鹽漬化土壤水分微波雷達(dá)反演與驗(yàn)證

王 學(xué)，劉全明，屈忠義，王麗萍，李相君，王耀強(qiáng)

（內(nèi)蒙古農(nóng)業(yè)大學(xué)水利與土木建筑工程學(xué)院，呼和浩特 010018）

鹽漬化土壤水分微波雷達(dá)反演與驗(yàn)證

王 學(xué)，劉全明※，屈忠義，王麗萍，李相君，王耀強(qiáng)

（內(nèi)蒙古農(nóng)業(yè)大學(xué)水利與土木建筑工程學(xué)院，呼和浩特 010018）

土壤介電常數(shù)是微波遙感進(jìn)行土壤含水率測(cè)量的物理基礎(chǔ)，尤其介電常數(shù)實(shí)部是必須解決的問題，土壤介電特性的研究顯得尤為重要。該文目的是試驗(yàn)與評(píng)價(jià)C波段RADARSAT-2 SAR（synthetic aperture radar）數(shù)據(jù)模擬土壤介電特性，進(jìn)而反演土壤水分的性能。以受鹽漬化影響較嚴(yán)重的內(nèi)蒙古河套灌區(qū)解放閘灌域?yàn)樵囼?yàn)區(qū)，首先回歸分析了介電常數(shù)實(shí)部與SAR四極化后向散射系數(shù)、地表粗糙度的復(fù)雜關(guān)系，并與Oh經(jīng)驗(yàn)?zāi)Ｐ蛯?duì)照，其決定系數(shù)R2為0.859 7，模擬精度較高；然后驗(yàn)證常用的2個(gè)介電常數(shù)模型，Dobson半經(jīng)驗(yàn)?zāi)Ｐ?、Hallikainen簡化實(shí)部經(jīng)驗(yàn)?zāi)Ｐ湍M的介電常數(shù)實(shí)部與實(shí)測(cè)值的決定系數(shù)R2分別為0.935 9、0.869，表明2個(gè)模型均能模擬地表土壤水分與介電常數(shù)實(shí)部的密切關(guān)系；最后構(gòu)建了Dobson模型、Hallikainen簡化實(shí)部模型反演土壤含水率的模型，并與統(tǒng)計(jì)回歸模型比照，其模擬數(shù)值與土壤實(shí)測(cè)值的決定系數(shù)R2分別為0.803 8、0.737 4、0.842 1，均方根誤差RMSE分別為5.2%、5.7%、5%。Dobson模型與統(tǒng)計(jì)回歸模型反演結(jié)果與實(shí)地土壤墑情分布較為吻合，具有良好的精度和適用性，從而建立了一個(gè)較為完整的土壤介電特性研究體系，為微波遙感監(jiān)測(cè)土壤水分奠定了基礎(chǔ)。

土壤水分；遙感；模型；土壤介電特性；Oh模型；Dobson模型；Hallikainen簡化實(shí)部模型；鹽漬化

0 引 言

土壤水分（即土壤含水率）在地表→大氣→地表的能量交換中扮演著極其重要的角色，在水資源合理利用、農(nóng)田灌溉以及旱澇災(zāi)害預(yù)報(bào)等農(nóng)業(yè)科學(xué)研究領(lǐng)域中具有重要的意義，也是研究者們長期密切關(guān)注的課題[1-4]，尤其是土壤水分在大尺度上的監(jiān)測(cè)具有重要意義。土壤水分傳統(tǒng)監(jiān)測(cè)方法是通過人工或觀測(cè)儀器在各個(gè)監(jiān)測(cè)點(diǎn)上獲得長周期具有較高精度的土壤水分信息。雖然能夠獲得觀測(cè)點(diǎn)上比較準(zhǔn)確的土壤水分信息，但這樣不僅費(fèi)時(shí)費(fèi)力而且難以將采集的點(diǎn)數(shù)據(jù)擴(kuò)展到面上，無法在大范圍內(nèi)有效反映土壤水分的時(shí)空變化情況。陸表的土壤水分含量可由可見光、熱紅外和微波遙感數(shù)據(jù)估算，光學(xué)遙感直接反演土壤水分有很多限制。微波遙感具有全天時(shí)、全天候和穿透能力強(qiáng)的特點(diǎn)，能夠獲取地表的時(shí)空信息,為全面觀測(cè)提供了可能。尤其主動(dòng)微波遙感可估算地表5 cm深度土層的土壤水分,成為獲取大尺度、長時(shí)間序列土壤水分的有效手段[5-6]。

