最美情侣中文字幕电影,在线麻豆精品传媒,在线网站高清黄,久久黄色视频

歡迎光臨散文網(wǎng) 會員登陸 & 注冊

衛(wèi)星數(shù)據(jù)處理(三)——SVD分析

2021-06-17 21:18 作者:Berton9407  | 我要投稿

(三)SVD分析

對于單一頻率或單色平面波,無論電場E還是磁場B都可以寫成:

C_0e%5E%7Bi%5Cleft(%20%5Coverrightarrow%7Bk%7D%5Ccdot%20%5Coverrightarrow%7Bx%7D-wt%20%5Cright)%7D.

其中,C_0代表初始時刻的E_0B_0。根據(jù)Faraday’s law,有:

%7B%5Cnabla%20%7D%5Ctimes%7BE%7D%3D-%5Cfrac%7B%5Cpartial%7BB%7D%7D%7B%5Cpartial%20t%7D.

則可以利用FFT變換式(%5Cnabla%3Dik%5Cfrac%7B%5Cpartial%7D%7B%5Cpartial%20t%7D%3D-iw)結(jié)合線性小擾動理論:

%7BE%7D%3D%7BE_0%7D%2B%7BE'%7D%3B%20%7BB%7D%3D%7BB_0%7D%2B%7BB'%7D

得到:%7Bk%7D%5Ctimes%20%7BE'%7D%3Dw%7BB'%7D。根據(jù)矢量叉乘的位置關(guān)系,看出波矢和磁場擾動有%7BB'%7D%5Ccdot%7Bk%7D%3D0。

MVA分析利用磁場分量的協(xié)方差矩陣得到特征值和特征向量,同時可以畫出磁場的矢端曲線圖(hodographs),用來表征波動的極化特征。但要注意的是,最小方差分析基于信號頻率極窄、波矢方向基本不隨頻率變化的假設(shè)。

對于波傳播特性的另一種分析是基于多維頻譜分析,不同于早期只用實部的McPherron et al. (1972)和只用虛部的Means (1972),Sanrolik et al. (2003)假設(shè)存在平面波的情況下,結(jié)合復(fù)數(shù)域的多維頻譜矩陣,結(jié)合Ladreiter et al. (1995)提出的奇異值分解(singular value decomposition)方法,提高最小化過程,得出更合理的傳播特性結(jié)果。主要過程:

  • 利用磁場信息經(jīng)過譜分析得到多維頻譜結(jié)果%5Cwidehat%7BB_i%7D%5Cleft(%20t%2Cf%20%5Cright)%20%5Cleft(%20i%3D1%2C%202%2C%203%20%5Cright)%20;

  • 對于特定頻率f_0和時刻t_0,都可以形成Hermitian頻譜矩陣S_%7B3%5Ctimes3%7D,其中元素s_%7Bij%7D%3D%5Cwidehat%7BB_i%7D%5Cwidehat%7BB_%7Bj%7D%5E%7B*%7D%7D;

  • 根據(jù)復(fù)數(shù)元素構(gòu)造矩陣A滿足(R(%5Ccdot)%E5%92%8CIm(%5Ccdot)分別代表實部和虛部):

