海表面盐度卫星微波遥感研究进展

海表面盐度SSS(Sea Surface Salinity)是研究大洋环流和海洋对全球气候影响的重要参量、是决定海水基本性质的重要因素之一。卫星微波遥感可以满足盐度研究过程中大范围、连续观测的需要,国际上统一的认识是选择频率为1.413 GHz的L波段作为盐度遥感的首选波段。目前,国外发展的海面盐度微波遥感反演算法主要有两种:基于海表发射率估算海表盐度的算法和基于贝叶斯定理提出的反演算法。影响盐度反演精度的因素主要有太空辐射、电离层法拉第旋转、大气、海面粗糙度等。其中,海面粗糙度对盐度反演影响很大,海面粗糙度处理模型可以分为3大类:理论算法(间接发射率模型、直接发射率模型)、经验算法、半经验半理论算法(Hollinger 半经验模型、WISE半经验模型、Gabarró模型)。SMOS卫星和Aquarius/SAC\|D卫星的成功发射,将海表面盐度遥感的反演精度控制在0.2 psu以内,通过改进反演算法,有望得到更高的反演精度。     

基于星载微波辐射计的海面风场对海表盐度反演影响研究

为了研究海面风场对海表盐度反演结果的影响,需要构建准确的海面风场影响下的海表亮温模型。将不同海面粗糙度模型计算的结果与Aquarius盐度计卫星产品中的海面粗糙度数据进行了比较,结果表明双尺度模型结合海面泡沫模型的计算结果与Aquarius卫星产品粗糙度数据一致性最好。基于此构建了海面风场影响下的海面亮温仿真模型以及双极化通道的盐度反演模型,研究了海面风场对L波段海面微波辐射特性的影响以及风场资料误差对盐度反演精度的影响。仿真结果表明,2m/s的风速误差对盐度反演结果的影响较大。在低温和大风速条件下达到1psu以上,尚不能满足目前盐度遥感的精度要求。20°的风向误差对在中小入射角条件下对盐度反演结果影响较小,对盐度遥感的月平均要求影响不大。     

巴拉望岛附近海域盐度遥感反演

针对利用光学遥感卫星进行近岸海域海表盐度的高精度反演需求,基于多元统计回归分析思想构建研究区海洋盐度反演模型,以巴拉望岛附近海域为研究区,分别以MERIS卫星和MODIS卫星影像为基础数据,对该海域的海表盐度进行反演,并利用实测Argo数据对反演结果进行精度评定和对比分析,剖析了研究区海水盐度的时空分布规律。结果表明:1)MERIS和MODIS 2种数据反演误差分别为-2.94%和1.86%,反演模型的相关系数R~2达到0.79,表明2种数据反演结果的精度较高,模型适用性较强;2)相比之下,MODIS数据反演效果优于MERIS数据;3)研究区域的海表盐度发生着季节性的变化,并且呈现出自西向东逐渐增加的趋势,这种现象可能与研究区气候、温度变化以及所处地理位置有内在联系。     

海面粗糙度引起的SMOS卫星亮温增益研究

海表面亮温是决定海表面盐度反演精度的关键因素。针对海表面粗糙度对海表面亮温增益的影响问题,修正了SMOS(Soil Moisture and Ocean Salinity)卫星亮温粗糙度模型。从SMOS卫星半经验半理论模型入手,利用2014年7月8日西北太平洋区域SMOS卫星L1C数据、L2级盐度产品数据、HY-2A(海洋二号)卫星SWH(有效波高)、U10(海面10m处风矢量)数据、Argo实测海表面盐度和海表面温度数据,使用非线性拟合、P检验和BP神经网络模型等方法,对平静海面亮温和海面粗糙度引起的亮温增益模型进行了算法修正。通过与Argo实测海表面盐度对比,评价了模型修正效果。模型修正后的水平极化亮温和垂直极化亮温反演出的海表面盐度的相对误差分别为0.005和0.004,优于模型修正前。     

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。     

基于Landsat TM图像的北京城市地表温度遥感反演研究

利用北京地区Landsat TM热红外波段数据，采用单通道算法反演得到北京地区地面温度分布图。从反演结果可以看出，北京城区地面温度比郊区地表温度高，郊区地表温度较低，密云水库、官厅水库等水体的温度最低，总体上北京城市热岛效应显著。地表比辐射率是通过Van经验公式反演得到，通过对比分析，表明该方法对自然地表的比辐射率反演效果较好。     

