基于对象和DEM的湿地信息提取——以洪河沼泽湿地为例

湿地资源遥感信息提取并满足一定的分类精度和效率,是湿地研究的关键问题之一。利用高分辨率的SPOT5影像数据,通过对研究区分别进行基于像素的监督分类和基于对象的分类,进行典型沼泽湿地专题信息提取,同时将面向对象分类与DEM辅助数据进行信息复合,探寻提升分类精度的方法。结果表明,基于对象和辅助数据DEM的信息提取,可以有效地改善遥感图像的分类精度。     

基于对象和DEM的湿地信息提取——以洪河沼泽湿地为例

湿地资源遥感信息提取并满足一定的分类精度和效率,是湿地研究的关键问题之一。利用高分辨率的SPOT5影像数据,通过对研究区分别进行基于像素的监督分类和基于对象的分类,进行典型沼泽湿地专题信息提取,同时将面向对象分类与DEM辅助数据进行信息复合,探寻提升分类精度的方法。结果表明,基于对象和辅助数据DEM的信息提取,可以有效地改善遥感图像的分类精度。     

基于决策树方法的淡水沼泽湿地信息提取——以三江平原东北部为例

湿地是生态系统的重要组成部分,及时、准确地获得湿地基础信息,对湿地的动态监测、保护与可持续利用及其它领域的研究具有重要意义。以三江平原东北部沼泽湿地为例,利用分类回归树(Classification and Regression Tree,CART)算法从训练样本数据集中挖掘分类规则,集成遥感影像的光谱特征、纹理特征和地学辅助数据建立研究区湿地信息提取的决策树模型。用实测的GPS样本点对分类结果进行精度验证,并与最大似然监督分类方法(Maximum Likelihood Classification,MLC)进行对比。结果表明,基于CART的决策树分类结果的总精度和Kappa系数分别为82.65%和0.7935,分类精度较MLC监督分类方法有明显提高,是内陆淡水沼泽湿地信息提取的有效手段。     

基于决策树分类技术的遥感影像分类方法研究

以河北唐山为研究区,应用Landsat ETM+影像数据和GIS数据,对决策树分类技术和传统计算机自动分类方法进行了比较。研究表明:决策树与传统自动分类方法相比,分类精度提高了18.29%,Kappa系数提高0.1878。在地形起伏的山区,应用DEM及其衍生数据等GIS数据作为辅助数据可以提高分类精度19.52%,Kappa系数提高0.281;反射率影像分类效果比原始DN值影像的分类效果好,分类精度提高15.86%;缨帽变换在压缩数据量的同时,分类精度有所降低。     

GIS支持下遥感图象中采矿塌陷地提取方法研究

采矿塌陷地动态监测是工矿区资源管理与环境保护的重要方面 ,遥感技术可在其中发挥重要作用 ,从遥感图象中提取采矿塌陷地是遥感应用于矿山资源环境监测的重要研究课题 .传统的提取方法主要基于光谱特征 ,精度与效率都难以满足应用要求 ,为了以较高的精度 ,从遥感图象中提取塌陷地 ,必须建立新的方法与模型 .将遥感技术与 GIS相结合进行专题信息提取是有效的途径之一 .本文根据研究区的特点 ,以具体应用为指导 ,遥感技术与 GIS相结合 ,提出了 GIS支持下的分层分类、基于 GIS变化区域识别的分类、基于 GIS和领域知识对遥感分类图象进行后处理、GIS支持下采矿塌陷地的直接提取等方法与模型 ,充分应用光谱特征、地学特征与信息、领域和专家知识及其他统计数据辅助进行遥感图象处理与专题信息提取 .这些方法在精度、效率等方面均较传统方法有较大提高 ,最大提取精度可达到 89% ,能够有效地对工矿区土地塌陷态势进行动态监测     

