Radar floating ice drifting monitoring and forecasting method

Abstract

The invention provides a radar floating ice drifting monitoring and forecasting method, which comprises the following steps of S1, acquiring radar sea ice original images, and identifying the same piece of sea ice on the two radar sea ice images; s2, performing cross-correlation analysis; s3, error correction; and S4, actual measurement radar sea ice image observation. According to the radar floating ice drifting monitoring and forecasting method, multiple pieces of sea ice on a radar sea ice image can be matched and tracked, real-time measurement of sea ice movement in the whole radar scanning area can be achieved, two-dimensional movement distribution of an ice drifting field is obtained, and the measurement precision and reliability are high.

Description

Technical Field

[0001] The invention relates to the technical field of sea ice monitoring. Background Art

[0003] The formation of sea ice can start from any layer of seawater, even the seabed. Ice formed below the water surface is called underwater ice, also known as submerged ice, and ice adhering to the seabed is called anchor ice. Since the density of deep ice is smaller than that of seawater, when they grow to a certain extent, they will float to the sea surface from different depths, making the ice on the sea surface continue to thicken. The Bohai Sea is shallow, and the formation of sea ice from the sea surface to the seabed is almost simultaneous.

[0004] Sea ice can be divided into floating ice and fixed ice according to its existence form. In 1959, my country formulated the first "Seaside Ice Observation Specifications" and began to carry out comprehensive routine sea ice observation work on the coast. In the early 1970s, ground radar sea ice observation work and receiving sea ice images from polar-orbiting meteorological satellites were carried out; since 1986, aerial remote sensing observation of sea ice has been started, and in 1989, airborne synthetic aperture radar was used to observe sea ice to reduce the disasters caused by icebergs. In addition to the research on radar sea ice echoes in my country, we use the sea ice radar backscatter coefficient to represent the change over time, eliminate the influence of distance factor and area factor, and greatly improve the accuracy of radar sea ice identification and classification. At present, the accuracy and reliability of the measurement of the drift motion distribution of sea ice in the entire radar scanning area still need to be improved. Summary of the invention

[0005] In order to improve the accuracy and reliability of the measurement of the drift motion distribution of sea ice in the entire radar scanning area, the present invention provides a radar ice floe drift monitoring and forecasting method.

[0006] The technical solution adopted by the present invention to achieve the above-mentioned purpose is: a radar ice drift monitoring and forecasting method, comprising the following steps:

[0007] Step S1: Collect the original radar sea ice images and identify the same piece of sea ice on the two radar sea ice images:

[0008] t2=t1+Δt

[0009] The displacement of each piece of sea ice in the interval Δt is the distance between the initial position of the sea ice at time t1 and the position of the sea ice at time t2;

[0010] Step S2: Cross-correlation analysis,

[0011] The signal corresponding to the radar sea ice image at time t1 is expressed as p(x,y)=I(x,y)+n1(x,y). When Δt is small, the signal corresponding to time t2 is expressed as q(x,y)=I(x-Δx,y-Δy)+n2(x,y), where n1(x,y) and n2(x,y) are noise signals. Due to the presence of noise, Δx and Δy cannot be directly extracted from p(x,y) and q(x,y).

[0012] Calculate the cross-correlation function of p(x,y) and q(x,y) and obtain the following formula (1):

[0013] r pq (τ x ,τ y )=∫∫p(x,y)q(x-τ x ,y-τ y )dxdy

[0014] =∫∫I(x,y)I(x-Δx-τ x ,y-Δy-τ y )dxdy (1)

[0015] Where: τ x and τ y They are the pixel coordinates of the X-axis and Y-axis of the two-dimensional radar chart;

[0016] The premise for formula (1) to be valid is that the image signal I(x, y) is uncorrelated with the noise signals n1(x, y) and n2(x, y), and their cross-correlation function is 0.

[0017] The autocorrelation function of the signal I(x,y) is expressed as:

[0018] r II (τ x ,τ y )=∫∫I(x,y)I(x-τ x ,y-τ y )dxdy (2)

[0019] Therefore, formula (1) can be transformed into: pq (τ x ,τ y )=r II (τ x -Δx,τ y -Δy)(3)

[0020] The autocorrelation function is an even function and reaches its maximum value at the origin, which satisfies the inequality: II (0,0)≥r II (τ x ,τ y), combined with formula (3), we can get formula (4):

[0021] r pq (τ x ,τ y )≤r pq (Δx,Δy) (4)

[0022] According to the position of the maximum value of the cross-correlation function, the average relative displacement of the sea ice in the area, that is, the displacement of the sea ice between time t1 and t2, is determined. If the acquisition interval Δt=t2-t1 of the image pair is known, the average speed of the sea ice moving within Δt is calculated using the basic definition of speed. The two radar sea ice images are divided into uniform grids. By calculating the cross-correlation function between each grid pair, the average relative displacement of the corresponding grid center position can be determined using the position of the maximum value of the cross-correlation function, thereby obtaining the displacement distribution of the sea ice in the entire radar scanning area, and then calculating the drift motion distribution of floating ice in the area.

