establishing the sediment remote sensing quantitative model, the sediment remote sensing quantitative model is applied to a satellite remote sensing image through ENVI (The Environment for Visualizing Images) software and spatial and temporal distribution of the sediment in the research region can be automatically drawn; and the method is a scientific and efficient river-mouth off-shore sediment research method.

Description

A kind of suspension bed sediment quantitative remote sensing method

Technical field

The present invention relates to remote sensing technology field, be specifically related to a kind of quantitative inversion to estuary and immediate offshore area suspension bed sediment thereof and remote sensing monitoring.

Background technology

Suspension bed sediment is one of most important water quality parameter in water body, and the number of its content and the rule of sedimentary movement thereof directly affect improvement and the flood passage safety in river mouth.Conventional suspension bed sediment assay method is by collection in worksite water sample, then filters water sample, dries calculating concentration after weighing.This method is time-consuming, effort, and is difficult to the requirement meeting large area monitoring.Remote sensing technology, with the feature of its macroscopic view, large area monitoring, can provide the remote sensing image in synchronization waters on a large scale, in monitoring water environment, therefore have its distinctive advantage.

Current, there is the research much belonging to domestically leading level suspension remote sensing aspect, but majority is the remote sensing monitoring of coarse resolution, the nonlinear relationship also not unification of model, and part research is only the qualitative analysis to silt distribution and flow field; And all estuary offshores as a whole, that carries out unifying studying rarely has report especially.

Summary of the invention

For the above-mentioned defect of prior art, the object of the present invention is to provide a kind of suspension bed sediment quantitative remote sensing method, it is by setting up quantitative remote sensing model and by this quantitative remote sensing models applying on satellite remote-sensing image, getting final product the spatial and temporal distributions of autodraft survey region estuary and immediate offshore area silt thereof.

The technical scheme that technical solution problem of the present invention adopts is as follows:

A kind of suspension bed sediment quantitative remote sensing method, it comprises the following steps:

Step S1, according to the time of Landsat satellite sensor through research area, within this time, utilize field spectroradiometer synchronously to carry out on-the-spot water spectral measurement in the estuary offshore region of described research area, carry out water sampling to described region, the water sample of collection carries out Remote Sensing of Suspended Sediment Concentration measurement simultaneously;

Step S2, the on-the-spot water spectral data utilizing field spectroradiometer to measure, in conjunction with Landsat satellite sensor spectral response functions, by the method for convolution described on-the-spot water spectral data equivalence in the indigo plant, green, red, closely red four wave bands of Landsat satellite sensor:

$x_i=\frac{\underset{{\mathrm{\λ}}_i={\mathrm{\λ}}_{imin}}{\overset{{\mathrm{\λ}}_{imax}}{\Σ}}f\left({\mathrm{\λ}}_i\right)r\left({\mathrm{\λ}}_i\right)}{\underset{{\mathrm{\λ}}_i={\mathrm{\λ}}_{imin}}{\overset{{\mathrm{\λ}}_{i\max }}{\Σ}}f\left({\mathrm{\λ}}_i\right)}---\left(1\right)$

Wherein: x is equivalent Remote Sensing Reflectance, i=1,2,3,4 represent indigo plant, green, red, closely red four wave bands respectively, and λ is wavelength, and f (λ) is spectral response functions, and r (λ) is the on-the-spot water spectral data that field spectroradiometer is measured, λ _imin, λ _imaxbe respectively the wavelength X of i-th wave band _iminimum value and maximal value, the i.e. minimum value in Landsat satellite sensor i-th waveband channels interval and maximal value;

Step S3, random the water sample sample of collection is divided into modeling sample and checking sample;

Step S4, carried out the foundation of sediment remote sensing quantitative model by modeling sample, it comprises the following steps:

Step S41, set up sediment remote sensing quantitative model:

y＝kx+b(2)

Wherein, k and b is model parameter, and x is equivalent Remote Sensing Reflectance data, and y is suspension bed sediment data;

