An improved constraint method in optimal estimation of CO<Subscript>2</Subscript> from GOSAT SWIR observations

We propose an algorithm that combines a pre-processing step applied to the a priori state vector prior to retrievals, with the modified damped Newton method (MDNM), to improve convergence. The initial constraint vector pre-processing step updates the initial state vector prior to the retrievals if the algorithm detects that the initial state vector is far from the true state vector in extreme cases where there are CO₂ emissions. The MDNM uses the Levenberg-Marquardt parameter γ, which ensures a positive Hessian matrix, and a scale factor α, which adjusts the step size to optimize the stability of the convergence. While the algorithm iteratively searches for an optimized solution using observed spectral radiances, MDNM adjusts parameters γ and α to achieve stable convergence. We present simulated retrieval samples to evaluate the performance of our algorithm and comparing it to existing methods. The standard deviation of our retrievals adding random noise was less than 3.8 ppmv. After pre-processing the initial estimate when it was far from the true value, the CO₂ retrieval errors in the boundary layers were within 1.2 ppmv. We tested the MDNM algorithm’s performance using GOSAT L1b data with cloud screening. Our preliminary validations comparing the results to TCCON FTS measurements showed that the average bias was less than 1.8 ppm and the correlation coefficient was approximately 0.88, which was larger than for the GOSAT L2 product.     

XCO_2 satellite retrieval experiments in short-wave infrared spectrum and ground-based validation

Based on the optimal estimation method, a satellite XCO2 retrieval algorithm was constructed by combining LBLRTM with VLIDORT. One-year GOSAT/TANSO observations over four TCCON stations were tested by our algorithm, and retrieval results were compared with GOSAT L2 B products and ground-based FTS measurements. Meanwhile, the influence of CO2 line mixing effect on retrieval was estimated, and the research showed that neglecting CO2 line mixing effect could result in approximately 0.25% XCO2 underestimation. The accuracy of XCO2 retrievals was similar to GOSAT L2 B products at cloud-free footprints with aerosol optical depth less than 0.3, and 1% accuracy of XCO2 retrievals can be reached based on the validation result with TCCON measurements.     

A high-accuracy method for simulating the XCO_2 global distribution using GOSAT retrieval data

A high-accuracy surface modeling(HASM)method based on the fundamental theorem of surfaces,is developed to simulate XCO_2 surfaces using the GOSAT retrieval XCO_2 data.Two tests are designed to investigate the simulation accuracy.The first test divides the existing satellite retrieval XCO2 data into training points and testing points,and simulates the XCO2 surface using the training points while computing the simulation error using the testing points.The absolute mean error(MAE)of the testing points is 1.189 ppmv,and the corresponding values of the comparison methods,Ordinary Kriging,IDW,and Spline are1.203,1.301,and 1.355 ppmv,respectively.The second test simulates the XCO_2 surface using all the satellite retrieval points and uses the TCCON(Total Carbon Column Observing Network)site observation values as the ture values.For the six typical TCCON sites,the HASM simulation MAE is 1.688 ppmv,and the satellite retrieval MAE at the same sites is 2.147 ppmv.These results indicate that HASM can successfully simulate XCO_2 surfaces based on satellite retrieval data.     

A high-accuracy method for simulating the XCO<Subscript>2</Subscript> global distribution using GOSAT retrieval data

A high-accuracy surface modeling (HASM) method based on the fundamental theorem of surfaces, is developed to simulate XCO₂ surfaces using the GOSAT retrieval XCO₂ data. Two tests are designed to investigate the simulation accuracy. The first test divides the existing satellite retrieval XCO₂ data into training points and testing points, and simulates the XCO₂ surface using the training points while computing the simulation error using the testing points. The absolute mean error (MAE) of the testing points is 1.189 ppmv, and the corresponding values of the comparison methods, Ordinary Kriging, IDW, and Spline are 1.203, 1.301, and 1.355 ppmv, respectively. The second test simulates the XCO₂ surface using all the satellite retrieval points and uses the TCCON (Total Carbon Column Observing Network) site observation values as the ture values. For the six typical TCCON sites, the HASM simulation MAE is 1.688 ppmv, and the satellite retrieval MAE at the same sites is 2.147 ppmv. These results indicate that HASM can successfully simulate XCO₂ surfaces based on satellite retrieval data.     

