CN116070735A

CN116070735A - Yellow sea green tide distribution area based on side length and azimuth difference rule and drift prediction initial field manufacturing method thereof

Info

Publication number: CN116070735A
Application number: CN202211544270.XA
Authority: CN
Inventors: 丁一; 高松; 黄娟; 高宁; 辛蕾; 王宁; 靳熙芳; 王炜荔
Priority date: 2022-12-01
Filing date: 2022-12-01
Publication date: 2023-05-05

Abstract

A yellow sea green tide distribution area based on side length and azimuth difference rule and a drift prediction initial field manufacturing method thereof utilize manpower to select vertexes at distribution range boundaries as prediction initial fields, and establish vertex extraction rules based on side length and azimuth difference constraint by utilizing the initial field analysis.

Description

Yellow sea green tide distribution area based on side length and azimuth difference rule and drift prediction initial field manufacturing method thereof

Technical Field

The application relates to the technical field of model design and prediction, in particular to a yellow sea green tide distribution area based on side length and azimuth difference rules and a drift prediction initial field manufacturing method thereof.

Background

The green tide distribution area information and drift trend are the main reference basis for green tide interception net layout and the scheduling of salvaging ships.

CN201911249340.7 discloses a green tide biomass forecasting method, device, equipment and medium. The method comprises the following steps: determining an initially constructed green tide biomass estimation model; wherein the green tide biomass estimation model comprises at least one undetermined parameter; and determining the value of at least one undetermined parameter included in the initially constructed green tide biomass estimation model according to the reference distribution data of the green tide biomass in the target area so as to obtain the green tide biomass estimation model after the determination, wherein the reference distribution data is determined according to the satellite remote sensing image.

CN202210253275.0 discloses a method for predicting medium-long term trend of yellow sea green tide, which comprises the following steps: a, determining a Huang Hailu tide research area (33-37 DEG N,119-123 DEG E), dividing the research area into two key areas for green tide generation and development by taking 35 DEG N as a boundary, and determining main factors affecting the growth and drift of the green tide as meteorological factors and ocean factors; b, acquiring green tide multisource monitoring data and meteorological and ocean element observation data of a key area; c, analyzing early weather influence factors and sea influence factors of three indexes of green tide satellite discovery time, green tide main body drifting direction and green tide maximum distribution area, and respectively establishing a prediction model; d, obtaining a meteorological element value and a marine element value of the early period of the green tide of the year to be predicted, and carrying out medium-long term trend prediction on indexes such as the satellite discovery time, the main body drift direction, the maximum distribution area and the like of the year Huang Hailu tide according to the prediction model established in the step c to obtain a prediction result.

Satellite remote sensing has the advantages of being instantaneous and large in range, is a main data source for yellow sea green tide monitoring, and is the only means for acquiring complete and comprehensive green tide information. Huang Hailu tide disaster response centering is to monitor green tide coverage information by using satellite images, manufacture a distribution area by using the green tide coverage information, and then use the top point of the distribution area as an initial field to predict the drift of green tide distribution.

The green tide distribution is defined as the outer envelope surface of the enteromorpha covered region, and two existing distribution region acquisition methods exist: the first is manual sketching, and satellite remote sensing monitoring personnel sketch a distribution surface along the coverage outer edge based on green tide coverage information. The manual sketching has the advantages that: the quantity and the size of the distributed polygon vertexes are moderate, the distributed polygon vertexes can be directly used as an initial field, the calculation time consumption of drift prediction is short, and the defects are that: the subjectivity is strong, no unified standard exists, and different monitoring staff sketch different distribution surfaces. The second method is a buffer area method, which is generated by using a buffer tool in an ArcGIS toolbox, and is characterized in that a coverage area is buffered outwards for a certain distance to form an envelope surface as a distribution area, the number of vertexes of the distribution area generated by the method is huge, often from tens of thousands to hundreds of thousands, sparse processing is needed, the existing sparse method of ArcGIS is to extract vertexes at equal point intervals, when the number of the extracted vertexes is equal to the number of manually sketched vertexes, the deformation of a distribution polygon is larger, and certain green coverage points are omitted, so that manual correction is needed. Therefore, the buffer area method is used for extracting the distributed vertexes, and has the advantages of automatically forming uniform and standard distributed areas, and having the defects of huge quantity of the distributed areas vertexes and no standard and scientific method for pumping.