[0023] It also includes step S3: error correction,

[0024] The motion vector of sea ice at the geodetic coordinate (x, y) is expressed as The following complex function is constructed from the components of the motion vector in two directions:

[0025] w(x,y)=v x (x,y)+iv y (x,y) (7)

[0026] Perform a two-dimensional Fourier transform on w(x,y) to obtain its frequency domain function W(u,v):

[0027] W(u,v)=∫∫w(x,y)e -i(ux+vy) dxdy (8)

[0028] Perform low-pass filtering on W(u,v) to obtain Right now:

[0029]

[0030] Among them, B is the bandwidth used in filtering. After completing the low-pass filtering operation, Perform a two-dimensional inverse Fourier transform to obtain the modified two-dimensional complex function. Use formula (7) to reversely construct v from the real part of the modified two-dimensional complex function. x (x,y), construct v from the imaginary part y (x,y), and obtain the corrected two-dimensional motion distribution.

[0031] The method also includes step S4: measuring radar sea ice image observation.

[0032] The step 2 also includes the calculation of the cross-correlation function, and the method is as follows:

[0033] The two-dimensional fast Fourier transform is used to calculate the cross-correlation function. Let p(x,y), q(x,y) and r pq (τ x ,τ y ) are P(u,v), Q(u,v) and R(u,v), that is, formula (5) holds:

[0034]

[0035] The frequency domain expression corresponding to formula (1) is: R(u,v)=P * (u,v)·Q(u,v)(6)

[0036] Among them, P * (u,v) is the conjugate function of P(u,v),

[0037] Using the cross-correlation property of Fourier transform and the fast algorithm FFT, two Fourier transforms and one inverse Fourier transform are performed to calculate the cross-correlation function r pq (τ x ,τ y ).

[0038] The radar ice floe drift monitoring and forecasting method of the present invention can match and track multiple pieces of sea ice on the radar sea ice image, can realize real-time measurement of sea ice movement in the entire radar scanning area, and obtain the two-dimensional movement distribution of the ice drift field, with high measurement accuracy and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a schematic diagram of the measurement principle of radar ice drift monitoring.

[0040] Figure 2 This is a schematic diagram of sea ice movement monitored by radar ice drift.

[0041] Figure 3 The present invention is a flow chart of the radar ice floe drift monitoring and forecasting method.

[0042] Figure 4 This is the original radar sea ice image.

[0043] Figure 5 It is the radar ice drift vector distribution map. DETAILED DESCRIPTION

[0044] The radar ice floe drift monitoring and forecasting method of the present invention is as follows Figure 3As shown, the specific steps are as follows:

[0045] Step S1: Collect radar sea ice original images.

[0046] Ice drift measurement mechanism: Ice drift measurement uses two-dimensional cross-correlation analysis in the spatial domain, which quickly realizes the two-dimensional Fourier transform (2D-FFT) and its specific implementation of Fourier transform.

[0047] By using a matching algorithm to identify the same piece of sea ice on two radar sea ice images, the movement displacement of sea ice on the radar sea ice image can be measured. Considering the pixel size of the radar sea ice image and the scale ratio of the actual environment, the absolute value of the movement displacement of sea ice on the sea surface can be calculated. The acquisition interval of the two radar sea ice images is known, and the average speed of sea ice drift is calculated. Figure 1 This is a schematic diagram of the principle of measuring the motion of a single piece of sea ice. By using this method, matching and tracking multiple pieces of sea ice on the radar sea ice image can achieve real-time measurement of the sea ice motion in the entire radar scanning area and obtain the two-dimensional motion distribution of the ice drift field.

[0048]

[0049]

[0050] v x and v y Represents the speed of the X-axis and Y-axis respectively.

[0051] Figure 2 The diagram is a schematic diagram of the movement of sea ice in a certain area. The solid circle in the figure represents the initial position of the sea ice at time t1, the dotted circle represents the position at time t2 (t2 = t1 + Δt), and the arrows represent the displacement of each piece of sea ice within the Δt interval. The radar sea ice image is processed using the cross-correlation analysis method to obtain the corresponding results.

[0052] Step S2: Cross-correlation analysis and its fast calculation,

[0053] Combination Figure 1 and Figure 2 , if the signal corresponding to the radar sea ice image at time t1 is expressed as p(x,y)=I(x,y)+n1(x,y), when Δt is small, the signal corresponding to time t2 should be expressed as q(x,y)=I(x-Δx,y-Δy)+n2(x,y), where n1(x,y) and n2(x,y) are noise signals. Due to the presence of noise, Δx and Δy cannot be directly extracted from p(x,y) and q(x,y).

