CN101644572A - Detection method of ocean eddy variation based on historical similarity cases - Google Patents

Abstract

The invention provides a detection technology of ocean eddy variation based on historical similarity cases, belonging to the information technical field. The method is mainly applied to quantitative variation detection of ocean eddy, and the implementation technical scheme thereof is as follows: establishing a historical case library combined with an eddy spatial-temporal characteristic relationship extracted by a rough set method on the basis of expression models of the ocean eddy cases, then calculating the similarity between the current cases and the historical cases to obtain the historical case which is most similar to the current cases, and finally detecting variation of the current eddy cases according to the situation of the historical cases. Compared with the eddy variation studied by the existing dynamic analysis method via ocean water masses and dynamics, the method is simpler and more flexible; and the historical case library can be dynamically updated with self-learning capability, thus being capable of quickly adapting to ocean eddy with complicated spatial-temporal characteristics so as to carry out more reasonable and more accurate variation detection.

Description

A kind of ocean eddy change detecting method based on the historical similarity case

Technical field

The present invention relates to a kind ofly utilize the historical similarity case that ocean eddy is changed to carry out quantitative detection method, to belong to areas of information technology.

Background technology

At present, different research groups have adopted several different methods research at the ocean eddy variation issue, summarize and get up to mainly contain two kinds of methods: the one, and utilize the ocean water body to study the static method of vortex indirectly; The 2nd, by the flow direction of research ocean current and the dynamics dynamic approach that flow velocity carries out the ocean eddy signature analysis.The latter mainly contains two kinds of approach again: first kind is directly according to survey data or remote-sensing inversion data, analyzes ocean eddy, ocean current in conjunction with conventional ocean analytical approach; Second kind is that ocean current and vortex are carried out numerical simulation.The former changes in the growth and decline by the research water body and its space distribution rule reflects the growth and decline and the attributive character of ocean eddy indirectly, concrete research and example can be found in following document, [1] Fan Liqun, Li Fengqi etc., 1989. the northern sea area water-mass analysis in the South Sea. Chinese Marine University's journal, (S1): 169-178; [2] Su Yusong .1996. Bohai Seas such as Li Fengqi, Huang, eastern sea-water type distribute and water system is divided. ocean journal, 18 (6): 1-6.The latter is subdivided into two class research approach: first kind of approach is directly according to survey data or remote-sensing inversion data, analyzes ocean eddy, ocean current in conjunction with conventional ocean analytical approach; Second kind of approach is that ocean current and vortex are carried out numerical simulation, such as to the abyssal circulation numerical simulation of Japanese Kuroshio, to the research in Kuroshio life-span, utilize the extended model of POM simulate summer East Sea circulation with relevant two vortexs etc., concrete research and example can be with reference to documents once: [3] GuoBing Huo, Tang Yuxiang, observation and analysis that the Lu Sai English .1995. East Sea in spring Kuroshio Frontal Eddy is revolved, the ocean journal, 17 (1): 13-23; [4] GuoBing Huo, the .1997. East Sea, Pueraria lobota people peak Kuroshio Frontal Eddy is spun on the effect in continental shelf water and the exchange of Kuroshio water. ocean journal, 1 (6): 1-11; [5] HOGAN P J, HURLBURT H is of different wind forcingon circulation in the Japan/East Sea.Proceddings of theCREAMS 99 InternationalSymposium.Fukuoka E.1999.Impact, Japan:Ky ū sh ū University, 124-127; [6] Li Rongfeng, Guo Dongjian, the numerical experiment of .1995. China sea in winter vortex and contrary wind ocean current is deposited in celebrating once. tropical ocean, 14 (2): 1-9; [7] Li Yanchu, the season in whirlpools, marine site, northeast, the .2003. South Sea such as Cai Wenli and year border change. tropical oceanography newspaper, 5 (22): 61～70; [8] Li Hui bird with red feathers, Zhao protects the numerical simulation of benevolence .2001. Bohai Sea, Huang, East Sea circulation in summer. Marine Sciences, 25 (1): 28-32; [9] blue strong, Hong Jieli, the seasonal variations feature in big of Lee .2006. THE WESTERN SOUTH CHINA SEA cold whirlpool in summer. Advances in Earth Science, 11 (21): 1145～1152.

Above-mentioned research never ipsilateral is carried out the simulation and the analysis of Changing Pattern and vortex parameter etc. to ocean eddy, has obtained a lot of valuable achievements.But these methods all also exist certain limitation separately.Such as numerical simulation, though the delta data of ocean eddy different time can be provided continuously, be subjected to effect of boundary conditions, different zones needs to transfer ginseng and operation again, and is more consuming time; Though the mathematical statistics computing method can provide certain regional spatial and temporal distributions and characteristics of motion to a certain extent, have no idea to realize the detection of specific vortex; Though and the quantitative detecting method of ocean remote sensing can go out ocean eddy space local feature and parameter information at certain data extract, can't analyze and detect the development trend of vortex.Therefore, the detection problem at ocean eddy changes only depends on above-mentioned a kind of method not detect exactly all sidedly, needs to seek a kind of new approaches that can comprehensive above-mentioned several different methods.And adopt historical similar cases to carry out the solution of current problem, have and simplify knowledge acquisition, improve problem detection efficient, improve and detect quality, carry out advantage such as method accumulation.Can not only be under the situation that can't understand the phenomenon pests occurrence rule, rely on abundant historical data to realize the change-detection of phenomenon, and can be integrated effectively about all in the past achievements in research of this phenomenon, organize the higher level detection of realization again by suitable case.Therefore, from the angle of methodology, adopt the historical similarity case ocean eddy to be changed to detect be a kind of problem oriented combined extraction method.

Summary of the invention

Technology of the present invention is dealt with problems: overcoming the deficiencies in the prior art, a kind of ocean eddy change detecting method based on the historical similarity case is provided, is that a kind of simply saving time can efficiently be realized the ocean eddy change detecting method.

