AIRS红外高光谱卫星数据反演卷云光学厚度和云顶高度

基于大气红外探测器L1B红外高光谱辐射观测资料,结合中分辨率成像光谱仪(moderate resolution imaging spectroradiometer, MODIS)云产品数据,利用通用大气辐射传输模式(combined atmospheric radiative transfer model,CART),根据模式模拟和AIRS实际观测亮温的亮温差,研究从AIRS红外波段1 070～1 135 cm-1高光谱数据反演卷云的光学厚度和云顶高度。将反演的卷云光学厚度与云顶高度作为输入参数模拟计算650～1 150 cm-1波段卷云大气顶的辐射亮温谱,并将模拟值与AIRS观测亮温谱进行了对比分析。将反演的卷云光学厚度和云顶高度和AIRS的760通道(900.56 cm-1,11.1 μm)的亮温以及MODIS卷云反射率进行了对比分析。最后将反演的卷云云顶高度和MODIS云顶高度进行了对比分析。研究结果表明：反演所使用的650～1 150 cm-1波段模式模拟和观测亮温谱吻合得很好,说明CART可以较好的模拟AIRS亮温谱。反演的卷云参数与AIRS在大气窗口区的760通道(900.56 cm-1,11.1 μm)的亮温的分布满足低亮温对应较大的卷云光学厚度和高云顶高度。反演的卷云参数和MODIS卷云的反射率分布满足高卷云光学厚度和云顶高度对应高卷云反射率。反演的卷云云顶高度和MODIS的卷云云顶高度之间线性相关系数相对较高,且都在8.5～11.5 km的概率较高,两者的概率分布趋势一致。说明CART可以用于反演卷云的性质,反演结果具有一定的可靠性。     

Aerosol extinction coefficient profile retrieval in the oxygen A-band considering multiple scattering atmosphere. Test case: SCIAMACHY nadir simulated measurements

A multiple scattering inversion procedure for the aerosol extinction coefficient profile retrieval and error assessment in the oxygen A-band, for passive remote sensing instruments, has been developed. The procedure has been applied to SCIAMACHY nadir simulated measurements to investigate its effectiveness in the troposphere.The inversion procedure consists of a multiple scattering Forward Model, an inversion method and a complete sensitivity and error assessment tool. The Forward Model is based on LIDORT code; the inversion method, the sensitivity study and the complete error assessment are based on Optimal Estimation. The sensitivity and error analysis has been derived to investigate the profile retrieval errors due to the uncertainty of different aerosol optical properties, molecular and surface parameters.The analysis confirms that the profile retrieval accuracy and vertical resolution are strongly dependent on the oxygen A-band spectral resolution. The moderately high SCIAMACHY spectral resolution (0.4 nm in the oxygen A-band) results in distinguishing a maximum of three aerosol layers in troposphere. The SCIAMACHY tropospheric aerosol profile retrieval is shown to be highly sensitive to aerosol optical properties as phase function and single scattering albedo. The sensitivity study reveals an improvement of information content increasing the solar zenith angle and decreasing the surface albedo. As regards the forward model, negligible errors occur as the number of streams exceeds 6.     

云层对卫星遥感大气CO的影响与修正方法研究

研究云层对卫星遥感大气CO精度的影响,并以SCIAMACHY观测为例,详细论述一种云层影响的修正方法。修正过程中,采用朗伯云模型,全面考虑云层覆盖率、云顶高度和地面反射率的影响。将云层影响修正前后的卫星CO反演结果与地基FTIR测量值进行对比,发现修正云层影响后,两者的测量相关性明显提高。实验结果表明,提出的云层影响修正方法可明显提高卫星遥感大气CO的精度。     

