神经网络反演散射计风场算法的研究

建立了一个神经网络反演卫星散射计海面风场的B-P算法,给出了一个神经网络反演风场的模型,并利用该反演算法和模型对实际卫星散射计数据进行了海面风场反演试验,对风向的多解性利用圆中数滤波方法进行排除.对神经网络训练和检验数据集分别采用ERS-1/2散射计数据和欧洲中期天气预报(ECMWF)提供的风场作为配准点数据.把反演的风速和风向与CMCD4和ECMWF的风场作了比较,它们吻合得比较好;研究表明神经网络反演海面风场是可行和高效的.     

HY-2B卫星散射计海面风场产品质量分析 *

星载微波散射计是获取全球海面风场信息的主要手段, HY-2B卫星散射计的成功发射为全球海面风场数据获取的持续性提供了重要保障。本文利用欧洲中期天气预报中心(European Center for Medium-Range Weather Forecasts, ECMWF)再分析风场数据、热带大气海洋观测计划(Tropical Atmosphere Ocean Array, TAO)和美国国家数据浮标中心(National Data Buoy Center, NDBC)浮标获取的海面风矢量实测数据, 对HY-2B散射计海面风场数据产品的质量进行统计分析。分析表明, HY-2B风场与ECMWF再分析风场对比, 在4~24m·s^-1风速区间内, 风速和风向均方根误差(root mean square error, RMSE)分别为1.58m·s^-1和15.34°; 与位于开阔海域的TAO浮标数据对比, 风速、风向RMSE分别为1.03m·s^-1和14.98°, 可见HY-2B风场能较好地满足业务化应用的精度要求(风速优于2m·s^-1, 风向优于20°)。与主要位于近海海域的NDBC浮标对比, HY-2B风场的风速、风向RMSE分别为1.60m·s^-1和19.14°, 说明HY-2B散射计同时具备了对近海海域风场的良好观测能力。本文还发现HY-2B风场质量会随风速、地面交轨位置等变化, 为用户更好地使用HY-2B风场产品提供参考。     

基于混合密度网络的NSCAT散射计海面风场反演

以NSCAT散射计数据为例,介绍了一种神经网络反演海面风场的方法.风速的反演是基于多层感知器网络;多解风向的反演是基于多层感知器网络和混合密度模型组合而成的混合密度网络,其中的核函数采用高斯函数的形式.通过与欧洲中期天气预报模式风场和现场浮标数据对比,证明了该神经网络反演海面风场的有效性.     

Global air-sea surface carbon-dioxide transfer velocity and flux estimated using ERS-2 data and a new parametric formula

Using data from the European remote sensing scatterometer（ERS-2） from July 1997 to August 1998,global distributions of the air-sea CO₂ transfer velocity and flux are retrieved.A new model of the air-sea CO₂ transfer velocity with surface wind speed and wave steepness is proposed.The wave steepness（5） is retrieved using a neural network（NN） model from ERS-2 scatterometer data,while the wind speed is directly derived by the ERS-2 scatterometer.The new model agrees well with the formulations based on the wind speed and the variation in the wind speed dependent relationships presented in many previous studies can be explained by this proposed relation with variation in wave steepness effect.Seasonally global maps of gas transfer velocity and llux are shown on the basis of the new model and the seasonal variations of the transfer velocity and llux during the 1 a period.The global mean gas transfer velocity is 30 cm/h after area-weighting and Schmidt number correction and its accuracy remains calculation with in situ data.The highest transfer velocity occurs around 60°N and 60°S,while the lowest on the equator.The total air to sea CO₂ llux（calculated by carbon） in that year is 1.77 Pg.The strongest source of CO₂ is in the equatorial east Pacific Ocean, while the strongest sink is in the 68°N.Full exploration of the uncertainty of this estimate awaits further data.An effectual method is provided to calculate the effect of waves on the determination of air-sea CO₂ transfer velocity and fluxes with ERS-2 scatterometer data.     

Validation and accuracy analysis of wind products from scatterometer onboard the HY-2B satellite

The Chinese marine dynamic environment satellite HY-2B was launched in October 2018 and carries a Ku-band scatterometer. This paper focuses on the accuracies of HY-2B scatterometer wind data during the period from November 2018 to May 2021. The HY-2B wind data are validated against global moored buoys operated by the U.S. National Data Buoy Center and Tropical Atmosphere Ocean, numerical model data by the National Centers for Environmental Prediction, and the Advanced Scatterometer data issued b...     