[0054] According to the definition of the cross-correlation function, the cross-correlation function of p(x, y) and q(x, y) is calculated, and the following formula (1) can be obtained:

[0055] r pq (τ x ,τ y )=∫∫p(x,y)q(x-τ x ,y-τ y )dxdy

[0056] =∫∫I(x,y)I(x-Δx-τ x ,y-Δy-τ y )dxdy (1)

[0057] Where: τ x and τ y They are the pixel coordinates of the X-axis and Y-axis of the two-dimensional radar chart.

[0058] The premise for formula (1) to be valid is that the image signal I(x, y) is uncorrelated with the noise signals n1(x, y) and n2(x, y), and their cross-correlation function is 0.

[0059] In addition, according to the definition of the autocorrelation function, the autocorrelation function of the signal I(x,y) is expressed as:

[0060] r II (τ x ,τ y )=∫∫I(x,y)I(x-τ x ,y-τ y )dxdy (2)

[0061] Therefore, formula (1) can be transformed into:

[0062] r pq (τ x ,τ y )=r II (τ x -Δx,τ y -Δy) (3)

[0063] The autocorrelation function is an even function and reaches its maximum value at the origin. [6、18、19] , that is, satisfying the inequality: r II (0,0)≥r II (τ x ,τ y ). Combining formula (3), we can get formula (4):

[0064] r pq (τ x ,τ y )≤r pq(Δx,Δy) (4)

[0065] The above cross-correlation-based ice drift measurement is to calculate the two-dimensional cross-correlation function of two radar sea ice images, and determine the average relative displacement of sea ice in the area, that is, the displacement of sea ice between time t1 and t2, according to the location of the maximum value of the cross-correlation function. If the acquisition interval of the image pair Δt=t2-t1 is known, the average speed of sea ice moving within Δt can be calculated using the basic definition of speed.

[0066] To improve the efficiency of cross-correlation analysis and shorten the calculation time, two-dimensional fast Fourier transform is used to calculate the cross-correlation function. pq (τ x ,τ y ) are P(u,v), Q(u,v) and R(u,v), that is, formula (5) holds:

[0067]

[0068] The frequency domain expression corresponding to formula (1) is:

[0069] R(u,v)=P * (u,v)·Q(u,v) (6)

[0070] Among them, P * (u,v) is the conjugate function of P(u,v).

[0071] In the specific implementation of cross-correlation analysis, the cross-correlation property of Fourier transform and the fast algorithm FFT are used. Only two Fourier transforms and one inverse Fourier transform are needed to calculate the cross-correlation function r. pq (τ x ,τ y ), which realizes the rapid calculation of the cross-correlation function. It avoids complex integral operations and greatly improves the calculation speed.

[0072] The two radar sea ice images are divided into uniform grids. The cross-correlation function between each grid pair is calculated. The position of the maximum value of the cross-correlation function can be used to determine the average relative displacement of the corresponding grid center position, thereby obtaining the displacement distribution of sea ice in the entire radar scanning area, and then calculating the drift motion distribution of floating ice in the area.

[0073] Step S3, error correction:

[0074] During the radar imaging process, the imaging quality of the radar system will be reduced due to various reasons such as sea breeze, wave action, changes in the state of the sea ice surface, and electronic noise interference, resulting in blanks or error vectors in the analysis results. The validity judgment of motion vectors, the repair of blank data, and the correction of error vectors are important factors affecting the accuracy and reliability of ice drift measurements.

[0075] By rationally combining the drift motion vectors of floating ice, a new two-dimensional complex function is cleverly constructed, and the error vector is corrected by low-pass filtering. The following is an effective method for correcting the error vector.

[0076] The motion vector of sea ice at the geodetic coordinate (x, y) is expressed as The following complex function is constructed from the components of the motion vector in two directions:

[0077] w(x,y)=v x (x,y)+iv y (x,y) (7)

[0078] Perform a two-dimensional Fourier transform on w(x,y) to obtain its frequency domain function W(u,v):

[0079] W(u,v)=∫∫w(x,y)e -i(ux+vy) dxdy (8)

[0080] Perform low-pass filtering on W(u,v) to obtain Right now:

[0081]

[0082] Among them, B is the bandwidth used in filtering, which can be appropriately selected according to actual conditions.

[0083] After completing the low-pass filtering operation, Perform a two-dimensional inverse Fourier transform to obtain the modified two-dimensional complex function. Using formula (7), the real part of the modified two-dimensional complex function is used to construct v x (x,y), construct v from the imaginary part y (x,y), and obtain the corrected two-dimensional motion distribution.