云计算的蚕豆虫害可见-近红外光谱分类

利用蚕豆叶片可见-近红外反射光谱结合导数光谱对健康、少量、大量虫害三种等级的实验样本进行光谱特征分析,并选择虫害检测最优波段。采用Hadoop,Spark和VMWare虚拟机搭建云计算平台,使用MLlib机器学习库实现人工神经网络(ANN)和支持向量机(SVM)分类算法,并对三种等级蚕豆叶片全波段和最优波段光谱进行分类建模与预测。结果表明ANN虫害光谱分类模型准确率优于SVM虫害光谱分类模型,并且在云平台上运行效率更高,同时全光谱波段的预测准确性高于最优波段。通过扩展光谱数据集,云计算技术在光谱数据挖掘中的计算效率有显著提升。云计算分类检测可以为作物生物胁迫光谱识别提供新的技术和方法。     

The physical parameterization of the top-of-atmosphere reflection function for a cloudy atmosphere—underlying surface system: the oxygen A-band case study

The paper is devoted to the physical parameterization of the top-of-atmosphere reflection function. The accuracy of the parameterization is checked against exact radiative transfer calculations in a cloudy atmosphere for various cloud-top-heights, cloud optical and geometrical thicknesses, solar illumination and surface reflection conditions. It was found that the error of approximation is smaller than 5% for most cases studied at the wavelength interval , which corresponds to the oxygen A-band. This band is routinely used in cloud-top-height retrievals. The model proposed can be used for cloud-top-height and cloud geometrical thickness retrievals. This allows to avoid a standard look-up-table retrieval scheme, involving complex numerical procedures.     

Three-dimensional polarized Monte Carlo atmospheric radiative transfer model (3DMCPOL): 3D effects on polarized visible reflectances of a cirrus cloud

A polarized atmospheric radiative transfer model for the computation of radiative transfer inside three-dimensional inhomogeneous mediums is described. This code is based on Monte Carlo methods and takes into account the polarization state of the light. Specificities introduced by such consideration are presented. After validation of the model by comparisons with adding-doubling computations, examples of reflectances simulated from a synthetic inhomogeneous cirrus cloud are analyzed and compared with reflectances obtained with the classical assumption of a plane parallel homogeneous cloud (1D approximation). As polarized reflectance is known to saturate for optical thickness of about 3, one could think that they should be less sensitive to 3D effects than total reflectances. However, at high spatial resolution (80 m), values of polarized reflectances much higher than the ones predicted by the 1D theory can be reached. The study of the reflectances of a step cloud shows that these large values are the results of illumination and shadowing effects similar to those often observed on total reflectances. In addition, we show that for larger spatial resolution (10 km), the so-called plane-parallel bias leads to a non-negligible overestimation of the polarized reflectances of about 7–8%.     

A fast infrared radiative transfer model for overlapping clouds

A fast infrared radiative transfer model (FIRTM2) appropriate for application to both single-layered and overlapping cloud situations is developed for simulating the outgoing infrared spectral radiance at the top of the atmosphere (TOA). In FIRTM2 a pre-computed library of cloud reflectance and transmittance values is employed to account for one or two cloud layers, whereas the background atmospheric optical thickness due to gaseous absorption can be computed from a clear-sky radiative transfer model. FIRTM2 is applicable to three atmospheric conditions: (1) clear-sky, (2) single-layered ice or water cloud, and (3) two simultaneous cloud layers in a column (e.g., ice cloud overlying water cloud). Moreover, FIRTM2 outputs the derivatives (i.e., Jacobians) of the TOA brightness temperature with respect to cloud optical thickness and effective particle size. Sensitivity analyses have been carried out to assess the performance of FIRTM2 for two spectral regions, namely the longwave (LW) band (587.3-1179.5 cm⁻¹) and the short-to-medium wave (SMW) band (1180.1-2228.9 cm⁻¹). The assessment is carried out in terms of brightness temperature differences (BTD) between FIRTM2 and the well-known discrete ordinates radiative transfer model (DISORT), henceforth referred to as BTD (F−D). The BTD (F−D) values for single-layered clouds are generally less than 0.8 K. For the case of two cloud layers (specifically ice cloud over water cloud), the BTD (F−D) values are also generally less than 0.8 K except for the SMW band for the case of a very high altitude (>15 km) cloud comprised of small ice particles. Note that for clear-sky atmospheres, FIRTM2 reduces to the clear-sky radiative transfer model that is incorporated into FIRTM2, and the errors in this case are essentially those of the clear-sky radiative transfer model.     

