基于BP神经网络的盐渍土盐分遥感反演模型研究

采用遥感技术和BP神经网络技术,结合野外实测的盐渍土光谱特征和实验室化验的土壤含盐数据,对盐渍土盐分的遥感反演进行了模型的设计与编程实现.BP神经网络模型的预测精度在62.5%,明显高于传统统计模型的预测精度,表明BP神经网络能较好地模拟土壤含盐量与光谱数据之间的关系,可用于建立土壤盐分遥感反演模型.     

基于主成分分析法的戈壁地表砾石粒径遥感估测模型研究

戈壁地表砾石粒径与遥感多光谱数据、植被指数及地学因子存在相关关系,但这些因子间可能存在着多重相关性,如利用这些因子直接建模估测戈壁地表砾石粒径,则可能出现病态模型。利用主成分分析法筛选因子,既可保留多个相关因子的主要信息,又可避免因子间共线性的问题,达到降维、简化模型的效果。因此,本文以新疆哈密市境内山前洪积扇戈壁地表砾石为研究对象,以2010年Landsat TM遥感影像及30 m分辨率DEM为基本数据源,采用主成分分析法,从选择的43个遥感及地学因子（主要包括影像各波段信息、DEM、NDVI、 GEMI,影像经K-T变换得到SBI、GVI、WVI三个分量,通过纹理分析得到的各个波段的均值、方差、信息熵、相关性及对比度等纹理因子,以及利用DEM提取的粗糙度等）中,筛选提取其主成分。结果表明,第一主成分至第五主成分的累计贡献率达98.0%,以前5个主成分作为自变量,借助SPSS软件中的多元回归分析功能,建立戈壁地表砾石粒径估测的回归模型,模型经方差分析及相关性检验,达到显著相关水平。基于建立的估测模型,进行了戈壁地表砾石粒径估测,经验证,实测值与估算值紧密相关。研究可帮助我们了解戈壁的特征,为戈壁区改造利用,认识沙粒迁移、沙漠扩展提供技术支持。     

基于TM数据的太湖叶绿素A浓度定量反演

利用TM(ETM)数据与准实时地面采样数据,建立太湖叶绿素浓度反演模型。结果表明,TM3/(TM1+TM4)与叶绿素A浓度的相关性最好,并以此建立了太湖叶绿素A浓度线性反演模型,但反演精度并不高,因此,建立了一个两层BP神经网络模型反演太湖的叶绿素A浓度,结果表明,神经网络模型的反演精度远高于线性反演模型,16个测试样本表明,神经网络模型反演的相对误差小于30%的有15个点,占总测试样本93.75%,而线性反演模型反演相对误差在30%以下的仅有3个点,这表明对于太湖这样一个光谱特征复杂的二类水体,可以利用神经网络模型反演水质参数。     

基于纹理特征的香格里拉县冷杉林郁闭度遥感估测研究

以香格里拉县2006年TM影像、2006年森林资源二类调查小班数据为信息源,结合研究区冷杉林地面实测标准地（30m×30m）数据,提取香格里拉县冷杉林TM影像分布信息及标准地纹理特征因子（共48个）,对各因子进行相关分析;利用主成分法对纹理特征因子进行因子分析,最终选出13个纹理特征因子利用偏最小二乘法进行回归建模并进行模型精度检验。根据回归估测模型以及提取出的冷杉林各波段纹理特征因子,进行研究区冷杉林郁闭度反演。结果表明,基于遥感影像纹理特征建立的郁闭度遥感估测模型,其RE=13.8%,RMSE=10.39,精度为83.3%。研究区冷杉林郁闭度反演可知冷杉林郁闭度多分布在0.6~0.7范围内,多为中度郁闭林地。     