Technical solution of the present invention: a kind of historical similarity case of utilizing is to the ocean eddy change detecting method, and step is as follows:

Step 1: set up ocean eddy change histories case and express model, case is at the problem in the real world, comprise problem characteristic and the result describes set, then case expression model is the abstract expression to real world problem characteristic and result, specifically refers to the record of the feature description and the result of variations of ocean eddy here;

Step 2: express on the model basis in described case, extract the ocean eddy case data according to original remote sensing image data, be divided into two kinds of forms of grid and vector, raster data is mainly as the case background picture, case is expressed with the planar data of vector, in order to building the storehouse, case data is attribute and a spatial data of describing ocean eddy;

Step 3: adopt rough set method to carry out the spatial relationship extraction that ocean eddy changes case;

Step 4: on the basis of step 2 and step 3, make up ocean eddy change histories case library, historical case is exactly to the record of problem takes place, comprise that problem characteristic and result describe, historical case library then is made up of above-mentioned case, i.e. the situation of ocean eddy development and change;

Step 5: the similarity of carrying out the ocean eddy case is calculated, it is the similarity of calculating case in target case and the historical case library that similarity is calculated, the target case is the case that extracts at current problem, here be the case that will detect the ocean eddy correspondence of development and change, similarity computing formula (1):

${\mathrm{Similarity}}_{\mathrm{Case}(i,j)}=\frac{w_1\×S_{r\left(\mathrm{Case}(i,j)\right)}+w_2\×S_{a\left(\mathrm{Case}(i,j)\right)}+w_3{S}_{s\left(\mathrm{Case}(i,j)\right)}}{\mathrm{\Σ}{w}_1+w_2+w_3}---\left(1\right)$

S in the formula _{R (Case (i, j))}Be case i, the likeness coefficient of spatial relationship between the j; S _{A (Case (i, j))}Be case i, characteristic attribute likeness coefficient between the j; S _{S (Case (i, j))}Be case i, the likeness coefficient of spatial shape between the j; w ₁, w ₂And w ₃Be respectively above-mentioned S _{R (Case (i, j))}, S _{A (Case (i, j))}And S _{S (Case (i, j))}Weight coefficient;

Step 6: ocean eddy changes the detection of case, and concrete steps are as follows,

(1) given similarity threshold value (for example can choose 0.7) is chosen the likeness coefficient of calculating in the step 5 all historical cases greater than the similarity threshold value;

(2) according to the historical case result who chooses, therefrom find out the pairing vortex case of the bigger a plurality of case results of similarity; Described historical case result is the development and change situation of ocean eddy from period to another period;

(3) according to the similarity size, similarity is divided into three scopes, be 0.9-1.0,0.8-0.9,0.7-0.8, result according to the historical case that is not all corresponding selection of similarity scope gives different weights, obtains their weighted mean value and composes to the target case, as the result of target case.

The construction method of case expression model is in the described step 1: the ocean eddy of structure changes case expression model and is Case _i={ S _i, SA _1i, SA _2i..., SA _Ji, SR _1i, SR _2i..., SR _1i, Vortex _T1i→ Vortex _T2i(1)

I=1,2 ... K; J=1,2 ... M; L=1,2 ... N; K, N, M is natural number;

$S_i=\{(x_i^1,y_i^1),(x_i^2,y_i^2),...,(x_i^m,y_i^m)\};$

I is the sequence number of case in the formula, i=1, and 2 ... K, K are natural number; S _iThe spatial shape that is case i is described set, m 〉=3 wherein, and m is a positive integer, and parentheses represent coordinate right, and it is right to have m, S _iBe the coordinate set of the planar object bounds of ocean eddy, its coordinate logarithm of different vortexs is different; SA _1i, SA _2i..., SA _Ji, j=1,2 ... M, M are natural number, the various ATTRIBUTE INDEX of expression case i, and a total M is individual; SR _1i, SR _2i..., SR _1i, l=1,2 ... N, N are natural number, the various spatial relationship indexs of expression case i and geographical environment, and a total N is individual; Vortex _T1i→ Vortext _T2iRepresent the result of historical case development and change, i.e. the concrete situation of these ocean eddy development and change, t1 wherein, t2 represents former and later two state times of ocean eddy case i, " → " expression from time t1 to time t2, Vortex _T1iExpression t1 is the result phase of vortex case i constantly, Vortex _T2iExpression t2 is the result phase of vortex case i constantly; Case Case _iIt is the set of above particular content.

The step that the spatial relationship that adopts rough set to carry out ocean eddy variation case in the described step 3 extracts is as follows: a. sets up the rough set expression that ocean eddy changes the priori spatial relationship, specifically provides with spatial relationship decision table form;

The first, choose the particular space relation that ocean eddy changes that influences, described particular space pass is distance relation, the direction relations between vortex and the topological relation between vortex between vortex;

The second, adopt GIS SPATIAL CALCULATION method to carry out the selected particular space relation of ocean eddy case and calculate;

The 3rd, spatial relationship decision table according to the particular space relational result structure ocean eddy case of calculating, row is represented the historical case of ocean dynamic swirl in the decision table, the preceding part of row is called conditional attribute, represent each spatial relationship index, last row are called decision attribute, the attribute of its value for detecting;

B. the continuous variable that ocean eddy is changed in the case spatial relationship decision table is carried out discretize;

C. utilize old attribute reduction algorithms that the spatial relationship decision table that step b obtains is carried out the spatial relationship yojan, extract the decisive spatial relationship that influences target case result, and then extract the spatial relationship decision rule.

In the described step 5, the formula particular content is (case i represents the target case in the following description, and case j represents historical case) as explained below:

S in the formula (1) _{A (Case (i, j))}Calculating see formula (2)

$S_{a\left(\mathrm{Case}(i,j)\right)}=\frac{\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{w}_k\×S_{k\left(\mathrm{Case}(i,j)\right)}}{\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{w}_k}---\left(2\right)$

W in the formula _kBe the similarity weight of k characteristic attribute, S _{K (Case (i, j))}Be case i, the likeness coefficient of k the characteristic attribute of j.

S in the formula (1) _{R (Case (i, j))}Calculating mainly comprise dimensional orientation, the similarity of space topological and space length is calculated.Computing formula is seen formula (3)

$S_{r\left(\mathrm{Case}(i,j)\right)}=\frac{w_{\mathrm{dir}}\×S_{\mathrm{dir}\left(\mathrm{Case}(i,j)\right)}+w_{\mathrm{top}}\×S_{\mathrm{top}\left(\mathrm{Case}(i,j)\right)}+w_{\mathrm{dis}}{S}_{\mathrm{dis}\left(\mathrm{Case}(i,j)\right)}}{\mathrm{\Σ}{w}_{\mathrm{dir}}+w_{\mathrm{top}}+w_{\mathrm{dis}}}---\left(3\right)$

W in the formula _Dir, w _TopAnd w _DisBe respectively weight coefficient; S _{Dir (Case (i, j))}Be two case i, the likeness coefficient of dimensional orientation relation between the j; S _{Top (Case (i, j))}Be case i, the likeness coefficient of spatial topotaxy between the j; S _{Dis (Case (i, j))}Be case i, the likeness coefficient of space length relation between the j.The position relation similarity is calculated the area target dimensional orientation similarity calculation method that adopts Goyal to propose in the formula (3), particular content list of references [10] fourth rainbow .2004, space similarity theory and Study of calculation model. Wuhan University's doctorate paper. computing formula is as follows:

$S_{\mathrm{dir}\left(\mathrm{Case}(i,j)\right)}=1-\frac{\mathrm{dist}(D_i,D_j)}{4}---\left(4\right)$

D in the formula _iFeeling the pulse with the finger-tip mark case i is with respect to the direction of reference target (be meant in the research sea area, taken place in the buffer zone scope of vortex i, apart from vortex nearest on the vortex i time); D _jRefer to the direction of historical case j with respect to reference target; Dist (D _i, D _j) be meant case i, the direction distance between the j.

