Efficient collection of training data for sub-pixel land cover classification using neural networks

Artificial neural networks (ANNs) are a popular class of techniques for performing soft classifications of satellite images. They have successfully been applied for estimating crop areas through sub-pixel classification of medium to low resolution images. Before a network can be used for classification and estimation, however, it has to be trained. The collection of the reference area fractions needed to train an ANN is often both time-consuming and expensive. This study focuses on strategies for decreasing the efforts needed to collect the necessary reference data, without compromising the accuracy of the resulting area estimates. Two aspects were studied: the spatial sampling scheme (i) and the possibility for reusing trained networks in multiple consecutive seasons (ii). Belgium was chosen as the study area because of the vast amount of reference data available. Time series of monthly NDVI composites for both SPOT-VGT and MODIS were used as the network inputs. The results showed that accurate regional crop area estimation (R² > 80%) is possible using only 1% of the entire area for network training, provided that the training samples used are representative for the land use variability present in the study area. Limiting the training samples to a specific subset of the population, either geographically or thematically, significantly decreased the accuracy of the estimates. The results also indicate that the use of ANNs trained with data from one season to estimate area fractions in another season is not to be recommended. The interannual variability observed in the endmembers’ spectral signatures underlines the importance of using up-to-date training samples. It can thus be concluded that the representativeness of the training samples, both regarding the spatial and the temporal aspects, is an important issue in crop area estimation using ANNs that should not easily be ignored.     

Spatial and semantic effects of LUCAS samples on fully automated land use/land cover classification in high-resolution Sentinel-2 data

In this study, we test the use of Land Use and Coverage Area frame Survey (LUCAS) in-situ reference data for classifying high-resolution Sentinel-2 imagery at a large scale. We compare several pre-processing schemes (PS) for LUCAS data and propose a new PS for a fully automated classification of satellite imagery on the national level. The image data utilizes a high-dimensional Sentinel-2-based image feature space. Key elements of LUCAS data pre-processing include two positioning approaches and three semantic selection approaches. The latter approaches differ in the applied quality measures for identifying valid reference points and by the number of LU/LC classes (7–12). In an iterative training process, the impact of the chosen PS on a Random Forest image classifier is evaluated. The results are compared to LUCAS reference points that are not pre-processed, which act as a benchmark, and the classification quality is evaluated by independent sets of validation points. The classification results show that the positional correction of LUCAS points has an especially positive effect on the overall classification accuracy. On average, this improves the accuracy by 3.7%. This improvement is lowest for the most rigid sample selection approach, PS₂, and highest for the benchmark data set, PS₀. The highest overall accuracy is 93.1% which is achieved by using the newly developed PS₃; all PS achieve overall accuracies of 80% and higher on average. While the difference in overall accuracy between the PS is likely to be influenced by the respective number of LU/LC classes, we conclude that, overall, LUCAS in-situ data is a suitable source for reference information for large scale high resolution LC mapping using Sentinel-2 imagery. Existing sample selection approaches developed for Landsat imagery can be transferred to Sentinel-2 imagery, achieving comparable semantic accuracies while increasing the spatial resolution. The resulting LC classification product that uses the newly developed PS is available for Germany via DOI: https://doi.org/10.15489/1ccmlap3mn39.     

Use of field reflectance data for crop mapping using airborne hyperspectral image

Recent developments in hyperspectral remote sensing technologies enable acquisition of image with high spectral resolution, which is typical to the laboratory or in situ reflectance measurements. There has been an increasing interest in the utilization of in situ reference reflectance spectra for rapid and repeated mapping of various surface features. Here we examined the prospect of classifying airborne hyperspectral image using field reflectance spectra as the training data for crop mapping. Canopy level field reflectance measurements of some important agricultural crops, i.e. alfalfa, winter barley, winter rape, winter rye, and winter wheat collected during four consecutive growing seasons are used for the classification of a HyMAP image acquired for a separate location by (1) mixture tuned matched filtering (MTMF), (2) spectral feature fitting (SFF), and (3) spectral angle mapper (SAM) methods. In order to answer a general research question “what is the prospect of using independent reference reflectance spectra for image classification”, while focussing on the crop classification, the results indicate distinct aspects. On the one hand, field reflectance spectra of winter rape and alfalfa demonstrate excellent crop discrimination and spectral matching with the image across the growing seasons. On the other hand, significant spectral confusion detected among the winter barley, winter rye, and winter wheat rule out the possibility of existence of a meaningful spectral matching between field reflectance spectra and image. While supporting the current notion of “non-existence of characteristic reflectance spectral signatures for vegetation”, results indicate that there exist some crops whose spectral signatures are similar to characteristic spectral signatures with possibility of using them in image classification.     