滨海湿地信息提取方法比较研究

以江苏省典型滨海湿地为研究对象,利用2005年5月26日的Landsat7 ETM+图像数据,在湿地特征及其遥感图像表征分析的基础上,逐步提高湿地信息的提取精度,通过对多光谱遥感图像特征向量的分析,总结出一些湿地信息提取的规则和方法。在滨海湿地光谱特征分析的基础上,对研究区的图像进行了非监督分类,利用湿地的光谱相应特征、纹理特征、主成分变换、归一化差异水体指数等特征和相应的知识规则,得到用于优化分类的知识规则,采用分层分类的方法对非监督分类的结果进行了优化,从而使提取结果的精度较原来有了很大程度的提高。还利用给予数据挖掘的分层分类法进行分类比较,通过建立误差矩阵和对比各种分类方法的分类精度,总结出一套分类精度较高的针对该研究区的湿地信息提取方法。     

结合地籍数据的高密度城区面向对象遥感分类 

利用高分辨率遥感影像和GIS辅助数据,对高密度城区进行面向对象的土地利用覆被分类研究。使用NAIP高分辨率航空遥感影像,在多尺度影像分割的基础上,针对特定地物选择合适的影像分割参数。采用决策树方法建立高密度城市地区的分类规则,并结合该地区地籍图数据作为辅助数据,逐步进行高密度城市地区地物信息提取。利用辅助数据进行面向对象的遥感分类效果优于单纯依靠遥感影像进行的分类,且有效提取了道路和复杂的房屋等信息,得到了理想的分类结果,其总分类精度从常规面向对象方法的84.08%提高到89.79%。利用辅助数据进行遥感分类提高了高分辨率遥感影像的分类精度,说明了利用辅助数据进行遥感分类方法的有效性。     

基于DEM的西北干旱区典型地貌类型坡度提取分析

由于DEM数据本身的多尺度因素,加之地形、地貌特征具有宏观性与区域分异性的特点,直接的信息提取往往很难达到预期的目的,同时,利用DEM进行坡度提取,精度具有很大的不确定性。选取西北干旱区典型地貌类型的平原和丘陵地作为试验区,利用1∶1万和1∶5万DEM,对不同比例尺及不同空间分辨率DEM的坡度提取结果进行对比,分析变化规律,并对坡度提取误差进行量化模拟,成果对西北干旱区进行生态环境建设具有一定的参考价值。     

应用遥感及地理信息系统进行植被制图

利用卫星遥感数据制作复杂地形环境的植被图面临的最主要问题是精度，单纯对遥感数据（ＴＭ或ＳＰＯＩ）进行监督或非监督分类的精度低于５０％。本文选择美国亚利桑那州Ｓａｎｔａ Ｃａｔａｌｉｎａ山脉的Ｐｕｓｃｈ Ｒｉｄｇｅ作为研究区，分析地理信息系统模型在改善植被分类精度中的作用。结果表明，通过结合辅助数据和应用地理信息系统模型，其精度可以从３７．４１％提高到７１．６７％（ＳＰＯＴ数据，非监督分类），或从５     

基于垂直植被指数的湿地植被类型提取研究——以长江口九段沙湿地为例

针对潮滩湿地植被的特点,利用2005年5月7日的SPOT 5数据,结合自动提取土壤线算法和近红外一红外二维光谱特征空间,获取土壤线参数,并利用垂直植被指数对九段沙湿地典型试验区进行植被信息提取.计算结果表明,土壤线计算方法简单可靠,利用垂直植被指数进行分类其精度达到86.5％,相比最大似然分类方法提高了5.7％.     

南方丘陵地区水稻种植面积遥感信息提取的可行性分析*

以浙江省为试验区,在地理信息系统支持下综合利用多种地理信息,探讨丘陵地区大面积提取水稻种植面积信息的可行性。开展了分类识别方法的比较试验及训练样点相对稳定性试验。针对丘陵地区的复杂地形,在数字化地形图的基础上,建立数字地形模型(DTM),并衍生出地面坡度等地貌因子的数字化图像,结合NOAA/AVHRR数据,进行分类。试验结果表明,传统的分类识别方法中,最大似然法的分类精度可满足业务化运行的要求;建立在混合像元分解基础上的模糊监督分类,有较高的分类精度和较好的稳定性,具有较强的适应性;地貌因子参与遥感影像的分类,不仅可以有效地提高丘陵地区水稻种植面积信息的提取精度,而且还可以使面积信息提取精度保持一定的稳定性,提高空间精度;为探讨丘陵地区水稻种植面积信息遥感提取的可靠性和客观性,在训练样点保持相对稳定的前提下,对1996年和1997年浙江省水稻种植面积进行测算,两年的数量精度均在92%以上。     