L波段多角度裸露地表土壤水分反演研究

土壤水分是气象预报、农情监测以及水文模型的重要参数之一,利用被动微波遥感技术可以有效获取土壤水分。欧空局（ESA）计划于2009年发射卫星SMOS(Soil Moisture and Ocean Salinity) ,其主要目的是监测全球范围内的土壤水分和海洋盐度变化。根据SMOS的设置情况,寻找精度较高的半经验模型以便为进一步的土壤水分反演提供简化模型。对于裸露地表,地表粗糙度、土壤介电常数等因素影响最终的微波发射率。运用Dobson半经验介电常数模型计算土壤的介电常数,将计算结果输入高级积分方程模型(AIEM)。通过AIEM模拟的数据库,利用回归关系建立了一个精度相对较高的L波段多角度半经验模型。     

基于深度学习的海表温度遥感反演模型

针对传统的海温反演模型参数拟合过程复杂且在不同海域适应性较差等问题,为进一步提高海表温度反演精度,简化反演过程,以渤海海域为研究区域,选取该海域晴空下的MODIS遥感影像并结合实测浮标数据,利用深层神经网络建立海表温度的遥感反演模型。对反演结果进行精度分析,分析结果表明模型精度较为良好,反演值与实测值决定系数为0.978,标准差为1.28℃,平均绝对误差为0.98℃,证明了基于深度学习的海表温度遥感反演的可行性。     

风云三号传感器红外通道海表比辐射率模拟

海表比辐射率（SSE）是从卫星被动遥感数据反演海表温度（SST）的关键参数之一,本文讲述了利用Wu andSmith（1997）发展的SSE模型,对搭载在风云三号（FY-3A）上的红外大气探空仪（IRAS）第8（12.47μm）、9（11.11μm）、19（3.98μm）和20（3.76μm）波段,中分辨率光谱成像仪（MERSI）第5波段（11.54μm）,可见光红外扫描辐射计（VIRR）第3（3.65μm）、4（11.00μm）和5（12.13μm）波段的SSE进行模拟计算。模拟结果表明,SSE是观测角度、波长和海表粗糙度（海面风速）的函数。模拟值与IRAS、MERSI和VIRR红外通道波谱响应函数进行卷积计算得到波段SSE。在观测角度小于60°时,波段SSE对海面风速不敏感,使用风速为8m/s的SSE替代其他风速条件下的SSE所引起的误差小于0.5%。另外,使用波段SSE计算卫星观测值所导致的误差不超过0.05K,可以忽略不计。波段SSE随观测角度变化的规律可以用高斯函数进行描述,拟合误差小于0.02%,相关系数为1.000。本文的模拟结果可以用于SST的精确反演。     

利用SMOS卫星数据反演海洋盐度方法研究

以星载微波遥感的辐射传输方程为基础,利用 SMOS(土壤湿度海洋盐度)卫星的L1C级亮温数据,通过与辐射传输模型模拟的亮温进行对比,评估及验证亮温的数据质量,建立了海洋盐度反演算法。通过分析2012年7月东南太平洋区域(45°~5°S,140°~90°W)的下降轨道数据,发现MIRAS亮温与模型模拟亮温之间总是存在几K的系统偏差,即OTT,因此提出了两种反演盐度的方法:一种是修正OTT偏差,使用入射角0°~55°的数据反演盐度;另一种是不修正OTT偏差,使用大入射角范围35°~55°的数据进行盐度反演。再通过利用MIRAS多角度信息,对亮温作二阶多项式拟合,减少随机噪声对反演的影响。最后采用最小二乘法,使得MIRAS的二阶拟合亮温与模型仿真亮温最接近,迭代反演盐度值。并将反演结果分别与欧空局的L2级盐度数据产品和Argo盐度数据进行比较,来验证反演算法。结果表明:修正OTT之后全角度数据反演的盐度值在50 km×50 km范围内、卫星过境前后5 d,与Argo浮标盐度匹配比较的均值为1.38 pss,标准差为0.35 pss;不修正OTT,直接利用大入射角范围35°~55°的MIRAS亮温反演盐度,与Argo盐度误差均值为0.03 pss,标准差为0.33 pss;同时欧空局的L2级盐度与Argo盐度误差均值为0.26 pss,标准差为0.38 pss。可见利用大入射角范围的反演方法很好地反演了海洋盐度。     

A geometrical optics model based on the non-Gaussian probability density distribution of sea surface slopes for wind speed retrieval at low incidence angles