Considering that the existing two methods have advantages and disadvantages in the distribution area and the initial field manufacture, the method combines the advantages of the ArcGIS buffer area analysis and the manual vertex selection, and establishes an automatic vertex sparseness method. According to the vertex sparsification method, the sparse points can fully keep the distributed shape, the number of points is reduced to be equivalent to the number of manually selected points, the drifting prediction calculated amount is greatly reduced, and the operation efficiency is improved.

Disclosure of Invention

Aiming at the problems in the prior art, the application provides a yellow sea green tide distribution area and a drift prediction initial field manufacturing method based on side length and azimuth difference sparse rules. The method for acquiring the initial field has three advantages: firstly, the number is few, and the prediction time is reduced, secondly, the fewer number can fully keep the shape of the distribution surface, thirdly, the method can automatically obtain the standard unified result, reduce the manpower, save the time and improve the emergency level. The method of the present application may serve business or emergency monitoring.

A yellow sea green tide distribution area and a drift prediction initial field manufacturing method based on side length and azimuth difference sparse rules comprise the following steps:

step 1, calculating a normalized vegetation index NDVI by using a satellite image, extracting green tide coverage points by using a threshold method,

the green tide disaster information extraction is completed semi-automatically by combining standard false color image B432 (R-nir G-rB-G) with normalized vegetation index NDVI (Normalized Difference Vegetation Index) threshold segmentation;

step 2, generating a distribution area by using a buffer space analysis method by using the coverage points, manufacturing a buffer area for the green tide coverage points with the radius of 1-5km, and combining the buffer areas to be used as the green tide distribution area;

step 3, sparse the distribution boundary by using vertex sparse rules based on constraint of side length-azimuth change between adjacent vertexes; considering that the reduced camber line distance can be taken to ensure the reduced fitness, the small change of the azimuth direction of the vertex is also beneficial to the maintenance of the fitness, the sample with the shortest camber line is changed by the same azimuth direction, and meanwhile, the azimuth direction change of the vertex is limited within 80 degrees, so that a polygonal simplified model is formed, and the formula is as follows:

wherein y represents the distance between adjacent vertexes, and x represents the tangential direction change of the adjacent vertexes;

and step 4, adding X, Y longitude and latitude coordinates to the sparse vertexes to form a drift prediction initial field.

Further, in step 1, images of the early green tide, the development period, the outbreak period and the decay period are acquired by using satellites.

Further, in step 1, the normalized vegetation index is based on unique spectral characteristics of the green algae red band and the near infrared band, and the formula is: ndvi= (R _nir -R _red )/(R _nir +R _red ) Wherein R is _nir 、R _red Is the reflectivity of near infrared and red wave bands;

further, in step 1, green tide information is extracted by using an NDVI threshold method aiming at a green tide area, image histogram analysis of the green tide area is carried out, an NDVI threshold is determined, and the threshold floats around 0; the NDVI threshold value is taken to be 0 under the normal condition, and the NDVI threshold value is finely adjusted under the condition of the interference of the cloud mist;

further, the degree of coincidence of the distributed polygons after vertex sparseness and the original distributed polygons is evaluated through total area error (overlap error) and algebraic area error (algebraic error). The total area error is defined as the difference of the reduced polygon and the original polygon divided by the original polygon area. Wherein the difference portion includes an increased area S _i And reducing the area S _d And (3) summing. Algebraic area error is defined as the difference between the area of the reduced polygon and the area of the original polygon divided by the area of the original polygon. The error calculation formula is as follows:

oe＝(S _i +S _d )/S ₀

ae＝(S-S ₀ )/S ₀

the method for acquiring the initial field has the following advantages:

first, the number of vertexes is small, and the prediction time is reduced.

Secondly, the two aspects of standardization and automation are considered, the advantages of the two methods of ArcGIS buffer analysis and manual vertex selection are combined, and an automatic vertex sparseness method is established.

Third, a smaller number can sufficiently maintain the shape of the distribution surface.

