An assessment of cloud masking schemes for satellite ocean colour data of marine optical extremes

One of the most important steps in utilizing ocean colour remote-sensing data is subtracting the contribution of the atmosphere from the signal at the satellite to obtain marine water-leaving radiance. To be carried out accurately, this requires clear-sky conditions, i.e. all clouds need to be excluded or masked from the data prior to atmospheric correction. The standard cloud mask used routinely in the processing of NASA global ocean colour data is based on a simple threshold applied to the Rayleigh-corrected top-of-atmosphere (TOA) radiance. The threshold is kept purposefully low to ensure high-quality processing at a global scale. As a consequence, the standard scheme can sometimes inadvertently mask important extreme optical events such as intense blue–green algal (cyanobacteria) blooms or the outflow of sediment-rich waters from some of the world’s largest rivers. However, the importance of these extreme conditions, both for ecological and hydrological applications, requires that they should be appropriately monitored. Therefore, an assessment of existing cloud masking schemes that could provide valuable alternatives was carried out. A new hybrid cloud mask was also proposed and similarly tested. The selected schemes were systematically assessed over a full annual cycle of satellite ocean colour data on three example regions: the Baltic Sea, the Black and Azov Seas, and the Amazon River delta. The results indicate that the application of alternative cloud masking schemes produces a significant increase in clear-sky diagnostics that varies with the scheme and the region. Major occurrences of extreme optical conditions, such as cyanobacteria blooms, or river deltas formerly excluded from any processing may be recovered, but some schemes may underestimate the amount of thin clouds potentially detrimental to ocean colour atmospheric correction.     

Spatial and temporal characteristics and removal methodology of suspended particulate matter speckles from Geostationary Ocean Color Imager data

Reviewing six years of Geostationary Ocean Color Imager (GOCI) suspended particulate matter (SPM) concentration images from 2011 to 2016 revealed unexpected and some enormously high or low values. These speckles are randomly scattered throughout the entire study area or congregated at a certain part, which has strongly restricted the scientific applications of GOCI data thus far. They can be classified into four types: isolated, near-cloud, patch-type, and slot-related speckles, based on spatial distribution and potential causes. These types are investigated. The speckles are induced by a moving-cloud during a complete observation of 8 bands for each slot of GOCI, partly by an incomplete cloud masking procedure near cloud edges, by relatively low reflectance of pixels corresponding to a cloud shadow, by imperfect atmospheric corrections related to water vapor after cloud passage, or by sensor-related troubles related to the effects of stray light. Spatial and temporal variabilities of the identified speckles are investigated and used to develop a methodology for their removal. The SPM concentration values of the error-free pixels, passing through the post-processing of the speckle removal procedure, are compared to those previously with speckles. As a result, the enormously large values of SPM, occupying 6.06% of the pixel numbers, are all eliminated. Typical seasonality of SPM, unrevealed using the speckled images, is clearly presented through the removal procedure. This study addresses the importance of preprocessing for speckle error in SPM concentrations from GOCI data and implies more reliable SPM data without speckles can be used in scientific application research.     

Discriminating sea ice from low-level water clouds in split-window,mid-wavelength IR imagery

A technique is demonstrated to enhance the contrast between sea ice and low-level water clouds. The approach uses the brightness temperature difference (BTD) feature from data collected in the split-window, mid-wavelength infrared (IR) region (i.e. two bands at 3.7 μm and 4.0 μm). These spectral data are available with Visible Infrared Imager Radiometer Suite (VIIRS) moderate-resolution bands M12 and M13, respectively. Under daytime conditions, the data collected in these bands contain energy that originates from both the sun and the Earth–atmosphere system. Due to the small wavelength difference between these, the terrestrial energy component in the bands is typically quite similar as are the surface reflectances for sea ice and ocean surfaces. Thus, the enhanced contrast between sea ice and water clouds, evident in a M12–M13 BTD image, results from differences in the solar energy, which decreases rapidly across this atmospheric window. Observed BTD values for water clouds can exceed 30K, while those for snow-ice fields are typically much smaller (e.g. 0–5K). Thus, water clouds appear bright in the image while sea ice, oceans, and most land surfaces are very dark. The enhanced contrast in the split-window, mid-wave IR BTD image makes it valuable for both image analysis and use in cloud algorithms. In addition, these images support the creation of manually generated cloud masks that have been shown useful for quantitatively evaluating the performance of automated cloud analysis algorithms and cloud forecast models. In this article, the value of 3.7 μm minus 4.0 μm BTD imagery for distinguishing between sea ice and low-level water clouds is shown using VIIRS data collected over the Beaufort Sea on 31 May 2012. Manually generated cloud masks, derived in part from these data, are then used to quantitatively evaluate the effectiveness of various cloud tests, including those used in the VIIRS cloud mask algorithm and the Moderate Resolution Imaging Spectroradiometer (MODIS) cloud mask algorithm. The results strongly suggest that split-window, mid-wavelength IR imagery provides valuable information to help differentiate between clouds and sea ice. It is concluded that collecting data in these mid-wavelength IR bands should be considered part of any future satellite sensor designed for environmental monitoring, especially over the polar regions.     

