【摘要】：高溫目標(biāo)溫度顯著高于常溫地物，，包括煤層自燃、土法煉焦、林火、草原火、油井火炬、火山噴發(fā)等，高溫目標(biāo)識(shí)別及溫度反演方法研究對(duì)遙感資源調(diào)查、環(huán)境監(jiān)測(cè)、災(zāi)害預(yù)警等具有重要的理論意義和實(shí)用價(jià)值。 相較于熱紅外數(shù)據(jù)，短波紅外波段數(shù)據(jù)更能精確識(shí)別和定位高溫目標(biāo)。本文以地表為朗伯體、常溫地物與高溫目標(biāo)成分和溫度是均一的、常溫地物與高溫目標(biāo)的輻射能量為線性疊加為前提，并基于普朗克函數(shù)、能量守恒等原理，建立地表處短波紅外混合像元溫度反演模型并進(jìn)行敏感性分析及參數(shù)估算，在此基礎(chǔ)上采用馬氏距離、因子分析法進(jìn)行高溫目標(biāo)的識(shí)別，對(duì)識(shí)別結(jié)果進(jìn)行溫度反演來(lái)驗(yàn)證模型的可行性并與熱紅外溫度反演結(jié)果相對(duì)比。主要研究成果如下： 1．由短波紅外高溫目標(biāo)溫度反演模型，確定了大氣透過(guò)率Tθ、常溫背景反射率ρ、高溫目標(biāo)發(fā)射率ε、高溫目標(biāo)面積百分比S和溫度T共5個(gè)關(guān)鍵參數(shù)，并采用控制變量法對(duì)各參數(shù)進(jìn)行敏感性分析。對(duì)參數(shù)的估算及求解中，大氣透過(guò)率Tθ可以通過(guò)輻射傳輸模型MODTRAN反演出來(lái)，常溫背景反射率ρ是由背景像元估算法獲取的，對(duì)于土法煉焦高溫目標(biāo)發(fā)射率ε是通過(guò)對(duì)野外采集的煤、炭、焦炭樣品的反射率反算獲取的，而其它類高溫目標(biāo)的發(fā)射率用近似黑體輻射代替，這些參數(shù)的求解及估算為溫度反演奠定了基礎(chǔ)。 2．短波紅外高溫目標(biāo)識(shí)別與溫度反演的模型的適用性研究結(jié)果表明：可識(shí)別的最低溫度為525K（高溫目標(biāo)充滿整個(gè)像元時(shí)）、最小面積百分比為0.0003。也就是說(shuō)，即使高溫目標(biāo)面積很小，但當(dāng)溫度達(dá)到一定程度，仍可被識(shí)別出來(lái)。 3．高溫目標(biāo)識(shí)別結(jié)果表明：馬氏距離多元截尾法、馬氏距離多類判別法和因子分析法用于高溫目標(biāo)識(shí)別，識(shí)別精度分別為82%、79%和85%，能夠?qū)⒏邷啬繕?biāo)有效的識(shí)別出來(lái)。 4．對(duì)典型高溫目標(biāo)土法煉焦、林火、油井火炬的溫度反演，結(jié)果表明：短波紅外溫度反演結(jié)果與實(shí)際溫度（500K±）相吻合，而基于輻射傳輸方程法的熱紅外溫度反演結(jié)果與背景較為相近，易混淆，即熱紅外遙感數(shù)據(jù)并未能較好地反映出高溫目標(biāo)特性。
[Abstract]:The temperature of high temperature target is significantly higher than that of normal temperature ground object, including coal seam spontaneous combustion, soil coking, forest fire, grassland fire, oil well torch, volcanic eruption, high temperature target recognition and temperature inversion, investigation of remote sensing resources and environmental monitoring. Disaster early warning has important theoretical significance and practical value. Compared with thermal infrared data, shortwave infrared band data can accurately identify and locate high-temperature targets. In this paper, the surface is the Lambert body, the composition and temperature of the ground object at room temperature and the target at high temperature are uniform, and the radiation energy between the ground object at room temperature and the target at high temperature is linearly superimposed as the premise, and based on the Planck function and the principle of energy conservation, The temperature inversion model of short-wave infrared mixed pixel on the surface is established and the sensitivity analysis and parameter estimation are carried out. On this basis, Mahalanobis distance and factor analysis are used to identify the high-temperature target. The feasibility of the model is verified by temperature inversion of the recognition results and compared with the results of thermal infrared temperature inversion. The main results are as follows: 1. The atmospheric transmittance T 胃, the ambient background reflectivity 蟻 and the high temperature target emissivity 蔚 are determined by the inversion model of the high temperature target temperature of short wave infrared. The target area percentage of high temperature (S) and temperature T (T) are five key parameters, and the sensitivity of each parameter is analyzed by using the control variable method. In the estimation and solution of the parameters, the atmospheric transmittance T 胃 can be retrieved by the radiative transfer model MODTRAN, and the ambient temperature background reflectivity 蟻 is obtained by the background pixel estimation algorithm. For the emissivity of high temperature targets in soil coking, the emissivity is obtained by inverse calculation of the reflectivity of coal, carbon and coke samples collected in the field, while the emissivity of other high temperature targets is replaced by approximate blackbody radiation. The solution and estimation of these parameters lay a foundation for temperature inversion. 2. The applicability of the model of shortwave infrared high temperature target recognition and temperature inversion shows that the minimum recognizable temperature is 525K (when the high temperature target is filled with the whole pixel), and the minimum area percentage is 0.0003. In other words, even if the target area of high temperature is very small, when the temperature reaches a certain level, it can still be recognized. 3. The results of high temperature target recognition show that the recognition accuracy of Markov distance multivariate truncation method, Markov distance multi-class discriminant method and factor analysis method is 82%, 79% and 85%, respectively. Can effectively identify high temperature targets. 4. The temperature inversion of typical high temperature target soil coking, forest fire and oil well torch shows that the results of short wave infrared temperature inversion are in good agreement with the actual temperature (500K 鹵). The thermal infrared temperature inversion results based on the radiative transfer equation method are close to the background, which is easy to be confused, that is, the thermal infrared remote sensing data do not reflect the characteristics of the high temperature target well.
【學(xué)位授予單位】：吉林大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：TP722.5