Fourth, by using the method, the standard unified result can be automatically obtained, the manpower is reduced, the time is saved, and the emergency level is improved.

The yellow sea green tide distribution area and the drift prediction initial field manufacturing method thereof can serve business or emergency monitoring.

Drawings

The foregoing and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, wherein:

FIG. 1 is a schematic diagram of distribution boundary vertex selection (A, small curvature boundary small curvature boundary, B, large curvature boundary large curvature boundary)

FIG. 2 is a schematic view of pentagons sparse to trilateral

FIG. 3 is a simplified distribution vertex and distribution overlay of 2021, 5, 17,6, 7, 9,8, 5, details are shown 8, 5, and the remaining dates are similar

FIG. 4 is a schematic diagram of the vertices and side lengths of a buffer analysis distribution polygon

FIG. 5 is a schematic diagram showing the result of manual selection of the distribution peaks of green tide for 5 months and 17 days

FIG. 6 is a schematic view of manually selecting the adjacent vertex azimuth difference and side length (side length)

FIG. 7 is a simplified result of a method, equidistant method, douglas-Peucker method of the present application for a distribution of 2021, 8, 5 days

FIG. 8 is a graph of algebraic area error of the method, equidistant method, and the Taglas-Preker method of the present application

FIG. 9 is a plot of overall area error for the method, equidistant method, and the Tiglas-Puck method of the present application

Detailed Description

It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application, and the terminology used is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments based on the present application.

Example 1

A yellow sea green tide distribution area based on side length and azimuth difference sparse rules and a drift prediction initial field manufacturing method thereof comprise the following steps:

for the data and method (data and method) section, satellite data is employed.

The HY-1D satellite emits in the year of 6 and 11 in 2020 to realize networking with the HY-1C double satellites, and can realize full coverage for 2 times in 3 days in the sea area where yellow sea green tide occurs.

The space resolution of the coastal zone imager carried by the HY-1C/D satellite is 50m, the width is 950km, and the coastal zone imager comprises four wave bands of blue (0.42 mu m-0.50 mu m), green (0.52 mu m-0.60 mu m), red (0.61 mu m-0.69 mu m) and near red (0.76 mu m-0.89 mu m), so that green tide information can be effectively monitored.

Has become one of the main sources of yellow sea green tide business/emergency monitoring since 2021.

The research area of the present application is the southern yellow Sea area (southwellow Sea). The sea area is full of north wind in winter and full of south wind in summer. The surface sea current of the sea area has the same mode as wind under the influence of wind, and is from north to south in summer and from north to south in winter. The western sea area provides excellent environment for the growth of Porphyra yezoensis by virtue of unique geological conditions and tide rules of the North-Suzhou radiation sand, so that the Porphyra yezoensis is a famous Porphyra yezoensis cultivation area. The laver raft frame provides conditions for the attachment growth of green algae, and after the laver is harvested, the green algae of the raft frame enters the sea to become a source of green algae, and floats to the north under the combined action of a wind flow field. The north shore of the research area is Shandong province, green tide drifts from south to north every year, and the end of 6 months reaches the sea area under jurisdiction from 7 months, so that the cultivation industry, the tourism industry, the port shipping and the marine ecological environment of the area are affected.

The present application selects images of early green tide, development stage, outbreak stage and decay stage 4 (see table below) for study.

TABLE 1 image information

Satellite/transmission	Imaging date and time	Monitoring elements	Use of the same
				HY-1D/C	2021-05-1712:42	Green tide	Model creation and evaluation
HY-1C/C	2021-06-0610:42	Green tide	Model application and evaluation
				HY-1C/C	2021-07-0910:41	Green tide	Model application and evaluation
HY-1C/C	2021-08-0510:39	Green tide	Model application and evaluation

Huang Hailu tide is similar to the spectrum of land green vegetation, and the vegetation index method adopted in the land vegetation information extraction is widely adopted in the green tide information extraction, wherein the normalized vegetation index NDVI (Normalized Difference Vegetation Index) is widely applied to the green tide information extraction.