    A%3D%5Cleft(%20%5Cbegin%7Barray%7D%7Bc%7D%0A%09R%5Cleft(%20S_%7B11%7D%20%5Cright)%20%5C%2C%5C%2C%20%20R%5Cleft(%20S_%7B12%7D%20%5Cright)%20%5C%2C%5C%2C%20%20R%5Cleft(%20S_%7B13%7D%20%5Cright)%5C%5C%0A%09R%5Cleft(%20S_%7B12%7D%20%5Cright)%20%5C%2C%5C%2C%20%20R%5Cleft(%20S_%7B22%7D%20%5Cright)%20%5C%2C%5C%2C%20%20R%5Cleft(%20S_%7B23%7D%20%5Cright)%5C%5C%0A%09R%5Cleft(%20S_%7B13%7D%20%5Cright)%20%5C%2C%5C%2C%20%20R%5Cleft(%20S_%7B23%7D%20%5Cright)%20%5C%2C%5C%2C%20%20R%5Cleft(%20S_%7B33%7D%20%5Cright)%5C%5C%0A%090%20%20%20-%5Cmathrm%7BIm%7D%5Cleft(%20S_%7B12%7D%20%5Cright)%20%5C%2C%5C%2C-%5Cmathrm%7BIm%7D%5Cleft(%20S_%7B13%7D%20%5Cright)%5C%5C%0A%09%5Cmathrm%7BIm%7D%5Cleft(%20S_%7B12%7D%20%5Cright)%20%5C%2C%5C%2C%20%20%20%200%20%20-%5Cmathrm%7BIm%7D%5Cleft(%20S_%7B23%7D%20%5Cright)%5C%5C%0A%09%5Cmathrm%7BIm%7D%5Cleft(%20S_%7B13%7D%20%5Cright)%20%5C%2C%5C%2C%20%20%5Cmathrm%7BIm%7D%5Cleft(%20S_%7B23%7D%20%5Cright)%20%5C%2C%5C%2C%20%20%20%20%200%5C%5C%0A%5Cend%7Barray%7D%20%5Cright)%20%3B

  • A進行奇異值分解得到特征值(%5Clambda%20_%7Bs1%7D%5Cgeqslant%20%5Clambda%20_%7Bs2%7D%5Cgeqslant%20%5Clambda%20_%7Bs3%7D)和特征向量e_%7Bsi%7D;

  • 橢率%5Cvarepsilon%20_s、平面波參量F_s和波矢-背景磁場夾角%5Ctheta_%7Bs-kB_0%7D表達式有:

    %5Cvarepsilon%20_s%3D%5Cfrac%7B%5Clambda%20_%7Bs2%7D%7D%7B%5Clambda%20_%7Bs1%7D%7D,F_s%3D1-%5Csqrt%7B%5Cfrac%7B%5Clambda%20_%7Bs3%7D%7D%7B%5Clambda%20_%7Bs1%7D%7D%7D,%5Ctheta%20_%7Bs-kB_0%7D%3D%5Ccos%20%5E%7B-1%7D%5Cleft(%20%7Be_%7Bs3%7D%7D%5Ccdot%20%5Cleft%5B%200%2C0%2C1%20%5Cright%5D%20%2F%7C%7Be_%7Bs3%7D%7D%7C%20%5Cright)%20。

其中,平面波參量近似為1時才滿足原假設(shè),若小于1則可考慮其不服從平面波假設(shè)的前提條件。另外,注意區(qū)分橢率表達式和歸一化約化磁螺度(normalized reduced magnetic helicity)%5Csigma_m,其更接近圓偏振度D_%7Bc2%7D的定義,且無法得到更多的波動性參量。

%5Csigma%20_m%5Cleft(%20t%2Cf%20%5Cright)%20%3D%5Cfrac%7B2%5Cmathrm%7BIm%7DS_%7B12%7D%5Cleft(%20t%2Cf%20%5Cright)%7D%7B%5Csum%7B%5Cmathrm%7Btr%7D%5Cleft%5B%20S%5Cleft(%20t%2Cf%20%5Cright)%20%5Cright%5D%7D%7D%3D%5Cfrac%7B2%5Cmathrm%7BIm%7D%5Cleft(%20S_%7B12%7D%20%5Cright)%7D%7BS_%7B11%7D%2BS_%7B22%7D%2BS_%7B33%7D%7D.

有別于Sanrolik et al. (2003)基本不考慮噪聲水平的情況,Taubenschuss & Sanrolik (2019)從原理上進行方法改進,包括100%極化的波動J_p、波動噪聲J_%7Bn1%7D和各向同性的儀器及背景噪聲J_%7Bn2%7D。滿足:

J%5Cequiv%20J_p%2BJ_n%3DJ_p%2BJ_%7Bn1%7D%2BJ_%7Bn2%7D%0A%5C%5C%0A%3D%5Cleft(%20%5Cbegin%7Bmatrix%7D%0A%09J_%7B11%7D%26%09%09J_%7B12%7D%26%09%090%5C%5C%0A%09J_%7B12%7D%5E%7B*%7D%26%09%09J_%7B22%7D%26%09%090%5C%5C%0A%090%26%09%090%26%09%090%5C%5C%0A%5Cend%7Bmatrix%7D%20%5Cright)%20%2B%5Cleft(%20%5Cbegin%7Bmatrix%7D%0A%09%5Ctilde%7Ba%7D%26%09%090%26%09%090%5C%5C%0A%090%26%09%09%5Ctilde%7Ba%7D%26%09%090%5C%5C%0A%090%26%09%090%26%09%090%5C%5C%0A%5Cend%7Bmatrix%7D%20%5Cright)%20%2B%5Cleft(%20%5Cbegin%7Bmatrix%7D%0A%09c%26%09%090%26%09%090%5C%5C%0A%090%26%09%09c%26%09%090%5C%5C%0A%090%26%09%090%26%09%09c%5C%5C%0A%5Cend%7Bmatrix%7D%20%5Cright)%20.

當(dāng)然,一般情況下,J_p并不是100%偏振的,由圓偏振和線偏振共同組成,基于去除元素值為0的矩陣J_%7Bp2%5Ctimes2%7D,其斯托克斯參量(Stocks parameters)可表示為:

S%3DJ_%7Bp11%7D%2BJ_%7Bp22%7D%3B%20Q%3DJ_%7Bp11%7D-J_%7Bp22%7D%3B%20U%3DJ_%7Bp12%7D%2BJ_%7Bp12%7D%5E%7B*%7D%3B%20V%3Di%5Cleft(%20J_%7Bp12%7D%5E%7B*%7D-J_%7Bp12%7D%20%5Cright)%20.

總偏振度、線偏振度和圓偏振度分別用D_%7Bp2%7D、D_%7Bl2%7DD_%7Bc2%7D計算,表達式分別為:

%5Cfrac%7B%5Csqrt%7BQ%5E2%2BU%5E2%2BV%5E2%7D%7D%7BS%7D,%5Cfrac%7B%5Csqrt%7BQ%5E2%2BU%5E2%7D%7D%7BS%7D%5Cfrac%7BV%7D%7BS%7D。對于圓偏振度的描述,可以看出J_%7Bp12%7D有正負性,可以用來表征極化方向,正代表右旋,負代表左旋。由此,可以在橢率計算中加入其符號變量,,替代相關(guān)的圓偏振信息。若考慮噪聲,Taubenschuss & Sanrolik (2019)給出以下計算過程:

  • 利用磁場信息經(jīng)過譜分析得到多維頻譜結(jié)果%5Cwidehat%7BB_i%7D%5Cleft(%20t%2Cf%20%5Cright)%20%5Cleft(%20i%3D1%2C%202%2C%203%20%5Cright)%20;

  • 對于特定頻率f_0和時刻t_0,都可以構(gòu)成復(fù)數(shù)頻譜矩陣J_%7B3%5Ctimes3%7D,元素計算同無噪聲下的S_%7B3%5Ctimes3%7D;

  • J_%7B3%5Ctimes3%7D采用奇異值分解得到特征值%5Clambda%20_%7B1%7D%5Cgeqslant%20%5Clambda%20_%7B2%7D%5Cgeqslant%20%5Clambda%20_%7B3%7D及其對應(yīng)的特征向量e_i;

  • J_%7B3%5Ctimes3%7D的實部再采用奇異值分解得到特征值%5Clambda%20_%7Br1%7D%5Cgeqslant%20%5Clambda%20_%7Br2%7D%5Cgeqslant%20%5Clambda%20_%7Br3%7D及其對應(yīng)的特征向量e_%7Bri%7D;

  • 橢率%5Cvarepsilon%20_J、平面波參量F_J和波矢-背景磁場夾角%5Ctheta_%7BJ-kB_0%7D表達式有:

    %5Cvarepsilon%20_J%3D%5Csqrt%7B%5Cfrac%7B%5Cleft(%20%5Clambda%20_%7Br2%7D-%5Clambda%20_2%20%5Cright)%7D%7B%5Cleft(%20%5Clambda%20_%7Br1%7D-%5Clambda%20_2%20%5Cright)%7D%7D%5Ccdot%20%5Cmathrm%7Bsign%7D%5Cleft(%20%5Cmathrm%7BIm%7D%5Cleft(%20J_%7Bp12%7D%20%5Cright)%20%5Cright)%20F_r%3D1-%5Csqrt%7B%5Cfrac%7B%5Clambda%20_%7Br3%7D%7D%7B%5Clambda%20_%7Br1%7D%7D%7D,%5Ctheta%20_%7Br-kB_0%7D%3D%5Ccos%20%5E%7B-1%7D%5Cleft(%20%7Be_%7Br3%7D%7D%5Ccdot%20%5Cleft%5B%200%2C0%2C1%20%5Cright%5D%20%2F%7C%7Be_%7Br3%7D%7D%7C%20%5Cright)%20。

此時,總偏振度D_%7BpJ%7D%3D%0A%5Cfrac%7B%5Cmathrm%7Btr%7D%5Cleft(%20J_p%20%5Cright)%7D%7B%5Cmathrm%7Btr%7D%5Cleft(%20J%20%5Cright)%7D%3D1-%5Cfrac%7B%5Cmathrm%7Btr%7D%5Cleft(%20J_n%20%5Cright)%7D%7B%5Cmathrm%7Btr%7D%5Cleft(%20J%20%5Cright)%7D%3D%5Cfrac%7B%5Clambda%20_1-%5Clambda%20_2%7D%7B%5Clambda%20_1%2B%5Clambda%20_2%2B%5Clambda%20_3%7D%5Cequiv%20D_%7Bp3e%7D。

其中,tr(%5Ccdot)代表矩陣的跡,D_%7Bp3e%7D由Eliis et al. (2005)經(jīng)八個三維蓋爾曼(Gell-Mann)矩陣替代二維泡利(Pauli)自旋矩陣得到的偏振結(jié)果。此外,還能根據(jù)D_%7BpJ%7D得到相關(guān)的信噪比(signa to noise ratio,SNR),滿足:

SNR%3D%5Cfrac%7B%5Cmathrm%7Btr%7D%5Cleft(%20J_p%20%5Cright)%20%2B%5Cmathrm%7Btr%7D%5Cleft(%20J_%7Bn1%7D%20%5Cright)%7D%7B%5Cmathrm%7Btr%7D%5Cleft(%20J_%7Bn2%7D%20%5Cright)%7D%3D%5Cfrac%7BJ_%7Bp11%7D%2BJ_%7Bp22%7D%2B2%5Ctilde%7Ba%7D%7D%7B3c%7D%3D%5Cfrac%7BD_%7BpJ%7DS%2B%5Cleft(%201-D_%7BpJ%7D%20%5Cright)%20S%7D%7B3c%7D.

基本上,事件要挑選盡量高的SNR(不小于10),這樣可以有效避免噪聲淹沒信號,得到的結(jié)果更加可靠。對于實際磁場信號的處理,通常先取一定長度的數(shù)據(jù)得到時間段內(nèi)較為可靠的背景磁場B_0,再進行Magnetic-Field Aligned磁場變換,此變換則可將擾動大致分在平行于背景磁場的壓縮(compressional)擾動和垂直于背景磁場的橫向(transverse)擾動,且背景磁場的單位矢量方向為[0,0,1]。接著,運用小波分析得到多維頻譜結(jié)果。此時,由于在小波分析的過程中,會有主動窗口的選擇效應(yīng),因此噪聲水平往往會在一定低的水平,此時SNR水平往往顯得突出,得到的平面波參量也更接近于1,符合平面波的前提假設(shè)條件,波動分析的結(jié)果也更加準(zhǔn)確。

衛(wèi)星數(shù)據(jù)處理(三)——SVD分析的評論 (共 條)

分享到微博請遵守國家法律
泰来县| 平顶山市| 舟山市| 和静县| 南通市| 尚志市| 万全县| 玉环县| 临沧市| 伊吾县| 阜康市| 英吉沙县| 和静县| 乾安县| 济南市| 玉田县| 清徐县| 介休市| 图木舒克市| 彝良县| 洪雅县| 广州市| 应用必备| 军事| 景洪市| 花垣县| 台东县| 通江县| 漳平市| 毕节市| 宜阳县| 织金县| 陆丰市| 民乐县| 晋宁县| 镇平县| 达州市| 阿拉善右旗| 拜城县| 惠安县| 进贤县|