A daytime cloud detection algorithm for FY-3A/VIRR data

Cloud detection is essential for the retrieval of atmospheric and surface parameters and it directly impacts the quality of many satellite geophysical products used in weather, climate and environmental research. In this article, a daytime cloud detection algorithm based on multi-spectral thresholds is proposed to discriminate clouds from clear skies for the visible and infrared radiometer (VIRR), which is a key instrument on board the Chinese FengYun-3A (FY-3A) polar-orbiting meteorological satellite, launched 27 May 2008. The VIRR has ten bands in the wavelengths 0.43–12.5 μm and provides global observations of atmosphere, ocean and land in the visible and infrared regions of the spectrum. In this algorithm, the underlying surface is divided into five ecological types: snow/ice, desert, coastal, land and water, and seven spectral bands of the VIRR are used to indicate a level of confidence that the VIRR is observing clear skies. This algorithm also utilizes the 1.6 μm band and the difference between the 1.38 and 1.6 μm bands to respectively detect water cloud and high cloud. An example of cloud detection and a comparison with an official cloud masking product are given; the results show that this algorithm performs well and is better than the official algorithm in cloud detection.     

Determination of the cloud fraction in the SCIAMACHY ground scene using MERIS spectral measurements

SCIAMACHY (SCanning Imaging Absorption spectroMeter for Atmospheric CartograpHY) is a passive remote sensing spectrometer observing backscattered radiation from the atmosphere and the Earth's surface, in the wavelength range between 240 and 2380 nm. The instrument is onboard ENVironmental SATellite (ENVISAT) which was launched on 1 March 2002. The Medium Resolution Imaging Spectrometer (MERIS) is also one of the 10 instruments onboard the ENVISAT satellite. MERIS is a 68.5° field-of-view nadir-pointing imaging spectrometer which measures the solar radiation reflected by the Earth in 15 spectral bands (visible and near-infrared). It obtains a global coverage of the Earth in three days. Its main objective is to measure sea colour and quantify ocean chlorophyll content and sediment, thus providing information on the ocean carbon cycle and thermal regime. It is also used to derive the cloud top height, aerosol and cloud optical thickness, and water vapour column. The ground spatial resolution of the instrument is 260 m × 290 m. This paper is aimed at determining the cloud fraction in SCIAMACHY pixels (typically, 30 km × 60 km ground scenes) using MERIS observations and number of thresholds for MERIS top-of-atmosphere reflectances and their ratios. Thresholds utilize the fact that clouds are bright white objects having similar reflectances in the blue and red. The MERIS cloud fraction has been derived for a number of SCIAMACHY states with area of 916 km × 400 km. The results are compared with correspondent cloud fractions obtained using SCIAMACHY polarization measurement devices (PMDs). Large differences are found between cloud fractions derived using SCIAMACHY and MERIS measurements. It is recommended to use highly spatially resolved MERIS observations instead of SCIAMACHY PMD measurements to retrieve cloud fractions in SCIAMACHY pixels. The improvements advised will enhance SCIAMACHY trace gas and cloud retrievals in the presence of broken cloud fields.     