The Remote Sensing of Suspended Sediment Concentration data that step S42, the equivalent Remote Sensing Reflectance utilizing step S2 to record and step S1 record, carry out correlation analysis, obtain the related coefficient P1 of the indigo plant in modeling sample, green, red, closely red four wave bands and suspension bed sediment, P2, P3, P4 respectively:

$P_i=\frac{\mathrm{\Σ}\left(x_{i}y\right)-\frac{{\mathrm{\Σx}}_i\mathrm{\Σ}y}{N}}{\sqrt{({{\mathrm{\Σx}}_i}^2-\frac{{\left({\mathrm{\Σx}}_i\right)}^2}{N})(\sqrt{{\mathrm{\Σy}}^2-\frac{{\left(\mathrm{\Σ}y\right)}^2}{N})}})}---\left(3\right)$

Wherein, P _ifor the related coefficient of i-th wave band of modeling sample, x _iy is the product of the equivalent Remote Sensing Reflectance of i-th wave band of a certain sample in modeling sample and the suspension bed sediment data of this sample, ∑ x _ifor the equivalent Remote Sensing Reflectance sum of i-th wave band of samples all in modeling sample, ∑ y is the suspension bed sediment data sum of all samples in modeling sample, and N is modeling sample number;

The equivalent Remote Sensing Reflectance data chosen and Remote Sensing of Suspended Sediment Concentration data are solved the model parameter k in formula (2) and b by step S43, the equivalent Remote Sensing Reflectance data choosing a value correspondence of maximum absolute value in related coefficient P1, P2, P3, P4 and Remote Sensing of Suspended Sediment Concentration data;

Step S5, obtain satellite image in research area of Landsat satellite sensor and described satellite image is carried out pre-service in ENVI software;

Step S6, under ENVI software, utilize band math instrument that the sediment remote sensing quantitative model that step S4 sets up is applied on the satellite image after step S5 process, inverting obtains the Remote Sensing of Suspended Sediment Concentration of research area.

Also comprise between described step S4 and step S5 and carry out precision test by the formula (2) of checking sample to solving model parameter k and b, the method of checking is: to choose in checking sample in each sample with related coefficient P1, P2, P3, the equivalent Remote Sensing Reflectance data that in P4, a value of maximum absolute value is corresponding substitute into the suspension bed sediment analogue value of trying to achieve each checking sample in formula (2) as independent variable, based on the suspension bed sediment data of each checking sample obtained in the suspension bed sediment analogue value of each checking sample and step S1, the precision of sediment remote sensing quantitative model is characterized with root-mean-square error RMSE:

$RMSE=\sqrt{\frac{\underset{j=1}{\overset{n}{\Σ}}{(y_j-y_j^{\′})}^2}{n}}---\left(4\right)$

Wherein, y _jfor the suspension bed sediment data of a jth checking sample, y ' _jfor the suspension bed sediment analogue value of a jth checking sample, n is checking number of samples.

In described step S43, the Regression Function of the equivalent Remote Sensing Reflectance data chosen and Remote Sensing of Suspended Sediment Concentration data separate SPSS software is solved model parameter k and b.

Pre-service in described step S5 comprises radiation calibration, atmospheric correction, inlays, land and water is separated and cutting.

The invention has the beneficial effects as follows: suspension bed sediment quantitative remote sensing method of the present invention, because acquisition and the pre-service of satellite image data are determined very much, quantitative remote sensing model is set up, quantitative model need only be applied on satellite remote-sensing image by ENVI software, the just spatial and temporal distributions of energy autodraft research area silt is science, efficiently river mouth offshore silt research method.

Accompanying drawing explanation

Fig. 1 is research area and field experiment locus schematic diagram;

Fig. 2 be sediment remote sensing inverting quantitative model set up curve;

Fig. 3 is the checking curve of sediment remote sensing inverting quantitative model;

Fig. 4 river mouth offshore True color synthesis remote sensing image and pre-service;

Fig. 5 the mouth of the Zhujiang River, sharpening door, Korea Spro Jiangkou offshore suspension bed sediment remote-sensing inversion result.