The SeaSat-A satellite scatterometer

This paper describes the methods used to develop performance requirements and design characteristics for the microwave scatterometer (SASS) ocean-surface wind sensor on the NASA SeaSat-A satellite. Wind vector measurement requirements from the SeaSat user community such as wind speed and direction accuracy, resolution cell size, grid spacing, and swath width formed the basis for defining instrument characteristics. The resulting scatterometer is designed for 14.6 GHz using four fan beam antennas to measure wind speed and direction over a 1000-km swath width with a resolution cell size50 times 50km. Results presented show scatterometer accuracy satisfies user requirements for wind speed from 4 m/s to greater than 24 m/s for the nominal SeaSat-A orbit of 790 km altitude,108deginclination, and 0.001 eccentricity.     

一种温度相关的HY-2A散射计地球物理模型函数

利用散射计测量海面后向散射系数, 并通过地球物理模型函数(geophysical model function, GMF)反演得到海面风场。目前散射计风场反演所采用的GMF一般只考虑雷达极化方式、雷达入射角、风速和相对风向对海面后向散射系数的影响, 而相关研究表明海表温度(sea surface temperature, SST)对Ku波段散射计风场反演具有不可忽略的影响。文章利用海洋二号A卫星散射计(Haiyang-2A Scatterometer, HY2A-SCAT)后向散射系数观测值、欧洲中期天气预报中心(European Center for Medium-Range Weather Forecasts, ECMWF )再分析风矢量和SST数据, 采用人工神经网络方法, 建立起一种SST相关的GMF (TNGMF)。对TNGMF进行分析后发现, 海面后向散射系数随着SST的增加而增加, 并且其增加幅度与雷达极化方式、风速有关。为了对比, 文章使用相同数据集和相同方法建立了不包含SST的GMF (NGMF), 将美国国家航天航空局散射计-2 (National Aeronautics and Space Administration Scatterometer-2, NSCAT2) GMF、TNGMF和NGMF分别用于HY2A-SCAT风场反演实验。试验结果表明, 采用NSCAT2 GMF、NGMF反演得到的风速在低温时系统性偏小, 在高温时系统性偏大; 而TNGMF可较好地纠正SST对风速偏差均值的影响, 从而提高反演风场质量。     

一种改进的扇形波束旋转扫描散射计海面风矢量反演算法

扇形波束旋转扫描散射计（RFSCAT）是约十年前才被提出来的一种新型星载微波散射计。与其它旋转扫描散射计类似,其星下点附近区域和刈幅边缘区域的风场反演误差相对较大。在本文设定的参数条件下,RFSCAT散射计刈幅边缘区域的风向反演精度相对于轨道中间区域降低了约9°。针对这一问题,本文为RFSCAT散射计提出了一种改进的风矢量反演算法。新算法的主要特征是,根据风向反演偏差直方图,在整个刈幅区域内,对模糊解风向取值区间进行自适应扩展,以获取并保留更多可能风向解。利用模拟的100条轨道的L2A数据,对新算法进行反演验证。实验结果证明,新算法能够有效改善RFSCAT散射计星下点附近区域和轨道刈幅边缘区域的风向反演精度。星下点和刈幅边缘上的风矢量单元的风向反演精度相对于标准的MLE算法分别提高了1.6°和9°。     

中法海洋卫星散射计近海岸海面风场反演研究

中法海洋卫星散射计(CSCAT)使用扇形波束旋转扫描体制,能够以多角度测量同一海面的雷达后向散射系数,并具有空间分辨率较高的特点。这为近海岸海面风场反演提供了新的机遇。本文介绍了CSCAT近海岸海面风场处理的主要流程和关键技术。特别地,在风场反演之前,利用一种矩形窗算术平均的方法将L1B级的高分辨率条带数据组合平均到相应的风矢量单元中,从而实现近海岸风场反演的快速预处理。通过对比CSCAT、欧洲先进散射计(ASCAT)以及美国QuikSCAT的近海岸风场,发现CSCAT风场的质量在离岸40 km以外区域具有良好的一致性,而在离岸40 km以内显著恶化。分析表明,CSCAT近海岸区域风场统计特征恶化的原因可能是由潜在的海冰污染引起的。总体而言,CSCAT的近海岸风场与模式背景风场和浮标风场都具有良好的一致性。     

AAFE RADSCAT 13.9-GHz measurements and analysis: Wind-speed signature of the ocean

About 10 years ago, the advanced application flight experiment radiometer scatterometer (AAFE RADSCAT) made its first successful measurements of ocean radar scattering cross section from a NASA C-130 aircraft. This instrument was developed as a research tool to evaluate the use of microwave frequency remote sensors (particularly radars) to provide wind-speed information at the ocean's surface. The AAFE RADSCAT flight missions and analyses helped establish the feasibility of the satellite scatterometer for measuring both wind speed and direction. Probably the most important function of the AAFE RADSCAT was to provide a data base of ocean normalized radar cross-section (NRCS) measurements as a function of the surface wind vector at 13.9 GHz. NRCS measurements over a wide parametric range of incidence angles, azimuth angles, and winds were obtained in a series of RADSCAT aircraft missions from 1973 to 1977. Presented herein are analyses of data from the 26 RADSCAT flights during which the quality of the sensor and the surface wind measurements were felt to be understood. Subsets of this data base were used to model the relationship between theKu-band radar signature and the ocean-surface wind vector. The models developed partly from portions of this data base, supplemented with data from the Seasat (JASIN Report), were used for inversion of the Seasat-A Satellite Scatterometer (SASS) radar measurements to vector winds. This paper summarizes results from a comprehensive analysis of the RADSCAT/ocean wind signature deduced from this complete data set.     

