基于分组分裂窗算法的MTSAT-1R地表温度反演

以黑河流域上游和中游为研究区,针对MTSAT-1R卫星数据,运用MODTRAN 4.0及晴空状态下的TIGR大气廓线数据,发展了根据地表比辐射率、大气水汽含量、传感器观测角度分组模拟的分裂窗算法,进行地表温度反演。分析了传感器噪声、地表比辐射率和大气水汽含量3个参数对该算法的影响,并结合模拟数据、地面观测数据及MODIS地表温度产品,对反演结果进行分析评价。结果表明:当传感器垂直观测或大气水汽含量小于2.5g/cm2时,反演精度在1K以内;反演结果与地面观测数据对比差异较小,在阿柔站RMSE为3.7 K(日)/1.4 K(夜),在盈科站RMSE为2.4K(日)/2.0K(夜);与MODIS地表温度产品比较,空间分布呈现出一致性。总之,分组分裂窗算法能较好地用于MTSAT-1R卫星数据进行地表温度反演。     

GF-5卫星模拟单通道SST反演

针对国内外热红外数据空间分辨率低、反演温度精度低的问题,提出了基于高分辨率GF-5卫星热红外通道修订的单通道海表温度(sea surface temperature,SST)反演算法。以大气传输模型软件(moderate resolution atmospheric transmittance and radiance code4.0,MODTRAN 4.0)与全球大气廓线数据(thermodynamic initial guess retrieval,TIGR)为基础,采用Jiménez-MuozSobrino's单通道算法和QIN单通道算法对GF-5卫星热红外数据进行SST的模拟反演研究。通过对比不同条件下的反演误差,结果表明,B11通道的反演精度高、效果最好,B12、B10通道次之,B09通道反演精度最低;传感器垂直向下观测即观测天顶角为0°时的SST反演精度最高。由于单通道算法反演误差较大,因此进行了二次修订,修订后效果有明显改善,误差在1K以内。     

基于MODIS的海表面温度反演系统设计与实现

海表面温度(SST)是重要的海洋物理参数,对海洋研究具有重要意义.通过遥感数据反演是目前获取SST数据的主要方法之一.遥感反演数据具有反映大区域尺度的海温连续分布状况,且数据获取及时稳定等特点.本文基于IDL构建了面向EOS/MODIS数据的SST反演系统,讨论了系统的功能和结构,给出了具体的数据反演处理流程,对系统中采用的SST反演算法和云检测方法进行了详细说明.最后以我国东海海域为例,将反演结果与多年观测资料对比,表明系统反演得到的SST分布规律与多年观测资料一致,同时还利用同步观测的浮标数据和NASA MODIS SST产品对反演结果进行验证,结果显示系统反演的SST平均误差为1℃.数据精度满足海洋预报和其他科学研究的需要.     

基于FY-3B微波成像仪的海面风速反演

风云三号B星微波成像仪的10.65、18.7、23.8和36.5 GHz频点对海表面粗糙度和介电特性比较敏感,能够用于海面地球物理参数的反演。为获得一种适用于全球大部分海域的海面风速反演算法,利用快速辐射传输模式和再分析大气廓线库模拟微波成像仪海面微波辐射特性,在此基础上建立了半经验反演算式,并利用浮标现场测量数据及WindSat全极化辐射计风速产品对半经验算法和经验算法分别进行了验证和对比。另外,通过分析风向对风速反演的影响,借助AT BV-TBH模型,尝试利用查找表算法对风向造成的晴空区域风速反演偏差进行初步校正。校正风向误差后,反演风速与浮标风速的均方根误差为0.9775 m/s。     

基于PMDB数据集的海表温度反演误差统计分析

红外辐射计在测量海表温度时,大气水汽含量是影响遥感反演精度的主要因素之一。利用1988年1月~2009年11月60°N~60°S的SST和水汽月平均资料及AVHRR准同步实测数据集PMDB,分析了海表温度及其卫星反演误差和海面水汽含量的相关关系。结果表明：海表温度和水汽含量的纬度带月均值具有明显的季节变化特征。纬度越高,遥感SST偏离实测SST越小;在中低纬海域,PFSST算法通常低估了海表温度,而高纬度海域则高估了海表温度;SST误差的气候态月均值在中高纬海域具有较好的季节分布特征,与水汽、SST的季节变化相关性较高。     