利用全极化微波辐射计资料反演台风境内海面风场

作为一种新兴的被动遥感技术,全极化微波辐射计不仅可以提供海面风速产品,还可以提供海面风向产品.以往利用全极化微波辐射计观测亮温进行海面风场反演仅在晴空条件下进行,本文通过对观测亮温结合台风区域海面风场的分布特征进行分析,验证了全极化微波辐射计具有在台风等恶劣天气条件下进行海面风场观测的能力.基于敏感性分析实验,确定使用6.8 GHz和10.7 GHz等低频通道组合可进行台风区域内海面风场反演.其中,海面风速反演使用基于统计的多元线性回归算法,同时对海面温度、大气水汽含量、云中液态水含量及降水强度等物理量进行反演计算,为海面风向反演做准备.海面风向反演使用物理统计法进行,借鉴散射计风向反演使用的最大似然估计法.通过在全极化辐射传输前向模型中加入降水对大气透过率的影响、设计第三和第四Stokes通道亮温环境影响修正函数,在实现台风区域内海面风向反演的同时减小了反演误差.通过对“云娜”台风境内海面风场进行数值计算,验证了本文反演算法的可行性,并对反演误差的空间分布特征进行了分析.将2004年各台风过程的海面风场反演结果与散射计风场产品进行对比,海面风速和海面风向反演的均方根误差分别为1.64 m·s^-1和18.02°.     

XCO2 satellite retrieval experiments in short-wave and infrared spectra with SCIATRAN model for Sahara Desert

The spectra of O_2 A-band(0.76 μm) and CO_2 near-infrared emissions(1.6 μm) are simulated by the SCIATRAN radiative transfer model(V3.1.23), and compared with those observed by GOSAT-FTS(Greenhouse gases Observing SATellite-Fourier Transform Spectrometer). Systematic deviations between the observed and simulated spectra are found to exist,especially for the O_2 A-band. The discrepancies are characterized by their mean differences averaged over the observed spectral ranges. A correction is applied to the observed GOSAT-FTS L1B(V141.141) spectra by scaling the spectral intensity measured by TANSO-FTS(Thermal and Near infrared Sensor for carbon Observation Fourier Transform Spectrometer) onboard GOSAT.The average columnar CO_2 concentrations(XCO_2) are retrieved from the observed and the corrected GOSAT-FTS spectra by using the SCIATRAN inversion algorithm. Compared with the GOSAT-FTS L2 XCO_2 data products retrieved from the observed spectra of GOSAT-FTS, the SCIATRAN retrievals from the corrected spectra show a much better agreement, with the relative error less than 1%. But the results of GOSAT TANSO-FTS(V161.160) show smaller residuals than GOSAT TANSO-FTS(V141.141) without mean residual correction. The results indicate that the mean residual correction would increase the precision of XCO_2 retrieval for spectra with systematic deviations.     

基于初至波走时层析成像的Tikhonov正则化与梯度优化算法

初至波走时层析成像是利用地震初至波走时和其传播的射线路径来反演地下介质速度的技术.该问题本质上是一个不适定问题,需要使用正则化方法并辅之以适当的最优化技巧.本文从数值优化的角度介绍了初至波走时层析成像的反演原理,建立了Tikhonov正则化层析成像反演模型并提出求解极小化问题的加权修正步长的梯度下降算法.该方法可以从速度模型的可行域中迭代找到一个最优解.数值试验表明,该方法是可行和有应用前景的.     

A comparison of atmospheric CO2 concentration GOSAT-based observations and model simulations