Embodiment

Below in conjunction with the drawings and specific embodiments, content of the present invention is described in further details.

Embodiment:

1, research area and water Saudi Arabia levy:

In preferred embodiment of the present invention, research area is the main estuary in Guangdong Province and immediate offshore area thereof.It mainly comprises, the Zhujiang River: total length 2214 kilometers, drainage area 453690 sq-km, and annual runoff is billion cubic meter more than 3300, occupies the second of national rivers water system; Han Jiang: Han Jiang basin is positioned at East Guangdong, southwest Fujian, the second largest basin of Guangdong Province except Pearl River Delta, drainage area 30112 sq-km, Han Jiang for many years mean sediment runoff is 693.22 ten thousand tons, Long-term Average Sediment Transport Modulus is 212.68 tons every square kilometre, Han Jiang mean annual sediment content 0.258 kilogram every cubic metre; Sharpening door: sharpening door is positioned at flood gulf, Zhuhai City, Guangdong Province people from enterprise stone, and be the main sea gate of Xijiang River runoff, annual runoff 923 billion cubic meter, accounts for 28.3% of Zhujiang River sea of faces yielding flow, annual sediment discharge 2314 ten thousand tons, account for that the Zhujiang River enters extra large total discharge of sediment 33%; Moyangjiang River: Moyangjiang River is positioned at the west and south, Guangdong Province, the basin total area 6091 square kilometres, long 199 kilometers of river, run-off is 82.1 billion cubic meters, and average water total resources is 86.5 billion cubic meters for many years; Mirror river: water system river, the South Sea, coastal maximum river, west of Guangdong Province Yuexi, catchment area is 9464 square kilometres, long 232 kilometers of river; Nine river, continents: river, nine divisions of China in remote antiquity, Beibu Bay of The South China Sea water system, total length 162 kilometers, total drainage area 3337 square kilometres.Accompanying drawing 1 is seen in study area.

2, synchronous (accurate synchronous) water body experiment

Utilize ASD field spectroradiometer, according to mensuration on the water surface, in Guangdong Province, the measurement of on-the-spot water spectral and water body sampling are carried out in main estuary offshore region, and water body experimental point is shown in Fig. 1.

3, suspension bed sediment laboratory measurement

The water sample gathered loads in water sample bottle, avoids sunlight to irradiate and within 24 hours, send laboratory back to and carries out Remote Sensing of Suspended Sediment Concentration mensuration, and measuring method is filtration, oven drying method.

4, in situ optic measurement satellite sensor Equivalent Calculation

The on-the-spot water spectral data utilizing ASD field spectroradiometer to measure, in conjunction with Landsat satellite sensor spectral response functions, spectroscopic data ASD measured by the method for convolution is equivalent in the indigo plant, green, red, closely red four wave bands of Landsat satellite sensor, and computing formula is as follows:

$x_i=\frac{\underset{{\mathrm{\λ}}_i={\mathrm{\λ}}_{imin}}{\overset{{\mathrm{\λ}}_{imax}}{\Σ}}f\left({\mathrm{\λ}}_i\right)r\left({\mathrm{\λ}}_i\right)}{\underset{{\mathrm{\λ}}_i={\mathrm{\λ}}_{imin}}{\overset{{\mathrm{\λ}}_{i\max }}{\Σ}}f\left({\mathrm{\λ}}_i\right)}---\left(5\right)$

In formula (5), x is equivalent Remote Sensing Reflectance, i=1,2,3,4 represent indigo plant, green, red, closely red four wave bands respectively, and λ is wavelength, f (λ) is spectral response functions, and r (λ) is the on-the-spot water spectral data that field spectroradiometer is measured, λ _imin, λ _imaxbe respectively the wavelength X of i-th wave band _iminimum value and maximal value, the i.e. minimum value in Landsat satellite sensor i-th waveband channels interval and maximal value.