利用地基红外高光谱发射率数据进行云参数反演(1): 云相态判别

云相态是气候模式中的重要参数,也是遥感反演过程中进行云滴有效半径、云水含量等微物理参数反演的重要前提。在研究了云层有效发射率光谱对云相态敏感性的基础上,提出了基于云层有效发射率光谱的云相态表达特征,包括800～900 cm-1区域的有效发射率斜率、900～1 000 cm-1区域的有效发射率斜率、上述两个区域的有效发射率斜率之差、862.1与989.8 cm-1的有效发射率之比、862.1与989.8 cm-1的有效发射率之差、1 900.1与2 029.3 cm-1的有效发射率之比、远红外窗区有效发射率平均值与900 cm-1有效发射率之比等7个特征。建立了利用支持向量机进行云相态判别的方法,开展了模拟数据验证试验,并利用遗传算法优化了支持向量机的径向基核函数参数和惩罚因子。将该方法用于处理ARM计划中SGP站点的AERI仪器获得的数据,得到的云相态判别结果与Shupe提出的多仪器综合判别结果进行了比较。结果表明,利用红外波段不同窗区的有效发射率光谱特征可以实现发射率低于0.95的云层的相态判别,建立的基于支持向量机的云相态判别方法与Shupe方法的总体判别结果较为一致,但有约30%的云层由于发射率较大而标记为不透明云。基于红外高光谱发射率数据的云相态判别技术充分考虑了光谱斜率、比值和差值等信息,是较为稳定有效的薄云相态判别方法。     

应用单波段亮度变化率获取无云干扰的卫星数据

在卫星遥感检测中,云对数据的反演带来了严重的干扰。精确地确定云区及减弱云的干扰是具有挑战性的问题。根据云的大气辐射特性给出了卫星图像亮度变化率计算公式、图像反射率的变化率计算公式和热红外波段图像亮度温度变化率的计算方法。利用单波段亮度变化率检测卫星数据中云干扰相对强度的新方法,快速、精确地获取云区位置,得到云区以外无云干扰的数据。根据各光谱波段具有不同的透射云层能力的特点,文章还给出了卫星各红外波段对云层透射能力的客观评价以及影响透射效果的因素。在此基础上利用红外波段对伪卷云透射图像数据完成了多波段数据融合。很好地恢复了可见光波段云区下面的真实数据。统计数据表明,融合后的云区图像数据与可见光波段相关性保持一致。     

A Monte Carlo method for 3D thermal infrared radiative transfer

A 3D Monte Carlo model for specific application to the broadband thermal radiative transfer has been developed in which the emissivities for gases and cloud particles are parameterized by using a single cubic element as the building block in 3D space. For spectral integration in the thermal infrared, the correlated k-distribution method has been used for the sorting of gaseous absorption lines in multiple-scattering atmospheres involving 3D clouds. To check the Monte-Carlo simulation, we compare a variety of 1D broadband atmospheric fluxes and heating rates to those computed from the conventional plane-parallel (PP) model and demonstrate excellent agreement between the two. Comparisons of the Monte Carlo results for broadband thermal cooling rates in 3D clouds to those computed from the delta-diffusion approximation for 3D radiative transfer and the independent pixel-by-pixel approximation are subsequently carried out to understand the relative merits of these approaches.     