基于高光谱数据的天山北坡积雪孔隙率反演研究

以新疆天山北坡中段典型流域季节性积雪为研究对象,基于高光谱遥感监测技术,分析了融雪期积雪孔隙率与光谱反射率的相关性。采用偏最小二乘法(PLS)对相关性较高的波段进行压缩,并提取贡献率最高的前四个主成分,以此用来确定神经网络的隐含节点数、输入层、输出层的初始权值,建立PLS-BP模型进行积雪孔隙率反演研究。结果表明:当隐含节点数为3,模型的线性确定相关系数(R²)较高为0.9159,RMSE为0.04,相对误差为0.23。与传统偏最小二乘回归(PLSR)、主成分回归(PCA)建模方法相比,精度较高,所建定量模型可用于高光谱遥感反演积雪孔隙率。     

长江口水域多光谱遥感水深反演模型研究

利用Landsat-7 ETM 遥感影像反射率和实测水深值之间的相关性可以探测水深。该文介绍单波段、双波段比值和多波段3种线性回归模型以及动量BP人工神经网络水深反演模型。选择长江口北港河道上段作为研究区,利用上述模型,分两种情况进行水深反演:一是以河道全部历史样本建模;二是将河道按自然水深划分为浅水区和深水区分别建模。结果表明:神经网络模型预测精度高于线性回归模型;水深分区后线性回归和神经网络模型预测误差均有所减小。     

基于半分析模型的渤海湾近岸海域悬浮泥沙浓度遥感反演

悬浮泥沙浓度是海洋质量和环境评价的重要参数之一,遥感技术可以实现对海洋水色要素基础信息实时、连续、大范围监测,是当前大尺度海洋生态环境监测的重要手段。然而,现今并没有一个通用的模型可以实现对所有复杂水体悬浮泥沙浓度的反演。该文基于野外站点观测、结合研究区水体光谱特性,构建了基于多波段准分析算法(QAA)的渤海湾近岸海域悬浮泥沙浓度遥感反演半分析模型,并利用Landsat TM遥感数据,对研究区悬浮泥沙浓度进行了反演,制作了悬浮泥沙浓度空间分布图。研究结果显示,该文构建的半分析模型与前人研究的经验统计模型相比,具有更高的精度和普适性,相对误差达到18.28%;研究区的悬浮泥沙浓度整体较低,平均泥沙浓度为27mg/L,且呈现近海大于远海、南部海域大于北部海域的规律。     

基于ETM遥感影像的二滩库区水深反演研究

在常规测量研究的基础上,通过分析ETM遥感影像不同波段对水体、水深的敏感性,选取二滩水电站库区上段37断面以上河段,基于GIS平台建立了水深和相应的ETM适宜波段间的比值法水深遥感反演模型,从而快速获取大范围水域水深信息,整体精度达到80%以上.遥感反演水深作为评价浅水区泥沙淤积及其发展趋势的经济可行的方法之一,为库区泥沙淤积研究提供了新的思路.     

基于环境一号卫星的内陆水体水质多光谱遥感监测

基于环境一号卫星多光谱数据,针对我国内陆水体开展遥感监测研究。该研究建立分地区分季节经验模型反演叶绿素a浓度,基于生物光学模型的近红外单波段方法反演悬浮物浓度,以悬浮物浓度间接反演透明度,最终基于叶绿素a和悬浮物获得基于遥感监测的内陆水体营养状态指数。以2009年6月巢湖同步观测试验验证本文的水质参数反演算法,表明叶绿素a浓度反演精度不如悬浮物浓度,两者反演精度符合水环境监测业务需求。     