Errors in scatterometer-radiometer wind measurement due to rain

An on-board microwave radiometer can correct measurement errors due to rain attenuation for a satellite-borne microwave scatterometer, thereby allowing more accurate determination of the ocean-surface wind vector. In clear weather or when area-extensive homogeneous clouds and rain are present, corrections can be made properly even if the radiometer footprint is much larger than that of the scatterometer. When frontal or cyclonic storms are present, so that the rain and cloud distributions are nonuniform across the footprints, substantial errors remain in the "corrected" scatterometer measurements (in some cases "corrections" actually increase the errors). Simple-geometry simulations are reported here for nonidentical overlapping scatterometer and radiometer footprints with large gradients of attenuation and wind speed. In addition, examples are presented for a hurricane observed by the SEASAT-1 oceanographic satellite. These simulations demonstrate the size of the errors in wind speed (and direction for the hurricane) remaining after "correction." At higher wind speeds, the error is sometimes twice as large as the actual wind speed. The worst errors occur when the scatterometer footprint overlaps two or more radiometer footprints and the attenuation in the scatterometer footprint differs greatly from those in parts of the radiometer footprints. The presence of such large errors is inherent in systems having independent scan patterns for radiometer and scatterometer and having large radiometer footprints. A true radiometer-scatterometer system, having identical coincident footprints comparable in size with typical rain cells, could overcome this problem.     

Features of marine wind fields over New Zealand waters from ERS‐1 scatterometer data.

Wind data from the ERS‐1 scatterometer have been processed for New Zealand waters. These show spatial features of marine wind fields which have previously been difficult to resolve using conventional surface‐based measurements. Winds across the western access to Cook Strait, delineated by a corridor between Farewell Spit and western Taranaki, were analysed and profiles of wind stress extracted. These show characteristic structures for south‐easterly events in which the stress steadily increases from Farewell Spit towards the Taranaki Coast. In westerly or north‐westerly events the structure is more uniform. The mean stress across this corridor has been compared to that calculated from surface‐based measurements at Farewell Spit and the Maui‐A oil and gas production platform off Cape Egmont. The Farewell spit data lead to underestimates of the stress, which partially reconciles previous attempts to model wind‐driven currents off the west coast of the South Island from these data. In these the currents were underestimated. The Maui‐A data are unbiased in westerly events but give overestimates in southeasterly winds. An improved estimate of the mean stress can be derived from using a combination of wind data from these two stations.     

A study of remote sensing of sea-surface wind field with 8-12 GHz broad band radar scatterometer

The paper is devoted to the results of the study on sea-surface wind field remote sensing using 8-12 GHz broad band radar scatterometer. Principles and techniques of remote sensing with broad band scatterometer are described. Variations of microwave scattering coefficients on sea surface with different frequencies and incident angles at different polarization and wind velocity are given and the correlationship between scattering coefficients and wind speeds. The result shows that the research is valuable.lt provides data and basis for remote sensing of the sea-surface wind fields with broad band radar scatterometer.     

基于SVR和LSTM算法反演散射计下有效波高研究

有效波高反演对于海洋工程及海洋环境安全具有重要意义.我国海洋二号(HY-2A)卫星载有散射计和高度计等获取海洋要素的仪器.散射计可获取海洋风场数据但无法直接获取有效波高数据,高度计可获取海洋有效波高数据但覆盖区域狭小.本文将散射计与高度计各自优势结合,利用支持向量回归(SVR)和长短期记忆(LSTM)智能算法反演散射计...     

Evaluation of ENVISAT ASAR data for sea surface wind retrieval in Hong Kong coastal waters of China

The C-band wind speed retrieval models, CMOD4, CMOD - IFR2, and CMOD5 were applied to retrieval of sea surface wind speeds from ENVISAT （European environmental satellite） ASAR （advanced synthetic aperture radar） data in the coastal waters near Hong Kong during a period from October 2005 to July 2007. The retrieved wind speeds are evaluated by comparing with buoy measurements and the QuikSCAT （quick scatterometer） wind products. The results show that the CMOD4 model gives the best performance at wind speeds lower than 15 m/s. The correlation coefficients with buoy and QuikSCAT winds are 0.781 and 0.896, respectively. The root mean square errors are the same 1.74 m/s. Namely, the CMOD4 model is the best one for sea surface wind speed retrieval from ASAR data in the coastal waters near Hong Kong.     