[0084] Step S4. Analysis of measured radar sea ice images,

[0085] From December 2001 to February 2023, a radar observation station was established on the JZ9-3 oil field (DRPW platform) in the Liaodong Bay ice zone. The ice-measuring radar was used to conduct long-term on-site real-time observations and complete the observation of drift ice around the platform from the initial ice period to the final ice period. Figure 4 and Figure 5Shown are the original image and vector distribution map from 09:59:07 to 10:02:58 on January 20, 2011.

[0086] The present invention is described by way of embodiments, and those skilled in the art will appreciate that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application belong to the protection scope of the present invention.

Claims

1. A radar ice floe drift monitoring and forecasting method, characterized in that: The following steps are involved: Step S1: Collect the original radar sea ice images and identify the same piece of sea ice on the two radar sea ice images: t2=t1+Δt The displacement of each piece of sea ice in the interval Δt is the distance between the initial position of the sea ice at time t1 and the position of the sea ice at time t2; Step S2: Cross-correlation analysis, The signal corresponding to the radar sea ice image at time t1 is expressed as p(x,y)=I(x,y)+n1(x,y). When Δt is small, the signal corresponding to time t2 is expressed as q(x,y)=I(x-Δx,y-Δy)+n2(x,y), where n1(x,y) and n2(x,y) are noise signals. Due to the presence of noise, Δx and Δy cannot be directly extracted from p(x,y) and q(x,y). Calculate the cross-correlation function of p(x,y) and q(x,y) and obtain the following formula (1): Where: τ x and τ y They are the pixel coordinates of the X-axis and Y-axis of the two-dimensional radar chart; The premise for formula (1) to be valid is that the image signal I(x, y) is uncorrelated with the noise signals n1(x, y) and n2(x, y), and their cross-correlation function is 0. The autocorrelation function of the signal I(x,y) is expressed as: r II (t x ,t y )=∫∫I(x,y)I(x-τ x ,y-t y )dxdy (2) Therefore, formula (1) can be transformed into: pq (τ x ,τ y )=r II (τ x -Δx,τ y -Δy) (3) The autocorrelation function is an even function and reaches its maximum value at the origin, which satisfies the inequality: II (0,0)≥r II (τ x ,τ y ), combined with formula (3), we can get formula (4): r pq (t x ,t y )≤r pq (Δx,Δy) (4) According to the position of the maximum value of the cross-correlation function, the average relative displacement of the sea ice in the area, that is, the displacement of the sea ice between time t1 and t2, is determined. If the acquisition interval Δt=t2-t1 of the image pair is known, the average speed of the sea ice moving within Δt is calculated using the basic definition of speed. The two radar sea ice images are divided into uniform grids. By calculating the cross-correlation function between each grid pair, the average relative displacement of the corresponding grid center position can be determined using the position of the maximum value of the cross-correlation function, thereby obtaining the displacement distribution of the sea ice in the entire radar scanning area, and then calculating the drift motion distribution of floating ice in the area.

2. The radar ice floe drift monitoring and forecasting method according to claim 1, characterized in that: It also includes step S3: error correction, The motion vector of sea ice at the geodetic coordinate (x, y) is expressed as The following complex function is constructed from the components of the motion vector in two directions: w(x,y)=v x (x,y)+iv y (x,y) (7) Perform a two-dimensional Fourier transform on w(x,y) to obtain its frequency domain function W(u,v): W(u,v)=∫∫w(x,y)e -i(ux+vy) dxdy (8) Perform low-pass filtering on W(u,v) to obtain Right now: Among them, B is the bandwidth used in filtering. After completing the low-pass filtering operation, Perform a two-dimensional inverse Fourier transform to obtain the modified two-dimensional complex function. Use formula (7) to reversely construct v from the real part of the modified two-dimensional complex function. x (x,y), construct v from the imaginary part y (x,y), and obtain the corrected two-dimensional motion distribution.

3. The radar ice floe drift monitoring and forecasting method according to claim 1, characterized in that: The method also includes step S4: measuring radar sea ice image observation.

4. The radar ice floe drift monitoring and forecasting method according to claim 1, characterized in that: The step 2 also includes the calculation of the cross-correlation function, and the method is as follows: The two-dimensional fast Fourier transform is used to calculate the cross-correlation function. Let p(x,y), q(x,y) and r pq (τ x ,τ y ) are P(u,v), Q(u,v) and R(u,v), that is, formula (5) holds: The frequency domain expression corresponding to formula (1) is: R(u,v)=P * (u,v)·Q(u,v)(6) Among them, P * (u,v) is the conjugate function of P(u,v), Using the cross-correlation property of Fourier transform and the fast algorithm FFT, two Fourier transforms and one inverse Fourier transform are performed to calculate the cross-correlation function r pq (τ x ,τ y ).

Images