基于高光谱信息的芦苇和香蒲地上干生物量反演方法研究

植物生物量是反映湿地生态系统功能与健康状况的重要指标,高光谱技术可以为定量监测挺水植物生物量提供快速、无损的数据采集与分析处理方法。利用Field Spec 3野外高光谱辐射仪,获取了北京市门城湖湿地公园内典型挺水植物芦苇(Phragmites australis)和香蒲(Typha angustifolia)的冠层高光谱数据,同步采集了对应区域两种挺水植物的样品,测定了其地上干生物量;分别采用原始光谱、微分光谱、植被指数、归一化和比值光谱指数、"三边"参数、吸收特征参数6种方法,分析了芦苇和香蒲的冠层高光谱数据与其地上干生物量关系的差异,选择与地上干生物量显著相关的高光谱特征参数,利用一元线性回归、逐步多元回归和偏最小二乘回归方法,建立反演模型;利用交叉检验中的3K-CV(Three K-fold Cross Validation)方法,对反演模型的估测精度进行了检验。结果表明,在通过6种光谱分析方法获得的高光谱特征参数中,与芦苇地上干生物量相关系数最大的高光谱特征参数为550~800 nm波段的吸收深度,相关系数为0.847;与香蒲地上干生物量相关系数最大的高光谱特征参数是由680 nm、735 nm两波段构建的归一化光谱指数,相关系数为0.953。从3种回归模型的反演结果来看,偏最小二乘回归模型的反演结果最好,其次是逐步多元回归模型,而一元线性回归模型的反演结果相对较差。3K-CV法的交叉检验结果显示,反演芦苇和香蒲地上干生物量的3种模型都取得了较理想的估测精度,最小估测精度为90.86%,最大为95.23%。利用挺水植物冠层高光谱信息,可以反演其地上干生物量。     

Feasibility study on the binary-parameter retrieval model of ocean suspended sediment concentration based on MODIS data

This paper brought out a new idea on the retrieval of suspended sediment concentration, which uses both the water-leaving radiance from remote sensing data and the grain size of the suspended sediment. A principal component model and a neural network model based on those two parameters were constructed. The analyzing results indicate that testing errors of the models using the two parameters are 0.256 and 0.244, while the errors using only water-leaving radiance are 0.384 and 0.390. The stability of the models with grain size parameter is also better than the one without grain size. This research proved that it is necessary to introduce the grain size parameter into suspended sediment concentration retrieval models in order to improve the retrieval precision of these models.     

渤海海洋悬浮泥沙二元特征参数MODIS遥感反演模型研究

This paper brought out a new idea on the retrieval of suspended sediment concentration, which uses both the water-leaving radiance from remote sensing data and the grain size of the suspended sediment. A principal component model and a neural network model based on those two parameters were constructed. The analyzing results indicate that testing errors of the models using the two parameters are 0.256 and 0.244, while the errors using only water-leaving radiance are 0,384 and 0.390. The stability of the models with grain size parameter is also better than the one without grain size. This research proved that it is necessary to introduce the grain size parameter into suspended sediment concentration retrieval models in order to improve the retrieval precision of these models.     

黄河入海悬沙季节输送变化与输送机制的数值研究(英文)

Based on sediment and discharge flux data for the Yellow River, realistic forcing fields and bathymetry of the Bohai Sea, a suspended sediment transport module is driven by a wave-current coupled model to research seasonal variations and mechanisms of suspended load transport to the Bohai Sea. It could be concluded that surface sediment concentration indicates a distinct spatial distribution characteristic that varies seasonally in the Bohai Sea. Sediment concentration is rather high near the Yellow River estuary, seasonal variations of which are controlled by quantity of sediment from the Yellow River, suspended sediment concentration reaches its maximum during summer and fall. Furthermore, sediment concentration decreases rapidly in other seas far from the Yellow River estuary and maintains a very low level in the center of the Bohai Sea, and is dominated by seasonal variations of climatology wind field in the Bohai Sea. Only a small amount of sediments imported from the Yellow River are delivered northwestward to the southern coast of the Bohai Bay. Majority of sediments are transported southeastward to the Laizhou Bay, where sediments are continuously delivered into the center of the Bohai Sea in a northeastward direction, and part of them are transported eastward alongshore through the Bohai Strait. 69% of sediments from the Yellow River are deposited near the river delta, 31% conveyed seaward, within which, 4% exported to the northern Yellow Sea through the Bohai Strait. Wind wave is the most essential contributor to seasonal variations of sediment concentration in the Bohai Sea, and the contribution of tidal currents is also significant in shallow waters when wind speed is low.     

