下载  >  课程资源  >  专业指导  > A simple “clear water” atmospheric correction algorithm for Landsat-5 sensors

A simple “clear water” atmospheric correction algorithm for Landsat-5 sensors 评分:

In this study, an atmospheric correction algorithm (improved “clear water” atmospheric correction algorithm for broad-band sensors, or “ICAB” algorithm) was constructed and tested, and the results are shown in this paper. The Rayleigh scattering contributions at three visible bands and one near-infr
International Journal of Remote Sensing period, thus polluting the lake water. The blue-green algae (Microcystis)bloom event of 85 spring 2007 in Taihu Lake caught the worlds'attention. Algae-polluted waters in the lake have adversely affected the lives of the several million local residents. Thus, there is an urgent need to effectively monitor and manage the water quality in Taihu Lake, and to better understand the optical, biological, and ecological processes and phenomena occurring in this lake(Wang, Shi, and Tang 2011) Recent advances in optical sensor technology have made it possible to study biogeochemical processes in aquatic environments at spatial and temporal scales, which was not possible previously(Clavano, Boss, and Lee 2007). These advances allowed Sci- entists to utilize satellite images to synoptically investigate large-scale surface features in Taihu Lake(Chen, Wen, and Xiao 2011b; Chen et al. 2010; Yang et al. 2006). However,95 the water-leaving reflectance detected by satellite sensors is fairly weak. For example, the component of the measured refectance, backscattering out of the water and transmitting to the top of the atmosphere, is less than 10% in the blue band, and typically much smaller AQ2 in the green and near-infrared bands( gordon and Castano 1989; Gordon and clark 1981) In order to obtain useful water-leaving information from Taihu Lake, one must remove the 100 atmospheric absorption and scattering effects. Thus, it is necessary to perform research on atmospheric correction in this area to effectively monitor and manage water colour in Taihu Lake 2.2, Data used The field measurements recorded from 27 to 2 8 October 2003 in Taihu lake. from 1 to 9 105 September 2009 in the Yellow River estuary and from 4 to 15 October 2010 in the Changjiang(Yangtze) River Estuary were used to construct an empirical spectral relation- ship for improving the limitations of the ' clear water'method in broadband sensors. The Landsat-TM images obtained on 28 October 2003 were used to evaluate the performance f the ICaB algorithm in Taihu Lake. Additionally, to ac the inft of110 Rayleigh scattering, the look-up table of Rayleigh and atmospheric parameters, including total ozone(262 Dobson units(DU)(1 DU =2.69 x 1020 molecules m 2)), zonal wind speeds at 10 m above the water surface(0.44 ms), and the atmospheric pressure at mean sea level(1023. 8 mb) on 28 October 2003 for Taihu Lake, was also collected from the official Nasa website 115 In general, a scale problem exists when comparing satellile-inferred and observed sur- face reflectance(zhao, Tamura, and Takahashi 2000). In the case ofLandsal-TM imageries the spatial resolution is 30 m and the size of one Landsat pixel is 30 m x 30 m. The mea- sured satellite reflectance value represents the average value of that area. However, when measuring surface reflectance using a spectrometer in the field, the value of only one loca- 120 tion may be obtained. therefore, in principle surface reflectance estimated from Landsat cannot be directly compared with an observed surface reflectance, unless the field observa- tion sites had been set up in a large homogeneous surface area(larger than 30 m x 30 m) thus the observation sites for the validation data sets were chosen in a large area (larger Lhan a landsat pixel size)in which a generally homogeneous water colour exists 125 Reflectance at top-of-atmosphere in tM band 2(R,(TM2)) is highly variable(Dekker, AQ3 VoS, and Peters 2002). Many factors contribute to the reflectance features found in tM band 2, mainly consisting of absorption by dissolved organic matter and chlorophyll-a, as well as backscattering by suspended particles( Gitelson et aL. 2008). In this study, the imagery