卫星红外高光谱反演大气水汽含量

利用高光谱大气红外探测仪AIRS模拟及观测数据,发展基于主成分分析技术的多层前馈神经网络反演算法,进行大气中水汽柱总量(IWV)的反演计算、模拟及实测验证。首先,基于全球晴空大气廓线训练样本SeeBorV4.0,利用快速辐射传输模式CRTM进行了辐射传输模拟计算,得到全球高光谱分辨率模拟辐亮度;其次,利用主成分分析技术对模式模拟和AIRS实测高光谱数据进行降维、去噪及去相关处理,并采用多层前向神经网络算法反演大气水汽柱总量;最后,利用数值试验、AIRS实测L1B数据及其水汽产品,对反演算法进行了验证。通过与AIRS官方大气产品的统计分析,本算法反演均方根误差为0.387 g/cm²,最大偏差为0.82 g/cm²,空间分辨率保留了AIRS像素原分辨率(比AIRS官方大气产品高3倍)。     

基于Landsat 8 TIRS数据的大气水汽含量反演劈窗算法

Landsat 8是2013年最新发射的Landsat卫星,携带了OLI和TIRS两个传感器,其中TIRS传感器获取了两个临近的热红外通道信息。劈窗协方差—方差比算法(SWCVR)是一种最通用的基于热红外的大气水汽含量反演方法,利用两个热红外通道(其中一个在大气窗口,另一个在大气吸收谱段)的吸收差异来反演大气水汽含量,该方法已经在MODIS等中低分辨率(1km)的热红外数据上得到很好的应用。将SWCVR算法移植到较高分辨率的Landsat 8TIRS数据上,并对水汽含量反演结果进行精度验证。气象数据验证结果表明,水汽含量的反演精度可以达到0.43g/cm~2。用MODIS水汽产品(MOD05)做交叉验证,反演的水汽含量和MOD05水汽含量的均方根误差(RMSE)为0.44g/cm~2,平均绝对误差(MAE)为0.34g/cm~2。总的来说,SWCVR算法应用于Landsat 8数据的水汽含量反演也能得到一个较高的精度。     

HJ-1A/B星地表反照率生成算法与检验

提出了适合环境与灾害监测预报小卫星-A、B星(简称HJ-1A/B星)CCD相机的大气订正算法,并基于不同地表特性和大气条件下的辐射传输模拟数据,建立HJ-1A/B星的窄波段向宽波段反照率转换的模型.利用多级灰阶靶标实测数据、敦煌检验场实测数据验证了大气订正算法以及转换模型的可靠性和精度,并将HJ-1A/B星影像数据计算的反照率产品与同时相的MODIS反照率产品进行对比分析.结果表明:文章提出的HJ-1A/B星CCD相机大气订正算法可有效校正大气影响;窄波段向宽波段反照率转换模型反演的反照率精度可靠;基于研究成果生成的HJ-1A/B星地表反照率与MODIS反照率产品一致性较好,满足后续遥感数据定量化模型研究的精度需要.     

利用MODIS数据反演南海南部海表温度及时空变化分析

南海海水表面温度对中国陆地的气候变化具有显著的影响。以南海南部海域为例,首先对MODIS基础数据进行几何校正及影像去云等预处理,利用辐射传输模型MODTRAN计算大气透过率,利用MODIS数据第31和32波段辐射亮度值计算亮度温度,采用劈窗算法反演南海南部海域海表温度,反演结果与产品及实测数据进行回归分析,采取决定系数（R ²）、误差平方和（SSE）及均方根误差（RMSE）进行拟合情况评价。决定系数（R ²）大于0.8,SSE、RMSE较小,其中反演结果与实测数据的SSE为1.025,RMSE为0.158,说明反演精度良好。研究表明:温度具有明显的区域和季节变化特征,秋冬较低,春夏较高,在空间上从离近岸向中心海域方向递减,海盆中心温度低。温度受气候的影响,与厄尔尼诺现象呈正相关,与拉尼娜现象呈负相关。     

AMSR-E的C/X双波段垂直极化亮温反演海面盐度

利用星载传感器观测数据反演的大气海洋参数进行综合研究是物理海洋学等海洋学科获取信息的重要手段。特别是全球尺度精细化的海面盐度信息,只有通过卫星观测才能有效获得。星载微波辐射计能够获取低频波段的亮温数据,具有一定的海面盐度反演能力。使用AMSR-E的C/X双波段垂直极化亮温数据和AMSR-E的L2b产品反演亚马孙河口海区和中国黄海东海海区的海面盐度,并将反演结果与WOA09气候态盐度数据进行比较。实验表明:使用星载微波辐射计的C/X双波段垂直极化亮温能够有效获得河口区域海面盐度的变化趋势,反演精度约为1.5psu,具有一定的应用价值,为将来我国盐度卫星的数据处理提供一定的帮助。     