Global geospatial data from Earth observation: status and issues

ABSTRACT

Data covering the whole of the surface of the Earth in a homogeneous and reliable manner has been accumulating over many years. This type of data became available from meteorological satellites from the 1960s and from Earth-observing satellites at a small scale from the early 1970s but has gradually accumulated at larger scales up to the present day when we now have data covering many environmental themes at large scales. These data have been used to generate information which is presented in the form of global data sets. This paper will give a brief introduction to the development of Earth observation and to the organisations and sensors which collect data and produce global geospatial data sets. Means of accessing global data sets will set out the types of data available that will be covered. Digital elevation models are discussed in a separate section because of their importance in georeferencing image data as well as their application to analysis of thematic data. The paper will also examine issues of availability, accuracy, validation and reliability and will look at future challenges.     

基于三维激光扫描数据的地铁隧道断面测量

地铁主体施工完毕后为了对线路进行调线调坡,需要对地铁隧道进行断面测量,三维激光扫描与常规测量方法相比具有非接触式测量,可高密度采集空间三维点云数据等特点,为地铁断面测量提供了新的途径。本文基于Leica Scanstation 2扫描仪分析了三维激光扫描点云数据采集步骤和数据处理流程,阐述了基于点云数据的地铁断面测量方法,分析了三维激光扫描技术在隧道断面测量中应用的可行性。研究结果表明,这种高密度、高精度的隧道断面能够满足地铁调线调坡等方面的技术要求。     

A high-precision heuristic model to detect home and work locations from smart card data

Smart card-automated fare collection systems now routinely record large volumes of data comprising the origins and destinations of travelers. Processing and analyzing these data open new opportunities in urban modeling and travel behavior research. This study seeks to develop an accurate framework for the study of urban mobility from smart card data by developing a heuristic primary location model to identify the home and work locations. The model uses journey counts as an indicator of usage regularity, visit-frequency to identify activity locations for regular commuters, and stay-time for the classification of work and home locations and activities. London is taken as a case study, and the model results were validated against survey data from the London Travel Demand Survey and volunteer survey. Results demonstrate that the proposed model is able to detect meaningful home and work places with high precision. This study offers a new and cost-effective approach to travel behavior and demand research.     

CASS在城镇地籍数据库建设中的应用

CASS地形地籍成图软件是基于AutoCAD平台技术的数字化测绘数据采集系统,而城镇地籍管理系统普遍采用GIS平台进行处理空间矢量化数据,另外地形、地籍属性等数据均需要无损导入GIS平台。结合某市1∶500城镇地籍数据库建设,浅析地籍数据库建设中利用CASS软件进行数据采集的方法,为国土管理部门提供参考。     

无人机航测技术在风电场测图中的应用

本文提出了一种利用无人机航空摄影测量技术完成风电场测图的方法,利用该方法可以实现测区1∶2000地形图数据的生产。首先根据航摄区域和航线规划布设控制点、像控点、检查点,利用飞马F300无人机进行航摄作业获取航测数据,然后对数据进行预处理,按照标准格式导入Context Capture软件建立三维倾斜模型,通过EPS2016立体量测软件基于三维倾斜模型完成地形图数据的采集,对于模型变形的区域,利用MapMatrix软件进行数据采集与校核。利用安徽宿州风电场测图项目验证了该方法满足1∶2000大比例尺地形图精度要求。     

Comparing methods for mapping canopy chlorophyll content in a mixed mountain forest using Sentinel-2 data

The Sentinel-2 Multi-Spectral Imager (MSI) has three spectral bands centered at 705, 740, and 783 nm wavelengths that exploit the red-edge information useful for quantifying plant biochemical traits. This sensor configuration is expected to improve the prediction accuracy of vegetation chlorophyll content. In this work, we assessed the performance of several statistical and physical-based methods in retrieving canopy chlorophyll content (CCC) from Sentinel-2 in a heterogeneous mixed mountain forest. Amongst the algorithms presented in the literature, 13 different vegetation indices (VIs), a non-parametric statistical approach, and two radiative transfer models (RTM) were used to assess the CCC prediction accuracy. A field campaign was conducted in July 2017 to collect in situ measurements of CCC in Bavarian forest national park, and the cloud-free Sentinel-2 image was acquired on 13 July 2017. The leave-one-out cross-validation technique was used to compare the VIs and the non-parametric approach. Whereas physical-based methods were calibrated using simulated data and validated using the in situ reference dataset. The statistical-based approaches, such as the modified simple ratio (mSR) vegetation index and the partial least square regression (PLSR) outperformed all other techniques. As such the modified simple ratio (mSR3) (665, 865) gave the lowest cross-validated RMSE of 0.21 g/m² (R² = 0.75). The PLSR resulted in the highest R² of 0.78, and slightly higher RMSE =0.22 g/m² than mSR3. The physical-based approach-INFORM inversion using look-up table resulted in an RMSE =0.31 g/m², and R² = 0.67. Although mapping CCC using these methods revealed similar spatial distribution patterns, over and underestimation of low and high CCC values were observed mainly in the statistical approaches. Further validation using in situ data from different terrestrial ecosystems is imperative for both the statistical and physical-based approaches' effectiveness to quantify CCC before selecting the best operational algorithm to map CCC from Sentinel-2 for long-term terrestrial ecosystems monitoring across the globe.     