星载激光雷达云和气溶胶分类反演算法研究

激光探测对于获取云和气溶胶的垂直廓线,研究大气中云和气溶胶的垂直分布特征以及对全球气候变化的影响意义重大。而星载大气激光雷达云气溶胶分类算法的研究,对于激光雷达数据的参数反演及应用极为重要。针对激光条件下探测的云和气溶胶特有的光学信息和空间分布,结合概率统计与机器学习算法,提出了一种对于云/气溶胶、云相态及气溶胶子类型识别的分类算法,实现了星载激光雷达的大气特征层快速、有效分类。算法采用中国地区2016年CALIOP的观测数据作为样本数据,主要由三部分组成：（1）基于激光探测的云和气溶胶层不同的光学特性以及地理空间分布特征,分别构建了云和气溶胶的γ₅₃₂,χ,δ,Z和lat的五维概率密度函数,以此为基础构建云气溶胶的分类置信函数,并基于此实现了云和气溶胶类型的反演;（2）选取支持向量机(SVM)作为随机朝向冰晶粒子(ROI)和水云分类的算法模型基础,结合云层的γ₅₃₂,χ,δ Z和云顶温度T的概率密度函数构建ROI,水平朝向冰晶粒子(HOI)和水云的分类置信函数以修正SVM误分的特征层以及筛选出水云中少部分的HOI冰云,获得云相态的分类结果;（3）以各气溶胶子类型的光学以及空间分布特性为基础,采用决策树策略的气溶胶子类型识别算法实现了对气溶胶子类型的区分,完成气溶胶子类型的识别。利用现有CALIOP观测结果作为样本数据构建分类数据库,避免了对于地面以及航测数据的依赖,而机器学习则大大简化了算法的实现过程,使得云气溶胶分类更加高效。算法结果与正交极化云气溶胶激光雷达垂直特征层分布数据(CALIPSO VFM)产品对比分析：云层有98.51%一致性,气溶胶有88.43%的一致性,且白天比夜间一致性高。对于云相态分类,可以有效区分出水云和冰云,其中二者水云一致性高达93.44%。在气溶胶子类型反演结果中,可以准确识别出大多数气溶胶特征层子类型。霾、沙尘以及晴空三种典型情况下的反演结果均与CALIOP VFM产品数据具有较好的一致性。其中,霾天的大部分煤烟型以及污染型（污染沙尘以及污染大陆）气溶胶反演结果与VFM具有较好的一致性。沙尘天也能够获得较好的沙尘以及污染沙尘的结果。晴空为数不多的气溶胶层也取得了较为一致的结果。对于实现的星载大气激光雷达特征层分类算法,针对CALIOP激光测量的云气溶胶层的分类进行了重要的改进,在保证一定精度的基础上,简化了算法,提高了数据处理的效率,在下一步工作中,将分别构建不同时段和季节的分类模型以及提高两种不同偏振特性的冰云和气溶胶子类型的分类精度。     

基于光谱特征的高光谱遥感影像检索

面向海量遥感信息管理中遥感影像检索的需求,以高光谱遥感信息为例对光谱特征应用进行了探讨,提出基于光谱特征的遥感影像检索包括基于点源示例的检索和基于面源示例的检索,检索中的关键问题是光谱特征提取与相似性度量。基于光谱曲线的相似性度量可以采用光谱角和光谱信息散度进行;基于光谱特征的检索可以通过提取反射和吸收光谱、光谱特征匹配的方法进行;而光谱曲线量化指标如中心矩、分维数和信息熵等的效果较差,不适宜应用于检索。     

Retrieval of tropospheric NO2 columns from satellite measurements in presence of cirrus: A theoretical sensitivity study using SCIATRAN and prospect application for the A-Train

A theoretical sensitivity study of the influence of cirrus cloud properties on tropospheric NO₂ columns retrieved from the spaceborne Ozone Monitoring Instrument (OMI) measurements is performed. It is conducted within the framework of the synergetic use of A-Train sensors to derive more representative trace gas products. We aim to study the potential effects of cirrus clouds on tropospheric NO₂ retrievals using a retrieval algorithm that, unlike the OMI Standard and DOMINO algorithms, does not correct for the effects of clouds. The sensitivity study is based on the radiative transfer code SCIATRAN that performs both simulations of top of atmosphere (TOA) reflectances as measured by an OMI-like band and tropospheric NO₂ column retrievals based on the differential optical absorption spectroscopy (DOAS) method. The results of the sensitivity study show that if a correction for cirrus clouds is not included in our simple retrieval that does not account for clouds in the first place, the tropospheric column can be underestimated by 55%. This underestimation depends strongly on cirrus parameters as, in order of importance, cloud fraction, cloud optical depth, asymmetry factor of cirrus cloud phase function and cloud top height. The perspective of the synergy between OMI and cloud information obtained from cloud-derived products of the A-Train is evaluated in two parts by applying a simple cloud correction scheme based on the independent pixel approximation (IPA). Firstly, we evaluated the tropospheric NO₂ column retrievals error caused by uncertainties in cirrus cloud properties. Secondly we studied the influence of subpixel cloud optical depth variability on NO₂ retrievals. From our simulations, it is demonstrated that the error will be reduced significantly if the cloud fraction is lower or equal to 0.5. In this case, the cloud fraction and the cloud optical depth must be known within accuracy less than 0.05% and 50%, respectively. The cloud top height and the asymmetry factor must be known within uncertainty of at least 1 km and less than 0.05, respectively. The latter result shows that the uncertainty of the asymmetry factor is a major source of error in the cloud correction for tropospheric NO₂ retrieval in the presence of cirrus.     