Satellite observations of atmospheric CO2 are able to truly capture the variation of global and regional CO2 concentration.The model simulations based on atmospheric transport models can also assess variations of atmospheric CO2 concentrations in a continuous space and time,which is one of approaches for qualitatively and quantitatively studying the atmospheric transport mechanism and spatio-temporal variation of atmospheric CO2 in a global scale.Satellite observations and model simulations of CO2 offer us two different approaches to understand the atmospheric CO2.However,the difference between them has not been comprehensively compared and assessed for revealing the global and regional features of atmospheric CO2.In this study,we compared and assessed the spatio-temporal variation of atmospheric CO2 using two datasets of the column-averaged dry air mole fractions of atmospheric CO2(XCO2)in a year from June 2009 to May 2010,respectively from GOSAT retrievals(V02.xx)and from Goddard Earth Observing System-Chemistry(GEOS-Chem),which is a global 3-D chemistry transport model.In addition to the global comparison,we further compared and analyzed the difference of CO2 between the China land region and the United States(US)land region from two datasets,and demonstrated the reasonability and uncertainty of satellite observations and model simulations.The results show that the XCO2 retrieved from GOSAT is globally lower than GEOS-Chem model simulation by 2 ppm on average,which is close to the validation conclusion for GOSAT by ground measures.This difference of XCO2 between the two datasets,however,changes with the different regions.In China land region,the difference is large,from 0.6 to 5.6 ppm,whereas it is 1.6 to 3.7 ppm in the global land region and 1.4 to 2.7 ppm in the US land region.The goodness of fit test between the two datasets is 0.81 in the US land region,which is higher than that in the global land region(0.67)and China land region(0.68).The analysis results further indicate that the inconsistency of CO2concentration between satellite observations and model simulations in China is larger than that in the US and the globe.This inconsistency is related to the GOSAT retrieval error of CO2 caused by the interference among input parameters of satellite retrieval algorithm,and the uncertainty of driving parameters in GEOS-Chem model.     

Comparison of SO<Subscript>2</Subscript> column retrievals from BRD and DOAS algorithms

Atmospheric SO₂ has a significant impact on the urban environment and global climate. Band Residual Difference Algorithm (BRD) and Differential Optical Absorption Spectroscopy (DOAS) were used respectively by NASA and ESA science team to derive SO₂ columns from satellite observations, but there are few studies on the comparison and validation of BRD and DOAS SO₂ retrievals under the same observation conditions. In this study, the radiative transfer model SCIATRAN was firstly used to validate the accuracies of BRD and DOAS SO₂ retrievals, and analyse the uncertainty of SO₂ retrieval caused by band selection, O3 absorption, aerosol, surface reflectance, solar and viewing zenith angle. Finally, BRD and DOAS algorithms were applied to the same radiances from satellite observations, and comparisons of BRD and DOAS SO₂ retrievals were conducted over volcanic eruption and North China. Results show that, for the case with low SO₂ columns, BRD SO₂ retrievals have higher retrieval accuracy than DOAS, but typical seasonal variation with high SO₂ column in winter and low in summer can be more clearly discernible in DOAS SO₂ retrievals than BRD from satellite observations. For the case with high SO₂ columns, the differences between BRD (310.8–314.4 nm) and DOAS (315–327 nm) retrievals are large, and the value and accuracy of BRD (310.8–314.4 nm) SO₂ retrievals are lower than those of DOAS (315–327 nm) retrievals. Compared with the SO₂ inputs in forward model, both BRD (310.8–314.4 nm) and DOAS (315–327 nm) SO₂ retrievals are underestimated for the case with high SO₂ columns. The selection of wavelength range can significantly affect the accuracy of SO₂ retrieval. The error of BRD SO₂ retrieval from 310.8–314.4 nm is lower than other bands in the ultraviolet spectral region (306–327 nm). The increase of wavelength in the ultraviolet spectral region 306–330 nm can reduce the underestimation of DOAS SO₂ retrievals in the case of high SO₂ column, but slight overestimation of SO₂ retrieval is found from the 315–327 nm range in the case of low SO₂ column. The values of BRD and DOAS SO₂ retrieval decrease with atmospheric O₃ column and aerosol optical depth increasing, but increase with surface reflectance increasing. Large solar zenith angle and viewing zenith angle can introduce more errors to the BRD and DOAS SO₂ retrievals. This study is important for the improvement of retrieval algorithm and the application of SO₂ products from satellite observations.     