南方丘陵地区水稻种植面积遥感信息提取的可行性分析

以浙江省为试验区，在地理信息系统支持下综合利用多种地理信息，探讨丘陵地区大面积提取水稻种植面积信息的可行性。开展了分类识别方法的比较试验及训练样点相对稳定性试验。针对丘陵地区的复杂地形，在数字化地形图的基础上，建立数字地形模型（ＤＴＭ），并衍生出地面坡度等地貌因子的数字化图像，结合ＮＯＡＡ／ＡＶＨＲＲ数据，进行分类。试验结果表明，传统的分类识别方法中，最大似然法的分类精度可满足业务化运行的要求；建立在混合像元分解基础上的模糊监督分类，有较高的分类精度和较好的稳定性，具有较强的适应性；地貌因子参与遥感影像的分类，不仅可以有效地提高丘陵地区水稻种植面积信息的提取精度，而且还可以使面积信息提取精度保持一定的稳定性，提高空间精度；为探讨丘陵地区水稻种植面积信息遥感提取的可靠性和客观性，在训练样点保持相对稳定的前提下，对１９９６年和１９９７年浙江省水稻种植面积进行测算，两年的数量精度均在９２％以上。     

Mapping plant communities in a local Arctic landscape applying a scanned infrared aerial photograph in a geographical information system

The vegetation of a 5km2 area in front of the Midtre Lovenbreen glacier, Northwest Spitsbergen, Svalbard, was mapped on the scale 1 10000. The main aim of the study was to develop a new method of vegetation classification based on a probability model, and apply the method on a digitized aerial colour infrared (CIR) photograph with a better ground resolution than provided by the Landsat and SPOT satellites. Large-scale data from different sources such as the CIR-aerial photograph, information layers derived from a digital elevation model (DEM) and vegetation sampling in the field have been integrated in a GIS. Probability models build the links between GIS data layers and plant communities resulting from classification of field data. Eight plant communities were defined by means of vegetation data and mapped automatically by classification of the CIR-photograph. Based on the probability model, maps were produced showing the actual and potential distribution of plant communities. The accuracy of the vegetation map was improved by including additional information from the DEM.     

Matlab 自组织竞争神经网络遥感图像分类——以福州市琅歧岛土地覆盖ö土地利用类型为例

以福州市琅歧岛土地覆盖/土地利用类型为例, 以遥感图像解译知识为基础, 使用TM、Aster的融合图像, 将DEM 因子作为待分类图像的波段加入其中, 构成新的待分类图像, 利用Matlab 平台构建自组织竞争神经网络, 在不依赖网络训练样本选取的前提下, 仿真的结果能真实的反映原始图像的特征, 分类总精度为91. 14% , Kappa 系数为0. 89, 实例证明自组织竞争神经网络分类方法是一种行之有效的分类方法。     

Towards operational monitoring of a northern wetland using geomatics-based techniques

This paper presents a geomatics-based approach for the operational monitoring of spatio-temporal changes in a northern wetland. It demonstrates how valuable, and otherwise unattainable, spatially distributed and timely baseline data can be obtained using basic remote sensing techniques. It further shows how these data can be combined to retrieve information pertaining to the hydro-ecological relationships in the wetland. The study was conducted in the Peace-Athabasca Delta (PAD), which is a large wetland complex in northeastern Alberta, Canada. Combinations of Radarsat Synthetic Aperture Radar (SAR) and optical satellite images (Landsat or SPOT multispectral) were used to generate a time-series of flood maps for the six-year period, 1996 to 2001. These maps clearly depict the extent of the 1996 and 1997 overland floods and the subsequent water level draw down. A flood duration map that shows how long each image cell was inundated was generated by combining the series of flood maps. The flood duration map highlights regions where the duration of flooding appears to be highest or lowest. Such maps are invaluable for any ecological change detection protocols that may be developed for this region. The general vegetation patterns were also mapped using multi-temporal SPOT-4 images from the summer season (May and August) of 2001 to an accuracy of 86%. By comparing the vegetation and flood duration maps, the relationship between vegetation patterns and duration of flooding could be examined. Results indicated that basins inundated for longer periods (3-5 years) were dominated by relatively more productive graminoid (grass-like) vegetation, whereas, areas flooded for less than two years were characterised by less productive shrub vegetation. Airborne scanning LiDAR (Light Detection and Ranging) data from the summer (June) of 2000 were also used to generate a Digital Elevation Model (DEM) of selected non-flooded areas. LiDAR DEM accuracy was satisfactory (Root Mean Square Error of 0.24 m) and it proved to be sufficiently detailed to detect the subtle topographic patterns in this relatively flat region. Comparison of the vegetation map to the DEM demonstrated that the shrubs were located in areas that were, on average, between 0.5 m (in south) and 0.73 m (in north) higher than the graminoid covered regions. Notwithstanding other parameters that influence the distribution of vegetation, these results indicate that flood duration and elevation are two important factors. The usefulness of these spatial databases recommends the timely generation of flood and vegetation maps in the continued monitoring of the changes and relationships in this delta.     