When formula (3) topological spatial relationship similarity was calculated, employing quantitative Analysis on the notion field figure of the topological relation of correspondence went out the similarity between the concrete topological relation.Can calculate the likeness coefficient of topological relation according to formula (5):

$S_{\mathrm{top}\left(\mathrm{Case}(i,j)\right)}=1-\frac{\mathrm{DisTopo}[\mathrm{Topo}\left(i\right),\mathrm{Topo}\left(j\right)]}{7}---\left(5\right)$

Topo in the formula (i) feeling the pulse with the finger-tip mark case i is with respect to the topological relation of reference target (be meant in the research sea area, taken place in the buffer zone scope of vortex i, apart from vortex nearest on the vortex i time); Topo (j) refers to the topological relation of historical case j with respect to reference target; DisTopo[Topo (i), Topo (j) is meant case i, the topological relation distance between the j.The space length similarity is calculated and is adopted formula (6) in the formula (3):

$S_{\mathrm{dis}\left(\mathrm{Case}(i,j)\right)}=1-\frac{|d_i-d_j|}{{\mathrm{Original}}_d}---\left(6\right)$

D in the formula _iFeeling the pulse with the finger-tip mark case i is with respect to the distance of reference target (be meant in the research sea area, taken place in the buffer zone scope of vortex i, apart from vortex nearest on the vortex i time); d _jRefer to the distance of historical case j with respect to reference target; d _i-d _jRepresent the range difference between two scene two targets, revise range difference between them on the occasion of, Original with absolute value sign _dBe meant poor with respect to the maximal value of reference target distance value and minimum value of each case in the case library.

S in the formula (1) _{S (Case (i, j))}Computing method depend on the concrete state that case presents.Learning (space) phenomenon when the ground of case correspondence is that spatial shape combination with point, line, surface or even more complicated occurs on spatial shape, need adopt different similarity calculation method at the different spaces form.Case to the wire space characteristics, adopt " based on the radius vector sequence similarity algorithm of center of gravity " document [11] Du Yunyan that specifically sees reference, Su Fenzhen, .2005. such as the Zhang sky are based on the ocean eddy characteristic information space Study on Similarity of reasoning by cases. tropical oceanography newspaper, 24 (3): 1～9; To having the space case of planar feature, adopt improved " based on the polygon similarity algorithm of mechanics ", concrete list of references [12] Fan Lingtao, Wu Siyuan, Chen Jian .2003. is based on the polygon similarity measure method of mechanics. Shanghai Communications University's journal, 37 (6): 874～877.

The present invention's advantage compared with prior art is:

(1) the present invention utilizes the historical case of vortex to survey the development and change situation of current vortex.Compare with method for numerical simulation, the present invention adopts and makes up historical case model and avoided the understanding of ocean eddy motion principle and genesis mechanism and probe into, and has simplified the knowledge acquisition process, and the raising problem solves efficient; With analyze the ocean eddy method according to remote sensing data in conjunction with conventional method of analysis and compare, the status data that the present invention not only can obtain remote sensing image can obtain relevant in history multidate data by historical case library simultaneously, and is that a kind of simply saving time can efficiently be realized the method for ocean eddy change-detection.

(2) the historical case library of the present invention's structure has self-learning ability, and promptly each new ocean eddy extracts to after the case, obtains its development and change situation, deposits it in historical case library as new case then.Dynamically updating of historical case library makes it to change the ocean eddy under more kinds of complex situations and makes detection, and this is non-existent in additive method.

Description of drawings

Fig. 1 is the FB(flow block) of the inventive method;

Fig. 2 is that the rough set of ocean eddy spatial relationship in the inventive method is expressed figure;

Data that Fig. 3 provides for the present invention and the vortex exemplary plot that on data, is identified, wherein Fig. 3 a is a sea level height difference SSHA data plot, and Fig. 3 b is an ocean current field flow speed data plot, and Fig. 3 c is a sea surface temperature SST data plot;

Fig. 4 is South Sea vortex case grid and the vector data figure that extracts in the embodiment of the invention 1, and wherein Fig. 4 a is the ocean current field data, and Fig. 4 b is that sea level height is poor, and Fig. 4 c is the sea surface temperature data, and Fig. 4 d is the vortex polar plot;

Fig. 5 learns the extraction flow process of accumulateing the spatial relationship rule in the phenomenon for ground.

Embodiment

As shown in Figure 1, the inventive method is embodied as: at first express model according to the historical case of ocean eddy, extract case data on the remotely-sensed data basis; Carry out accumulateing in the ocean eddy extraction of spatial relationship afterwards by rough set method, ocean eddy spatial relationship rough set is expressed and is seen accompanying drawing 2; Make up ocean eddy on this basis and change case library; By similarity calculating ocean eddy is changed to make at last and analyze and detect.

Embodiment 1

The ocean eddy that occurs year February in November, 2003 to 2009 with South China Sea district (its span is approximately 0 °～23 ° N, 99 °～121 ° E, and area is about 3,500,000 square kilometres) is a research object.

The source book of this example adopts US Naval Research Laboratory (Navy Research Laboratory, ocean essential field (sea level height difference SSHA, sea surface temperature SST and the ocean current field Current) data that the global layering ocean numerical model of 1/32 * 1/32 degree that NRL) provides simulates, this pattern and multiple satellite data assimilate, mainly comprise and utilize ENVISAT, GFO and JASON-1 etc. carry out the assimilation of sea level height difference SSHA, utilize the IR data to carry out the assimilation of SST.Constructed case then gets access to through expert's identification from these three key element field data in this example, and Fig. 3 has provided data and the example of the South Sea vortex that identified on data.