For the green tide information extraction part, because the research area is positioned in the offshore area, the water body environment is complex, in order to avoid the enteromorpha information from leaking as much as possible and reduce the influence of human factors, the result is more approximate to a true value, and the enteromorpha disaster information extraction is completed semi-automatically by combining the standard false color image B432 (R-nir G-rB-G) with the normalized vegetation index NDVI (Normalized Difference Vegetation Index) threshold segmentation.

In the pseudo-color image, the large-scale algal bloom is displayed in red, the contrast ratio between the large-scale algal bloom and water is stronger than that between the true color image and the water, the visual interpretation of green tide is facilitated, and the normalized vegetation index is based on the unique spectral characteristics of green algae red wave bands and near infrared wave bands, and the formula is as follows: ndvi= (R _nir -R _red )/(R _nir +R _red ) Wherein R is _nir 、R _red Is the reflectivity of near infrared and red wave bands.

The normalized vegetation index can effectively enhance green tide information and improve the contrast ratio of green tide and water, so that the green tide area can be effectively identified by combining the green tide information and the water.

And then, extracting green tide information aiming at the green tide area by using an NDVI threshold method. The green tide information extraction is carried out aiming at the green tide area, so that cloud and fog interference information can be effectively removed, and the accuracy of the green tide information is improved. And carrying out green tide region image histogram analysis, and determining an NDVI threshold value, wherein the threshold value is floating near 0. The NDVI threshold is typically taken to be 0, and is fine-tuned in the presence of cloud-to-cloud interference.

Step 2, generating a distribution area by using a buffer space analysis method by using the coverage points;

the green tide distribution area is manufactured by using Buffer analysis tools in ArcToolboxes in ArcGIS desktop software based on green tide coverage points, and the Buffer radius is set to be 3km.

The number of polygon tops of the formed distribution area is often tens of thousands when the distribution area of the enteromorpha is smaller, and often tens of thousands when the distribution area of the enteromorpha is largest.

If the method is directly used as a distributed initial field to carry out drift prediction operation, the calculation amount is large, and the time is too long to meet the emergency requirement.

The sparse processing is needed, the number of the sparse vertexes is as small as possible, and the shape of the original polygon can be effectively maintained.

Step 3, sparse distribution boundaries are realized by using vertex sparse rules based on adjacent vertex side length-azimuth difference constraint;

when the vertexes are simplified and selected, two principles are followed, firstly, boundary arcs between two adjacent sample vertexes are as long as possible, the arc length can effectively reduce the number of the simplified vertexes, and secondly, the line segment formed by two adjacent vertexes has good coincidence degree with the simplified front boundary, so that the difference between the front and the rear of the distribution polygon simplification is small, and the high simplification precision is kept. Analysis shows that the two principles are contradictory, the excessive arc length of adjacent vertexes leads to excessive deformation after the distribution polygon is simplified, the boundary length of the adjacent vertexes with too severe shape is kept to be too small, the number of the vertexes of the polygon after the simplification is too large, and the calculated amount of drift prediction is correspondingly too large.

Attention is focused on regions of large curvature and small regions. In the boundary region with small curvature, the side length between sample points is as long as possible, so that the simplification rate is improved; in the boundary region with large curvature, the importance is attached to the degree of coincidence before and after simplification, and the simplification error is reduced. In fig. 1, a graph a is a region with small curvature, A0 is a selected sample point, A1 is a point to be selected, a side length between the points is longer, a boundary curvature in B is large, B0 is a selected point, and B1B2B3 is a point to be selected, where in order to keep the coincidence of a simplified rear boundary and a simplified front boundary, only a point B1 with a short distance can be selected.

It is also understood that emphasis is placed on relatively straight boundary and curved boundary vertex selection. For a boundary that is relatively straight (relatively straightboundary), the arc line between adjacent vertexes is longer, as in fig. 1A, A0 is a selected vertex, A1 is a vertex to be selected, an arc line A0A1 is longer, tangential directions of the two vertexes are almost the same, the change of the directions is small, and the coincidence degree of the line segment A0A1 and the arc line A0A1 is high. For a boundary (Boundarywith large curvature) with a large curvature, the arc distance between adjacent vertexes is smaller, and if the arc between vertexes is too long, the tangential azimuth angle change is large, so that the deformation is large. In fig. 1B, the vertexes B0, B1, B2, and B3 are selected as points to be selected, and B1 is selected, so that the arc line B0B1 is shorter, the angle α1 of the tangential angles of the two vertexes is smaller, the line segment B0B1 matches with the arc line B0B1 well, and B1 is a selectable point. And B2 is selected, the azimuth angle change is larger than alpha 2, the matching degree of the line segment B0B2 and the arc line B0B2 is poor, and the point B2 is unsuitable. B3 is selected, the azimuth change is larger, so that the selected vertexes form polygons which cannot contain algae coverage points, and B3 is not acceptable in vertex selection.