Mapping fallow lands in Vietnam's north-central mountains using yearly Landsat imagery and a land-cover succession model

The objective of this article is to investigate whether it is possible to use Landsat data together with ancillary data and temporal context to accurately identify land covers found in the fallow areas of Montane Mainland Southeast Asia's (MMSEA's) difficult-to-map swidden landscapes. A rule-based non-parametric hybrid classification method that integrates knowledge about the vegetation regrowth patterns in these landscapes with analysis of Landsat imagery is developed. The method is applied to three upland districts of the Nghe An Province, Vietnam. The results show that the hybrid classification approach, with an overall accuracy of 90%, is superior to using a traditional maximum likelihood classifier, which generated an overall accuracy of 68%. The hybrid classification results indicate that the landscape is dominated by bush and bamboo, while the maximum likelihood classification suggests a landscape that is predominantly grass covered. The hybrid classification results are in agreement with local knowledge and information from fieldwork-based reports and articles on swidden systems in the study area and other parts of MMSEA.     

基于机载LiDAR的高寒山区遥感高程数据精度评估

遥感高程数据是获取缺资料地区DEM（Digital elevation models）数据的重要手段。然而,由于高寒山区实地高程测量稀少,难以对多源遥感DEM数据进行统一验证。ICESat-2等新的遥感高程数据在高寒山区也缺乏相应的精度评估。针对此问题,以青藏高原东北缘的冰沟流域作为研究区,采用机载航空遥感获取的大范围LiDAR（Light Detection And Ranging）DEM数据对新产品ICESat-2 ATL06（Ice, Cloud, and Land Elevation Satellite-2, Land Ice Height）、ALOS DEM（12.5 m分辨率）以及新版本SRTM V3（SRTM Arc-Second Global 1 V003）、ASTER GDEM V3（ASTER Global DEM）进行验证,并分析地形因子与均方根误差RMSE的关系。研究结果表明：ICESat-2 ATL06数据在高寒山区的RMSE为0.747 m。由于其较高的精度,可用于验证缺资料地区的其他遥感高程数据。其他遥感高程数据的精度都相对较低,ALOS 12.5 m数据的RMSE为5.284 m;ASTER GDEM V3版本的RMSE为9.903 m。实验所采用的4种遥感高程数据与机载LiDAR DEM均具有较高的相关性,相关系数在0.998与1.000之间。实验还揭示了坡度是影响遥感DEM精度的主要因素。除ICESat-2 ATL06外,其他高程数据的RMSE均随坡度的增大先减小再增大,且都存在一个最佳坡度值。鉴于地形复杂多样的冰沟流域具有青藏高原高寒山区的典型特征,多源遥感DEM数据在该区域的验证结论具有较好的代表性,可为相似地区DEM数据的使用和评估提供重要的知识补充。     

Deriving terrain and textural information from stereo RADARSAT data for mountainous land cover mapping

A new procedure is proposed for land cover classification in a mountainous area using stereo RADARSAT-1 data. The method integrates a few types of information that can be extracted from the same stereo RADARSAT images: (1) the Digital Elevation Model (DEM) generated from the stereo RADARSAT images; (2) terrain information (elevation, slope and aspect) extracted from the derived DEM; and (3) textural information derived from the same RADARSAT images. An Artificial Neural Network (ANN) classifier is applied for the land cover classification. Performance of the proposed method is evaluated using a mountainous study area in Southern Argentina, where there is a lack of up-to-date information for environmental monitoring. The results show that the integration of textural and terrain information can greatly improve the accuracy of the classification using the ANN classifier. It demonstrates that stereo RADARSAT images provide valuable data sources for land cover mapping, especially in mountainous areas where cloud cover is a problem for optical data collection and topographical data are not always available.     