5, suspension bed sediment remote sensing model is set up and checking

According to the water sample total number of samples (40 samples) gathered, Stochastic choice 25 composition of sample modeling samples, carry out the foundation of sediment remote sensing quantitative model, and 15 remaining composition of sample checking samples, for the precision test of sediment remote sensing quantitative model.

The concrete grammar of modeling is:

5.1, sediment remote sensing quantitative model is set up:

y＝kx+b(6)

Wherein, k and b is model parameter, and x is equivalent Remote Sensing Reflectance data, and y is suspension bed sediment data;

5.2, the Remote Sensing of Suspended Sediment Concentration data recorded by Remote Sensing of Suspended Sediment Concentration measurement in the equivalent Remote Sensing Reflectance that utilizes formula (5) to calculate (only using the equivalent Remote Sensing Reflectance of each wave band in checking sample here) and step 1 (in like manner, here the Remote Sensing of Suspended Sediment Concentration data in checking sample are only used), carry out correlation analysis, obtain the related coefficient P1 of the indigo plant in modeling sample, green, red, closely red four wave bands and suspension bed sediment, P2, P3, P4 respectively:

$P_i=\frac{\mathrm{\Σ}\left(x_{i}y\right)-\frac{{\mathrm{\Σx}}_i\mathrm{\Σ}y}{N}}{\sqrt{({{\mathrm{\Σx}}_i}^2-\frac{{\left({\mathrm{\Σx}}_i\right)}^2}{N})(\sqrt{{\mathrm{\Σy}}^2-\frac{{\left(\mathrm{\Σ}y\right)}^2}{N})}})}---\left(7\right)$

In formula (7), P _ifor the related coefficient of i-th wave band of modeling sample, x _iy is the product of the equivalent Remote Sensing Reflectance of i-th wave band of a certain sample in modeling sample and the suspension bed sediment data of this sample, ∑ x _ifor the equivalent Remote Sensing Reflectance sum of i-th wave band of samples all in modeling sample, ∑ y is the suspension bed sediment data sum of all samples in modeling sample, and N is modeling sample number.

5.3, choose equivalent Remote Sensing Reflectance data corresponding to a value of maximum absolute value in related coefficient P1 that formula (7) calculates, P2, P3, P4 and Remote Sensing of Suspended Sediment Concentration data, the Regression Function of the equivalent Remote Sensing Reflectance data chosen and Remote Sensing of Suspended Sediment Concentration data separate SPSS software is solved model parameter k and b.

In step 5.3, such as, when the maximum absolute value of P2, the equivalent Remote Sensing Reflectance data chosen are the x in all checking samples ₂value (i.e. the equivalent Remote Sensing Reflectance of green wave band), Remote Sensing of Suspended Sediment Concentration data are the Remote Sensing of Suspended Sediment Concentration value (laboratory records) of each checking sample, then the Regression Function of SPSS software is utilized to solve model parameter k and b, can certainly be realized by graphing method (x-axis is equivalent Remote Sensing Reflectance data, and y-axis is Remote Sensing of Suspended Sediment Concentration data).Finally try to achieve parameter k=1740.3, b=-32.55, establish suspension bed sediment remote sensing model as shown in Figure 2, the expression formula of this model this be written as:

y＝1740.3x-32.55(8)。

5.4, carry out precision test by checking sample to formula (8), the method for checking is: choose and verify that in sample, in each sample, the equivalent Remote Sensing Reflectance data corresponding with a value (such as P2 is maximum) of maximum absolute value in related coefficient P1, P2, P3, P4 (then correspond to each x verified in sample ₂value) substitute into the suspension bed sediment analogue value of trying to achieve each checking sample in formula (8) as independent variable, based on the suspension bed sediment data of each checking sample obtained in the suspension bed sediment analogue value of each checking sample and step 3, characterize the precision of sediment remote sensing quantitative model with root-mean-square error RMSE:

$RMSE=\sqrt{\frac{\underset{j=1}{\overset{n}{\Σ}}{(y_j-y_j^{\′})}^2}{n}}---\left(9\right)$

In formula (9), y _jfor the suspension bed sediment data of a jth checking sample, y ' _jfor the suspension bed sediment analogue value of a jth checking sample, n is checking number of samples, and 1≤j≤n, the result as shown in Figure 3.