associated with Rt(TM2)was zoomed in at the pixel-visual level, where the pixel size may 130 be discriminated by the visual determination method. These procedures were performed Chen et al COLOUR FIGURE 12).4°E N 人 Taihu lake INP+ 7F14H Fi Transects of R,(TM2)at five synchronous field stations AQ24 using the Environment for Visualizing Images(ENVI) image processing software. The AQ4 pixels associated with R,(TM2)around five synchronous field stations were selected; the corresponding transects are shown in Figure 1. It was found that the r,(tM2) around the field station was very uniform, which also supports the authors previous determination in135 bio-optical experiments that the observation sites of a validation data set should be chosen in a large area in which a generally homogeneous water colour exists 23. Field measurements The field measurements of three data sets were performed from 10: 00 to 14: 00 local tir At each station. normalized water-leaving reflectance measurements were recorded from a 140 boat. The reflectance was measured with a spectroradiometer with 5 fibre-optic, covering AQ5 the spectral range of 350-2500 nm(Spectral Devices, Boulder, CO, ASD). Although data AQ6 were collected in the range of 350-2500 nm, with a spectral resolution of 3 nm(full width at half-maximum, FWHM) and a 1. 4 nm sampling interval for the 350-1050 nm spectral range(ASD1999), the main data used in this study were those in the range of 400-900 nm,145 which is the generally used wavelength for water colour remote sensing(Deng and Li 2003 Gons, Auer, and Effler 2008: Ouaidrai and Vermote 1999). During the recording of the measurements, the tip of the optical fibre was kept x l m above the water surface by means of a 3 m long, hand-held black pole. Radiance of both the water surface (Lsw(2))and a standard grey board (Lp(n) was measured(i is the wavelength). Ten curves were acquired 150 for each target. In order to effectively avoid the interference of the ship with the water surface and the infuence of direct solar radiation, the instrument was positioned at an angle B of 90-135 to the plane of the incident radiation and away from the Sun, as shown AQ7 in Figure 2(Mueller et al. 2003; Mueller and Fargion 2002). The viewing angle of the water AQ8 surface, a, was controlled between 30 and 45 with the vertical direction. In this way most 155 AQ9 of the direct sunlight was eliminated while the impact of the ship's shadow was minimized (Mueller et al. 2003; Mueller and Fargion 2002). Immediately after measuring the water radiance, the spectroradiometer was rotated upwards by 90 120 to measure skylight. The AQ10 International Journal of Remote Sensing ASD Airl Figure 2. Viewing geometry of the above-water measurement. AQ25 view azimuth angle in this measurement was kept the same as that when measuring the water radiance(Mueller and Fargion 2002) Remote-sensing reflectance Rrs(1) was calculated as follows Rrs (n) (λ) where Lw()is the water-leaving radiance and Edo+(n) is the total incident radiant flux of the water surface. The terms in Equation(1)were further calculaled as follows Lw()= Lsw(A)-rlsky(i) (2) Op where Lsky()represents the diffuse radiation from the sky, which contains no information about the waters properties and hence must be eliminated; r is the radiation from the 1 sky reflected at the air-water interface, the value of which depends on the solar azimuth, measurement geometry, wind speed, and surface roughness; and pp()is the reflectance of the grey plate. In this study, r is calculated by assuming that the water is a blackbody at wavelengths from 1000 to 1020 nm(Hale and Querry 1973)and by assuming that r is wavelength-independent Doxaran et al. 2002 ). The remote-sensing reflectance curves 170 obtained from Equation(1)are shown in Figure 3 AQ32 2.4. ICAB algorithm The radiance received by a sensor at the lop-of-atmosphere in a spectral band centred on a wavelength A, L(), may be divided into the following components: Lpath(A), the radiance generated along the optical path by scattering in the atmosphere and by specu-175 ar reflection of atmospherically scattered light(skylight) from the water surface; Lg(i) the contribution arising from specular reflection of