Retrieval of the physical properties of an anelastic solid half space from seismic data

In recent years, due to the rapid development of computation hard- and software, time domain full-wave inversion, which makes use of all the information in the seismograms without appealing to linearization, has become a plausible candidate for the retrieval of the physical parameters of the earth's substratum. Retrieving a large number of parameters (the usual case in a layered substratum comprising various materials, some of which are porous) at one time is a formidable task, so full-wave inversion often seeks to retrieve only a subset of these unknowns, with the remaining parameters, the priors, considered to be known and constant, or sequentially updated, during the inversion. A known prior means that its value has been obtained by other means (e.g., in situ or laboratory measurement) or simply guessed (hopefully, with a reasonable degree of confidence). The uncertainty of the value of the priors, like that of data noise, and the inadequacy of the theoretical/numerical model employed to mimick the seismic data during the inversion, is a source of retrieval error. We show, on the example of a homogeneous, isotropic, anelastic half-plane substratum configuration, characterized by five parameters: density, P and S wavespeeds and P and S quality factors, when a perfectly-adequate theoretical/numerical model is employed during the inversion and the data is free of noise, that the retrieval error can be very large for a given parameter, even when the prior uncertainty of another single parameter is very small. Furthermore, the employment of other load and response polarization data and/or multi-offset data, as well as other choices of the to-be-retrieved parameters, are shown, on specific examples, not to systematically improve(they may even reduce) the accuracy of the retrievals when the prior uncertainty is relatively-large. These findings, relative to the recovery, via an exact retrieval model processing noiseless data obtained in one of the simplest geophysical configurations, of a single parameter at a time with a single uncertain prior, raises the question of the confidence that can be placed in geophysical parameter retrievals: 1) when more than one parameters are retrieved at a time, and/or 2) when more than one prior are affected by uncertainties during a given inversion, and/or 3) when the model employed to mimick the data during the inversion is inadequate, 4) when the data is affected by noise or measurement errors, and 5) when the parameter retrieval is carried out in more realistic configurations.     

MULTI-OFFSET ACOUSTIC INVERSION OF A LATERALLY INVARIANT MEDIUM: APPLICATION TO REAL DATA1

The aim of seismic inversion methods is to obtain quantitative information on the subsurface properties from seismic measurements. However, the potential accuracy of such methods depends strongly on the physical correctness of the mathematical equations used to model the propagation of the seismic waves. In general, the most accurate models involve the full non-linear acoustic or elastic wave equations. Inversion algorithms based on these equations are very CPU intensive. The application of such an algorithm on a real marine CMP gather is demonstrated. The earth model is assumed to be laterally invariant and only acoustic wave phenomena are modelled. A complete acoustic earth model (P-wave velocity and reflectivity as functions of vertical traveltime) is estimated. The inversion algorithm assumes that the seismic waves propagate in 2D. Therefore, an exact method for transforming the real data from 3D to 2D is derived and applied to the data. The time function of the source is estimated from a vertical far-field signature and its applicability is demonstrated by comparing synthetic and real water-bottom reflections. The source scaling factor is chosen such that the false reflection coefficient due to the first water-bottom multiple disappears from the inversion result. In order to speed up the convergence of the algorithm, the following inversion strategy is adopted: an initial smooth velocity model (macromodel) is obtained by applying Dix's equation to the result of a classical velocity analysis, followed by a smoothing operation. The initial reflectivity model is then computed using Gardner's empirical relationship between densities and velocities. In a first inversion step, reflectivity is estimated from small-offset data, keeping the velocity model fixed. In a second step, the initial smooth velocity model, and possibly the reflectivity model, is refined by using larger-offset data. This strategy is very efficient. In the first step, only ten iterations with a quasi-Newton algorithm are necessary in order to obtain an excellent convergence. The data window was 0–2.8 s, the maximum offset was 250 m, and the residual energy after the first inversion step was only 5% of the energy of the observed data. When the earth model estimated in the first inversion step is used to model data at moderate offsets (900 m, time window 0.0–1.1 s), the data fit is very good. In the second step, only a small improvement in the data fit could be obtained, and the convergence was slow. This is probably due to the strong non-linearity of the inversion problem with respect to the velocity model. Nevertheless, the final residual energy for the moderate offsets was only 11%. The estimated model was compared to sonic and density logs obtained from a nearby well. The comparison indicated that the present algorithm can be used to estimate normal incidence reflectivity from real data with good accuracy, provided that absorption phenomena play a minor role in the depth interval considered. If details in the velocity model are required, large offsets and an elastic inversion algorithm should be used.     