(1) express model according to newly-built case, adopt the spatial shape of a certain moment ocean eddy and corresponding time-space attribute feature thereof to express at this test, particular content is as follows:

Case _i＝{ID _i，P _i，A _i，A _1i，A _2i，…A _5i，F _1i，F _2i，L _oi，L _ai，Dir _1i，Dir _2i，Dir _3i，Dis _1i，Top _1i，S _pi，S ₁，S ₂，S ₃.??Vortex _t1(A ₂，Dir ₂，S _p)→Vortex _t2(A ₂，Dir ₂，S _p)}?????(7)

I=1,2 ... K; T1, t2 represent the state time; The change procedure of " → " expression vortex from time t1 to time t2.Wherein, specifically comprise the vortex case and indicate number (ID), the girth (P) of vortex figure spot, figure spot area (A), vortex type (A ₁), intensity (A ₂), this is the residing state (A of vortex constantly ₃), time in stage (A ₄), this vortex process duration (A ₅).The generation development of vortex often has substantial connection with its region physics marine environment and other oceanographic phenomena, and based on available research achievements, the present invention has selected 4 marine environment indexs, is vortex center zone sea surface temperature (F ₁), center and the peripheral temperature difference (F ₂), the geographic longitude in vortex center zone (Lo), geographic latitude (L _a); Next is the spatial shape index level yardstick (S of vortex ₁), long axis length (S ₂), minor axis length (S ₃); Be the speed (S that vortex moves at last _p).More than be total to M=16 ATTRIBUTE INDEX.Select 5 spatial relationship indexs simultaneously: the geographic orientation (Dir of vortex open region ₁), the mobile orientation (Dir of vortex main shaft ₂), the direction in space relation (Dir in the buffer set scope of vortex and research marine site target case (perhaps claiming the target vortex) between the nearest vortex of distance objective case time of having taken place ₃), space length relation (Dis ₁) and spatial topotaxy (Top ₁), be total to N=5 spatial relationship index like this.The transformation results that finally will express and will detect is: to the intensity of next this vortex of the moment, moving direction and speed are surveyed.In conjunction with the case library that makes up, wherein case number K=250.

(2) ocean eddy data extract.Extract the ocean eddy case data according to original remote sensing image data, be divided into two kinds of formatted datas of grid and vector, raster data is mainly as the case background picture, case is expressed with the planar data of vector, the boundary coordinate set in planar zone is the spatial shape S of case, and different vortex coordinates are to the value difference of m.Specifically as shown in Figure 4, wherein Fig. 4 a is the ocean current field data, and Fig. 4 b is that sea level height is poor, and Fig. 4 c is the sea surface temperature data, and Fig. 4 d is the vortex polar plot.

(3) based on rough set method vortex case spatial relationship is extracted.Related to a plurality of geographical environment indexs in the expression model of formula (7) and dimensional orientation concerns index, when therefore making up, needed earlier these indexs of ocean eddy case are extracted at the ocean eddy case library in research sea area.

Learn the spatial relationship rule extraction of phenomenon with rough set with carrying out, its necessary condition is that the various spatial relationships that phenomenon is learned on ground are carried out quantitative expression, and converting the form of rough set method data processing effectively to, Fig. 2 learns the idiographic flow that phenomenal space concerns that rough set is represented with being.Learn the rough set of phenomenal space relation as seen from the figure and represent following several steps of branch: 1. spatial relationship is chosen; 2. spatial relationship quantitative description; 3. construct the spatial relationship decision table.After learning the spatial relationship of phenomenon with two-dimentional form, promptly can analyze and learn with extracting the main space relation rule of phenomenon with the method for rough set with representing.

The extraction of spatial relationship rule mainly comprises following several steps.Fig. 5 learns the extraction flow process of accumulateing the spatial relationship rule in the phenomenon with rough set method with carrying out.As shown in Figure 5, adopt rough set method to learn when accumulateing the spatial relationship rule extraction in the phenomenon, mainly be divided into following a few step: 1. the rough set of spatial relationship is expressed with carrying out; 2. utilize the discretization method of rough set theory that the decision table that obtains is carried out discretize; 3. the spatial relationship decision table that the old attribute reduction algorithms of utilizing rough set is learned phenomenon to the ground of discretize carries out the spatial relationship yojan, and forms last spatial relationship decision rule table; To last spatial relationship rule, also need computer memory to concern the coverage and the degree of confidence of decision rule.

The geographical environment index of vortex correspondence mainly adopts the grid analysis operator of GIS to carry out, the employing of obtaining of dimensional orientation index is obtained based on the direction relations computing method between the area target of raster data, adopts ArcMapVBA to realize according to above-mentioned algorithm programming during specific implementation.

(4) make up historical case library.This example is chosen 50 the typical vortex evolution process in South China Sea district and is carried out the case library structure.Determined down in order to simplify the quantity of case library in case quantity, simultaneously do not lose ocean eddy again and develop information really, example of the present invention for each ocean eddy procedure definition 5 typical states (generation, development and stabilization, weaken and wither away), each typicalness is corresponding to a historical case.Historical case library sees Table 1 (50 corresponding 250 cases of process).Delegation represents a case stage in the table, and every five-element constitute a process case, is listed as each index in the corresponding case expression model.In order to carry out this method validation, choose 10 vortex processes as test cases (50 cases are the target case that we will detect, and table is slightly) from historical case library at random, carry out the detection and the precision evaluation of voorticity, moving direction and rate travel.

Zone, table 1 South Sea 2003-2009 ocean eddy changes case library

Annotate: the unit of field is respectively in this table: intensity (A ₂): centimetre; Vortex center zone sea surface temperature (F ₁): degree; The girth (P) of vortex figure spot: rice, figure spot area (A) square metre.

(5) after historical case library makes up, carry out the extraction and the result of variations of the historical case of similarity according to similarity computing formula and algorithm and survey.Before carrying out global similarity calculating, need respectively the weight of each index in the similar calculating with spatial relationship of attribute to be set.Wherein, ATTRIBUTE INDEX is determined the influence degree of each index to the ocean eddy testing result according to this field achievement in research in the past, be provided with as follows during calculating: w _P: (0.05), w _A: (0.05), w _A1: (0.1), w _A2: (0.1), w _A3: (0), w _A4: (0.15), w _A5: (0.05), and the pairing environment Background Field of ocean eddy is described the index weight and is provided with as follows: w _F1: (0.05), w _F2: (0.05), w _Lo: (0.05), w _La: (0.05), w _Dir1: (0.025), w _Dir2: (0.025), w _Dir3: (0.025), w _Dis1: (0.05), w _Top1: (0.05), w _Sp: (0.05), w _S1: (0.025), w _S2: (0.025), w _S3: (0.025).