Land-cover classification using multi-temporal GF-1 wide field view data

ABSTRACT

Vegetation is an important land-cover type and its growth characteristics have potential for improving land-cover classification accuracy using remote-sensing data. However, due to lack of suitable remote-sensing data, temporal features are difficult to acquire for high spatial resolution land-cover classification. Several studies have extracted temporal features by fusing time-series Moderate Resolution Imaging Spectroradiometer data and Landsat data. Nevertheless, this method needs assumption of no land-cover change occurring during the period of blended data and the fusion results also present certain errors influencing temporal features extraction. Therefore, time-series high spatial resolution data from a single sensor are ideal for land-cover classification using temporal features. The Chinese GF-1 satellite wide field view (WFV) sensor has realized the ability of acquiring multispectral data with decametric spatial resolution, high temporal resolution and wide coverage, which contain abundant temporal information for improving land-cover classification accuracy. Therefore, it is of important significance to investigate the performance of GF-1 WFV data on land-cover classification. Time-series GF-1 WFV data covering the vegetation growth period were collected and temporal features reflecting the dynamic change characteristics of ground-objects were extracted. Then, Support Vector Machine classifier was used to land-cover classification based on the spectral features and their combination with temporal features. The validation results indicated that temporal features could effectively reflect the growth characteristics of different vegetation and finally improved classification accuracy of approximately 7%, reaching 92.89% with vegetation type identification accuracy greatly improved. The study confirmed that GF-1 WFV data had good performances on land-cover classification, which could provide reliable high spatial resolution land-cover data for related applications.     

Large-scale mapping of the riverbanks,mud flats and salt marshes of the Scheldt basin,using airborne imaging spectroscopy and LiDAR

For maintaining the tidal waterways in the Scheldt basin, including the rivers Rupel and Durme and a large part of the Nete catchment, and for ecological monitoring of the mud flats, salt marshes and riverbank vegetation, the Flemish government needs detailed maps of these rivers and their bank structures. These maps indicate not only vegetation types, plant associations and sediment types but also hard structures, such as quays, locks, sluices and roads. Different remote sensing techniques were used to collect the data necessary to produce the required detailed maps. During the months of July and August 2007 an airborne flight campaign took place to collect hyperspectral and LiDAR data of the Scheldt basin and the Nete catchments. These rivers have a total length of about 240 km. The Airborne Imaging Spectrometer for Applications (AISA) Eagle sensor acquired hyperspectral data in 32 spectral bands covering the visible/near-infrared (VIS/NIR) part of the electromagnetic spectrum with a ground resolution of 1 m. A multiple binary classification algorithm based on Fisher's linear discriminant analysis (LDA) was used to map the salt marshes and riverbank vegetation. Ground truth information, that is vegetation and sediment types, together with their geographical locations collected around the time of the flight campaign, was used to train the classifier in the later classification step. Laser scanning was performed using the Riegl LMS-Q560. The LiDAR dataset obtained had a resolution of at least 1 point per m² and was used to produce a digital elevation model (DEM) that contains all elements of the terrain. From this DEM a digital terrain model (DTM) was derived by applying appropriate filtering techniques. The elevation models were used primarily to derive information on the height, slope and aspect of the banks and dikes, but they also served as expert knowledge in the classification of the mud flats and bank vegetation.

Overall, this work illustrates how airborne hyperspectral and LiDAR data can be used to derive highly detailed maps of the sediments, vegetation and hard structures along tidal rivers in large river basins. It also shows how large datasets can be handled in an expert system, in combination with different classification techniques, to produce the required result and accuracy.