Cloud 3D effects on broadband heating rate profiles: I. Model simulation

Solar broadband heating directly drives the atmospheric and ocean circulations, and is largely determined by cloud spatial 3-diminesional (3D) structures. To study the cloud 3D effects on radiation, a 3D broadband Monte-Carlo radiative transfer model, along with an Independent Pixel/Column Approximation (IPA) method, is used to simulate radiation and heating rate of three typical cloud fields generated by cloud resolving models (CRM). A quantitative and statistical estimation of cloud 3D effects has been developed to investigate the impact of cloud 3D structures on both heating rate strength, STD_Bias, and vertical distribution, CorrCoef. The cloud 3D structures affect some clouds more in heating rate strength and others more in vertical distribution. It is crucial to use the combination of CorrCoef and STD_Bias for better quantitative evaluation of the 3D effects. Furthermore, there is no simple way to define a critical resolution (or average radius), within which the IPA heating rate profiles closely represent the true 3D heating rate profiles. The critical radius (or resolution) strongly depends on solar incident angle as well as cloud vertical distribution. Also, the critical radii for clear-sky columns are larger than for cloudy columns, although the corresponding STD_Bias for clear-sky columns are smaller than for cloudy columns. Analysis based on two different statistical average methods illustrates that the cloud 3D effects due to the dimensionality difference between the 3D clouds (circle average) and 2D clouds (line average) significantly impact on the heating rate profiles.     

ETM图像地物反射率光谱反演

用搭载在卫星Landsat-7上的多光谱成像仪ETM记录的来自地气系统的辐亮度光谱图像,反演地物反射率光谱,在ETM 1波段,反射率反演的不确定度基本上保证在17%以内,2,3波段在10%以内,优于目前常用的简易算法。用反演的地物辐亮度光谱合成的sRGB真彩色图像,清晰度欠佳,未作经验调整时明显偏色,用于航天对地真彩色摄影这一反演精度还不够,但用于目视判读明显优于原图像。     

海雾的遥感光学辐射特性

为实现海雾的卫星自动监测,利用大量AVHRR3/NOAA17卫星观测数据,比较不同目标物(海雾区、不同云系及晴空下垫面)在可见光、近红外波段的反射辐射特性差异,统计分析了海雾的遥感光学辐射特性.结果表明,海雾区三通道反射率满足RCh1RCh3a>RCb2,甚至出现RCh3a>RCh1>RCh2不同于其它目标物(RCh1>RCh2>RCh3a.运用Streamer辐射传输模式模拟了海雾、不同云系在卫星高度处的三通道反射率特性,从理论上进一步验证了海雾所具有的光学辐射特性,同时指出,Ch3a的反射率对粒子粒径的响应明显,即粒子粒径越小,Ch3a的反射率越高.通过米氏散射理论对这一现象进行原因分析.     