介電常數(shù)是描述電磁場與物質(zhì)相互作用關(guān)系的一個(gè)宏觀參量[7-8]，土壤含水量不同，其介電特性就明顯不同，進(jìn)而使得散射系數(shù)和亮溫度不同，這就是微波遙感進(jìn)行土壤含水量反演的物理基礎(chǔ)，土壤介電特性研究尤為重要[9-10]。宋書藝等[11-13]通過對(duì)土壤介電特性進(jìn)行研究，改進(jìn)介電常數(shù)測(cè)量方法，提高測(cè)量精度，但是測(cè)量方法復(fù)雜，及時(shí)能夠進(jìn)行反演工作，但所需參數(shù)較多模型實(shí)用性較小；曾江源等[14-16]均對(duì)土壤介電特性、土壤介電常數(shù)與含水量關(guān)系進(jìn)行了系統(tǒng)的研究，但是這些學(xué)者的研究的研究僅用同極化數(shù)據(jù)，未考慮四極化情況，導(dǎo)致模擬精度普遍較低。趙昕等[17-19]建立水分反演模型時(shí)，只引入后向散射系數(shù)，未考慮土壤地表粗糙度因素，或只考慮相關(guān)長度L、均方根高度S中的一種，導(dǎo)致水分反演精度較低；郭曼等[20-21]進(jìn)行介電模型研究，進(jìn)而反演水分工作時(shí)，只是正向研究介電模型，將介電常數(shù)代入模型參與計(jì)算，沒有進(jìn)行介電模型反演水分的思路。本文從以下幾點(diǎn)出發(fā)：首先根據(jù)Oh模型，分析土壤介電特性，建立土壤介電常數(shù)反演模型；然后分析研究現(xiàn)有土壤介電模型，結(jié)合實(shí)測(cè)數(shù)據(jù)，確定適用于本試驗(yàn)區(qū)的介電模型；再將介電模型進(jìn)行逆向推理，得到介電常數(shù)水分反演模型；最后根據(jù)介電水分反演，結(jié)合土壤地表粗糙度、介電常數(shù)，建立統(tǒng)計(jì)回歸水分反演模型。

1 試驗(yàn)區(qū)概況與數(shù)據(jù)獲取

1.1 試驗(yàn)區(qū)概況

試驗(yàn)區(qū)位于河套灌區(qū)解放閘灌域內(nèi)，地處內(nèi)蒙古自治區(qū)巴彥淖爾市杭錦后旗境內(nèi)，東經(jīng)106°43′－107°15′、北緯40°48′－40°59′，北靠陰山，東郊臨河市，南望鄂多斯高原，西與磴口接壤，是典型引黃河水灌溉的旗縣。海拔1 033~～1 055 m，屬溫帶高原型、大陸性氣候，全年平均氣溫6.3～7.7 ℃，干燥少雨，全年平均降雨量為139.4 mm，而平均蒸發(fā)量達(dá)2 070.4 mm，兼于解放閘灌域復(fù)雜的土壤水鹽環(huán)境系統(tǒng)，使其成為理想的試驗(yàn)區(qū)域。在研究區(qū)域內(nèi)設(shè)置100個(gè)采樣點(diǎn)，數(shù)據(jù)采集時(shí)間為春季4月份，此時(shí)灌區(qū)為春灌前的裸露地表無植被覆蓋，因此進(jìn)行水分反演工作時(shí)無需考慮植被的影響。如圖1為試驗(yàn)區(qū)雷達(dá)影像及采樣點(diǎn)分布。

1.2 Radarsat-2 C波段影像數(shù)據(jù)

作為當(dāng)今世界十分先進(jìn)的SAR系統(tǒng)，Radarsat-2具有成像模式多、分辨率高、成像幅寬大、視角范圍廣等特點(diǎn)，可在大范圍內(nèi)快速成像，減少衛(wèi)星過境時(shí)間。與此同時(shí)還有多種極化方式可供選擇，提高了對(duì)目標(biāo)物進(jìn)行精細(xì)刻畫的能力[22]。本研究使用C波段Radarsat-2 的HH+HV+VH+VV精細(xì)全極化模式的雷達(dá)影像，軌道號(hào)43 459，幅寬（km）25×25，分辨率8 m，入射角30.42°，影像數(shù)據(jù)為SLC格式，其中H代表水平極化方式，V代表垂直極化方式，二者組結(jié)合。

通過雷達(dá)影像處理軟件ENVI SARscape來處理獲取的Radarsat-2的雷達(dá)數(shù)據(jù)，數(shù)據(jù)處理主要包括以下內(nèi)容：數(shù)據(jù)聚焦、多視處理、斑點(diǎn)濾波、地理編碼和輻射定標(biāo)、幾何校正、提取后向散射系數(shù)。如表1中所示部分采樣點(diǎn)數(shù)值。

圖1 試驗(yàn)區(qū)雷達(dá)影像圖Fig.1 Radar image of experimental area

表1 根據(jù)雷達(dá)影像獲取的樣點(diǎn)數(shù)據(jù)（部分樣點(diǎn)）Table 1 Sample data obtained by radar image (partial samples)