6, Landsat satellite image obtains and pre-service

Satellite remote sensing date is adopted to be that Landsat satellite sensor data planned by the Landsat (Landsat) of the NASA provided by United States Geological Survey (UnitedStatesGeologicalSurvey) herein.In order to monitor the secular variation of river mouth offshore suspension bed sediment, research selection 1988,2013 Korea Spro Jiangkou data, 1987,2013 thes mouth of the Zhujiang River, sharpening gated data, with cloudless or that cloud is few data for general data.The Landsat satellite sensor number that research uses is as shown in table 1.

The Landsat satellite sensor data list that table 1 research uses

Data prediction: the four scape satellite remote sensing date quality that research adopts are higher, in the cloudless covering in offshore region, river mouth.Radiation calibration, atmospheric correction have been carried out successively to data, have inlayed, land and water be separated and cutting, after satellite remote-sensing image pre-service, see Fig. 4.

7, satellite image sediment remote sensing inverting

Under ENVI software, utilize " BandMath (band math) " instrument that formula (8) is applied on the satellite image after step 6 process, inverting obtains the Remote Sensing of Suspended Sediment Concentration of main estuary.

Korea Spro Jiangkou, the mouth of the Zhujiang River, sharpening door suspension bed sediment remote-sensing inversion the results are shown in Figure 5; The spatial and temporal distributions of the mouth of the Zhujiang River, sharpening door, Korea Spro Jiangkou offshore suspension bed sediment can be analyzed according to remote-sensing inversion result.

Should be understood that, application of the present invention is not limited to above-mentioned citing, for those of ordinary skills, can be improved according to the above description or convert, and all these improve and convert the protection domain that all should belong to claims of the present invention.

Claims (4)

1. a suspension bed sediment quantitative remote sensing method, is characterized in that, it comprises the following steps:

Step S1, according to the time of Landsat satellite sensor through research area, within this time, utilize field spectroradiometer synchronously to carry out on-the-spot water spectral measurement in the estuary offshore region of described research area, carry out water sampling to described region, the water sample of collection carries out Remote Sensing of Suspended Sediment Concentration measurement simultaneously;

Step S2, the on-the-spot water spectral data utilizing field spectroradiometer to measure, in conjunction with Landsat satellite sensor spectral response functions, by the method for convolution described on-the-spot water spectral data equivalence in the indigo plant, green, red, closely red four wave bands of Landsat satellite sensor:

$x_i=\frac{\underset{{\mathrm{\λ}}_i={\mathrm{\λ}}_{imin}}{\overset{{\mathrm{\λ}}_{imax}}{\Σ}}f\left({\mathrm{\λ}}_i\right)r\left({\mathrm{\λ}}_i\right)}{\underset{{\mathrm{\λ}}_i={\mathrm{\λ}}_{imin}}{\overset{{\mathrm{\λ}}_{i\max }}{\Σ}}f\left({\mathrm{\λ}}_i\right)}---\left(1\right)$

Wherein: x is equivalent Remote Sensing Reflectance, i=1,2,3,4 represent indigo plant, green, red, closely red four wave bands respectively, and λ is wavelength, and f (λ) is spectral response functions, and r (λ) is the on-the-spot water spectral data that field spectroradiometer is measured, λ _imin, λ _imaxbe respectively the wavelength X of i-th wave band _iminimum value and maximal value, the i.e. minimum value in Landsat satellite sensor i-th waveband channels interval and maximal value;