In this study, a large amount of data from precipitation radar (PR) and National Data Buoy Center (NDBC) buoys are collocated for the development and validation of a Geometrical Optics Model, in order to retrieve wind speed at small incidence angles. The omni-directional model is developed based on the combination of the quasi-specular scattering theory and non-Gaussian probability density distribution of ocean surface slope, and can be applied at incidence angles as high as 15°. There are four parameters included in the proposed model: the effective Fresnel reflection coefficient, the mean square slope, and the two coefficients associated with the kurtosis of the sea surface slope distribution. Using one half of the collocated data, the dependence of the four parameters on the in situ wind speed is acquired. The results show that the effective Fresnel reflection coefficient has a decrease relative to that obtained in previous studies. We combine the proposed model with the maximum-likelihood estimation (MLE) technique to retrieve the ocean surface wind speed at the 10 m height. The retrieved wind speeds are then validated against those measured by the NDBC buoys. The comparison shows that the root mean square error (RMSE) and bias between the model retrievals and buoy observations are 1.54 m s^–1 and 0.1 m s^–1, respectively, revealing high agreements in the wind speed estimations. The results of this study indicate that the proposed model and the PR measurements at low incidence angles can provide reasonably accurate estimates of the surface wind speeds within the range of 0–20 m s^–1.     

A new algorithm for microwave radiometer remote sensing of sea surface salinity without influence of wind

The microwave radiation of the sea surface, which is denoted by the sea surface brightness temperature, is not only related with sea surface salinity (SSS) and temperature (SST), but also influenced by sea surface wind. The errors of wind detected by satellite sensor have significant influences on the accuracy of SSS retrieval. The effect of sea surface wind on sea surface brightness temperature, i.e. ΔT_h,v , and the relations among ΔT_h,v , wind speed, sea surface temperature, sea surface salinity and incidence angle of observation are investigated. Based on the investigations, a new algorithm depending on the design of a single radiometer with dual polarizations and multi‐incidence angles is proposed. The algorithm excludes the influence of sea surface wind on SSS retrieval, and provides a new method for remote sensing of SSS.     

Radar imaging of Kelvin arms of ship wakes

The wave pattern generated by a moving ship is formed by two dominant features: the turbulent wake and a 'V'-shaped pattern trailing the ship, consisting of the two Kelvin arms. In this paper we investigate the radar imaging mechanism of Kelvin arms, which are formed by the cusp waves. A composite surface model for the radar backscattering at the ocean surface is used. The radar signatures of Kelvin arms can be attributed to tilt and hydrodynamic modulation of Bragg waves by the cusp waves. The proposed model allows the computation of the normalized radar backscattering cross-section (NRCS) as a function of radar frequency, polarization, incidence angle, wind speed and direction, and wavelength, direction, and slope of the cusp waves. By using this imaging model, radar signatures of cusp waves are calculated for several spaceborne Synthetic Aperture Radars (SARs): (1) the SEASAT L-band HH-polarized SAR, (2) the ERS-1/-2 VV-polarized SAR, (3) the RADARSAT C-band HH-polarized SAR, and (4) the X-, C- and L-band multipolarization SARs of the Space Radar Laboratory flown on the space shuttle during the SIRC/X-SAR mission in 1994. The results of the simulations are compared with SEASAT and SIR-C/X-SAR imagery of ship wake patterns. It is shown that the dependence of the observed radar signatures of Kelvin arms on radar look direction is consistent with the proposed imaging theory and that the measured relative mean NRCS values induced by Kelvin arms can be fairly well reproduced by the proposed model. The simulations indicate that ship wake signatures should be more clearly visible on SEASAT L-band SAR than on ERS-1/-2 or RADARSAT C-band SAR images. The radar signatures of Kelvin arms are strongest at low wind speeds and are not very sensitive to wind direction.     

SMOS first data analysis for sea surface salinity determination

Soil Moisture and Ocean Salinity (SMOS), launched on 2 November 2009, is the first satellite mission addressing sea surface salinity (SSS) measurement from space. Its unique payload is the Microwave Imaging Radiometer using Aperture Synthesis (MIRAS), a new two-dimensional interferometer designed by the European Space Agency (ESA) and operating at the L-band frequency. This article presents a summary of SSS retrieval from SMOS observations and shows initial results obtained one year after launch. These results are encouraging, but also indicate that further improvements at various data processing levels are needed and hence are currently under investigation.     