1.3 地表參數(shù)獲取

本研究采用安捷倫微波網(wǎng)絡(luò)分析儀，通過同軸探針法進(jìn)行采樣點(diǎn)的土樣介電常數(shù)測(cè)量。如表1中所示為部分采樣點(diǎn)的介電常數(shù)實(shí)部值。野外用厘米格網(wǎng)的剖面板測(cè)量地表粗糙度，計(jì)算獲取均方根高度S與相關(guān)長度L的數(shù)值。地表粗糙度反演模型的初期研究只有均方根高度S或相關(guān)長度L之一參與模型運(yùn)算，不能得到較好的反演結(jié)果，科研人員通過對(duì)S和L進(jìn)行組合來表示地表粗糙度，如Zribi等[23]利用S與L組成ZS=S2/L，他們采用的組合參數(shù)在模型反演中均取得了理想效果，本文將利用組合參數(shù)ZS進(jìn)行反演建模。用地溫計(jì)對(duì)地溫進(jìn)行3次實(shí)時(shí)測(cè)量，并取均值；用激光粒度儀Helos/B對(duì)采樣點(diǎn)土樣進(jìn)行土壤顆粒測(cè)量，獲得黏粒C與砂粒S的百分比含量。用100 cm3的環(huán)刀取樣測(cè)量土樣土壤容重。烘干法獲取土樣重量含水量，并轉(zhuǎn)換為體積含水率。

2 研究方法與結(jié)果

2.1 土壤介電常數(shù)特性

微波遙感的散射、輻射能量是介電常數(shù)的函數(shù)[24]，Oh等[25]得到了HH/VV、HV/VV與介電常數(shù)、地表粗糙度的經(jīng)驗(yàn)?zāi)Ｐ?。因本次使用C波段Radarsat-2影像，通過兩個(gè)通道得到HH和VV同極化的數(shù)值較大，較為準(zhǔn)確。而HV與VH交叉極化數(shù)據(jù)較小，故使用同極化后向散射系數(shù)比的Oh模型進(jìn)行介電常數(shù)與雷達(dá)后向散射系數(shù)的關(guān)系研究。

式中是法向入射的菲涅爾反射系數(shù)；θ=30.42°為雷達(dá)入射角；k為雷達(dá)波數(shù)；S為均方根高度；σHH、σVV為同極化后向散射系數(shù)。

利用采樣點(diǎn)的后向散射、介電常數(shù)及地表粗糙度數(shù)據(jù)，按照Oh模型進(jìn)行介電常數(shù)反推計(jì)算，其模擬與實(shí)測(cè)數(shù)據(jù)的決定系數(shù)R2=0.820 9，具有較高的精度，如圖2所示?？赏ㄟ^插值的方法得到介電常數(shù)的空間分布。

圖2 Oh模型介電常數(shù)模擬與實(shí)測(cè)值擬合分析Fig.2 Fitting analysis of Oh model dielectric constant simulated value and measured value

2.2 土壤介電常數(shù)模型

雖然土壤中各成分的介電常數(shù)組成了土壤介電常數(shù)，但水的介電常數(shù)起到了主導(dǎo)作用，所以影響土壤介電常數(shù)的最主要因素是土壤水分，此外頻率f、溫度T和土壤砂粒S、黏粒C等也會(huì)對(duì)介電常數(shù)產(chǎn)生影響[26]。因此，土壤介電常數(shù)模型應(yīng)充分考慮各個(gè)因素的影響。目前的介電模型主要分為理論模型、半經(jīng)驗(yàn)?zāi)Ｐ?、?jīng)驗(yàn)?zāi)Ｐ汀?/p>

2.2.1 Dobson模型

常用的Dobson模型是利用5種不同土壤類型的實(shí)測(cè)數(shù)據(jù)建立的1.4～18 GHz一個(gè)半經(jīng)驗(yàn)的土壤介電常數(shù)模型[27]，其形式簡單、應(yīng)用方便，只需輸入簡單參數(shù)即可。其模型公式：

式中ρb是土壤容重；ρs是土壤比重，一般取ρs=2.66；εs為土壤中固態(tài)物質(zhì)介電常數(shù)，εs=(1.01+0.44ρs)2?0.062≈4.7；a是一個(gè)常數(shù)a=0.65；β是與土壤類型即土壤砂土質(zhì)量百分?jǐn)?shù)和黏土質(zhì)量百分?jǐn)?shù)有關(guān)的復(fù)數(shù)參數(shù)；mv是土壤的體積含水量；εfw為純水的介電常數(shù)；f為入射電磁波頻率。

利用采樣點(diǎn)地表參數(shù)代入Dobson模型獲取土壤介電常數(shù)實(shí)部模擬值，與實(shí)測(cè)介電常數(shù)的決定系數(shù)為0.935 9，如圖3所示。可見Dobson模型適用于本試驗(yàn)區(qū)的介電特性模擬。

圖3 Dobson模型模擬數(shù)值與實(shí)測(cè)數(shù)值擬合分析Fig.3 Fitting analysis of Dobson model simulates value and measured value

2.2.2 Hallikainen模型

Hallikainen等[28]在1.4～18 GHz的頻率范圍內(nèi)測(cè)得不同含水率、不同土壤質(zhì)地的介電常數(shù)。在數(shù)據(jù)分析的基礎(chǔ)上，建立了以土壤質(zhì)地和含水量為輸入變量的經(jīng)驗(yàn)公式，其通式為：