Numerical simulation on seasonal transport variations and mechanisms of suspended sediment discharged from the Yellow River to the Bohai Sea

Based on sediment and discharge flux data for the Yellow River, realistic forcing fields and bathymetry of the Bohai Sea, a suspended sediment transport module is driven by a wave-current coupled model to research seasonal variations and mechanisms of suspended load transport to the Bohai Sea. It could be concluded that surface sediment concentration indicates a distinct spatial distribution characteristic that varies seasonally in the Bohai Sea. Sediment concentration is rather high near the Yellow River estuary, seasonal variations of which are controlled by quantity of sediment from the Yellow River, suspended sediment concentration reaches its maximum during summer and fall. Furthermore, sediment concentration decreases rapidly in other seas far from the Yellow River estuary and maintains a very low level in the center of the Bohai Sea, and is dominated by seasonal variations of climatology wind field in the Bohai Sea. Only a small amount of sediments imported from the Yellow River are delivered northwestward to the southern coast of the Bohai Bay. Majority of sediments are transported southeastward to the Laizhou Bay, where sediments are continuously delivered into the center of the Bohai Sea in a northeastward direction, and part of them are transported eastward alongshore through the Bohai Strait. 69% of sediments from the Yellow River are deposited near the river delta, 31% conveyed seaward, within which, 4% exported to the northern Yellow Sea through the Bohai Strait. Wind wave is the most essential contributor to seasonal variations of sediment concentration in the Bohai Sea, and the contribution of tidal currents is also significant in shallow waters when wind speed is low.     

南黄海辐射沙洲邻近海域表层悬浮颗粒物浓度遥感反演

利用2003年春、秋季南黄海辐射沙洲邻近海域水体表观光谱测量与同步水体取样数据,分析该海域不同季节表层水体遥感反射率光谱响应特性。结合常用水色传感器波段光谱响应模拟,建立春、秋季表层水体总悬浮颗粒物浓度(TSM)与悬浮泥沙浓度(SSC)的最适单波段、双波段(波段比值)和3波段统计反演模式。结果表明,不同季节水体上述3类算法的最优波段均不相同,欲寻求适合所有季节水体的统一反演算法,仅能选择次优波段。通过比较,3种不同反演模式中三波段模式更适于反演近岸水体悬浮颗粒物浓度。     

Diagnostic experiments for transport mechanisms of suspended sediment discharged from the Yellow River in the Bohai Sea

Five diagnostic experiments with a 3D baroclinic hydrodynamic and sediment transport model ECOMSED in couple with the third generation wave model SWAN and the Grant–Madsen bottom boundary layer model driven by the monthly sediment load of the Yellow River, were conducted to separately diagnose effects of different hydrodynamic factors on transport of suspended sediment discharged from the Yellow River in the Bohai Sea. Both transport and spatio-temporal distribution of suspended sediment concentration in the Bohai Sea were numerially simulated. It could be concluded that suspended sediment discharged from the Yellow River cannot be delivered in long distance under the condition of tidal current. Almost all of sediments from the Yellow River are deposited outside the delta under the condition of wind-driven current, and only very small of them are transported faraway. On the basis of wind forcing, sediments from the Yellow River are mainly transported north-northwestward, and others which are first delivered to the Laizhou Bay are continuously moved northward. An obvious 3D structure characteristic of sediment transport is produced in the wind-driven and tide-induced residual circulation condition. Transport patterns at all layers are generally consistent with circulation structure, but there is apparent deviation between the depth-averaged sediment flux and the circulation structure. The phase of temporal variation of sediment concentration is consistent with that of the bottom shear stress, both of which are proved to have a ten-day cycle in wave and current condition.     