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 徐希孺,柳欽火,陳家宜;遙感陸面溫度[J];北京大學(xué)學(xué)報(bào)(自然科學(xué)版);1998年Z1期

2 紀(jì)宏金;地球化學(xué)背景與異常劃分的多元方法[J];長(zhǎng)春地質(zhì)學(xué)院學(xué)報(bào);1988年03期

3 紀(jì)宏金;多元正態(tài)總體假設(shè)檢驗(yàn)在礦化帶識(shí)別中的應(yīng)用[J];長(zhǎng)春地質(zhì)學(xué)院學(xué)報(bào);1991年03期

4 孔祥生,苗放,劉鴻福,董宇陽(yáng);遙感技術(shù)在監(jiān)測(cè)和評(píng)價(jià)土法煉焦污染源中的應(yīng)用[J];成都理工大學(xué)學(xué)報(bào)(自然科學(xué)版);2005年01期

5 毛克彪,覃志豪,劉偉;用MODIS影像和單窗算法反演環(huán)渤海地區(qū)的地表溫度[J];測(cè)繪與空間地理信息;2004年06期

6 覃志豪,Zhang Minghua,Arnon Karnieli,Pedro Berliner;用陸地衛(wèi)星TM6數(shù)據(jù)演算地表溫度的單窗算法[J];地理學(xué)報(bào);2001年04期

7 丁鳳;徐涵秋;;TM熱波段圖像的地表溫度反演算法與實(shí)驗(yàn)分析[J];地球信息科學(xué);2006年03期

8 戎志國(guó);劉誠(chéng);孫涵;馬輪基;盧乃錳;劉京晶;張玉香;鐘仕全;張艷;張鵬;張甲?;李亞君;張行清;馬瑞升;王君華;;衛(wèi)星火情探測(cè)靈敏度試驗(yàn)與火情遙感新探測(cè)通道選擇[J];地球科學(xué)進(jìn)展;2007年08期

9 王麗美;姜永濤;孫林;王春香;;MODIS分裂窗算法反演地表溫度[J];地理空間信息;2011年06期

10 時(shí)艷香,紀(jì)宏金,陸繼龍,馬力,段國(guó)正;水系沉積物地球化學(xué)分區(qū)的因子分析方法與應(yīng)用[J];地質(zhì)與勘探;2004年05期

本文編號(hào)：2473982