將此模型改動(dòng)變成以下公式

將S、C、mv、Smv、Cmv、這8項(xiàng)看作獨(dú)立變量，其中S砂土百分比、C為黏土百分比含量。將a0、a1、a2、b0、b1、b2、c0、c1、c2這9項(xiàng)看作待求的待定系數(shù)，其目的是將原來的非線性問題轉(zhuǎn)化為線性問題。

利用采樣點(diǎn)數(shù)據(jù)建模并驗(yàn)證，發(fā)現(xiàn)模擬與實(shí)測(cè)值的決定系數(shù)R2=0.869，如圖5所示。

圖4 Hallikainen簡化實(shí)部模型模擬與實(shí)測(cè)值擬合分析Fig.4 Fitting analysis of simulation value of Hallikainen simplified real part model and measured value

2.3 土壤水分反演模型

多年來國內(nèi)外學(xué)者對(duì)土壤介電常數(shù)進(jìn)行了大量的實(shí)驗(yàn)研究，在試驗(yàn)數(shù)據(jù)的基礎(chǔ)上，依據(jù)介電混合的思想，發(fā)展了多種土壤介電常數(shù)模型[29]。Dobson模型、Hallikainen簡化實(shí)部模型模經(jīng)過他們模擬數(shù)據(jù)與實(shí)測(cè)數(shù)據(jù)相關(guān)性分析，表明他們具有較高的相關(guān)性。因此本文對(duì)Dobson模型、Hallikainen簡化實(shí)部模型進(jìn)行公式變形，得到土壤水分反演模型。

2.3.1 Dobson水分反演模型

上文研究表明Dobson模型能夠較好反應(yīng)介電常數(shù)與土壤含水率關(guān)系，故對(duì)Dobson模型進(jìn)行變形而得到土壤水分反演公式。將Dobson模型公式變?yōu)椋?/p>

對(duì)公式（6）進(jìn)行對(duì)數(shù)變形，得到土壤水分反演模型：

將70個(gè)采樣點(diǎn)的參數(shù)代入公式（7）得到Dobson模型反推含水模擬數(shù)值，公式（7）中所用的介電常數(shù)為Oh模型反演介電常數(shù)得到的數(shù)值。通過30個(gè)數(shù)據(jù)對(duì)Dobson模型反推含水模擬數(shù)值與土壤實(shí)測(cè)含水值的相關(guān)性分析，得到?jīng)Q定系數(shù)R2=0.803 8,均方根誤差RMSE=0.052，如圖5所示。

圖5 Dobson模型反推土壤含水率模擬數(shù)值與實(shí)測(cè)值的擬合Fig.5 Fitting analysis of Dobson model inversion value and measured value of soil moisture content

2.3.2 Hallikainen水分反演模型

上文研究表明Hallikainen簡化實(shí)部模型也能較好反映土壤介電常數(shù)與土壤含水率關(guān)系，故對(duì)簡化實(shí)部模型進(jìn)行反推，得到Hallikainen簡化實(shí)部水分反演模型：

將采樣點(diǎn)的參數(shù)代入公式得到Hallikainen簡化實(shí)部模型反推含水模擬數(shù)值，其決定系數(shù)R2=0.737 4，均方根誤差RMSE=0.057，如圖6所示。

圖6 Hallikainen簡化實(shí)部模型反推土壤含水率模擬值與實(shí)測(cè)值擬合分析Fig.6 Fitting analysis of soil moisture content simulation value of Hallikainen simplified real part model and measured value

2.3.3 統(tǒng)計(jì)回歸水分反演模型

根據(jù)AIEM正演模型雷達(dá)入射角、頻率、均方根高度、相關(guān)長度、介電常數(shù)、水分，改變其中任一變量都會(huì)對(duì)后向散射系數(shù)產(chǎn)生影響[30]。通過對(duì)AIEM模型進(jìn)行機(jī)理特征分析發(fā)現(xiàn)：對(duì)頻率變化后向散射系數(shù)響應(yīng)圖進(jìn)行分析，同極化的后向散射系數(shù)隨著頻率的增大而增大；對(duì)入射角變化后向散射系數(shù)響應(yīng)圖進(jìn)行分析，同極化的后向散射系數(shù)隨著入射角的增大而減??；過對(duì)均方根變化后向散射系數(shù)響應(yīng)圖進(jìn)行分析，同極化的后向散射系數(shù)先隨著入射角的增大而增大，到達(dá)某一數(shù)值后，后向散射系數(shù)呈減小的趨勢(shì)；對(duì)相關(guān)長度變化后向散射系數(shù)響應(yīng)圖進(jìn)行分析，同極化的后向散射系數(shù)隨著相關(guān)長度的增大而減??；土壤水分變化后向散射系數(shù)響應(yīng)圖進(jìn)行分析發(fā)現(xiàn)，同極化的后向散射系數(shù)隨著土壤水分的增大而增大。在使用雷達(dá)影像數(shù)據(jù)時(shí)其入射角、頻率是固定值，因此本文用四極化后向散射系數(shù)HH、HV、VH、VV及其組合HH/VV、HV/VH，組合地表粗糙度ZS以及Oh模型反演介電常數(shù)ε建立經(jīng)驗(yàn)回歸模型，結(jié)果如公式（9）所示。其模型反演與實(shí)測(cè)值決定系數(shù)R2達(dá)0.8421，均方根誤差RMSE=0.05，圖7所示。