Symbol attribute is definite in the computing formula (7), if vortex type (A ₁) and the residing state (A of this moment vortex ₃), when calculating similarity, when inquiry when the value on this generic attribute is equal to or belong to the value of case on this generic attribute, its similarity is 1, otherwise similarity is 0.Determine the numeric type attribute: the girth (P) of vortex figure spot, figure spot area (A), intensity (A ₂), time in stage (A ₄), this vortex process duration (A ₅), vortex center zone sea surface temperature (F ₁), center and the peripheral temperature difference (F ₂), the geographic longitude in vortex center zone (Lo), geographic latitude (L _a), taken place in vortex and the survey region one buffer set scope apart from the space length (Dis between the nearest vortex of vortex time ₁), the speed (S that vortex moves _p); And the spatial shape index of vortex: horizontal scale (S ₁), long axis length (S ₂), minor axis length (S ₃), when calculating similarity, obtain the absolute value of difference of desired value of the correspondence of two cases earlier, carry out distance divided by the method for pairing all the desired value spans of this index then and standardize, deduct the similarity that this value is this index with 1 then.

Girth (P) similarity computing formula as vortex figure spot is: $S_{P\left(\mathrm{Case}(i,j)\right)}=1-\frac{|P_i-P_j|}{|{\mathrm{Max}}_P-{\mathrm{Min}}_P|}$

S _{P (Case (i, j))}Be the girth similarity of target case i and historical case j, P _iBe the girth of target case i, P _jBe the girth of historical case j, Max _PBe the maximal value of all case girths in the case library, Min _PMinimum value for all case girths in the case library.Other several numeric type property value similarity calculation method are identical.

Mobile orientation (the Dir of vortex main shaft ₂), and taken place in vortex and the survey region one buffer set scope apart from the direction in space between the nearest vortex of this vortex time relation (Dir ₃) calculate with formula (4).

Taken place in vortex and the survey region one buffer set scope apart from the spatial topotaxy (Top between the nearest vortex of this vortex time ₁) calculate with formula (5).

Between concern S _{R (Case (i, j))}Computing formula as follows:

$S_{r\left(\mathrm{Case}(i,j)\right)}=\frac{\begin{array}{c}{S}_{\mathrm{Dir}1\left(\mathrm{Case}(i,j)\right)}*{\mathrm{<figure-callout\; id="1"\; label="w\; Dir"\; filenames="A2009100866900021H2.png"\; state="\{\{state\}\}">w</figure-callout>}}_{\mathrm{Dir}}+S_{\mathrm{Dir}2\left(\mathrm{Case}(i,j)\right)}*w_{\mathrm{Dir}2}+S_{\mathrm{Dir}3\left(\mathrm{Case}(i,j)\right)}*w_{\mathrm{Dir}3}\\ +S_{\mathrm{Dis}1\left(\mathrm{Case}(i,j)\right)}*{\mathrm{<figure-callout\; id="1"\; label="w\; Dis"\; filenames="A2009100866900021H2.png"\; state="\{\{state\}\}">w</figure-callout>}}_{\mathrm{Dis}}+S_{\mathrm{Top}1\left(\mathrm{Case}(i,j)\right)}*w_{\mathrm{Top}1}\end{array}}{{\mathrm{<figure-callout\; id="1"\; label="w\; Dir"\; filenames="A2009100866900021H2.png"\; state="\{\{state\}\}">w</figure-callout>}}_{\mathrm{Dir}}+w_{\mathrm{Dir}2}+w_{\mathrm{Dir}3}+{\mathrm{<figure-callout\; id="1"\; label="w\; Dis"\; filenames="A2009100866900021H2.png"\; state="\{\{state\}\}">w</figure-callout>}}_{\mathrm{Dis}}+{\mathrm{<figure-callout\; id="1"\; label="w\; Top"\; filenames="A2009100866900021H2.png"\; state="\{\{state\}\}">w</figure-callout>}}_{\mathrm{Top}}}$

Characteristic attribute S _{A (Case (i, j))}Computing formula as follows:

$S_{a\left(\mathrm{Case}(i,j)\right)}=\frac{\begin{array}{c}{S}_{P\left(\mathrm{Case}(i,j)\right)}*w_P+S_{A\left(\mathrm{Case}(i,j)\right)}*w_A+S_{A1\left(\mathrm{Case}(i,j)\right)}*w_{A1}+S_{A2\left(\mathrm{Case}(i,j)\right)}*w_{A2}\\ +S_{A3\left(\mathrm{Case}(i,j)\right)}*w_{A3}+A_{A4\left(\mathrm{Case}(i,j)\right)}*w_{A4}+S_{A5\left(\mathrm{Case}(i,j)\right)}*w_{A5}+S_{F1\left(\mathrm{Case}(i,j)\right)}*w_{F1}\\ +S_{F2\left(\mathrm{Case}(i,j)\right)}*w_{F2}+S_{\mathrm{Lo}\left(\mathrm{Case}(i,j)\right)}*w_{\mathrm{Lo}}+S_{\mathrm{La}\left(\mathrm{Case}(i,j)\right)}*w_{\mathrm{La}}+S_{\mathrm{Sp}\left(\mathrm{Case}(i,j)\right)}*w_{\mathrm{Sp}}\\ +S_{S1\left(\mathrm{Case}(i,j)\right)}*w_{S1}+S_{S2\left(\mathrm{Case}(i,j)\right)}*w_{S2}+S_{S3\left(\mathrm{Case}(i,j)\right)}*w_{S3}\end{array}}{w_P+w_A+w_{A1}+w_{A2}+w_{A3}+w_{A4}{+w}_{A5}+w_{F1}+w_{F2}+W_{\mathrm{Lo}}+w_{\mathrm{La}}+w_{\mathrm{Sp}}+w_{S1}+w_{S2}+w_{S3}}$

Because spatial shape we adopt three indexs to describe, the attribute similarity computing method are adopted in their calculating, so w in the formula (1) ₃=0, S _{S (Case (i, j))}Calculating merge to S _{A (Case (i, j))}In.

Result of calculation above last comprehensive is obtained the similarity of target case i and historical case j.