Step S3, random the water sample sample of collection is divided into modeling sample and checking sample;

Step S4, carried out the foundation of sediment remote sensing quantitative model by modeling sample, it comprises the following steps:

Step S41, set up sediment remote sensing quantitative model:

y＝kx+b(2)

Wherein, k and b is model parameter, and x is equivalent Remote Sensing Reflectance data, and y is suspension bed sediment data;

The Remote Sensing of Suspended Sediment Concentration data that step S42, the equivalent Remote Sensing Reflectance utilizing step S2 to record and step S1 record, carry out correlation analysis, obtain the related coefficient P1 of the indigo plant in modeling sample, green, red, closely red four wave bands and suspension bed sediment, P2, P3, P4 respectively:

$P_i=\frac{\mathrm{\Σ}\left(x_{i}y\right)-\frac{{\mathrm{\Σx}}_i\mathrm{\Σ}y}{N}}{\sqrt{({{\mathrm{\Σx}}_i}^2-\frac{{\left({\mathrm{\Σx}}_i\right)}^2}{N})(\sqrt{{\mathrm{\Σy}}^2-\frac{{\left(\mathrm{\Σ}y\right)}^2}{N})}})}---\left(3\right)$

Wherein, P _ifor the related coefficient of i-th wave band of modeling sample, x _iy is the product of the equivalent Remote Sensing Reflectance of i-th wave band of a certain sample in modeling sample and the suspension bed sediment data of this sample, ∑ x _ifor the equivalent Remote Sensing Reflectance sum of i-th wave band of samples all in modeling sample, ∑ y is the suspension bed sediment data sum of all samples in modeling sample, and N is modeling sample number;

The equivalent Remote Sensing Reflectance data chosen and Remote Sensing of Suspended Sediment Concentration data are solved the model parameter k in formula (2) and b by step S43, the equivalent Remote Sensing Reflectance data choosing a value correspondence of maximum absolute value in related coefficient P1, P2, P3, P4 and Remote Sensing of Suspended Sediment Concentration data;

Step S5, obtain satellite image in research area of Landsat satellite sensor and described satellite image is carried out pre-service in ENVI software;

Step S6, under ENVI software, utilize band math instrument that the sediment remote sensing quantitative model that step S4 sets up is applied on the satellite image after step S5 process, inverting obtains the Remote Sensing of Suspended Sediment Concentration of research area.

2. suspension bed sediment quantitative remote sensing method according to claim 1, it is characterized in that, also comprise between described step S4 and step S5 and carry out precision test by the formula (2) of checking sample to solving model parameter k and b, the method of checking is: to choose in checking sample in each sample with related coefficient P1, P2, P3, the equivalent Remote Sensing Reflectance data that in P4, a value of maximum absolute value is corresponding substitute into the suspension bed sediment analogue value of trying to achieve each checking sample in formula (2) as independent variable, based on the suspension bed sediment data of each checking sample obtained in the suspension bed sediment analogue value of each checking sample and step S1, the precision of sediment remote sensing quantitative model is characterized with root-mean-square error RMSE:

$RMSE=\sqrt{\frac{\underset{j=1}{\overset{n}{\Σ}}{(y_j-y_j^{\′})}^2}{n}}---\left(4\right)$

Wherein, y _jfor the suspension bed sediment data of a jth checking sample, y ' _jfor the suspension bed sediment analogue value of a jth checking sample, n is checking number of samples.

3. suspension bed sediment quantitative remote sensing method according to claim 1, it is characterized in that, in described step S43, the Regression Function of the equivalent Remote Sensing Reflectance data chosen and Remote Sensing of Suspended Sediment Concentration data separate SPSS software is solved model parameter k and b.

4. suspension bed sediment quantitative remote sensing method according to claim 1, is characterized in that, the pre-service in described step S5 comprises radiation calibration, atmospheric correction, inlays, land and water is separated and cutting.