Retrieval of remotely sensed sea surface salinity using MODIS data in the Chinese Bohai Sea

Salinity dominates seawater density and directly affects physical and biochemical processes. Having a reliable retrieval model is essential to providing frequent and accurate sea surface salinity (SSS) data for marine research. Remote-sensing techniques provide alternatives for SSS data retrieval with its advantages of wide area surveys and real-time monitoring. In the present study, inverse relationship between SSS and coloured dissolved organic matter (CDOM) concentration in the Chinese Bohai Sea was verified. Thus, four simple band ratios of the original remote-sensing reflectance (R_rs) used to retrieve the CDOM concentration were compared and tested during SSS retrieval. R_rs (531)/R_rs (551) performed best among the four given band ratios. The model employed here can be applied to derive SSS with a root mean square error (RMSE) of 0.26 practical salinity units (psu) (R² = 0.76). A calibration model was verified using a discrete dataset of the measured SSS and was tested further during mapping of SSS in the Chinese Bohai Sea during 2010–2014. The yielded spatial patterns of SSS were satisfactory and an inverse relationship between SSS and the Yellow River discharge was confirmed.     

NEWS SECTION

Abstract

The action balance equation is solved numerically using the method of characteristics. The algorithm can handle any functional form of the current velocity and the source function. It therefore allows a comparison of the hydrodynamic parts of the models of Alpers and Hennings (1984), Shuchman el al. (1985) and Holliday et al. (1986,1987), all describing the imaging mechanism of mapping bottom topography with side-looking airborne radar (SLAR) and synthetic aperture radar (SAR). The effects of advection and second-order terms, neglected in the original work of Alpers and Hennings (1984), are included and studied. It is shown that advection is important, notably at L-band for features smaller than 1 km, such as sand waves, while the second-order terms shift the modulation of the wave spectrum upward. All models studied give similar results for L-band and X-band, showing that the dependence of the relaxation rate on wave number and wind speed variations due to variations in current velocity has only a small influence. The predicted modulations at L-band are of the order of 10 per cent, in agreement with the Seasat data. At X-band, however, the predicted modulations are an order of magnitude smaller than at L-band. This disagrees with the experimental data which seem to indicate that modulations at X-band are of the same order of magnitude as those at L-band. Advection is important for the positioning of modulations in the radar backscatter relative to the bottom topography. However, the positional accuracy of existing experimental data is not good enough to allow comparison with theoretical predictions.     

SAR-derived wind fields at the coastal region in the East/Japan Sea and relation to coastal upwelling

The relationship between the modification of synthetic aperture radar (SAR) wind field and coastal upwelling was investigated using high-resolution wind fields from Advanced Land Observing Satellite (ALOS) Phased Array type L-band synthetic aperture radar (PALSAR) imagery and sea-surface temperature (SST) from National Oceanic and Atmospheric Administration/Advanced Very-High-Resolution Radiometer (NOAA/AVHRR) data. The retrieved SAR wind speeds seem to agree well with in situ buoy measurements with only a relatively small error of 0.7 m s^?1. The SAR wind fields retrieved from the east coast of Korea in August 2007 revealed a spatial distinction between near and offshore regions. Low wind speeds of less than 3 m s^?1 were associated with cold water regions with dominant coastal upwelling. Time series of in situ measurements of both wind speed and water temperature indicated that the upwelling was induced by the wind field. The low wind field from SAR was mainly induced by changes in atmospheric stability due to air–sea temperature differences. In addition, wind speed magnitude showed a positive correlation with the difference between SST and air temperature (R² = 0.63). The dependence of viscosity of water on radar backscattering at the present upwelling region was negligible since SAR data showed a relatively large backscattering attenuation to an SST ratio of 1.2 dB °C^?1. This study also addressed the important role of coastal upwelling on biological bloom under oligotrophic environments during summer.     

New sea surface salinity retrieval methods based on SMOS data

The quality of brightness temperatures (T_Bs) provided by the Soil Moisture and Ocean Salinity (SMOS) satellite is assessed and validated by comparing them with simulated T_Bs. T_B simulations are computed using the default transfer model implemented by the European Space Agency (ESA) level 2 ocean salinity processor, with auxiliary data taken from The European Centre for Medium Range Weather Forecasts (ECMWF). The ascending SMOS data of the western Pacific Ocean region (0 N–30 N, 120 E–150 E) in July and August 2012 are analysed, and biases of several kelvin are observed between the averaged SMOS T_B and simulated T_B, which strongly affect the accuracy of sea surface salinity (SSS) retrieval. Two methods are proposed in this article to deal with the biases. The first is that the biases are corrected, and the full range of SMOS multiangular T_B is used for retrieval. The second is that the biases are not corrected, and only T_B at angles of incidence in the range of 35–55 are used. Then all measurements of a given Stokes parameter in a pixel are regressed with respect to angles of incidence using a second-order polynomial fit in order to reduce noise in the T_B. Finally, a least squares iterative process is used to retrieve SSS using the fitted T_B. The accuracy of the retrieved SSS using the above two methods is estimated by comparing them with the Array for Real-Time Geostrophic Oceanography SSS and the ESA level 2 SSS.