The similarity computing formula is specific as follows:

S _{(Case(i，j))}＝S _{r(Case(i，j))}*w ₁+S _{a(Case(i，j))}*w ₂

w ₁Be spatial relationship weight, w ₁=0.4; w ₂Be characteristic attribute weight, w ₂=0.6.

Because calculated amount is very big, all processes all realize with program.

The threshold value that similarity extracts in the present invention's test is made as 70%, obtain the final tache that enters detection after the historical similar cases, three scopes according to the division of similarity threshold value, need give different weights to the historical case that extracts, size according to similarity is given different values respectively, specifically be set to: when the similarity value [0.7,0.8) time, weight is 0.2; [0.8,0.90) time be 0.3; Be 0.5 when [0.9,1].

After retrieving the similar cases set of similarity in the defined threshold scope, just can be according to these similar cases, the weighted mean value by obtaining corresponding detection is as the value of the detection of target case.For example to detect target case i, calculate by top similarity, obtain similarity [0.7,0.8) in the scope be historical case j1, similarity [0.8,0.90) in the scope be historical case j2, similarity is [0.9, that 1] scope is interior is historical case j3, and the voorticity of target case i (A2) detected value computing formula is so:

S _A2(Case(i))＝S _A2(Case(j1))*0.2+S _A2(Case(j2))*0.3+S _A2(Case(j3))*0.5

Moving direction (Mdir) the detected value computing formula of target case i is:

S _{Mdir(Case(i))}＝S _{Mdir(Case(j1))}*0.2+S _{Mdir(Case(j2))}*0.3+S _{Mdir(Case(j3))}*0.5.

Rate travel (Msp) the detected value computing formula of target case i is:

S _Msp(Case(i))＝S _{Msp(Case(j1))}*0.2+S _{Msp(Case(j2))}*0.3+S _{Msp(Case(j3))}*0.5.

Concrete result of calculation sees Table 2 respectively, table 3 and table 4.Each row is represented an ocean eddy process that will detect in the table, and what " detected value " row were corresponding is the detected value of intensity, moving direction and the rate travel of certain this ocean eddy of stage, and accuracy of detection is meant the order of accuarcy that detected value is compared with actual value.

Table 2 intensity detection result and accuracy table

Annotate: intensity (A ₂) unit be rice.

Table 3 moving direction testing result and accuracy table

Case No.	Detected value (development)	+ actual value	Accuracy of detection	Detected value (stablizing)	Actual value	Accuracy of detection	Detected value (slackening)	Actual value	Accuracy of detection	Detected value (extinction)	Actual value	Accuracy of detection	Mean accuracy
														??11	North (76.51)	Northeast (55.35)	??88.24％	North (77.76)	North (97.35)	??89.12％	North (90.58)	North (78.95)	??93.54％	North (107.95)	Northwest (125.73)	??90.12％	??90.26％
??16	North (74.67)	North (111.44)	??79.57％	North (78.76)	Northeast (39.37)	??78.12％	North (98.12)	Northwest (143.24)	??74.93％	North (95.47)	Northeast (45.57)	??72.28％	??76.23％
														??20	Northwest (112.63)	Northwest (120.48)	??95.58％	North (70.62)	North (91.51)	??88.39％	North (89.29)	Northwest (152.48)	??64.89％	Northeast (57.77)	Northeast (55.79)	??98.9％	??86.94％
??27	North (69.97)	North (111.33)	??77.02％	North (80.07)	North (80.61)	??99.7％	North (92)	East (15.1)	??57.28％	North (99.07)	West (169.37)	??60.94％	??73.74％
														??33	North (87.18)	Northeast (44.99)	??76.56％	North (83.74)	Northwest (146.34)	??65.22％	North (83.45)	Northwest (114.09)	??82.98％	North (112.32)	Northwest (128.41)	??91.06％	??78.96％
??59	Northwest (121.05)	Northeast (56.32)	??64.04％	North (84.24)	Northwest (139.63)	??69.23％	Northeast (65.51)	Northeast (33.91)	??82.44％	North (89.99)	Northeast (62.89)	??84.94％	??75.16％
														??62	??*	Northwest (146.92)	???*	??*	Northwest (113.26)		North (83.08)	East (18.64)	??64.2％	???*	Northwest (128.31)	??*	??*
??73	Northeast (49.69)	North (84.13)	??80.87％	Northeast (45.59)	Northwest (131.58)	??52.22％	North (110.92)	East (22.1)	??50.66％	North (81.89)	North (72.71)	??94.9％	??69.66％
														??77	Northeast (66.27)	North (103.03)	??79.68％	Northwest (155.02)	North (108.9)	??74.38％	Northwest (116.31)	Northeast (35.79)	??55.27％	West (104.92)	Northeast (46.64)	??67.62％	??69.24％
??80	North (92.94)	Northwest (129.54)	??79.67％	North (73.71)	East (21.87)	??71.2％	North (74.12)	Northwest (119.12)	??75％	Northwest (112.5)	North (83.4)	??83.83％	??77.43％
														Mean accuracy			??80.14％			??76.40％			??70.78％			??82.73％	??77.51％

Annotate: * represents this test cases in historical case library in this table, and no similarity case under given condition is not then carried out the detection of moving direction to it.Orientation angles unit is degree.

Table 4 rate travel testing result and accuracy table

Annotate: * represents this test cases in historical case library in this table, and no similarity case under given condition is not then carried out the detection of moving direction to it.The unit of rate travel is rice/sky.

At 10 procedural test cases, its overall mean intensity accuracy of detection is 85.56%, majority reaches more than 80%, vortex moving direction and precision are low slightly, some case does not find similar cases when given threshold value 70%, the testing result great majority of the similar cases that finds are more than 80%, and average moving direction accuracy of detection is 88%, and average rate travel accuracy of detection is 84.66%.Because the process case limited amount of ocean eddy, some special case can not find similar cases, and perhaps the case accuracy of detection is lower, and this problem can be resolved along with the expansion of case quantity.

By test as can be known, the inventive method can be carried out the detection by quantitative and the analysis of ocean eddy based on the large tracts of land data of long-time sequence, and method simple and flexible, practical is seen also from test findings and can be satisfied application demand.

Embodiment 2

Ocean eddy with in November, 2003 to the 2009 year appearance in February of Gulf Stream zone is a tested object.

The method of its structure case library and computing method are with example 1.This example is chosen 120 the typical vortex evolution process in Gulf Stream zone and is carried out the case library structure.

In order to carry out the inventive method checking, choose 20 vortex processes as test cases from historical case library at random.The structure of case and spatial relationship extraction process such as example 1.After building case and case library, the extraction and the result of variations of carrying out the historical case of similarity according to similarity computing formula and algorithm detect.