圖7 經(jīng)驗(yàn)回歸模型土壤含水率反演值與實(shí)測(cè)值擬合分析Fig.7 Fitting analysis of empirical regression model inversion value and measured value of soil moisture content

將剩余的30個(gè)數(shù)據(jù)代入經(jīng)驗(yàn)回歸模型，計(jì)算其相對(duì)誤差，并計(jì)算其模擬精度。部分?jǐn)?shù)據(jù)見表2所示。

表2 經(jīng)驗(yàn)回歸模型土壤含水率反演精度檢驗(yàn)Table 2 Soil moisture content inversion accuracy test of regression model

2.3.4 水分反演模型對(duì)比

最后使用 ENVI軟件最大似然法對(duì)Dobson模型、Hallikainen簡化實(shí)部模型反推含水率數(shù)值與統(tǒng)計(jì)回歸模型結(jié)果分類，得到3種土壤墑情分布圖（圖8所示）。根據(jù)3種模型反演的土壤墑情在空間分布存在明顯差異。

圖8 不同模型反演的土壤含水率結(jié)果Fig.8 Soil moisture content inversion results of different models

3種模型反演的土壤墑情統(tǒng)計(jì)結(jié)果如表3所示。

表3 土壤含水率統(tǒng)計(jì)結(jié)果Table 3 Statistic results of soil moisture content

表3統(tǒng)計(jì)了Dobson模型反推含水率模型、Hallikainen簡化實(shí)部模型反推含水率模型、經(jīng)驗(yàn)回歸模型模擬的不同墑情等級(jí)占比，從統(tǒng)計(jì)結(jié)果可以看出Dobson反推水分模型與統(tǒng)計(jì)回歸模型所占比重基本相等，其主要原因在于Hallikainen簡化實(shí)部模型未考慮地表粗糙度影響。

3 結(jié)論與討論

1）通過Oh模型反演介電常數(shù)值，能夠?yàn)榻殡姵?shù)模型反推含水值提供數(shù)據(jù)的支持。通過對(duì)常用的Dobson模型和Hallikainen簡化實(shí)部模型的驗(yàn)證。發(fā)現(xiàn)2種模型都能較好地反映土壤介電常數(shù)與土壤含水的密切關(guān)系，尤其是Dobson模型的效果更好。

2）經(jīng)Dobson模型、Hallikainen簡化實(shí)部模型反推含水率模型驗(yàn)證發(fā)現(xiàn)兩者均可用于土壤水分反演，且Dobson模型與統(tǒng)計(jì)回歸經(jīng)驗(yàn)?zāi)Ｐ凸πл^為一致，具有較高的精度與適用性，而Hallikainen簡化模型模擬水分的效果劣于前兩者。

3）Dobson模型、Hallikainen簡化實(shí)部模型反推含水率模型、統(tǒng)計(jì)回歸經(jīng)驗(yàn)?zāi)Ｐ腿叩耐寥缐勄榉植?，?.2～0.3范圍內(nèi)所占比重較多，說明3種反演水分模型都能夠較好的反映試應(yīng)驗(yàn)區(qū)的土壤水分分布情況。

本文推薦的經(jīng)驗(yàn)?zāi)Ｐ鸵蕾囉诘乇碓囼?yàn)參數(shù)，具有區(qū)域的限制性。如何從理論模型如AIEM物理模型出發(fā)研究各參數(shù)間的機(jī)理關(guān)系建模，以擴(kuò)大土壤水分反演模型的普適性是今后研究的重點(diǎn)。

[1] Jagdhuber T, Hajnsek I, Papathanassiou K P. An iterative generalized hybrid decomposition for soil moisture retrieval under vegetation cover using fully polarimetric SAR[J]. IEEE Journal of Selected Topics in Applied Earth Observations and. Remote Sensing. 2015, 8: 3911－3922.

[2] Jacome A, Bernier M, Chokmani K, et al. Monitoring volumetric surface soil moisture content at the La Grande Basin Boreal Wetland by radar multi polarization data[J]. Remote Sens, 2013, 5: 4919－4941.

