澳大利亚东南部森林山火HJ卫星遥感监测

以2009年2月发生在澳大利亚东南部的森林山火为研究对象,利用HJ-1B遥感影像识别森林山火,分析HJ-1B在林火灾害事故中的监测能力,通过对HJ-1B IRS B07设计参数及数据特点进行分析,提出适用于HJ-1B卫星林火监测的归一化火点指数(Ku)算法.研究表明:Ku值大于0.40为潜在可能的火点像元,云耀斑和地表虚假高温点是影响林火监测的主要噪声.由于HJ-1B没有获取到研究区域未着火前的影像数据,利用MODIS(Moderate-resolution Imaging Spectroradiometer)空间分辨率为250 m的通道1和通道2计算植被指数,其结果能较好的应用于HJ-1B林火监测算法中.通过对比分析HJ-1B林火监测结果和MODIS林火产品MOD14认为,HJ-1B能更好的监测出澳大利亚东南部森林火灾,反映出火灾的局部空间分布和细节特征.     

基于随机森林的航天器电信号多分类识别方法

针对航天器电特性信号数据存在数据量大、特征维数高、计算复杂度大和识别率低等问题,提出基于主成分分析(PCA)的特征提取方法和随机森林(RF)算法,对原始数据进行降维,提高计算效率和识别率,实现对航天器电信号数据的快速、准确识别分类。随机森林算法在处理高维数据上具有优越的性能,但是考虑到时间复杂度问题,利用主成分分析方法对数据进行压缩和降维,在保证准确率的同时提高了计算效率。实验结果表明:与其他算法相比,针对航天器电特性信号数据,本文方法在准确率、计算效率和稳定性等方面均显示出优异的性能。     

Evaluating the statistical performance of less applied algorithms in classification of worldview-3 imagery data in an urbanized landscape

In recent decade, analyzing the remotely sensed imagery is considered as one of the most common and widely used procedures in the environmental studies. In this case, supervised image classification techniques play a central role. Hence, taking a high resolution Worldview-3 over a mixed urbanized landscape in Iran, three less applied image classification methods including Bagged CART, Stochastic gradient boosting model and Neural network with feature extraction were tested and compared with two prevalent methods: random forest and support vector machine with linear kernel. To do so, each method was run ten time and three validation techniques was used to estimate the accuracy statistics consist of cross validation, independent validation and validation with total of train data. Moreover, using ANOVA and Tukey test, statistical difference significance between the classification methods was significantly surveyed. In general, the results showed that random forest with marginal difference compared to Bagged CART and stochastic gradient boosting model is the best performing method whilst based on independent validation there was no significant difference between the performances of classification methods. It should be finally noted that neural network with feature extraction and linear support vector machine had better processing speed than other.     

A comparative study of decision tree,random forest,and convolutional neural network for spread-F identification

Ionospheric spread-F (SF) is a commonly observed phenomenon of electron density perturbation in the F-layer. The ionospheric irregularities structure has an adverse effect on the propagation of electromagnetic waves in the ionosphere. The automatic identification of ionospheric spread-F and statistical study of the formation of spread-F are of great significance to the study of the physical mechanism of ionospheric inhomogeneity and for prediction of ionospheric irregularities. In this paper, we describe and implement three automatic identification methods of spread-F based on machine learning: decision tree, random forest, and convolutional neural network (CNN). The performance of these automatic identification methods was verified using a large set of test data. Results show that the accuracy of all three methods on identifying ionograms with spread-F exceeded 90%. After comparing the results of the three methods, we found that the decision tree method was the simplest and with the structure easiest to be understood, and it required the shortest interpretation time. In terms of the identification results, the random forest method provided better results than the decision tree method, and the CNN method was the best at accurately identifying ionograms with spread-F.     

一种基于改进后的孤立森林机器学习方法在飞行器控制系统试验电源数据包络分析中的应用

为了弥补当前试验数据包络分析方法存在的缺陷,提出并实现了一套新的数据包络分析方法。针对控制系统设计方式相同的不同飞行器多条数据一致性判断问题,采用改良后的孤立森林方法进行数据包络分析,快速找出异常电源电压数据。这种方法还可以推广到类似其他采样数据的数据包络分析场景。在此基础上开发了数据包络分析软件,并进行了多次验证试验。结果表明,提出的数据包络分析方法能有效判断出飞行器控制系统电源所涉及链路中的隐性或显性问题。     

基于深度森林的卫星ACS执行机构与传感器故障识别

针对卫星姿态控制系统（ACS）闭环回路的故障难以辨识的问题,引入深度森林算法,实现执行机构与传感器故障识别。首先针对可获取的少量卫星姿态控制系统遥测数据,结合系统动力学特性,研究合适的特征选择和特征提取方法,再结合深度森林算法进行故障信息学习与辨识,建立故障预测模型,实现执行机构故障与传感器故障的识别。半物理仿真结果表明：在存在气浮台干扰力矩、卫星转动惯量未知、飞轮非线性特性、闭环故障传播等多种不利因素情况下,深度森林算法对于执行机构和传感器故障具有高效的识别能力。     