时间域航空电磁2.5维非线性共轭梯度反演

对于时间域航空电磁法二维和三维反演来说,最大的困难在于有效的算法和大的计算量需求.本文利用非线性共轭梯度法实现了时间域航空电磁法2.5维反演方法,着重解决了迭代反演过程中灵敏度矩阵计算、最佳迭代步长计算、初始模型选取等问题.在正演计算中,我们采用有限元法求解拉式傅氏域中的电磁场偏微分方程,再通过逆拉氏和逆傅氏变换高精度数值算法得到时间域电磁响应.在灵敏度矩阵计算中,采用了基于拉式傅氏双变换的伴随方程法,时间消耗只需计算两次正演,从而节约了大量计算时间.对于最佳步长计算,二次插值向后追踪法能够保证反演迭代的稳定性.设计两个理论模型,检验反演算法的有效性,并讨论了选择不同初始模型对反演结果的影响.模型算例表明:非线性共轭梯度方法应用于时间域航空电磁2.5维反演中稳定可靠,反演结果能够有效地反映地下真实电性结构.当选择的初始模型电阻率值与真实背景电阻率值接近时,能得到较好的反演结果,当初始模型电阻率远大于或远小于真实背景电阻率值时反演效果就会变差.     

Evaluating the potential of a blended passive microwave‐interactive multi‐sensor product for improved mapping of snow cover and estimations of snow water equivalent

The retrieval of Snow Water Equivalent (SWE) from remote sensing satellites continues to be a very challenging problem. In this paper, we evaluate the accuracy of a new SWE product derived from the blending of a passive microwave SWE product based on the Advanced Microwave Sounding Unit (AMSU) with a multi‐sensor snow cover extent product based on the Interactive Multi‐sensor Snow and Ice Mapping System (IMS). The microwave measurements have the ability to penetrate the snow pack, and thus, the retrieval of SWE is best accomplished using the AMSU. On the other hand, the IMS maps snow cover more reliably due to the use of multiple satellite and ground observations. The evolution of global snow cover from the blended, the AMSU and the IMS products was examined during the 2006 snow season. Despite the overall good inter‐product agreement, it was shown that the retrievals of snow cover extent in the blended product are improved when using IMS, with implications for improved microwave retrievals of SWE. In a separate investigation, the skill of the microwave SWE product was also examined for its ability to correctly estimate SWE globally and regionally. Qualitative evaluation of global SWE retrievals suggested dependence on land surface temperature: the lower the temperature, the higher the SWE retrieved. This temperature bias was attributed in part to temperature effects on those snow properties that impact microwave response. Therefore, algorithm modifications are needed with more dynamical adjustments to account for changing snow cover. Quantitative evaluation over Slovakia in central Europe, for a limited period in 2006, showed reasonably good performance for SWE less than 100 mm. Sensitivity to deeper snow decreased significantly. Copyright © 2007 John Wiley & Sons, Ltd.     

基于遗传算法优化神经网络权值的大坝结构损伤识别

针对传统 BP 神经网络存在着容易陷入局部极小点、训练时间太长等缺点,本文采用基于浮点编码的遗传算法,对 BP 神经网络的初值空间进行了遗传优化。用基于浮点编码的遗传算法来优化 BP 神经网络的权值,可得到最佳初始权值矩阵,并按误差前向反馈算法,沿负梯度搜索进行网络学习。文中以混凝土重力坝结构作为算例,用结构的模态频率变化作为网络的输入向量,结构的损伤位置作为输出向量,对网络进行了训练。仿真结果表明：遗传 BP 神经网络的收敛和诊断能力优于传统 BP 神经网络,可有效地运用到大坝结构的健康诊断与损伤识别中。     

基于非局部全核变分方法的稀疏角多能CT重建

稀疏角采样与减小X射线源电流可有效降低多能谱CT低辐射剂量,然而会导致投影数据不足且包含较大噪声,重建图像会严重降质。针对这一问题,本文对传统全核变分（TNV）正则化方法进行推广,利用非局部梯度向量构成的雅克比矩阵的低秩特性,提出非局部全核变分（NLTNV）正则化方法。该方法用单个正则项同时建模能谱CT图像的结构相似性、梯度域稀疏性与非局部自相似性3种先验信息,能恢复稀疏角度投影含较大噪声（剂量较低）时图像的结构特征,并且有效缓解了用多正则项建模多能谱CT图像不同先验信息所导致的正则化参数过多问题。此外,基于NLTNV的重建模型为凸优化模型,保证了算法的稳定性与收敛性。实验结果表明,与TNV正则化方法相比,本方法显著提升重建图像的整体质量。      