Feasibility of using Geostationary Ocean Colour Imager (GOCI) data for land applications after atmospheric correction and bidirectional reflectance distribution function modelling

Korea's Geostationary Ocean Colour Imager (GOCI) has very high temporal resolution as well as wide spatial coverage. There is thus great interest in testing its applicability for monitoring land areas in addition to ocean areas. GOCI has eight spectral bands, from blue to near-infrared. These bands can be sensitive to vegetation change, but their wavelength ranges are slightly different from those of the extensively studied Moderate Resolution Imaging Spectroradiometer (MODIS). This study examines whether GOCI data can be applied for land monitoring and how GOCI data should be processed so as to reflect the spectral characteristics of land surfaces as detected by polar-orbit satellite sensors. Several image processing steps were performed for the GOCI data, including atmospheric correction and semi-empirical bidirectional reflectance distribution function modelling, before the results were compared with the MODIS land-surface product. Among the four GOCI normalized difference vegetation index (NDVI) products tested in this study, the GOCI NDVI with viewing-angle-adjusted reflectance showed the best agreement with MODIS NDVI calculated from normalized reflectance, with the lowest root mean square error of 0.126. Additionally, its temporal trends over forest and mixed vegetation areas were similar to those of MODIS NDVI during the study period from September to December.     

Extending the MODIS 1 km ocean colour atmospheric correction to the MODIS 500 m bands and 500 m chlorophyll-a estimation towards coastal and estuarine monitoring

National and regional obligations to control and maintain water quality have led to an increase in coastal and estuarine monitoring. A potentially valuable tool is high temporal and spatial resolution satellite ocean colour data. NASA's MODIS-Terra and -Aqua can capture data at both 250 m and 500 m spatial resolutions and the existence of two sensors provides the possibility for multiple daily passes over a scene. However, no robust atmospheric correction method currently exists for these data, rendering them unusable for quantitative monitoring applications. Therefore, this paper presents an automatic and dynamic atmospheric correction approach allowing the determination of ocean colour. The algorithm is based around the standard MODIS 1 km atmospheric correction, includes cloud masking and is applicable to all of the visible 500 m bands. Comparison of the 500 m ocean colour data with the standard 1 km data shows good agreement and these results are further supported by in situ data comparisons. In addition, a novel method to produce 500 m chlorophyll-a estimates is presented. Comparisons of the 500 m estimates with the standard MODIS OC3M algorithm and to in situ data from a near-coast validation site are given.     

A novel algorithm of cloud detection for water quality studies using 250 m downscaled MODIS imagery

ABSTRACT

This study is part of a project aimed at developing an automated algorithm for algal bloom detection and quantification in inland water bodies using Moderate Resolution Imaging Spectroradiometer (MODIS) imagery. An important step is to adequately detect and exclude clouds and haze because their presence affects chlorophyll-a (chl-a) estimations. Currently available cloud masking products appear to be ineffective in turbid coastal waters. The purpose of this study is to develop a cloud masking algorithm based on a probabilistic algorithm (Linear Discriminant Analysis) and designed for water bodies by using MODIS images downscaled at a 250 m spatial resolution (MODIS-D-250). Confusion matrix shows that the new cloud mask algorithm yields very satisfactory results, enabling water classification for heavy turbid conditions with a mean kappa coefficient (κ) of 0.993 and a 95% confidence interval ranging from 0.990 to 0.997. The model also shows a very low commission error (sensitive to the presence of haze), which is essential for accurate water quality monitoring, knowing that the presence of clouds/haze/aerosols leads to major issues in the estimation of water quality parameters. The cloud mask model applied on MODIS-D-250 images improves the sensitivity to haze and the classification of turbid waters located at the edge of urban areas better than the operational MODIS products, and it clearly shows an improvement of the spatial resolution (250 m spatial resolution) compared to other cloud mask algorithms (500 m or 1 km spatial resolution), leading to an increase in exploitable data for water quality studies.     