direct sunlight from the water surface (Sun's glitter); Lwe(2), the contribution arising from sunlight and skylight reflecting from individual whitecaps on the water surface; and lw(), the desired water-leaving radiance Chen et al 0.09 0.08 0.07 0.06 0.05 0.03 0.02 0.01 0 400450500550600650700750800850900 Wavelength(nm) Figure 3. Spectral curves collected from Taihu Lake, Yellow River Estuary, and Changjiang river Estuary, China, respectively AQ26 Lt(入)=Lpa()+T(2)Lg()+1(2)Lwe()+t(x2)Lw() (4) In this equation, T() and t(h)are the direct and diffused transmittance of the atmo-180 sphere, respectively. Near the Sun's glitter pattern, T(A)Lg(n)is so large that the imagery AQ1 virtually useless and must be discarded. Away from the glitter pattern, I()Lg() becomes negligibly small (Gordon and Voss 1999; Hu et al. 2001). The whitecap contributions, t()Lwc(), can be estimated using the surface wind speed (Siegel et al. 2000), and then he largest of the remaining terms, and most dillicult lo estimate, is Lpath(2). Neglecting 185 the contribution of T()Lg(n) and t(aLwe(i), and converting to reflectance, the equation becomes(Chen, Fu, and Zhang 2011a) R1(入)=R1(1)+R1()+R(λ)+t(入)Rr(), (5) where Rr(), Ra(A), and Rra(i) represent the contributions from atmospheric scattering that result from air molecules(Rayleigh), aerosols, and raylcigh-acrosol interactions, respec- tively. In this equation, the R()at all the visible and niR wavelengths depends on the 190 atmospheric molecular composition of the atmosphere, the Sun, and the viewing geometry AQ12 (Gordon and Castano 1989; Gordon and Voss 1999), and to a lesser extent on water-surface roughness( Gordon and Wang 1992). Hence the Rr() may be computed accurately without the use of remotely sensed data(Lavender et al. 2005). Therefore, at niR bands(wave- length= ANIR)in 'clear waters, Rrs is approximated to zero (Gordon and Franz 2008;195 Gordon and Voss 1999; Zhang 2009)and then the contribution of ra(NIR)+ rraanir may be estimated (with its sum defined as Ras(NiR))as follows: AQ33 ras (anir)=ra (anir ) Rra (anir Rt (anir-Rr(anir) (6) AQ13 If Ras(n)at two different niR bands are available, then the contributions of ras ()in the visible may be extrapolated using the formula suggested by gordon and Wang(1994) exp [n(1. Furthermore, n is the Angstrom exponent and may be estimated from the given Ras (a)at 200 wo different Nir bands. However, there is only one nir band for most broadband sensors International Journal of Remote Sensing such as Landsat, CBERS, HJ-A, and so on, so this equation cannot be used to extrapolate the value of Ras (a) from niR band to visible bands. To improve the limitation of the clear water'method in broadband sensors, an empirical spectral relationship is constructed by field measurements and is assumed to be of the following form Rs(λ2)=aRs(λ1)+b where Rrs( 1) and Rrs(2)are the respective remote-sensing reflectances at two visi ble bands measured just above the water-air surface; a and b are empirical coefficients AQ14 Combining Equations (5)and( 8) yields the following Rt(21)-R1(入1) b Rt(22)-R(入2)aRas(入1)Ras(22) t(1) (λ1) t (n)=exp 0.5τr()+τoz(入)+(1 )Ta() (10) cos Bo where 0o is the solar zenith angle; to(i)is the ozone optical depth; F is the probabil ot the photo ol scattering, will be scattered through an 210 angle of <90; and oo is the aerosol single-scattering albedo However, in practice the term AQ15 (1 oFta()is always unknown. Fortunately, as discussed by Gordon and Clark(1981) and Gordon et al. (1999), the term(I-( F)is usually less than 1/ 6 for most aerosol AQ16 models, and hence for an aerosol optical thickness of < 06(probably the upper limit for which is the reasonable event to attempt atmospheric correction of ocean colour imagery 215 Gordon and Castano 1989)the error introduced by ignoring this term completely is only 10% for small solar zenith angles, which is accurate enough for the purposes of this study Gordon and Voss 1999; Gordon and Wang 1994; Hu and Carder 2002). Hence, t(i) may be computed approximately by completely ignoring the aerosol component, giving the following 0.5T(入)+τo(2) t()=exp In this study, at any surface pressure P, the Rayleigh optical depth tr(n) is as follows (Hansen and Travis 1974) P Ur (12) where tro(1) represents the rayleigh optical depth at the standard atmospheric pressure Po, of 1013. 