一种新的地球物理反演方法——模拟原子跃迁反演法

详细研究了一般地球物理反问题的迭代优化求解过程与物理学中原子跃迁过程的对应关系,建立了反演问题中模型空间、初始模型、局部极值模型、最优化模型等与原子的态空间、定态、激发态、基态等的对应关系. 在此基础上,模拟了物理学中原子从激发态向基态跃迁的物理过程,建立了一种与原子跃迁过程相对应的非线性随机跃迁数学模型和模型解跃迁搜索准则,导出了适用于一般地球物理资料的模拟原子跃迁的非线性反演算法. 用理论测试函数对这种新的反演方法进行了数值试验,结果表明该方法具有解不依赖于初始模型、收敛速度快等优点.     

利用ENVISAT/AATSR资料反演黄土高原陇东地区地表温度

由于ENVISAT/AATSR资料不同角度热辐射亮度值之间存在较高的相关性从而导致较大误差的产生,本文尝试避开这种误差源,只选取天底观测数据对黄土高原陇东地区整层大气水汽含量及地表温度进行反演.与MODIS整层大气水汽含量产品对比验证表明,本文结果与MODIS产品有一定差异,但是可以直接用于大气透过率的估算.结合野外观测数据对地表温度反演结果的检验表明,最大绝对误差为4.0 ℃,平均相对误差为5.0％,因此,该算法在黄土高原陇东地区应用比较成功.     

大地电磁数据的Occam反演改进

Occam 反演以其稳定的收敛性,在大地电磁（MT）数据解释中有广泛的应用。但是其每次迭代均用一维搜索方法求拉格朗日因子μ值,需要许多次正演计算,速度非常慢。在讨论了 Occam 反演中数据拟合差随μ变化的基础上,本文采用了μ值在一定步长下逐次递减的求取方法,每次迭代只需一次正演,极大地提高了计算速度。另外,反演求得光滑模型而非最光滑模型,分辨率更高。理论及实际数据的反演试算均表明,和 Occam 反演相比,反演依然稳定,但速度更快,结果也更真实。     

EFFECTS OF FOREST ON THE SNOW PARAMETERS DERIVED FROM MICROWAVE MEASUREMENTS DURING THE BOREAS WINTER FIELD CAMPAIGN

Passive microwave data have been used to infer the areal snow water equivalent (SWE) with some success. However, the accuracy of these retrieved SWE values have not been well determined for heterogeneous vegetated regions. The Boreal Ecosystem–Atmosphere Study (BOREAS) Winter Field Campaign (WFC), which took place in February 1994, provided the opportunity to study in detail the effects of boreal forests on snow parameter retrievals. Preliminary results reconfirmed the relationship between microwave brightness temperature and snow water equivalent. The pronounced effect of forest cover on SWE retrieval was studied. A modified vegetation mixing algorithm is proposed to account for the forest cover. The relationship between the microwave signature and observed snowpack parameters matches results from this model.     

An efficient non-linear least-squares 1D inversion scheme for resistivity and IP sounding data

Non-linear least-squares inversion operates iteratively by updating the model parameters in each step by a correction vector which is the solution of a set of normal equations. Inversion of geoelectrical data is an ill-posed problem. This and the ensuing suboptimality restrict the initial model to being in the near vicinity of the true model. The problem may be reduced by introducing damping into the system of equations. It is shown that an appropriate choice of the damping parameter obtained adaptively and the use of a conjugate-gradient algorithm to solve the normal equations make the 1D inversion scheme efficient and robust. The scheme uses an optimal damping parameter that is dependent on the noise in the data, in each iterative step. The changes in the damping and relative residual error with iteration number are illustrated. A comparison of its efficacy over the conventional Marquardt and simulated annealing methods, tested on Inman's model, is made. Inversion of induced polarization (IP) sounding is obtained by inverting twice (true and modified) DC apparent resistivity data. The inversion of IP data presented here is generic and can be applied to any of the IP observables, such as chargeability, frequency effect, phase, etc., as long as these observables are explicitly related to the DC apparent resistivity. The scheme is used successfully in inverting noise-free and noisy synthetic data and field data taken from the published literature.