The historical case library (table 1) that makes up; Voorticity testing result and accuracy of detection see Table 2; Vortex moving direction testing result and accuracy of detection see Table 3; Vortex rate travel testing result and precision see Table 4.

Table 1 Gulfstream zone 2003-2009 ocean eddy changes case library

??OID	??ID	??A 1	??A 2	...	??F 1	??...Dir 1	...	??P???????????????????A
									??1	??1	Cold whirlpool	??0.33822873	...	??24.55	... north	...	??132773.20797??1250020680.7092
??2 ??3	??1 ??1	Cold whirlpool, cold whirlpool	??0.42887987 ??0.53610540	... ...	??23.2403 ??23.2403	... northwest ... do not have	... ...	??207622.68522??3237803541.8164 ??289012.95190??6288361198.6091
									??4	??1	Cold whirlpool	??0.62191519	...	??21.9716	... the northwest	...	??319913.49635??7719337320.1788
??5	??1	Cold whirlpool	??0.35343211	...	??20.6284	... do not have	...	??442521.64082??11266185344.815
									??6	??2	Cold whirlpool	??0.29210488	...	??16.5612	... the northwest	...	??232242.58191??3061229560.8371
??7	??2	Cold whirlpool	??0.41190232	...	??173	... the west	...	??386500.37919??9850710684.6714
									??8	??2	Cold whirlpool	??0.29567105	...	??18.9866	... do not have	...	??174137.98905??2204484611.2327
??9	??2	Cold whirlpool	??0.29483771	...	??19.509	... do not have	...	??278338.62887??5869369317.1917
									??10	??2	Cold whirlpool	??0.20513824	...	??17.1955	... do not have	...	??199590.88807??2653259642.6448
??...	??...	??...	??...	...	??...		...	??...???????????...
									??596 ??597	??120 ??120	Warm whirlpool, warm whirlpool	??0.37122133 ??0.47012503	... ...	??27.2328 ??27.6433	... the southeast ... the southeast	... ...	??351602.55651??7890189311.2719 ??443459.80063??14950656301.162
??598	??120	Warm whirlpool	??0.47776438	...	??26.0015	.. do not have	...	??452975.02334??15340543992.138
									??599	??120	Warm whirlpool	??0.39343765	...	??25.8522	... south	...	??409114.60473??10578454337.269
??600	??120	Warm whirlpool	??0.32810657	...	??256657	... south	...	??347154.52417??7701269368.6922

Annotate: the unit of field is respectively in this table: intensity (A ₄): rice; Vortex center zone sea surface temperature (F ₅): degree; The girth (P) of vortex figure spot: rice, figure spot area (A) square metre.

Table 2 intensity detection result and accuracy table

Annotate: intensity (A ₂) unit be rice.

Table 3 moving direction testing result and accuracy table

Annotate: orientation angles unit is degree.

Table 4 rate travel testing result and accuracy table

Annotate: the unit of rate travel is rice/sky.

By the test to above 20 target cases, voorticity and rate travel accuracy of detection reach more than 80% mostly, and moving direction is low slightly, more than 70%; Need to prove that the accuracy of detection of moving direction here is meant the accuracy of detection of concrete orientation angle value.Because the target case randomly draws, if the target case is special case, in historical case library, just be difficult to the case that finds similarity higher so, then accuracy of detection will be lower, and the case accuracy of detection is relevant with choosing of target case.

Claims (4)

1, utilize the historical similarity case that ocean eddy is changed the method that detects, it is characterized in that step is as follows:

Step 1: set up ocean eddy and change case expression model, case is at the problem in the real world, comprise problem characteristic and the result describes set, then case expression model is the abstract expression to real world problem characteristic and result, specifically refers to the record of the feature description and the result of variations of ocean eddy here;

Step 2: express on the model basis in described case, extract the ocean eddy case data according to original remote sensing image data, be divided into two kinds of forms of grid and vector, raster data is mainly as the case background picture, case is expressed with the planar data of vector, in order to building the storehouse, case data is attribute and a spatial data of describing ocean eddy;

Step 3: adopt rough set method to carry out the spatial relationship extraction that ocean eddy changes case;

Step 4: on the basis of step 2 and step 3, make up ocean eddy change histories case library, historical case is exactly to the record of problem takes place, comprise that problem characteristic and result describe, historical case library then is made up of above-mentioned case, i.e. the situation of ocean eddy development and change;

Step 5: the similarity of carrying out the ocean eddy case is calculated, it is the similarity of calculating case in target case and the historical case library that similarity is calculated, the target case is the case that extracts at current problem, is the case of wanting the ocean eddy correspondence of change detected situation here.Similarity computing formula (1):

$\mathrm{Similarit}{y}_{\mathrm{Case}(i,j)}=\frac{w_1\&times;S_{r\left(\mathrm{Case}(i,j)\right)}+w_2\&times;S_{a\left(\mathrm{Case}(i,j)\right)}+w_3{S}_{s\left(\mathrm{Case}(i,j)\right)}}{\mathrm{\&Sigma;}{w}_1+w_2+w_3}---\left(1\right)$

S in the formula _{R (Case (i, j))}Be case i, the likeness coefficient of spatial relationship between the j; S _{A (Case (i, j))}Be case i, characteristic attribute likeness coefficient between the j; S _{S (Case (i, j))}Be case i, the likeness coefficient of spatial shape between the j; w ₁, w ₂And w ₃Be respectively above-mentioned S _{R (Case (i, j))}, S _{A (Case (i, j))}And S _{S (Case (i, j))}Weight coefficient;

Step 6: ocean eddy changes the detection of case, and concrete steps are as follows,

(1) given similarity threshold value is chosen the likeness coefficient of calculating in the step 5 all historical cases greater than the similarity threshold value;

(2) according to the historical case result who chooses, therefrom find out the pairing vortex case of the bigger a plurality of case results of similarity; Described historical case result is the development and change situation of ocean eddy from period to another period;

(3) according to the similarity size, similarity is divided into three scopes, be 0.9-1.0,0.8-0.9,0.7-0.8, result according to the historical case that is not all corresponding selection of similarity scope gives different weights, obtains their weighted mean value and composes to the target case, as the result of target case.