25 mb. In this study, tr()is calculated using Gordon et al. ' s method( Gordon et aL. 1988). Due to the fact that the Landsat-TM is a broadband scanning radiometer, the225 tr(n)estimated by this method should be an average weighted by the Landsat-TM sensor spectral response functions, as follows r(M:) (13) J. Ch kor(A)represents the ozone absorption coefficient defined so that the ozone spectral 3o o.7> where TMi refers to the ith band of the Landsat sensor, f(i) is the landsat/TM sensor spectral response function for band and s is the sample count of spectral response funclions optical depth for an ozone concentration in du (Dobson units or milliatmosphere cm)is as follows( Gordon and Voss 1999) (DU) 1000 where DUis the ozone concentration in Dobson units. If there is at least one pixel in the study area covered by 'clear waters, the Ras for tM band 4, ras(tM4), may be estimated235 using the clear water'method advised by Gordon and Voss(1999) Ras(TM4)≈R1(TM4)一R+(TM4) (15) Substituting Equations(7)and(15)into Equation(10) yields the same expression, but the result also includes the unknown parameter n 1(x1)-R(k1)+b-RA(2)-R(x2)=Aa(TM4)8(1,TM4 t(1) t(入2) t(1) Ras(TM4)E(λ2,TM4) (16) t(λ2) here r(i and R (n2)are the remote-sensing reflectance of the clear water'observed at the top-of-atmosphere at wavelengths A1 and A2, respectively. Therefore, as long as the 240 clear water'pixel is known, Ras(TM4)and n may be estimated from Equations (14)and (15) using the non-linear regression method, respectively. Afterwards, the water-leaving reflectance may also be estimated by the following formula Rrs(.)=Rt(.)-Rr(a)-Ras(TM4)exp [n(INIR-2)1 t(λ) 3. Discussion and results 3.I. Construction of the empirical spectral relationship for the /CAB algorithm 245 To match the bandwidth of Landsat-TM, the ASD measurements and remote-sensing reflectance were aggregated using the Landsat-TM sensor spectral response functions before model calibration(Chen, Wen, and Xiao 2011b). In order to determine the optimal empirical spectral relationship and eliminate the limitations of the 'clear water method used in broadband sensors the correlation coefficient matrix of the visible and nir bands 250 of Landsat sensors was constructed and is shown in Table 1. According to Table 1, the correlations in the remote-sensing reflectances between different bands produce differ- ent correlation coefficients, e.g. the correlation coefficient of 485 nm versus 560 nm is the highest, and that of 48.5 nm versus 825 nm is the lowest. Clearly, the empirical relationship of 485 nm versus 560 nm is greater than others as an empirical spectral rela-255 tionship for the ICAB algorithm. Figure 4 reveals that the remote-sensing reflectance at 485 nm is linearly dependent on the reflectance at 560 nm; the correlation coefficien for this relation may reach 0.9427(n= 87). Consequently, it is reasonable to choose AQ19 Rrs(tmi)= 1. 5147Rrs(TM2 )as an empirical spectral relationship for the ICab algorithm The empirical spectral relationship is constructed using the three respective independent 260 International Journal of Remote Sensing Table 1. Correlation coefficients between remote-sensing reflectance values in the different bands of the Landsat-TM Sensor Wavelength (nm) 485 560 660 830 0.9427 0.6847 04415 560 0.9427 0.8077 0.5316 0.6847 0.8077 0.7778 0.4415 0.5316 0.7778 0.08 1.514 0.07 R2=09427 0.06 E E0.05 COLOUR 0.04 FIGURE 足0.03 0.02 0.01 0.015 0.025 0.03 0.035 0.045 R、(485mm)(sr Figure 4. Empirical relationship between Rrs(485 nm) and Rrs (560 nm) data sets collected from Yellow River Estuary, Changjiang River Estuary, and Taihu Lake The applicability of the ICAB algorithm is mainly dependent on the empirical spectral rela Lionship thus it is possible to apply the ICab algorithm developed for Taihu Lake, China, to other areas, such as Changjiang River Estuary and Yellow River estuary. Variations in chlorophyll-a, which is the major cause of variations in water colour spec-265 tra in inland lakes and coastal regions, have the greatest effect between 400 and 470 nm, in terms of