Validation of gridded data set of global surface wind/wind-stress vector field

Gridded data of global surface wind/wind-stress vectors, called J-OFURO v2, were obtained from satellite scatterometer (QuikSCAT/SeaWinds) data for the decade from August 1999 to July 2009. The data were validated by comparing with (1) in situ measurements from moored buoy observations, (2) a data product from the same scatterometer using a different gridding procedure (IFREMER), and (3) data products based on numerical models (NCEP-1 and NCEP-2). The results on averaging all buoy data revealed lower mean differences, lower root-mean-square (RMS) differences, and higher correlations for the wind/wind-stress components for the J-OFURO v2 product than the other data products. The statistical values calculated for each buoy station showed tendencies of decreasing reliability with increasing latitude in the mid-latitude region, while the reliabilities in the equatorial areas were low. We performed intercomparisons between the J-OFURO v2 data and the other data sets to clarify discrepancies among different wind products in open ocean regions with few moored buoys. We determined that the meridional wind components from the NCEP products exhibited poleward deviations compared to data from the J-OFURO v2 product. Relatively high mean differences, high RMS differences, and low correlations were found in the equatorial ocean for the NCEP products. Striped features were spatially correlated with buoy locations in the equatorial Pacific, which suggested that the reliability of the NCEP products was governed by buoy locations in this region.     

基于散射计风场数据的台风强度诊断方法——以海洋二号卫星数据为例

本文提出了一种基于散射计风场数据的台风定强方法。该方法定义了台风所在海区的一个圆形区域上的风速平均值来对台风进行定强,并使用海洋二号卫星散射计历史数据对该方法进行了验证。验证结果表明,该台风定强方法可以克服散射计反演台风风速过低的问题。与台风强度监测的历史记录比较的结果显示,在台风强度低于"强台风"这一强度等级时,该方法能够有效估算台风强度。     

Comparison of sea surface wind field measured by HY-2A scatterometer and WindSat in global oceans

In this study, we present a comprehensive comparison of the sea surface wind ?eld measured by scatterometer(Ku-band scatterometer) aboard the Chinese HY-2 A satellite and the full-polarimetric radiometer WindSat aboard the Coriolis satellite. The two datasets cover a four-year period from October2011 to September 2015 in the global oceans. For the sea surface wind speed, the statistical comparison indicates good agreement between the HY-2 A scatterometer and WindSat with a bias of nearly 0 m/s and a root mean square error(RMSE) of 1.13 m/s. For the sea surface wind direction, a bias of 1.41° and an RMSE of 20.39° were achieved after excluding the data collocated with opposing directions. Furthermore,discrepancies in sea surface wind speed measured by the two sensors in the global oceans were investigated.It is found that the larger dif ferences mainly appear in the westerlies in the both hemispheres. Both the bias and RMSE show latitude dependence, i.e., they have signi?cant latitudinal ?uctuations.     

Comparing satellite and meteorological data on wind velocity over the Black Sea

Wind-velocity data obtained from in situ measurements at the Golitsyno-4 marine stationary platform have been compared with QuikSCAT scatterometer data; NCEP, MERRA, and ERA-Interim global reanalyses and MM5 regional atmospheric reanalysis. In order to adjust wind velocity measured at a height of 37 m above the sea surface to a standard height of 10 m with stratification taken into account, the Monin–Obukhov theory and regional atmospheric reanalysis data are used. Data obtained with the QuikSCAT scatterometer most adequately describe the real variability of wind over the Black Sea. Errors in reanalysis data are not high either: the regression coefficient varies from 0.98 to 1.06, the rms deviation of the velocity amplitude varies from 1.90 to 2.24 m/s, and the rms deviation of the direction angle varies from 26° to 36°. Errors in determining the velocity and direction of wind depend on its amplitude: under weak winds (<3 m/s), the velocity of wind is overestimated and errors significantly increase in determining its direction; under strong winds (>12 m/s), its velocity is underestimated. The influence of these errors on both spatial and temporal estimates of the characteristics of wind over the Black Sea is briefly considered.     

Central swath mapping by a future satellite-borne fan-beam microwave scatterometer for inferring global ocean wind fields

The SeaSat-A satellite scatterometer is a microwave sensor designed to provide a capability for mapping the global ocean surface wind speed and direction. Four fan beams whose major axes are oriented atpm45andpm135degto the flight vector cover a swath width of 1900 km, but a central region remains that is inadequately mapped. In this paper, two additional fan beams are suggested for a future scatterometer that provide a more complete coverage of the central region.