Determination of sea surface temperature at large observation angles using an angular and emissivity-dependent split-window equation

This paper proposes an angular and emissivity-dependent split-window equation that permits the determination of the sea surface temperature (SST) to a reasonable level of accuracy for any observation angle, including large viewing angles at the image edges of satellite sensors with wide swaths. This is the case of the MODIS radiometer both on EOS Terra/Aqua platforms, with observation angles of up to 65° at the surface, for which the split-window equation has been developed in this study. The algorithm takes into account the angular dependence of both the atmospheric correction (due to the increase of the atmospheric optical path with angle) and the emissivity correction (since sea surface emissivity (SSE) decreases with observation angle). Angular-dependent coefficients have been estimated for the atmospheric terms, and also an explicit dependence on the SSE has been included in the algorithm, as this parameter has values different to a blackbody surface for off-nadir angles, the SSEs also being dependent on surface wind speed. The proposed algorithm requires as input data at-sensor brightness temperatures for the split-window bands (31 and 32 of MODIS), the observation angle at each pixel, an estimate of the water vapor content (which is provided by the MODIS MOD07/MYD07 products) and accurate SSE values for both channels. The preliminary results show a good agreement between SSTs estimated by the proposed equation for off-nadir viewings of MODIS-Terra images and in situ SST measurements, with a root-mean square error (RMSE) of about ± 0.3 K, for which the MODIS SST product gives an RMSE larger than ± 0.7 K.     

Optimized split-window coefficients for deriving surface temperatures from inland water bodies

Large inland water bodies constituting lakes, reservoirs and inland-seas are excellent proxy indicators for climate change. Using thermal infrared satellite data, a recent study found that a global set of inland water bodies showed significant warming in seasonal nighttime Lake Surface Water Temperatures (LSWTs) between 1985 and 2009. Split-window land surface temperature (LST) retrievals are typically tuned for a broad range of land surface emissivities and global atmospheric conditions, and are not optimized for inland water body surfaces, whereas split-window sea-surface temperatures (SSTs) are only tuned for a single emissivity (water), but over ocean atmospheres. Over inland water bodies, these two approaches can lead to region dependent errors in LSWTs, spurious trends, and inconsistencies between sensors in the long-term temperature record of inland water bodies. To address this issue, the primary goal of this paper was to develop a methodology for deriving a set of optimized split-window coefficients, individually tuned for the regional atmospheric conditions of 169 globally distributed, saline and freshwater inland water bodies from multiple satellite sensors including the Moderate Resolution Imaging Spectroradiometer (MODIS) on Terra and Aqua; Along Track Scanning Radiometer (ATSR) including ATSR-1, ATSR-2, AATSR; and Advanced Very High Resolution Radiometer (AVHRR-3). The new Inland Water-body Surface Temperature (IWbST) v1.0 algorithm was applied to Terra MODIS and Advanced Along Track Scanning Radiometer (AATSR) data and validated with in situ water temperature data from sites with widely contrasting atmospheric conditions: Lake Tahoe in California/Nevada, a high-elevation cool and dry site, and the Salton Sea in California, a low-elevation warm and humid site. Analysis showed improved accuracy in LSWTs in terms of bias and RMSE when compared to the standard MODIS LST and AATSR SST products. For example, the IWbST RMSE at Salton Sea was reduced by 0.4 K when compared to the operational MODIS product. For the AATSR data, the IWbST RMSE was reduced by 0.36 K at Tahoe and 0.29 K at Salton Sea when compared to results obtained using the operational AATSR split-window coefficients. The IWbST improvements are significant in relation to the current accuracy of water temperature retrievals from space (< 0.5 K), and will enable the derivation of long-term, accurate LSWTs consistently across multiple sensors for climate studies.     