霾光谱特性分析与卫星遥感识别算法

近年来,我国频繁发生的灰霾污染事件和常态性的高细颗粒物浓度(PM_2.5),已经引起了全世界范围的广泛关注。卫星遥感能够对大气污染进行快速准确地监测。然而在大气遥感领域具有代表性的中分辨率成像光谱仪(NASA/MODIS),其云监测和暗像元反演算法通常把灰霾当做薄云、雾或地表亮目标处理,无法反演霾天的气溶胶光学厚度(aerosol optical depth, AOD)。笔者研究了云、雾、霾、地表覆盖等不同像元在可见光、近红外以及红外通道的光谱特性。基于MODIS数据,参考相关的云监测和气溶胶反演算法,选取多个对灰霾敏感的光谱通道,计算表观反射率和亮度温度。针对不同波段,分别探讨了霾与薄云、低层云、雾、浓密植被和地表亮目标等像元之间的光谱差异,统计灰霾分布的阈值区间,并设计基于MODIS卫星遥感数据的灰霾识别自动处理流程。通过对2008年华北平原春夏两个重霾事件进行测试,该算法的霾分布监测结果与卫星真彩图具有较好的一致性。基于北京和香河AERONET站点观测的高AOD数据,验证了本算法的霾识别率接近80%,在一定程度能够弥补MODIS标准气溶胶算法用于灰霾天的不足。最后,分析了灰霾识别过程中的主要误差来源,并提出了基于霾纹理特征,以及其他辅助数据支撑的改进方法。     

氧气A带吸收系数的温度依赖关系研究

氧气A带是理想的大气要素反演通道,吸收系数是重要的参数之一,它影响到反演结果的精度.结合HITRAN2012数据库和大气温度廓线图,分析氧气A带吸收系数的影响因素,推出各因素与温度的依赖关系,确定吸收系数随温度的变化.结果表明,氧气A带谱线半宽度受温度依赖系数影响较小,而受温度影响较大.线型因子随温度产生了两种变化,在谱线半宽度以外的谱线位置上,随温度的增大,函数值减小,而在中心频率到谱线半宽度的谱线位置上,随温度的升高而增大.谱线线强对温度具有强依赖关系.利用逐线积分算法计算氧气A带吸收系数,同时考虑了谱线半宽度的压力展宽效应和谱线线强及半宽度对温度的依赖关系,得出氧气A带吸收系数对温度的依赖关系主要来源于线强的温赖关系,尤其是中心频率处温度影响较大;而Lorentzian线型函数的温赖关系不明显.利用布鲁克光谱仪在1 cm-1下测量63 m处氧气A带的吸收光谱,与理论模型在同等条件下的透过率比较,误差小于0.83%,验证了温度校正模型的正确性.     

基于最优估计理论、联合星载主被动传感器资料的液态云微物理特性反演研究

针对液态云微物理特性精确反演的迫切需求, 综合主被动传感器的探测优势, 联合CloudSat雷达反射率和Aqua光学厚度资料, 提出基于最优估计理论的液态云微物理参数反演算法.通过假设粒子谱服从对数正态分布, 基于前向物理模式建立测量变量与反演变量的函数关系, 借助谱分布参数的先验信息、通过算法迭代得到谱参数的最优解, 进而利用前向物理模式反演液态云微物理参数, 并根据误差传递理论计算反演不确定度.通过设计反演方案, 基于实测个例数据并与CloudSat官方发布产品和经验算法反演结果对比验证.结果表明: 基于最优估计理论、联合主被动传感器资料的液态云微物理参数反演结果与官方发布产品一致性较好, 弥补了经验算法误差大、扩展性差的不足, 对于开展国内星载和机载W波段毫米波雷达液态云微物理参数反演研究具有重要的借鉴意义. 关键词： CloudSat Aqua 液态云 最优估计理论     

基于三维重建理论的目标光谱散射特性研究

根据三维重建理论,基于目标的多角度视图,重建了目标表面的三维点云。利用德洛奈三角剖分法结合可见性原理,得到了目标的曲面和曲面面元的法线方向。根据粗糙面散射理论和目标表面的双向反射分布函数(BRDF),结合大气传输软件Modtran计算的某时间、地点的背景光谱辐射亮度,数值分析了目标光谱散射亮度分布特性。以覆盖车衣的汽车为例,重建的三维几何模型误差为4.11%,数值计算了目标在三个波段的光谱散射亮度分布。上述方法可以进一步用于卫星和其他空间目标的光谱辐射、散射特性研究。