[3] 劉全明，成秋明，王學(xué)，等. 河套灌區(qū)土壤鹽漬化微波雷達(dá)反演[J]. 農(nóng)業(yè)工程學(xué)報(bào)，2016，32(16)：109－114. Liu Quanming, Cheng Qiuming, Wang Xue, Li Xiangjun. Soil salinity inversion in Hetao Irrigation district using microwave radar[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2016, 32(16): 109－114. (in Chinese with English abstract)

[4] 馬洪章，劉素美，彭愛華，等. L 波段主被動(dòng)微波協(xié)同反演裸土土壤水分[J].農(nóng)業(yè)工程學(xué)報(bào)，2016，32(19)：133－138. Ma Hongzhang, Liu Sumei, Peng Aihua, Sun Lin, Sun Genyun. Active and passive cooperative algorithm at L-Band for bare soil moisture inversion[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2016, 32(19): 133－138. (in Chinese with English abstract)

[5] 陳晶，賈毅，余凡. 雙極化雷達(dá)反演裸露地表土壤水分[J].農(nóng)業(yè)工程學(xué)報(bào)，2013，29(10)：109－115.Chen Jing, Jia Yi, Yu Fan. Soil moisture inversion by radar with dual-polarization[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2013, 29(10): 109－115. (in Chinese with English abstract)

[6] Dubois P C, Van Zyl J, Engman T. Measuring soil moisture with imaging radars[J]. IEEE Transactions on Geoscience and Remote Sensing, 1995, 33(4): 915－926.

[7] 劉偉. 植被覆蓋地表極化雷達(dá)土壤水分反演與應(yīng)用研究[D].北京:中國科學(xué)院遙感應(yīng)用研究所，2005. Liu Wei. Study on Soil Moisture Inversion and Application with Polarization Radar in Vegetated Area[D]. Beijing: Institute of Remote Sensing Applications, Chinese Academy of Sciences, 2005. (in Chinese with English abstract)

[8] Fung A K, Chen K S. An update on the IEM surface back scattering model[J]. IEEE Transactions on Geoscience and Remote Sensing, 2004, 1(2): 75－77.

[9] Notarnicol A, C. A bayesian change detection approach for retrieval of soil moisture variations under different roughness conditions[J]. IEEE Geoscience and Remote Sensing Letters. Lett., 2014, 11: 414－418.

[10] Gorrab A, Zribi M, Baghdadi N, et al. Retrieval of both soil moisture and texture using Terra SAR-X images[J]. Remote Sens., 2015, 7: 10098－10116.

[11] 宋書藝. 濱海土壤微波介電特性研究[D]. 杭州：浙江大學(xué)，2012. Song Shuyi. Study on Microwave Dielectric Property of Coastal Soil[D]. Hangzhou: Zhejiang Uiversity, 2012. (in Chinese with English abstract)

[12] 夏明耀，陳志雨. 空間波法測(cè)地物介電常數(shù)的反演方法[J].電子與信息學(xué)報(bào)，1999，21(2)：252－257. Xia Mingyao, Chen Zhiyu. Inverse formulations for permittivity measurement of ground materials using space-wave method[J]. Journal of Electronics, 1999, 21(2): 252－257. (in Chinese with English abstract)

[13] 賈明權(quán). 典型地物微波介電特性實(shí)驗(yàn)研究[D].成都：電子科技大學(xué)，2008. Jia Mingquan. Experimental Research on Characteristics of Microwave Dielectric Properties of Typical Terrestrial Objects[D]. Chengdu: University of Electronic Science and Technology, 2008. (in Chinese with English abstract)

[14] 曾江源，李震，陳權(quán)，等. SAR土壤水分反演中的介電常數(shù)實(shí)部簡化模型[J]. 紅外與毫米波學(xué)報(bào)，2012，31(6)：556－562. Zeng Jiangyuan, Li Zhen, Chen Quan, et al. A simplified model of the real part of the soil complex permittivity for soil moisture estimation from SAR image[J]. J.Infrared Millim. Waves, 2012, 31(6): 556－562. (in Chinese with English abstract)

[15] 雷磊，塔西甫拉提·特依拜，丁建麗，等. 干旱區(qū)鹽漬土介電常數(shù)特性研究與模型驗(yàn)證[J]. 農(nóng)業(yè)工程學(xué)報(bào)，2013，29(16)：125－133. Lei Lei, Tashpolat·Tiyip, Ding Jianli, et al. Constant characteristic and model verification of saline soil dielectric in arid area[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2013, 29(16): 125－133. (in Chinese with English abstract)

[16] 劉華貴，曾健. 土壤介電常數(shù)—含水量關(guān)系模型的比較[J].土工基礎(chǔ)，2011，25(2)：58－64. Liu Huagui, Zeng Jian. Deformation regularity during shallow buried tunnel construction in urban[J]. Soil Engineering and Foundation, 2011, 25(2): 58－64. (in Chinese with English abstract)

[17] 趙昕，黃妮，宋現(xiàn)鋒，等. 基于Radarsat 2與Landsat 8協(xié)同反演植被覆蓋地表土壤水分的一種新方法[J]. 紅外與毫米波學(xué)報(bào)，2016，35(5)：519－616. Zhao Xin, Huang Ni, Song Guanfeng, et al. A new method for soil moisture inversion in vegetation-covered area based on Radarsat 2 and Landsat 8[J]. J. Infrared Millim Waves, 2016, 35(5): 519－616. (in Chinese with English abstract)

[18] 李彪. 基于Radarsat-2雷達(dá)影像和BP人工神經(jīng)網(wǎng)絡(luò)的土壤墑情監(jiān)測(cè)研究[D]. 呼和浩特：內(nèi)蒙古農(nóng)業(yè)大學(xué)，2015. Li Biao. The Study of Monitoring Soil Moisture Based on Radarsat-2 Remote Sensing Images and BP Artificial Neural Network[D]. Hohhot: Inner Mongolia Agricultural University, 2015. (in Chinese with English abstract).