Land surface temperature and emissivity retrieval by integrating MODIS data onboard Terra and Aqua satellites

This paper presents an algorithm to retrieve land surface temperature (LST) and emissivity by integrating MODIS (Moderate Resolution Imaging Spectroradiometer) data onboard Terra and Aqua satellites. For a study area, there will be four pairs of day and night observations by MODIS onboard two satellites every day. Solar zenith angle, view zenith angle, and atmospheric water vapour have first been taken as independent variables to analyse their sensitivities to the same infrared channel measurements of MODIS on both Terra and Aqua satellites. Owing to their similar influences on the same MODIS band from Terra and Aqua satellites, four pairs of MODIS data from Terra and Aqua satellites can be thought of as MODIS measurement on a satellite at different viewing angles and viewing time. Comparisons between the retrieved results and in-situ measurements at three test sites (Qinghai Lake, Poyang Lake and Luancheng in China) indicate that the root mean square (rms) error is 0.66 K, except for the sand in Poyang Lake area. The rms error is less than 0.7 K when the retrieved results are compared with Earth Observing System (EOS) MODIS LST data products using the physics-based day/night algorithm. Emissivities retrieved by this algorithm are well compared to EOS MODIS emissivity data products (V5). The proposed algorithm can therefore be regarded as complementary and an extension to the EOS physics-based day/night algorithm.     

Retrieval of high-resolution sea surface temperature data for Sendai Bay, Japan, using the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER)

We studied sea surface temperature (SST) retrieval algorithms for Sendai Bay, using output from the thermal-infrared channels of the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) on board Terra. While the highest resolutions of other satellite SST products are about 1 km, the ASTER thermal-infrared channels provide 90-m spatial resolution. To develop the ASTER algorithm, we employed statistical methods in which SSTs retrieved from the thermal-infrared measurements were tuned against the Moderate Resolution Imaging Spectroradiometer (MODIS) SST product with a 1-km spatial resolution. Terra also carries a MODIS sensor, which observed the same area as the ASTER sensor at the same time. The MODIS SST was validated around Sendai Bay, revealing a bias of −0.15 °C and root mean-square difference (RMSD) of 0.67 °C against in situ SSTs. Taking into account the spatial-resolution difference between ASTER and MODIS, match-up was generated only if the variability of ASTER brightness temperatures (T₁₃) was small in a pixel of MODIS SST (MP). The T₁₃ within one MP was about 121 pixels. The standard deviation (σ₁₃) of T₁₃ was calculated for each cloud-free MP, and the threshold of σ₁₃ for choosing match-up MPs was decided by analyzing the σ₁₃ histogram of one ASTER image. The 15 synchronous pairs of ASTER/MODIS images are separated into two groups of 8 pairs called set (A) and 7 pairs called set (B). Using the common procedure, the match-ups are generated for set (A) and set (B). The former is used for developing the ASTER Multi-Channel SST (MCSST) algorithm, and the latter for validation of the developed ASTER SST. Analysis of the whole 15 pairs indicated that ASTER SST does not depend on the satellite zenith angle. We concluded that, using Akaike's information criterion with set (A) match-ups, the multiple regression formula with all five thermal-infrared channels was adequate for the ASTER SST retrieval. Validation of ASTER SST using match-up set (B) indicated a bias of 0.101 °C and RMSD of 0.455 °C.     

Validation and application of MODIS-derived SST in the South China Sea

Using a variety of in situ sea surface temperature (SST) data sets in the South China Sea, we validate the satellite-derived SST from the Moderate Resolution Imaging Spectroradiometer (MODIS). Analysis of a large number of match-up samples during 2008–2012 shows that the MODIS SSTs have biases ranging from –0.19°C to –0.34°C and standard deviation (STD) errors ranging from 0.58°C to 0.68°C. Specifically, mean biases are all negative but there are smaller cool biases in daytime than those in night-time. The monthly validation analysis shows that the biases exhibit apparent seasonal variations. The biases in daytime have relatively small magnitudes in spring and summer, while the negative biases in night-time are most apparent in summer. On the other hand, the time series of MODIS SSTs may exhibit an evident diurnal variation for some months, which roughly agrees with the in situ SST measurements. This study also highlights that the MODIS SSTs under cloud-free conditions are effective at detecting the high-frequency and small-scale oceanic features, such as the localized diurnal variation, oceanic front, and coastal upwelling.     