2, the historical similarity case of utilizing according to claim 1 is characterized in that the ocean eddy change detecting method: the construction method of case expression model is in the described step 1: the ocean eddy of structure changes case expression model and is Case _i={ S _i, SA _1i, SA _2i..., SA _Ji, SR _1i, SR _2i..., SR _Li, Vortex _T1i→ Vortex _T2i(1)

I=1,2 ... K; J=1,2 ... M; L=1,2 ... N; K, N, M is natural number;

$S_i=\{(x_i^1,y_i^1),(x_i^2,y_i^2),...,(x_i^m,y_i^m)\};$

I is the sequence number of case in the formula, i=1, and 2 ... K, K are natural number; S _iThe spatial shape that is case i is described set, m 〉=3 wherein, and m is a positive integer, and parentheses represent coordinate right, and it is right to have m, S _iBe the coordinate set of the planar object bounds of ocean eddy, its coordinate logarithm of different vortexs is different; SA _1i, SA _2i..., SA _Ji, j=1,2 ... M, M are natural number, the various ATTRIBUTE INDEX of expression case i, M altogether; SR _1i, SR _2i..., SR _Li, l=1,2 ... N, N are natural number, the various spatial relationship indexs of expression case i and geographical environment, N altogether; Vortex _T1i→ Vortex _T2iRepresent the result of historical case development and change, i.e. the concrete situation of these ocean eddy development and change, t1 wherein, t2 represents former and later two state times of ocean eddy case i, " → " expression from time t1 to time t2, Vortex _T1iExpression t1 is the result phase of vortex case i constantly, Vortex _T2iExpression t2 is the result phase of vortex case i constantly; Case Case _iIt is the set of above particular content.

3, the historical similarity case of utilizing according to claim 1 is characterized in that the ocean eddy change detecting method: the step that the spatial relationship that described step 3 adopts rough set to carry out ocean eddy variation case extracts is as follows:

A. set up the rough set of ocean eddy variation priori spatial relationship and express, specifically provide with spatial relationship decision table form;

The first, choose the particular space relation that ocean eddy changes that influences, described particular space pass is distance relation, the direction relations between vortex and the topological relation between vortex;

The second, adopt GIS SPATIAL CALCULATION method to carry out the selected particular space relation of ocean eddy case and calculate;

The 3rd, spatial relationship decision table according to the particular space relational result structure ocean eddy case of calculating, row is represented the historical case of ocean dynamic swirl in the decision table, the preceding part of row is called conditional attribute, represent each spatial relationship index, last row are called decision attribute, the attribute of its value for detecting;

B. the continuous variable that ocean eddy is changed in the case spatial relationship decision table is carried out discretize;

C. utilize old attribute reduction algorithms that the spatial relationship decision table that step b obtains is carried out the spatial relationship yojan, extract the decisive spatial relationship that influences target case result, and then extract the spatial relationship decision rule.

4, the historical similarity case of utilizing according to claim 1 is characterized in that the ocean eddy change detecting method: in the described step 5, case i represents the target case in the following description, and case j represents historical case,

S _{A (Case (i, j))}Calculating see that formula is:

$S_{a\left(\mathrm{Case}(i,j)\right)}=\frac{\underset{k=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_k\&times;S_{k\left(\mathrm{Case}(i,j)\right)}}{\underset{k=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_k}---\left(2\right)$

W in the formula _kBe the similarity weight of k characteristic attribute, S _{K (Case (i, j))}Be case i, the likeness coefficient of k the characteristic attribute of j;

S _{R (Case (i, j))}Calculating comprise dimensional orientation, the similarity of space topological and space length is calculated, formula is as follows:

$S_{r\left(\mathrm{Case}(i,j)\right)}=\frac{w_{\mathrm{dir}}\&times;S_{\mathrm{dir}\left(\mathrm{Case}(i,j)\right)}+w_{\mathrm{top}}\&times;S_{\mathrm{top}\left(\mathrm{Case}(i,j)\right)}+w_{\mathrm{dis}}{S}_{\mathrm{dis}\left(\mathrm{Case}(i,j)\right)}}{\mathrm{\&Sigma;}{w}_{\mathrm{dir}}+w_{\mathrm{top}}+w_{\mathrm{dis}}}---\left(3\right)$

W in the formula _Dir, w _TopAnd w _DisBe respectively weight coefficient; S _{Dir (Case (i, j))}Be two case i, the likeness coefficient of dimensional orientation relation between the j; S _{Top (Case (i, j))}Be case i, the likeness coefficient of spatial topotaxy between the j; S _{Dis (Case (i, j))}Be case i, the likeness coefficient of space length relation between the j;

Position relation similarity computing formula is as follows in the formula (3):

$S_{\mathrm{dir}\left(\mathrm{Case}(i,j)\right)}=1-\frac{\mathrm{dist}(D_i,D_j)}{4}---\left(4\right)$

D in the formula _iTarget refers to the direction of case i with respect to reference target, and described reference target is meant in the research sea area, has taken place in the buffer zone scope of vortex i, apart from vortex nearest on the vortex i time; D _jRefer to the direction of historical case j with respect to reference target; Dist (D _i, D _j) be meant case i, the direction distance between the j;

When formula (3) topological spatial relationship similarity is calculated, calculate the likeness coefficient of topological relation according to formula (5):

$S_{\mathrm{top}\left(\mathrm{Case}(i,j)\right)}=1-\frac{\mathrm{DisTopo}[\mathrm{Topo}\left(i\right),\mathrm{Topo}\left(j\right)]}{7}---\left(5\right)$

Topo in the formula (i) feeling the pulse with the finger-tip mark case i is with respect to the topological relation of reference target, and described reference target is meant in the research sea area, has taken place in the buffer zone scope of vortex i, apart from vortex nearest on the vortex i time; Topo (j) refers to the topological relation of historical case j with respect to reference target; DisTopo[Topo (i), Topo (j) is meant case i, the topological relation distance between the j;

The space length similarity is calculated and is adopted formula (6) in the formula (3):

$S_{\mathrm{dis}\left(\mathrm{Case}(i,j)\right)}=1-\frac{|d_i-d_j|}{\mathrm{Origina}{l}_d}---\left(6\right)$

D in the formula _iFeeling the pulse with the finger-tip mark case i is with respect to the distance of reference target, and described reference target is meant in the research sea area, has taken place in the buffer zone scope of vortex i, apart from vortex nearest on the vortex i time; d _jRefer to the distance of historical case j with respect to reference target; d _i-d _jRepresent the range difference between two scene two targets, revise range difference between them on the occasion of, Original with absolute value sign _dBe meant poor with respect to the maximal value of reference target distance value and minimum value of each case in the historical case library.

Images