[19] 李成鋼. 冬小麥微波散射特性及參數(shù)反演研究[D].成都：電子科技大學(xué)，2013. Li Chengang. The Characteristics of Microwave Scattering of Winter Wheat and the Study of Parametric Inversion[D]. Chengdu: University of Electronic Science and Technology, 2013. (in Chinese with English abstract)

[20] 郭曼. 基于SAR數(shù)據(jù)的稀疏植被覆蓋條件下的地表土壤水分反演研究[D]. 烏魯木齊：新疆大學(xué)，2012. Guo Mang. Study on Soil Moisture Inversion of Sparse Vegetation Cover Conditions Based on SAR Data[D]. Urumqi: Xinjiang University, 2012. (in Chinese with English abstract)

[21] 鄭磊. 基于微波遙感的裸露地表土壤水分反演研究[D].呼和浩特：內(nèi)蒙古農(nóng)業(yè)大學(xué)，2014. Zheng Lei. Research on Bare Surface Soil Moisture Inversion Based on The Microwave Remote Sensing[D]. Hohhot: Inner Mongolia Agricultural University, 2015. (in Chinese with English abstract)

[22] RADARSAT-2 衛(wèi)星介紹[Z]//中國科學(xué)院對(duì)地觀測(cè)與數(shù)字地球科學(xué)中心·用戶簡訊，北京：2008(73): 1－12.

[23] Zribi, Dechambre. Surface soil moisture estimation from the synergistic use of the (multi-incidence and multi-resolution) active microwave ERS Wind Scatterometer and SAR data[J]. Remote Sensing of Environment, 2003, 86(1): 30－41.

[24] Owe M, De Jeu R, Holmes T. Multisensor historical climatology of satellite-derived global land surface moisture[J]. Journal of Geophysical Research: Earth Surface (2003－2012), 2008, 113(F01002): 1－17.

[25] Oh Y. Quantitative and retrieval of soil moisture content and surface roughness from multipolarized radar observations of bare soil surfaces[J]. IEEE Transactions on Geoscience and Remote Sensing ,2004, 42(4): 596－601.

[26] 張廷軍，晉銳，高峰. 凍土遙感研究進(jìn)展—可見光、紅外及主動(dòng)微波衛(wèi)星遙感方法[J]. 地球科學(xué)進(jìn)展，2009，24(9)：963－972. Zhang Tingjun, Jing Rui, Gao Feng. Over view of thesatellite remote sensing of frozen ground: visible thermal infrared and radar sensor[J]. Advances in Earth Science, 2009, 24(9): 963－972. (in Chinese with English abstract)

[27] 梁志剛，陳云敏，陳贇. 利用同軸電纜電磁波反射技術(shù)測(cè)定非飽和土的含水率[J]. 巖土工程學(xué)報(bào)，2006，28(2)：191－195. Liang Zhigang, Chen Yunming, Chen Yun. Measurement of water content of unsaturated soil by TDR technique[J]. Chinese Journal of Geotechnical Engineering, 2006, 28(2): 191－195. (in Chinese with English abstract)

[28] Hallikainen M T, Ulaby F T, Dobson M C, et al. Microwave dielectric behavior of wet soil: part I:Empiricalmodels and experimental observations[J]. IEEE Transactions on Geoscience and Remote Sensing, 1985, GE-23(1): 25－34.

[29] 陳權(quán)，曾江源，李震，等. 遙感監(jiān)測(cè)介電常數(shù)與土壤含水率關(guān)系模型[J]. 農(nóng)業(yè)工程學(xué)報(bào)，2012，28(12)：171－175. Chen Quan, Zeng Jiang Yuan, Li Zhen, et al. Relationship model of soil moisture and dielectric constant monitored with remote sensing[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2012, 28(12): 171－175. (in Chinese with English abstract)

[30] 任鑫. 多極化—多角度SAR土壤水分反演算法研究[D].北京：中國科學(xué)院遙感應(yīng)用研究所，2003. Ren Xin. A Surface Moisture Inversion Technique Using Multi-Polarization and Multi-Angle Radar Images[D]. Beijing: Institute of Remote Sensing Applications, Chinese Academy of Sciences, 2003. (in Chinese with English abstract)

Inversion and verification of salinity soil moisture using microwave radar

Wang Xue, Liu Quanming※, Qu Zhongyi, Wang Liping, Li Xiangjun, Wang Yaoqiang
(1. Water Conservancy and Civil Engineering College, Inner Mongolia Agricultural University, Hohhot 010018, China)