MODIS数据地表温度反演劈窗算法比较

针对MODIS数据,分析比较了QIN和Wan-Dozier两种劈窗算法地表温度(LST)反演精度和误差分布。首先利用辐射传输模型MODTRAN4.0,结合TIGR大气廓线数据,评价两种算法绝对精度,然后基于误差传递理论分析评价二者的总精度,最后对两种算法的LST反演结果进行比较。研究表明针对所有廓线数据,两种算法绝对精度相差不大,但Wan-Dozier算法绝对精度受地表温度和水汽含量变化的影响程度要大于QIN算法;两种算法总精度相差不大,且主要误差源均为算法绝对精度和地表比辐射率精度,QIN算法反演结果对地表比辐射率的敏感性要略高于Wan-Dozier算法;两种算法得到研究区LST分布情况基本一致,均可表现空间LST分布差异,其中水体和裸土的LST反演结果差异较大,城镇和植被平均温度差异在0.5 K以内。     

Preliminary estimate of calibration of the moderate resolution imaging spectroradiometer thermal infrared data using Lake Titicaca

The absolute radiometric accuracy of Moderate Resolution Imaging Spectroradiometer (MODIS) thermal infrared (TIR) data was evaluated with in situ data collected in a vicarious calibration field campaign conducted in Lake Titicaca, Bolivia during May 26 and June 17, 2000. The comparison between MODIS TIR data produced by the version 2.5.4 Level-1B code and the band radiances calculated with atmospheric radiative transfer code MODTRAN4.0 based on lake surface kinetic temperatures measured by five IR radiometers deployed in the high-elevation Lake Titicaca and the atmospheric temperature and water vapor profiles measured by radiosondes launched on the lake shore on June, 15 2000, a calm clear-sky day, shows good agreements in bands 31 and 32 (within an accuracy of 0.4%) in the daytime overpass case. Sensitivity analysis indicates that the changes on the measured atmospheric temperature and water vapor profiles result in negligible or small effects on the calculated radiances in the atmospheric window bands (bands 20–23, 29, and 31–32). Therefore, comparisons for these bands were made for cases when lake surface temperature measurements were available but no radiosonde data were available and in subareas of 10×16 pixels where there was no in situ measurement but MODIS brightness temperatures in band 31 vary within ±0.15 K by using the validated band 31 to determine lake surface temperatures through the MODTRAN4.0 code. Comparisons and error analysis show that the specified absolute radiometric accuracies are reached or nearly reached in MODIS bands 21, 29, and 31–33 and that there is a calibration bias of 2–3% in bands 20, 22, and 23. The error analysis also shows that the radiosondes cannot provide accurate atmospheric temperature and water vapor profiles to estimate the calibration accuracies in the atmospheric sounding bands (bands 24–25, 27–28, and 34–36) at the specified 1% level and that the calibration accuracy in the ozone band 30 cannot be estimated without in situ measurements of ozone.     

Quality assessment and validation of the MODIS global land surface temperature

This paper presents an evaluation of the Earth Observing System (EOS) Moderate Resolution Imaging Spectroradiometer (MODIS) thermal infrared bands and the status of land surface temperature (LST) version-3 standard products retrieved from Terra MODIS data. The accuracy of daily MODIS LST products has been validated in more than 20 clear-sky cases with in situ measurement data collected in field campaigns in 2000–2002. The MODIS LST accuracy is better than 1°C in the range from ?10 to 50°C. Refinements and improvements were made to the new version of MODIS LST product generation executive code. Using both Terra and Aqua MODIS data for LST retrieval improves the quality of the LST product and the diurnal feature in the product due to better temporal, spatial and angular coverage of clear-sky observations.     

Sea surface temperature GOES-8 estimation approach for the Brazilian coast

This work presents a quite consistent procedure for estimation of sea surface temperature (SST) using data from the new generation of the Geostationary Operational Environmental Satellite (GOES). The SST methodology is based on the classical split-window equation. The regional split-window coefficients (A ₀, A ₁, A ₂ and A ₃) are estimated by an algorithm regression taking as dependent variable three datasets, i.e. the SST derived from National Oceanic and Atmospheric Administration (NOAA)-14 polar-orbiting satellite and from buoys of Pilot Research Moored Array in Tropical Atlantic (PIRATA) and National Programme of Buoys (PNBOIA). This work shows that the main advantage of the GOES-8 SST algorithm, in comparison with the multi-channel sea surface temperature (MCSST) procedure using Advanced Very High Resolution Radiometer (AVHRR) data, is the high frequency sampling imagery (each half-hour) which permits a daily image with much less quantity of cloud contamination. The algorithm results using AVHRR/NOAA-14 as input dataset for the regression show that the accuracy of the GOES-8 SST algorithm is better than 1.0°C for all Brazilian coast. For regional estimation, the accuracy has been improved to around 0.5°C. Also, the accuracy of GOES-8 SST is better than 0.7°C using in situ SST collected from moored and drifting buoys.