冰灾对车八岭山地常绿阔叶林区系和群落的影响

研究了2008年年初南方冰雪冻雨灾害对车八岭常绿阔叶林不同区系的冠层植被所造成的影响.结果显示,研究样地内温带分布的乔木受损比例高于热带分布的乔木受损比例.X~2检验说明不同区系间的冠层乔木个体受损比例差异显著(P<0.01);东亚分布及其变型受损个体比例最小,热带亚洲至热带非洲分布及变型、东亚和北美间断分布及其变型、热带亚洲分布及其变型受损个体比例较大.冠层乔木都以重度受损为主,中度受损和轻度受损个体数都远小于前者.CCA多元分析显示,株数密度、平均胸高断面积、坡度、海拔各自对不同物种的受损乔木的分布有显著影响(Monte Carlo test,P=0.010 0).优势种米槠Castanopsis carlesii受损植株的分布与群落的平均胸高断面积显著正相关;亚优势种栲Castanopsis fargesii、鸭公青Neolitsea chuii、尾尖叶柃Eurya acuminata受损植株的分布与坡度显著负相关;亚优势种荷木Schima superb与海拔显著正相关.总之,在相同的生境中能够占据林冠优势层的区系类型往往受灾更严重.     

星载激光雷达森林探测进展及趋势

森林是陆地上组成结构最复杂的生态系统,星载激光雷达兼具高垂直分辨率的优势和大范围数据获取的特点,在大区域尺度的森林参数定量反演方面具有独特的优势。从全波形激光雷达、光子计数激光雷达、成像激光雷达的技术特点、数据处理和森林参数提取等方面,总结了国内外森林探测激光雷达卫星载荷的发展、参数设置及应用潜力,以服务我国星载激光雷达森林探测系统的建设和发展。结合激光雷达辐射传输模型,探讨适合森林探测的激光雷达传感器参数方案,并以ICESat卫星的地球科学测高系统(GLAS)和先进地形激光测高系统(ATLAS) 数据为例,探讨星载激光雷达的森林参数反演方法,最后总结梳理了现有星载载荷的优劣并给出后续载荷发展的建议。     

最优特征选择下多层次分割的城市道路提取

针对道路提取过程中特征维数过高的问题,提出了一种基于ReliefF过滤式和Wrapper封装式的特征选择方法.将粒子群优化算法(PSO)作为Wrapper的搜索算法,优化过的随机森林算法(OPRF)作为Wrapper的分类器构成PSO_OPRF封装式子集评估器,对ReliefF预选后的特征子集进行评估,降低特征维度,选...     

GDP spatialization in Ningbo City based on NPP/VIIRS night-time light and auxiliary data using random forest regression

Accurate spatial distribution information on gross domestic product (GDP) is of great importance for the analysis of economic development, industrial distribution and urbanization processes. Traditional administrative unit-based GDP statistics cannot depict the detailed spatial differences in GDP within each administrative unit. This paper presents a study of GDP spatialization in Ningbo City, China based on National Polar-orbiting Partnership (NPP)/Visible Infrared Imaging Radiometer Suite (VIIRS) night-time light (NTL) data and town-level GDP statistical data. The Landsat image, land cover, road network and topographic data were also employed as auxiliary data to derive independent variables for GDP modelling. Multivariate linear regression (MLR) and random forest (RF) regression were used to estimate GDP at the town scale and were assessed by cross-validation. The results show that the RF model achieved significantly higher accuracy, with a mean absolute error (MAE) of 109.46 million China Yuan (CNY)·km⁻² and a determinate coefficient (R² = 0.77) than the MLR model (MAE = 161.8 million CNY·km⁻², R² = 0.59). Meanwhile, by comparing with the estimated GDP data at the county level, the town-level estimated data showed a better performance in mapping GDP distribution (MAE decreased from 115.1 million CNY·km⁻² to 74.8 million CNY·km⁻²). Among all of the independent variables, NTL, land surface temperature (T_s) and plot ratio (PR) showed higher impacts on the GDP estimation accuracy than the other variables. The GDP density map generated by the RF model depicted the detailed spatial distribution of the economy in Ningbo City. By interpreting the spatial distribution of the GDP, we found that the GDP of Ningbo was high in the northeast and low in the southwest and formed continuous clusters in the north. In addition, the GDP of Ningbo also gradually decreased from the urban centre to its surrounding areas. The produced GDP map provides a good reference for the future urban planning and socio-economic development strategies.