黄河入海泥沙悬移输送机制的控制试验(英文)

Five diagnostic experiments with a 3D baroclinic hydrodynamic and sediment transport model ECOMSED in couple with the third generation wave model SWAN and the Grant-Madsen bottom boundary layer model driven by the monthly sediment load of the Yellow River, were conducted to separately diagnose effects of different hydrodynamic factors on transport of suspended sediment discharged from the Yellow River in the Bohai Sea. Both transport and spatio-temporal distribution of suspended sediment concentration in the Bohai Sea were numerially simulated. It could be concluded that suspended sediment discharged from the Yellow River cannot be delivered in long distance under the condition of tidal current. Almost all of sediments from the Yellow River are deposited outside the delta under the condition of wind-driven current, and only very small of them are transported faraway. On the basis of wind forcing, sediments from the Yellow River are mainly transported north-northwestward, and others which are first delivered to the Laizhou Bay are continuously moved northward. An obvious 3D structure characteristic of sediment transport is produced in the wind-driven and tide-induced residual circulation condition. Transport patterns at all layers are generally consistent with circulation structure, but there is apparent deviation between the depth-averaged sediment flux and the circulation structure. The phase of temporal variation of sediment concentration is consistent with that of the bottom shear stress, both of which are proved to have a ten-day cycle in wave and current condition.     

黄河入海泥沙输运及沉积过程的数值模拟

以利津站代表的黄河入海径流和泥沙数据驱动ECOMSED模型,对黄河入海泥沙悬移输运过程的逐月时空变化、输送通量以及海底沉积效应进行了数值模拟实验。分析结果表明,在忽略再悬浮作用条件下,黄河入海泥沙的输运扩散过程具有明显的季节变化规律,且这种变化具有年际相似性。黄河泥沙入渤海后总体朝向辽东湾西侧海岸扩散,而主要沉降区域是黄河口附近,且随着距离的增大,沉积通量迅速降低。模拟沉积速率一般在0.5~0.1 mm/年左右,与实际调查结果非常接近。海底地形等高线向渤海海盆西部、渤海湾南部,以及渤海海峡方向突出,也反映了泥沙通量的输送方向。从黄河入海泥沙悬移扩散过程的季节变化特征及其海底沉积效应来看,渤海海域泥沙悬移输运过程受潮汐动力、余流和和底层流场等因子的制约。除了黄河河口地区以外,各月悬浮泥沙高浓度区基本一致,集中分布在潮流能量最强的海域,潮流水平动能的大小与悬沙浓度大小分布基本一致。泥沙悬移输运方向与模拟获得的渤海三维风驱-潮致Lagrange余流的方向具有明显的相关关系,泥沙扩散的方向和强度明显受余流方向和强度的控制。     

年代际尺度海岸演化机制与动力-统计建模方法

发展中长时间尺度的海岸演化预测方法与机理模型,是目前国际海岸演化研究领域关注的焦点和最具挑战性课题。本文系统性地回顾了近半个世纪以来不同时间尺度海岸演化模式的研究发展历程和成果,从动力机制和时间尺度两个方面对海岸演化机理模型存在的主要建模瓶颈问题进行了系统总结。通过对渤海湾西岸近一个半世纪以来海岸演化机制的初步分析,提出了以泥沙收支、环流输送、波浪掀沙以及波气候变化等要素作为年代际尺度海岸演化主要驱动因子,基于驱动力概化与统计升尺度有机衔接的年代际尺度海岸演化机理模型的动力—统计建模方法框架新思路,为推动海岸演化动力学理论体系和中长时间尺度海岸演化预测方法创新,提供了重要科学依据和借鉴。     

A model for the retrieval of suspended sediment concentrations in Taihu Lake from TM images

A study was conducted in the Taihu Lake with the aim of deriving a model for the retrieval of suspended sediment (SS) concentrations from Landsat TM images and in situ sampled data. The correlation between suspended sediment concentrations of lake and the reflectance obtained from the TM images is significant. By TM images and in situ sampled data in summer and winter, we obtained a comparative uniform model for the retrieval of suspended sediment concentrations in the Taihu Lake, that is lnSS = a*(R3/R1) + b, where lnSS is the natural logarithm of the suspended sediment concentration, R1 and R3 are the reflectance coincident with the 1st band and the 3rd band in TM images, a and b are the regression coefficients. Furthermore, we analysed the errors particularly to make sure the model is valid. The model is accurate to within 0.33(RMSE), suggesting that this model may be applicable to predict suspended sediment in the Taihu Lake from TM image throughout the year.