Soil dielectric constant is the physical basis for soil moisture simulation based on microwave remote sensing, and especially the real part of the dielectric constant is of great significance to the research of the soil dielectric characteristics. Main aim of this study was to investigate capability of C-band RADARSAT-2 SAR (synthetic aperture radar) data applied in the soil dielectric characteristics monitoring and the soil moisture inversion over agricultural fields. Bare area of Jiefangzha sub-district of Hetao Irrigation District in Inner Mongolia of China was selected as the study region, which was influenced by soil salinization seriously. In order to achieve above purposes, an image of Radarsat-2 SAR was bought in April 2016, which has a kind of four fine polarization SLC (single look complex) format, covering an area of 25 km × 25 km with 8 meter ground resolution. Taking spatial uneven distribution of the saline soil into account, 100 sampling points were designed in the study area, and soil digging depth was 10 cm. Hand-held GPS (global positioning system) receiver was used to record coordinates of the sampling points. The experiment data included the soil dielectric real constant, surface roughness, surface temperature, percentages of clay and sand particles, soil bulk density and soil moisture. Agilent microwave network analyzer was used to measure the real part value of soil dielectric constant with coaxial probe method. Surface roughness was measured using centimeter grid profile plate to calculate the value of RMS (root mean square) height and the correlation length, and then composite roughness was got to represent the surface roughness in later research. Real-time ground temperature of the sampling points was measured by geothermometer. Particle analysis was fulfilled with laser particle size analyzer named Helos/B, obtaining the percentage content of clay and sand particles. Soil bulk density was measured by ring cutter. Soil moisture was measured by way of drying. SAR scape module of ENVI software was mainly used to perform the radar image processing, including radiometric calibration, geometric correction, slant range turning and filtering. Four polarization back scatter coefficient values corresponding to the sampling points were extracted based on previous results by spatial analysis module of ArcGIS software. In order to analyze complex relationship between the real part of the dielectric constant with SAR four polarization back scattering coefficients and surface roughness, firstly Oh empirical model was established, for which the relative relationship was significant between simulated and measured soil moisture, and the value of R2was 0.8209. Results showed that Oh model can offer precise real part value of the dielectric constant to inverse the soil moisture based on the soil dielectric model by means of the remote sensing and surface roughness data. Then Dobson semi-empirical dielectric models and simplified Hallikainen real part experience model were verified, and the R2between the measured and simulated real part values was 0.935 9 and 0.869 respectively, which indicated that the 2 models can simulate close relationship of the surface soil moisture and the real part of the dielectric constant. Finally Dobson model and Hallikainen simplified real part soil moisture inversion model were constructed. Compared with the statistical regression model, it looked like that relative relationship between simulated and measured value was significant, and the value of R2was 0.803 8, 0.737 4, and 0.842 1, respectively, for the former 2 models and the statistical regression model, the RMSE (root mean square error) value was 5.2%, 5.7%, and 5% respectively. The inversion results of Dobson model and statistical regression model were similar with the field soil moisture distribution, so they had good precision and applicability. Without considering the surface roughness, the simulation result of Hallikainen simplified real part model was then slightly worse than the other 2 models. The soil dielectric characteristics researching system and the moisture retrieval models established in this study can promote the application of the microwave remote sensing in the soil moisture monitoring.

soil moisture; remote sensing; models; soil dielectric properties; Oh model; Dobson model; Hallikainen simplified real part model; salinization

10.11975/j.issn.1002-6819.2017.11.014

S152.7; P628.2

A

1002-6819(2017)-11-0108-07

王 學(xué)，劉全明，屈忠義，王麗萍，李相君，王耀強(qiáng). 鹽漬化土壤水分微波雷達(dá)反演與驗(yàn)證[J]. 農(nóng)業(yè)工程學(xué)報(bào)，2017，33(11)：108－114.

10.11975/j.issn.1002-6819.2017.11.014 http://www.tcsae.org

Wang Xue, Liu Quanming, Qu Zhongyi, Wang Liping, Li Xiangjun, Wang Yaoqiang. Inversion and verification of salinity soil moisture using microwave radar[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2017, 33(11): 108－114. (in Chinese with English abstract) doi：10.11975/j.issn.1002-6819.2017.11.014 http://www.tcsae.org

2017-01-03

2017-03-11

國家自然科學(xué)基金項(xiàng)目（51249007、51569018、51169016）；內(nèi)蒙古自然科學(xué)基金項(xiàng)目（2013MS0609）；“十三五”國家重點(diǎn)研發(fā)計(jì)劃項(xiàng)目（216YFC0501301）

王 學(xué)，男，山東濟(jì)南人，主要從事定量遙感反演理論及應(yīng)用研究。呼和浩特 內(nèi)蒙古農(nóng)業(yè)大學(xué)水利與土木建筑工程學(xué)院，010018。

Email：sdzqwx@126.com

※通信作者：劉全明，男，內(nèi)蒙古四子王旗人，副教授，博士，主要從事測(cè)繪工程教育與定量遙感反演理論及應(yīng)用研究。呼和浩特 內(nèi)蒙古農(nóng)業(yè)大學(xué)水利與土木建筑工程學(xué)院，010018。Email：nndlqm@sina.com