基于类内拟合的遥感影像薄云雾校正方法

光学遥感观测极易受到云雾影响,降低数据质量并限制其后续应用潜力。由此,提出了一种基于类内拟合的遥感影像薄云雾校正方法。首先,利用滑动窗口逐波段地搜索局部最小值,称之为暗目标,通过拟合不同波段的暗目标样本估计出薄云雾辐射的相关性。基于此,联合云雾波段相关性与成像模型,生成不含云雾干扰的合成假彩色影像,利用K均值分类自动得到地表覆被类型。利用地类信息,进一步选取晴空区像元获取不同地类在不同波段对间的线性关系。最后,将上述两种线性关系进行联立,求解出各地表类型在不同波段上的值,从而完成影像校正。通过模拟与真实实验对方法有效性和场景适用性进行测试,并从定性目视与定量评估两方面对结果进行检验。实验结果表明：提出方法可有效去除薄云雾干扰,适用于不同地表覆被类型场景,获得高光谱保真的校正地表。     

Vicarious calibration of GOCI for the SeaDAS ocean color retrieval

A key on-orbit calibration step for satellite remote sensing of ocean color is the vicarious calibration. This establishes the final gains for each spectral band on the sensor that minimize bias in the retrieved ocean color signal. The vicarious calibration is specific to the instrument and the atmospheric correction algorithm. The vicarious calibration gains for the Geostationary Ocean Color Imager (GOCI) are presented here, which were derived to optimize the performance of NASA’s standard atmospheric correction algorithm as implemented in the l2gen code and distributed through the SeaDAS open-source software package. Following NASA’s protocols, the near-infrared (NIR) bands were calibrated first, and the visible bands were then calibrated relative to this fixed NIR calibration. The gain for the 745-nm NIR band was derived using a fixed aerosol model, which was chosen based on the Angstrom Coefficients derived from MODIS on Aqua (MODISA). For the vicarious gains of the visible bands, two sources for the target water-leaving radiances were tested: matchups from MODISA and climatological data from SeaWiFS. A validation analysis using AERONET-OC data shows an improvement in sensor performance when compared with results using the current vicarious gains and results using no vicarious calibration. Good agreement was found in vicarious gains derived using both concurrent MODISA and climatological SeaWiFS as vicarious calibration data sources. These results support the use of a concurrent sensor for the vicarious calibration when in situ data are not available and demonstrate that using climatology from a well-calibrated sensor like SeaWiFS for the vicarious calibration is a valid alternative when it is not possible to use a concurrent sensor or in situ data. We recommend using the gains derived from concurrent GOCI matchups with MODISA for GOCI processing in SeaDAS/l2gen.     

Ocean colour and SST measurements with Chinese FY-1C satellite

Image processing techniques have been utilized for enhanced visual displays of cloud imagery for clouds over the Indian subcontinent during July, month of the summer monsoon. Indian Satellite (INSAT-1D) bi-channel (infrared and visible band) data were used for contrast manipulation and false colour composite (FCC) production. Two combinations of the infrared and visible bands were used to form the third band required for the FCC method. It is shown here that for the third band, a new combination (IR2/VIS) is better than a previously suggested combination (2VIS-IR) for identifying clouds of different optical thickness during the monsoon season. Various colour combinations have also been studied experimentally in order to produce efficient displays of clouds which conform to common perceptions of their natural colour.     

特征点维度的静态场景三维重建运动目标剔除

基于语义分割的图像掩膜方法常用来解决静态场景三维重建任务中运动物体的干扰问题,然而利用掩膜成功剔除运动物体的同时会产生少量无效特征点.针对此问题,提出一种在特征点维度的运动目标剔除方法,利用卷积神经网络获取运动目标信息,并构建特征点过滤模块,使用运动目标信息过滤更新特征点列表,实现运动目标的完全剔除.通过采用地面图像和航拍图像两种数据集以及DeepLabV3、YOLOv4两种图像处理算法对所提方法进行验证,结果表明特征点维度的三维重建运动目标剔除方法可以完全剔除运动目标,不产生额外的无效特征点,且相较于图像掩膜方法平均缩短13.36%的点云生成时间,减小9.93%的重投影误差.     

