遗传算法优化的BP神经网络城市不透水层百分比估算

本研究利用基于遗传算法优化的BP神经网络算法估算城市不透水层百分比。首先,将像元中各端元组分与BP神经网络的节点相对应进行BP网络建模,遗传算子建模;其次,对样本进行网络训练,先通过GA算法得到全局近最优网络权重集,然后用梯度下降算法训练网络,直到找到能充分反映特征空间中的数据分布模式的局部最优网络权重集;然后,训练好的网络被应用于整个影像用来估算城市不透水层覆盖百分比。在此基础上,对北京市地表不透水层百分比进行估算。试验结果表明,本研究所用的方法能有效利用中高分辨率遥感影像数据估算城市不透水层百分比。     

Improving impervious surface estimation: an integrated method of classification and regression trees (CART) and linear spectral mixture analysis (LSMA) based on error analysis

Classification and regression tree (CART) has been widely implemented to estimate impervious surface, an important indicator of urbanization and environmental quality. Although the CART algorithm gains higher overall accuracy than linear regression models, only very few studies have noticed that reliability of CART is affected by systematic errors. Especially, CART typically overestimates impervious surfaces in low-density urban areas and underestimates them in high-density urban areas. The primary objective of this study is to develop an improved integrated method to estimate impervious surface with higher accuracy by reducing the systematic errors of CART. This improved method was applied to three urban areas, Chicago (United States), Venice (Italy), and Guangzhou (China) to examine its effectiveness. When compared with the conventional CART, overall mean average error (MAE) and root mean square error (RMSE) of improved method are decreased by 22.64% and 20.93%, respectively, and R² rises from 0.9 to 0.96. In high-density impervious surfaces, where intensely developed urban area is located, the MAE and RMSE for the improved method are 0.066 and 0.088, respectively, largely improved from 0.100 to 0.130. Since accurate estimation of high-density impervious surfaces is the fundamental issue for monitoring and understanding the urban environment, the improved method demonstrated in this study is significant.     

基于光谱混合分解的流域不透水面提取及其动态分析——以于桥水库为例

不透水面不仅是城市非点源污染的主要来源,还是流域生态环境变化的主要因素之一。不透水面的数量、位置、几何形状、分布格局以及透水率与不透水率的比值,均影响着流域的水文环境,因此成为研究热点。本文以天津于桥水库流域为例,综合遥感(RS)与地理信息系统(GIS)技术,从流域尺度上研究1984~2013年间不透水面覆盖度的变化。在ENVI 5.1软件支持下,利用遥感影像获取1984,1994,2004和2013年4个时相的不透水面信息。采用修正后的归一化水体指数剔除水体信息,排除水体对不透水面提取精度的影响。运用线性光谱混合分析法(Linear Spectral Mixture Analysis,LSMA),提取流域不透水面覆盖度。结果表明:流域内不透水面覆盖度大多集中在1~5级,植被覆盖程度较高。近30年间不透水面比例逐年增加,2013年比1984年增加了2.802%,呈线性增长。中等分辨率的遥感影像适合流域尺度的不透水面提取的结果可作为流域水文及规划管理的重要基础性数据。     

“全域-局部”不透水面信息遥感分步提取模型

提出"全域-局部"遥感信息分布提取模型,通过计算和整合影像局部范围内的空间和光谱特征来优化全域上光谱混淆较大像元的提取精度。模型分为两个主要计算步骤:"全域"前分类与"局部"后分类;"全域"前分类将仅划分出满足一定精度阈值标准的像元,而"局部"后分类则在此部分分类结果基础上,进一步发掘和计算已分类像元所蕴含的信息来辅助对全域未分类像元的提取。在不透水面专题提取过程中,采用支持向量机SVM作为前分类器,通过控制精度阈值所对应的分类后验概率产生部分分类结果;采用调节最小距离分类器作为后分类器,根据一定的权重整合像元局部范围内的空间与光谱信息,代替了传统的全域光谱信息来优化分类。实验采用TM5影像以及所对应的NLCD(National Land Cover Data)标准不透水面产品作为测试集,"全域-局部"模型对应单一SVM模型的提取精度由80.31%提高为82.73%,局部后分类器精度较单一SVM模型由54.27%提高到59.94%。实验证明该模型具有较明显的精度提升且能够较好地解决不透水面与裸土混淆的问题,并得到空间形态上更为完善的不透水面提取结果。     

Characterization of defoliation conditions within a boreal Montane forest ecosystem

Landsat Thematic Mapper (TM) digital data acquired in the fall of 1988 and digital elevation model (DEM) data were evaluated for use in characterizing the spatial distribution of defoliation conditions within the boreal montane spruce‐fir ecosystem in the Black Mountains (Mtns.) of North Carolina. Correlation coefficients between the TM waveband data and field estimates of defoliation taken from 22 one hectare field plots were low (0.10 to ‐0.54). TM band four (near infrared) was the only waveband found to be significantly correlated with needle loss. Defoliation was also shown to be significantly correlated with digital elevation and aspect data. A multivariate linear regression model predicting percent defoliation was developed from the TM, elevation, and aspect data for the 21 field plots. In addition to having a high R2(0.85) the model was shown to reliably predict defoliation conditions throughout the ecosystem. A color‐coded classified image depicting the spatial distribution of defoliation conditions within the study site was generated by applying the model to the TM and DEM data.     