change in both shape and magnitude of the absorption spectrum( Garver and Sieg 1997; Tzortziou et al. 2007). Some small changes based on the ratio of photoprotectant to hotosynthetic carotenoids may have some bearing on the 480-560 nm spectral range but this is restricted to tropical or subtropical high light regimes (Smyth et al. 2006). Therefore, 270 the 485: 560 ratio is relatively unaffected by changes in chlorophyll-a and may be cho- sen as the pairing to construct the empirical spectral relationship for the ICab algorithm However, one limitation of this empirical spectral relationship is that the bio-optical data sets contained only a narrow range of optical properties of natural turbid waters, being col- lected from Yellow River Estuary, Changjiang River Estuary, and Taihu Lake. The results 275 are insufficient to completely evaluate the performances of the algorithms in other waters th different bio-optical prope Lion of algorithms based on more in situ measurements of waters with dilterent optical rties 3.2. Atmospheric correction of landsat imagery 280 3. 2.1. Rayleigh scattering contribution at the top-of-atmosphere Given the surface atmospheric pressure (to determine the value of T ()) and the sur- face wind speed (to define the roughness of the water surface), reflectance contributed 10 Chen et al by rayleigh scattering may be computed accurately, while even accounting for polarization effects(Gordon et al. 1988; Gordon and Wang 1992). In general, the reflectance associ- ated with Rayleigh scattering may be removed using the look-up table method suggested285 by gordon and Voss(1999)and Gordon and Wang(1994) Although the look-up table can be obtained from the SeaDAS software package pro vided by NASA, the rayleigh scattering contribution calculated from this look-up table cannot be directly used for broadband sensors such as Landsat-5, due to the differences in 290 spectral response functions and different band central wavelengths between Landsat-5 and MODIS Sensors. In this study, the look-up table method was used to compute the rayleigh scattering contribution and T ()of the nine visible and three nir bands of modis sen- sors by inputting the atmospheric parameters, including observed geometric and zonal winds speeds a 0 m above the water surface and the atmospheric pressure at mean sea 295 AQ34 level collected from the official Nasa website(ftp: \oceans. gsfc. nasa. gov)on 28 October 2003. Afterwards, an empirical relationship between rr(n) and tr(A) was constructed for estimating the Rr(a) of the three visible bands and one nir band of a Landsat-5 sensor Figure 5 shows the empirical relationship between Rr()and tr()regressed from the values of the nine visible and three nir bands of modis sensors calculated previously. 300 It was found that Rr(a) is quite linearly dependent on the value of tr(n): the coefficient of determination(r)for this may reach 0.9988. Accordingly, the Rayleigh scattering contri bution of Landsat-5 imageries recorded on 28 October 2003 can be calculated using this linear empirical relationship as long as the tr() of the three visible bands and one Nir band of the Landsat-5 sensor is known. For this study, the tr of the three visible bands and305 one Nir band of the Landsat-5 sensor may be estimated by Equation (13) 3.2.2. Aerosol scattering contribution at the top-of-atmosphere In the studies carried out by Gordon and Clark(1981): Gordon and Castano(1989)and Gordon and Voss(1994), the aerosol optical depth and aerosol type showed insignificant AQ20 or only slight changes at the range of below 50-100 km above the water surface, which 310 reveals that the aerosol scattering contribution is approximately unchanged in spatial, as long as the study area is small enough. This assumption is widely used in water colour remote sensing. Space does not permit a full review of this topic here instead the reader is referred to articles regarding this issue, such as those of Gordon et al. (1988); Gordon and Voss(1999); Gordon and Wang (1994): and Wang, Son, and Shi(2009). In this study,315 y=0.0911x+0.0002 0.025 R=0.9988 0015 COLOUR 0.01 FIGURE 0.005 0.05 0. 0.15 0.2 0.25 0.3 0.35 Figure 5. Empirical relationship between tr and R(a)