A dynamic global cloud layer for virtual globes

We describe a technique to merge multiple environmental satellite data sets for an hourly updated, near real-time global depiction of cloud cover for virtual globe applications. A global thermal infrared composite obtained from merged geostationary- (GEO) and low-Earth-orbiting (LEO) satellite data is processed to depict clear and cloudy areas in a visually intuitive fashion. This GEO-plus-LEO imagery merging is complicated by the fact that each individual satellite observes a single ‘snapshot’ of the cloud patterns, each taken at different times, whereas the underlying clouds themselves are constantly moving and evolving. For the cloudy areas, the brightness and transparency are approximated based upon the cloud top temperature relative to the local radiometric surface temperatures (corrected for surface emissivity variations) at the time of the satellite observation. The technique clearly defines and represents mid- to high-level clouds over both land and ocean. Due to their proximity to the Earth's surface, low-level clouds such as stratocumulus and stratus clouds will be poorly represented with the current technique, since warmer temperatures in this case do not correspond to higher cloud transparency. Overcoming this problem requires the introduction of multispectral channel combinations.     

Estimation of sea surface currents based on ocean colour remote-sensing image analysis

Accurate quantification of the sea surface current speeds with high spatial resolution and in near real-time is beneficial for many applications of physical oceanography, and can be implemented by processing ocean colour remote-sensing imagery data. The robust optical flow (ROF) approach applied to sub-image processing is described in detail in this article, and compared with a conventional maximum cross-correlation (MCC) block matching algorithm with sub-pixel operation. ROF results obtained from Geostationary Ocean Colour Imager (GOCI) imagery are shown to be in good agreement with sea surface currents derived from Ocean Surface Current Analyses-Real time (OSCAR) data, providing a validation of the ROF method.     

面向半球尺度数据的云检测方法

目的 云覆盖着地球上空大部分区域,在地球水循环、地气系统能量平衡和辐射传输过程中有着重要的作用,同时云也是天气气候中最重要、最活跃的因子之一;此外,云覆盖地表信息,导致影像配准、融合等处理过程的很多问题,所以云检测十分重要。方法 基于2015年发射的深空气候观测台（DSCOVR）卫星搭载的地球彩色成像相机（EPIC）数据,针对EPIC数据波段范围较广和影像数据是半球尺度的特点,以云指数法作为基础,提出一种新的面向半球尺度数据的云检测方法。首先,分析EPIC数据各个波段的波段特征,尤其是紫光波段,然后根据云在不同波段的反射特性,以指数的形式完成波段组合进行云检测,再与SVM（support vector machine）云检测法和可见光云检测法进行比较,最后利用EPIC L2产品对所获得的云分布图和统计云量值进行结果验证,以正确率、漏检率、误检率和Kappa系数作为参考标准完成精度评定。结果 实际EPIC夏季（2017年7月）和冬季（2017年1月）数据的实验结果表明,本文方法的正确率均高于91%,Kappa系数大于0.9;其他方法的正确率均低于89%,且Kappa系数在0.8左右,均小于0.9。所以本文能够有效地检测到薄云（即使在冬季）,且云量和云的分布都最为接近实际。结论 在EPIC影像的云检测过程中,本文方法从云分布图和云量结果两个方面都优于可见光云检测法和SVM云检测法,经EPIC L2产品验证,本文方法有效、可靠,且能够快速获得半球范围内云的分布情况,有助于对全球云的动态研究和自然天气预测。     

On the continuity of quantifying floating algae of the Central West Atlantic between MODIS and VIIRS

Studying abundance and distributions of floating macroalgae such as pelagic Sargassum calls for long-term continuous and consistent observations from multiple satellite sensors. Previous studies mainly relied on observations from the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Medium Resolution Imaging Spectrometer (MERIS). As a follow-on sensor, the Visible Infrared Imager Radiometer Suite (VIIRS) also has the appropriate spectral bands to detect and quantify floating macroalgae. Based on previous works on MODIS, this study presents an improved procedure to extract floating algae pixels from VIIRS Alternative Floating Algae Index (AFAI) imagery, with image filtering used to suppress noise and adjusted thresholds used to mask sun glint, clouds, and cloud shadows. The overall extraction accuracy is about 85%. Simultaneous daily observations from MODIS and VIIRS over the Central West Atlantic (CWA) show consistent spatial patterns, but VIIRS estimations of the algae coverage (in km²) are consistently lower than MODIS (around – 19% mean relative difference or MRD), possibly due to lower sensitivity of the VIIRS near-infrared (NIR) bands than the corresponding MODIS bands. Similarly, at monthly scale VIIRS also shows lower coverage than MODIS, and their difference (around – 29% MRD) is larger than the difference between MODIS-Aqua and MODIS-Terra estimates (around – 14% MRD). Despite these differences, the spatial and temporal patterns between VIIRS and MODIS observed algae distributions match very well at all spatial and temporal scales. These results suggest that VIIRS can provide continuous and consistent observations of floating algae distributions and abundance from MODIS as long as their differences are accounted for, thus assuring continuity in the future. Furthermore, once Sargassum biomass per unit Sargassum area is determined from field measurements, conversion of these area estimates to Sargassum biomass is straightforward.     