Estimation of impervious surfaces of Beijing,China, with spectral normalized images using linear spectral mixture analysis and artificial neural network

The objective of this article is to evaluate the effectiveness of various algorithms for estimating impervious surfaces. Linear spectral mixture analysis (LSMA) and multi-layer perceptron (MLP) network using original and spectral normalized images were applied to two ASTER images acquired on 31 August and 9 April 2004, respectively. Accuracy assessment was performed with a Quickbird image. Root-mean-square errors (RMSEs) were calculated and compared. Results indicated that LSMA with original images provided the poorest results. RMSE was 14.8% for the August image and 22.4% for the April image. Results from LSMA with normalized images improved significantly with RMSE of 12.6% for the August image and 18.9% for the April image. The MLP modelling with original images generated slightly better results with RMSE of 12.2% and 18.4% for each image. The MLP modelling of normalized images provided the best estimation, yielding a RMSE of 12.1% for the August image and 18.2% for the April image.     

Estimating time parameters of overland flow on impervious surfaces by the particle tracking method

Abstract

Travel time and time of concentration T_c are important time parameters in hydrological designs. Although T_c is the time for the runoff to travel to the outlet from the most remote part of the catchment, most researchers have used an indirect method such as hydrograph analysis to estimate T_c. A quasi two-dimensional diffusion wave model with particle tracking for overland flow was developed to determine the travel time, and validated for runoff discharges, velocities, and depths. Travel times for 85%, 95% and 100% of particles arrival at the outlet of impervious surfaces (i.e. T_t85, T_t95, and T_t100) were determined for 530 model runs. The correlations between these travel times and T_c estimated from hydrograph analysis showed a significant agreement between T_c and T_t85. All the travel times showed nonlinear relationships with the input variables (plot length, slope, roughness coefficient, and effective rainfall intensity) but showed linear relationships with each other.

Editor D. Koutsoyiannis; Associate editor S. Grimaldi     

NIGHT-FROST MODULATION OF NEAR-SURFACE SOIL-WATER ION CONCENTRATION AND THERMAL FIELDS

The modulation of the soil-water ion concentration and thermal fields resulting from night frost is illustrated with data from temperature-electric potential probes in the upper 15 cm of a sandy loam soil. The relative soil-water ion concentration, calculated from soil electric potential observations, aids in the detection of these modulation effects produced by non-conductive heat-mass transfer processes. The apparent thermal diffusivity, calculated from soil-temperature data, further facilitates the detection and analysis of these non-conductive processes. This new technology demonstrates the distorting effects of soil-water advection to the freezing zone, “ice purification” of soil water in the freezing zone and other coupled mass-heat transport effects produced by night frost. The conversion of electric potential observations to a relative index of soil-water ion concentration and the calculation of the apparent thermal diffusivity from the temperature data matrix are demonstrated as effective methods for the detection and analysis of the magnitude and direction of coupled-flow processes. [Key words: electric potential, frost, soil chemistry, soil temperature, thermal diffusivity.]     

An Object Extraction Approach for Impervious Surface Classification with Very-High-Resolution Imagery

Detailed land cover maps provide important information for research and decision-making but are often expensive to develop and can become outdated quickly. Widespread availability of aerial photography provides increased accessibility of high-resolution imagery and the potential to produce high-accuracy land cover classifications. However, these classifications often require expert knowledge and are time consuming. Our goal was to develop an efficient, accurate technique for classifying impervious surface in urbanizing Wake County, North Carolina. Using an iterative training technique, we classified 111 nonmosaicked, very-high-resolution images using the Feature Analyst software developed by Visual Learning Systems. Feature Analyst provides object extraction classifications by analyzing spatial context in relation to spectral data to classify high-resolution imagery. Our image classification results were 95 percent accurate in impervious surface extraction, with an overall total accuracy of 92 percent. Using this method, users with relatively limited geographic information system (GIS) training and modest budgets can produce highly accurate object-extracted classifications of impervious and pervious surface that are easily manipulated in a GIS.     

砾石土渗透稳定特性试验研究

目前砾石土被广泛地用作高土石坝中的防渗体,国内已建成的187 m高的瀑布沟堆石坝就是用它作心墙的防渗材料。介绍了瀑布沟心墙土料的渗透稳定试验结果,并论述了确定砾石土渗透稳定性能的的主要因素是细粒含量,提出了确定粗、细粒料的区分粒径、计算最优细料含量以及用细料含量判别渗透稳定性的方法。试验结果表明,工程中常以小于5 mm粒径含量不小于35%、小于0.1 mm的粒径含量不小于18%,以判别砾石土能否作为高土石坝防渗体的标准是可行的,无需另外要求小于0.005 mm的粒径含量大于10%。