The improvement of an object-oriented classification using multi-temporal MODIS EVI satellite data

Abstract

This paper investigates the contribution of multi-temporal enhanced vegetation index (EVI) data to the improvement of object-based classification accuracy using multi-spectral moderate resolution imaging spectral-radiometer (MODIS) imagery. In object-oriented classification, similar pixels are firstly grouped together and then classified; the produced result does not suffer the speckled appearance and closer to human vision. EVI data are from the MODIS sensor aboard Terra spacecraft. 69 EVI data (scenes) were collected during the period of three years (2001–2003) in a mountainous vegetated area. These data sets were used to study the phenology of the land cover types. Different land cover types show distinct fluctuations over time in EVI values and this information might be used to improve object-oriented land cover classification. Two experiments were carried out: one was only with single date MODIS multispectral data, and the other one including also the 69 EVI images. Eight classes were distinguished: temperate forest, tropical dry forest, grassland, irrigated agriculture, rain-fed agriculture, orchards, lava flows and human settlement. The two classifications were evaluated with independent verification data, and the results showed that with multi-temporal EVI data, the classification accuracy was improved 5.2%. Evaluated by McNemar's test, this improved was significant, with significance level p=0.01.     

Big location‐based social media messages from China's Sina Weibo network: Collection,storage, visualization,and potential ways of analysis

China's social media platform, Sina Weibo, like Twitter, hosts a considerable amount of big data: messages, comments, pictures. Collecting and analyzing information from this treasury of human behavior data is a challenge, although the message exchange on the network is readable by everyone through the web or app interface. The official Application Programming Interface (API) is the gateway to access and download public content from Sina Weibo and is used to collect messages for all mainland China. The nearby_timeline() request is used to harvest only messages with associated location information. This technical note serves as a reference for researchers who do not speak Mandarin but want to collect data from this rich source of information. Ways of data visualization are presented as a point cloud, density per areal unit, or clustered using Density‐Based Spatial Clustering of Applications with Noise (DBSCAN). The relation of messages to census information is also given.     

A new supervised classifier exploiting spectral-spatial information in the Bayesian framework

Conventional machine learning methods are often unable to achieve high degrees of accuracy when only spectral data are involved in the classification process. The main reason of that inaccuracy can be brought back to the omission of the spatial information in the classification. The present paper suggests a way to combine effectively the spectral and the spatial information and improve the classification’s accuracy. In practice, a Bayesian two-stage methodology is proposed embodying two enhancements: i) a geostatistical non-parametric classification approach, the universal indicator kriging and ii) the smooth multivariate kernel method. The former provides an informative prior, while the latter overcomes the assumption (often not true) of independence of the spectral data. The case study reports an application to land-cover classification in a study area located in the Apulia region (Southern Italy). The methodology performance in terms of overall accuracy was compared with five state-of-the-art methods, i.e. naïve Bayes, Random Forest, artificial neural networks, support vector machines and decision trees. It is shown that the proposed methodology outperforms all the compared methods and that even a severe reduction of the training set does not affect seriously the average accuracy of the presented method.     