Alignment of mountain lee waves viewed using NOAA AVHRR imagery,MST radar,and SAR

Recent radar observations of mountain waves in the troposphere and lower stratosphere above Aberystwyth (52.4°N, 4.0°W) indicate that, on average, the wave alignment is related more closely to the wind direction within the boundary layer than to the alignment of mountain ridges. This is investigated using independent data NOAA AVHRR imagery of both mountain-wave clouds and convective cloud streets, combined with surface synoptic wind measurements. The mountain-wave cloud bands are found to be aligned not at exactly 90° to the surface wind but rotated a further 18° clockwise. Similarly, in an important backup test, the cloud streets are found not to be parallel to the surface wind but rotated 12° clockwise, which agrees with over 30 years of observations, most recently of wind rows on the ocean by synthetic aperture radar (SAR). Because the wind rotates, on average, clockwise with increasing height in the northernhemisphere boundary layer, the mountain-wave clouds will be at 90° to the wind direction in the middle of the boundary layer. Therefore, the satellite images independently confirm earlier mesosphere-stratosphere-troposphere (MST) radar observations. Mountain lee waves may corrupt SAR measurements of surface wind above the ocean, so knowledge of their alignment is useful; two examples of lee waves modulating the sea roughness west of Aberystwyth are discussed.     

基于帧差和小波包分析算法的运动目标检测

提出了一种在镜头不动的情况下基于累积帧差分割和小波包分析融合技术的运动目标检测方法.这种方法可分为四步:使用改进的累积帧差算法和阈值分割算法完成目标区域的分割,并获得初始运动模板;利用小波包分析算法提取出单帧图像的边缘信息并获得细化的目标区域边缘图;根据初始运动模板和空域边缘图像的融合得到更精确的运动目标模板;最后结合原序列图像检测出完整的运动目标.实验结果表明:这种方法可以有效地从对比度较小和噪声较大的视频序列中较精确地检测出完整的运动目标.     

基于帧差和小波包分析算法的运动目标检测

提出了一种在镜头不动的情况下基于累积帧差分割和小波包分析融合技术的运动目标检测方法。这种方法可分为四步:使用改进的累积帧差算法和阈值分割算法完成目标区域的分割,并获得初始运动模板;利用小波包分析算法提取出单帧图像的边缘信息并获得细化的目标区域边缘图;根据初始运动模板和空域边缘图像的融合得到更精确的运动目标模板;最后结合原序列图像检测出完整的运动目标。实验结果表明:这种方法可以有效地从对比度较小和噪声较大的视频序列中较精确地检测出完整的运动目标。     

Nighttime polar cloud detection with MODIS

Cloud detection is the first step in studying the role of polar clouds in the global climate system with satellite data. In this paper, the cloud detection algorithm for the Moderate Resolution Imaging Spectrometer (MODIS) is evaluated with model simulations and satellite data collocated with radar/lidar observations at three Arctic and Antarctic stations. Results show that the current MODIS cloud mask algorithm performs well in polar regions during the day but does not detect more than 40% of the cloud cover over the validation sights at night. Two new cloud tests utilizing the 7.2 μm water vapor and 14.2 μm carbon dioxide bands, one new clear-sky test using the 7.2 μm band, and changes to the thresholds of several other tests are described. With the new cloud detection procedure, the misidentification of cloud as clear decreases from 44.2% to 16.3% at the two Arctic stations, and from 19.8% to 2.7% at the Antarctic station.