...展开详情
2013-08-22 上传 大小:811KB
举报 收藏
分享
精通C#游戏编程 C# Game Programming: For Serious Game Creation英文版

I wa nt to help you make your game. Everyone has a great game idea, but the path from initial idea to finished prod uct is not a clear one. There are an intimi dating number of progr amming languages, libraries, and prod uction methods. Even experienced game developers often fail to realize their vi

立即下载
html+css+js制作的一个动态的新年贺卡

该代码是http://blog.csdn.net/qq_29656961/article/details/78155792博客里面的代码,代码里面有要用到的图片资源和音乐资源。

立即下载
Camtasia 9安装及破解方法绝对有效

附件中注册方法亲测有效,加以整理与大家共享。 由于附件大于60m传不上去,另附Camtasia 9百度云下载地址。免费自取 链接:http://pan.baidu.com/s/1kVABnhH 密码:xees

立即下载
电磁场与电磁波第四版谢处方 PDF

电磁场与电磁波第四版谢处方 (清晰版),做天线设计的可以作为参考。

立即下载
压缩包爆破解密工具(7z、rar、zip)

压缩包内包含三个工具,分别可以用来爆破解密7z压缩包、rar压缩包和zip压缩包。

立即下载
算法第四版 高清完整中文版PDF

《算法 第4版 》是Sedgewick之巨著 与高德纳TAOCP一脉相承 是算法领域经典的参考书 涵盖所有程序员必须掌握的50种算法 全面介绍了关于算法和数据结构的必备知识 并特别针对排序 搜索 图处理和字符串处理进行了论述 第4版具体给出了每位程序员应知应会的50个算法 提供了实际代码 而且这些Java代码实现采用了模块化的编程风格 读者可以方便地加以改造

立即下载
jdk1.8下载

jdk1.8下载

立即下载
DroidCamX 6.5 电脑端和手机端(2018年版本)

DroidCamX 6.5 适配安卓8.0和win10系统。让你的安卓手机变成摄像头。

立即下载
身份证号对应籍贯表大全(共6456条)

身份证号对应籍贯表大全(共6456条),可以很方便查出身份证对应的籍贯,方便工作、项目使用

立即下载
DirectX修复工具V3.7在线修复版

DirectX修复工具(DirectX Repair)是一款系统级工具软件,简便易用。本程序为绿色版,无需安装,可直接运行。 本程序的主要功能是检测当前系统的DirectX状态,如果发现异常则进行修复。程序主要针对0xc000007b问题设计,可以完美修复该问题。本程序中包含了最新版的DirectX redist(Jun2010),并且全部DX文件都有Microsoft的数字签名,安全放心。 本程序为了应对一般电脑用户的使用,采用了傻瓜式一键设计,只要点击主界面上的“检测并修复”按钮,程序就会自动完成校验、检测、下载、修复以及注册的全部功能,无需用户的介入,大大降低了使用难

立即下载
同济大学线代第六版PDF高清扫描版

同济大学的线代第六版PDF高清扫描版 要考数学3的同学可以下载看下 上传记录里面还有考数3的其他资源 有需要的可以自行下载

立即下载
高等数学第七版(同济大学)下册pdf

高等数学第七版(同济大学)下册教材pdf (PS:高等数学第七版上下册均有,因上传文件容量有限,因此分为两次上传,请有需要上册的朋友点开我的资源下载页进行下载)

立即下载
Spring相关的外文文献和翻译(毕设论文必备)

Spring相关的外文文献和中文译文,毕业设计论文必备。SSM框架可使用。

立即下载
中国大学MOOC课件爬取(含视频)

实现对中国大学MOOC上的视频、文档、附件进行爬取的Python源码,无GUI、未打包exe,支持多进程、断点续传、文件结构同网页中显示结构。PS:此处为1.5.6版本,欢迎大家加我交流或者提建议(可直接获取最新版本)

立即下载
方方格子注册机

方方格子注册机,适用于方方格子所有的系列,全部系列均可以完美注册

立即下载
wifi密码字典包1G

1个G的wifi密码字典,跑包必备,目前大部分路由都关闭了wps,就算没关,也都有防pin,跑包虽然麻烦,但拥有一个强大的字典,成功率会大大提高。

立即下载
DroidCamX 专业版破解版6.7最新版

DroidCamX 专业版破解版6.7最新版,已经包含PC端和Android端

立即下载
PC版免费翻墙软件(电脑版VPN)

自己收藏的两款 翻墙软件可以使用,免费的并且是,但是只是PC版的 ,下载解压后 =》安装=》 右键'已管理员身份' 身份运行 即可使用,挺稳定的,希望对你有用。

立即下载
PC版VPN软件-Two free PC wall climbing software

两款免费的PC版 vpn软件,个人使用挺稳定的,特拿分享,希望对您有帮助!下载后 请以管理员方式进行运行

立即下载
13个scratch游戏源码

scratch小游戏合集,适合小朋友自学。包括贪吃蛇,大鱼吃小鱼,打砖块,走迷宫,格斗,飞机大战等

立即下载