Therefore, the distributed polygon generated by the buffer analysis is manually selected to have the number of vertexes as small as possible, and the distributed shape can be effectively maintained. And then analyzing the distance between adjacent vertexes and the rule of the included angle between each inflection point and two adjacent points, and finally establishing the green tide distribution polygonal vertex suction model.

The main factors considered in the selection of artificial sample vertexes are the edge length between vertexes and the local curvature of the boundary, and two corresponding characteristic quantities are established for the edge length: the first is the side length (side length) between adjacent sample vertices, such as the side length between vertices in fig. 1, the size of the side length determines the size of the reduction rate. The second is the direction difference of adjacent sample vertexes, the feature can effectively represent the local curvature of the boundary, and can determine the coincidence degree of the simplified front and rear boundaries, wherein the direction of the sample vertexes is defined as the direction of the tangential line of the point to the direction of the next candidate vertex. As ray B in FIG. 1B ₀ b ₀ Is oriented in the direction B ₀ The dot orientation direction.

To evaluate how well the sparse vertices fit the original distribution polygons, the present application uses the total area error (overlap error), and algebraic area error (algebraic error) to evaluate the simplification. The small error indicates small change after the distribution polygon is simplified, good simplifying effect, large error and bad simplifying effect. The total area error is defined as the difference of the reduced polygon and the original polygon divided by the original polygon area. Wherein the difference portion includes an increased area S _i And reducing the area S _d And (3) summing. Algebraic area error is defined as the difference between the area of the reduced polygon and the area of the original polygon divided by the area of the original polygon. The error calculation formula is as follows:

oe＝ (S _i +S _d )/S ₀ (1)

ae＝ (S-S ₀ )/S ₀ (2)

S ₀ to simplify the front polygonal area, S is to simplify the rear polygonal area, S _i To increase the area of the region after simplification, S _d The area of the region is reduced for simplification.

Taking fig. 2 as an example, the pentagon ABCDE becomes triangle ACD after vertex sparseness, and the area S is increased here _i Is S _△ADE Reduce the area S _d Is S _△ABC Pentagonal area S _ABCDE Then the total area error oe is calculated as (S _△ABC +S _△ADE )/S _ABCDE . In an extreme case, if only 1 or 2 vertices remain after the thinning, the polygon vertices are lost after the thinning, and the area change rate oe is 1.

In order to evaluate whether the number of vertexes effectively improves the prediction calculation efficiency after the sparsification, the green tide drift prediction mode of the center service operation is utilized to calculate the operation time of the original initial field and the sparsification initial field, and the comparison analysis is carried out.

The green tide coverage area of the scene image is extracted by using an NDVI threshold method and combining with expert experience, and the distribution area of the green tide is obtained by using a buffer area space analysis method with the buffer radius r set to be 3km.

The green tide coverage and distribution at 5, 17,6, 7, 9,8, 5 days 2021 is shown in fig. 3. The coverage and distribution areas of the four date image extraction are respectively as follows: 5.8/12242km ² ，890/44954km ² ，1210/42501km ² ，57/8615km ² The number of vertices is 56763, 66439, 101881, 35510.

Taking a distribution area of 2021, 5 months and 17 days as an example, analyzing and buffering to analyze corner features of the distribution, calculating the side length and the vertex corresponding angle alpha of the distribution area, if the angle is larger than 180 degrees, converting the angle into 360 degrees to alpha, finding that the angles are larger than 175 degrees, focusing on the range of 178 degrees to 180 degrees, and the distance is in the range of 0-0.2km, wherein most of the angles are in the range of 0-0.1, as shown in fig. 4.

Second, a distribution polygon reduced model

1) Vertex samples and features

The distribution range polygon of 2021, 5 and 17 days is counted, and the number of vertexes is 56763. The vertices are manually selected, the shape of the polygon which is selected as few as possible but is originally distributed is kept, the side length between the vertices of the boundary with large curvature is as small as possible, and the side length between the vertices is increased by a proper amount on the boundary with small curvature. The number of the selected top points is 287, so that the number of boundary top points is greatly reduced, and the degree of coincidence between the polygon formed by the selected top points and the polygon in the distribution range is high. And calculating the azimuth difference and the side length of the adjacent selected vertexes according to the vertexes manually selected by the expert. FIG. 5 is a schematic diagram of manual selection of the peak of the green tide distribution on day 5 and day 17.

2) Polygonal simplified model based on azimuth difference and side length

As can be seen from fig. 6, by manually selecting vertices, the direction of the adjacent vertices changes, and the distribution rule of the radian length between two points can be summarized as follows:

1) The change of the azimuth direction between the adjacent vertexes is small, the arc line is longer, and the change of the azimuth direction is large, and the arc line is shorter. The arc length between adjacent vertices is almost larger than the buffer radius by 3km.

2) There is no strict correspondence between the azimuth change and the arc length between two adjacent points, but the minimum original side length corresponding to the azimuth change at 0-60 degrees can be expressed by a formula of y= -0.0333x+5.x is the adjacent selected vertex azimuth change angle, and y is the corresponding minimum original side length.

3) The distance between two adjacent vertexes with the azimuth change of more than 60 degrees is 3km-4km.

Based on the analysis of the three rules, the reduced fitness can be ensured by taking a small arc distance, the small change of the azimuth direction of the vertex is also beneficial to the maintenance of the fitness, and the sample with the shortest arc is used for changing the azimuth direction of the vertex, and meanwhile, the change of the azimuth direction of the vertex is limited within 80 degrees, so that a polygonal simplified model is formed. The formula is as follows:

wherein y represents the distance between adjacent vertexes, and the tangential direction difference of the adjacent vertexes changes.

For distribution and initial field production flow thereof

The distribution and initial field generation scheme of the present application is divided into three steps: firstly, extracting green tide coverage points, calculating a normalized vegetation index NDVI by using satellite images, and then extracting the green tide coverage points by using a threshold method; the second step is generation of a distribution area, wherein a coverage point is utilized to generate the distribution area by adopting buffer space analysis, the third step is generation of a drift prediction initial field, and a polygon simplifying model (azimuth difference-distance constraint model) researched by the application is utilized to simplify the distribution polygon to be used as the drift prediction initial field of the distribution area. In this step, the distribution areas of a scene image are often not one, so that each distribution area is simplified according to the order, one of the distribution areas is firstly selected with the vertex number of 1, then the subsequent points are judged in turn, the second vertex is selected according to the model rule, then the newly selected vertex is used as the current point, the third vertex is selected, and so on until the vertex of the distribution area is completely selected.

For application and precision assessment

The polygon simplifying model is applied to simplify four-scene image distribution polygons of 5 months 17 days, 6 months 6 days, 7 months 9 days, 8 months 5 days, and the result is shown in figure 3. In the figure, the black boundary is the buffer formation distribution, and the red boundary is the simplified result, so that the coincidence degree of the simplified boundary and the buffer analysis distribution boundary is high. The simplified vertices are 357, 391, 602, 225, respectively, and the reduction rates are 159, 170, 169, 158, respectively. The total error in the area error is 0.79%,0.25%,0.42%,0.65%; the algebraic error is-0.07%, 0.05%,0.13%, and-0.10%. Overall, the simplification rate is high and the error is small.

Comparing with the existing method, the method for simplifying the polygon with interval point taking method

The interval method (Vertical distance limit method), the track-bridge method (track-bridge), the Douglas-Peucker method (Douglas-Peucker), wherein the track-bridge method and the diaphragm method are not suitable for the boundary with small fluctuation of adjacent vertexes, and even if the given limit difference is small, all middle vertexes forming a large bending boundary can be deleted, so that serious distortion of the line element form is caused, and the buffer forming boundary has small fluctuation of adjacent vertexes, so that the track-bridge method and the diaphragm method are not suitable. For this purpose, the present application uses the method of the present application to conduct a comparative analysis with the spaced dot extraction method and the dawshare-pock method.

According to the simplification rate of the method with the same simplification result, the equidistant parameters and the simplified parameters of the Douglas-Preker method are adjusted, and the equidistant parameters of the 5.17,6.6,7.9,8.5 date are respectively as follows: 4.1km,4.22km,4.16km,4.19km, the parameters of the Ticaged Lawster method are respectively: 0.48km,0.46km,0.515km,0.48km. For the four-scene image distribution polygon simplification, the 8-month 5-day result is shown in fig. 7, and other date results are similar to the 8-month 5-day result and are not shown here. In fig. 7, the equidistant method simplified boundary is shown as blue, the matching degree with the black distribution polygon is not good enough, especially in the boundary part with large curvature, the douglas-pock method and the method of the application are respectively shown as red and black, and the matching degree is good before simplification.

The overall error and algebraic error of the simplified areas of the three methods are calculated as shown in fig. 8-9. The method is generally superior to the method and the two methods in terms of algebraic error, the method of the application is generally superior to an equidistant method, the method of the application is superior to the method and the two methods in terms of total error, the difference between the method and the method of the application is smaller, and the error of the equidistant method is larger.

For drift prediction analysis

Huang Juan and the like establish a green tide drift prediction model (Huang Juan and the like, 2011) based on a Lagrange particle tracking method for yellow sea green tide disasters, and integrate the model into a yellow sea green tide disaster emergency remote sensing monitoring and prediction early warning system (Cao Conghua and the like, 2017) for business application. To clarify the effect of the distribution reduction operation on the predicted time, the time taken by the system for 72 hours of drift prediction before and after distribution reduction was counted, respectively (see table 2). The time for distribution drift prediction after four-stage simplification is obviously reduced, particularly, the distribution of the most serious outbreak period of the green tide disaster in 7 months and 9 days is realized, the drift prediction time is saved by about 27 minutes, and the precious time is saved for the establishment of the green tide disaster interception and salvage scheme.

TABLE 2 green tide profile simplified front-to-back drift prediction run time

Therefore, the application of the interval model can be given, and the interval model is completely adequate for the extraction of the initial field.

Claims

1. A yellow sea green tide distribution area based on side length and azimuth difference rules and a drift prediction initial field manufacturing method thereof comprise the following steps:

the green tide disaster information extraction is completed semi-automatically by combining standard false color image B432 (R-nirG-rB-g) with normalized vegetation index NDVI (NormalizedDifferenceVegetationIndex) threshold segmentation;

step 3, sparse the distributed boundaries by using a distance-angle constraint-based vertex sparse rule (DACR, distance-angle constraintrule); considering that the reduced camber line distance can be taken to ensure the reduced fitness, the small change of the azimuth direction of the vertex is also beneficial to the maintenance of the fitness, the sample with the shortest camber line is changed by the same azimuth direction, and meanwhile, the azimuth direction change of the vertex is limited within 80 degrees, so that a polygonal simplified model is formed, and the formula is as follows:

wherein y represents the distance between adjacent vertexes, and the tangential direction difference of the adjacent vertexes is changed;

and step 4, adding XY longitude and latitude coordinates to the sparse vertexes to form a drift prediction initial field.

2. The method of claim 1, wherein in step 1, satellite is used to capture images of the early green tide, the development period, the outbreak period, and the decay period.

3. The method of claim 1, wherein in step 1, the normalized vegetation index is based on unique spectral characteristics of the green algae red band and the near infrared band, and the formula is: ndvi= (R _nir -R _red )/(R _nir +R _red ) Wherein R is _nir 、R _red Is the reflectivity of near infrared and red wave bands.

4. A method according to any one of claims 1 to 3, wherein in step 1, green tide information is extracted for a green tide region by using an NDVI threshold method, and a green tide region image histogram analysis is performed to determine an NDVI threshold value, the threshold value being floating around 0; the NDVI threshold is typically taken to be 0, and is fine-tuned in the presence of cloud-to-mist interference.