Automated high resolution mapping of coffee in Rwanda using an expert Bayesian network

African highland agro-ecosystems are dominated by small-scale agricultural fields that often contain a mix of annual and perennial crops. This makes such systems difficult to map by remote sensing. We developed an expert Bayesian network model to extract the small-scale coffee fields of Rwanda from very high resolution data. The model was subsequently applied to aerial orthophotos covering more than 99% of Rwanda and on one QuickBird image for the remaining part. The method consists of a stepwise adjustment of pixel probabilities, which incorporates expert knowledge on size of coffee trees and fields, and on their location. The initial naive Bayesian network, which is a spectral-based classification, yielded a coffee map with an overall accuracy of around 50%. This confirms that standard spectral variables alone cannot accurately identify coffee fields from high resolution images. The combination of spectral and ancillary data (DEM and a forest map) allowed mapping of coffee fields and associated uncertainties with an overall accuracy of 87%. Aggregated to district units, the mapped coffee areas demonstrated a high correlation with the coffee areas reported in the detailed national coffee census of 2009 (R² = 0.92). Unlike the census data our map provides high spatial resolution of coffee area patterns of Rwanda. The proposed method has potential for mapping other perennial small scale cropping systems in the East African Highlands and elsewhere.     

Production of global land cover data – GLCNMO

Abstract

Global land cover is one of the fundamental contents of Digital Earth. The Global Mapping project coordinated by the International Steering Committee for Global Mapping has produced a 1-km global land cover dataset – Global Land Cover by National Mapping Organizations. It has 20 land cover classes defined using the Land Cover Classification System. Of them, 14 classes were derived using supervised classification. The remaining six were classified independently: urban, tree open, mangrove, wetland, snow/ice, and water. Primary source data of this land cover mapping were eight periods of 16-day composite 7-band 1-km MODIS data of 2003. Training data for supervised classification were collected using Landsat images, MODIS NDVI seasonal change patterns, Google Earth, Virtual Earth, existing regional maps, and expert's comments. The overall accuracy is 76.5% and the overall accuracy with the weight of the mapped area coverage is 81.2%. The data are available from the Global Mapping project website (http://www.iscgm.org/). The MODIS data used, land cover training data, and a list of existing regional maps are also available from the CEReS website. This mapping attempt demonstrates that training/validation data accumulation from different mapping projects must be promoted to support future global land cover mapping.     

Stakeholder analysis of the governance framework of a national SDI dataset – whose needs are met in the buildings and address register of the Netherlands?

ABSTRACT

National spatial data infrastructures are key to achieving the Digital Earth vision. In many cases, national datasets are integrated from local datasets created and maintained by municipalities. Examples are address, building and topographic information. Integration of local datasets may result in a dataset satisfying the needs of users of national datasets, but is it productive for those who create and maintain the data? This article presents a stakeholder analysis of the Basisregistratie Adressen en Gebouwen (BAG), a collection of base information about addresses and buildings in the Netherlands. The information is captured and maintained by municipalities and integrated into a national base register by Kadaster, the Cadastre, Land Registry and Mapping Agency of the Netherlands. The stakeholder analysis identifies organisations involved in the BAG governance framework, describes their interests, rights, ownerships and responsibilities in the BAG, and maps the relationships between them. Analysis results indicate that Kadaster and the municipalities have the highest relative importance in the governance framework of the BAG. The study reveals challenges of setting up a governance framework that maintains the delicate balance between the interests of all stakeholders. The results provide guidance for SDI role players setting up governance frameworks for national or global datasets.     

Estimation of discharges at river mouth with MODIS image

The transport of the sediment, carried in suspension by water, is central to hydrology and the ecological functioning of river floodplains and deltas. River discharge estimation is useful for demonstrating this information. In this study, we extracted MODIS reflectance values from a pixel near the river mouth after carrying out the simple atmospheric correction method, then applied single regression analysis to reflectance values and the in situ discharge of Naka River in Tokushima prefecture and Monobe River in Kochi prefecture, Japan. MODIS images and in situ data were taken from January through December, 2004. As a result, both in Naka River and Monobe River, robustly positive relationships between the discharges observed in situ and remotely sensed MODIS reflectance data in the region of river mouth were found throughout the year. In addition, we estimated monthly and annual average discharge from the MODIS reflectance with the regression formula. As a result, in situ average discharge was well estimated.     

云模式下地质信息用户行为采集体系实践

在分析了地质信息化现状的基础上,针对实际应用中无法获取用户行为情况的问题,提出了云模式下地质信息用户的行为采集体系。首先基于Logstash、ElasticSearch、Kibana构建了技术原型,并提出采集内容、分类以及制定采集接口规范。从技术方法实现角度阐述了系统开发环境、云容器集群节点部署、大数据检索分析等关键技术,并介绍了数据从采集获取、管理存储、分析与可视化统计的流程路线。以"地盒"产品数据采集为例,实现了该平台用户访问、地质分析功能模块使用情况、商店访问、资讯浏览量等信息的采集与统计分析。该体系能够支撑互联网用户行为信息采集需求,可实时统计出用户的爱好,了解用户的习惯,为地学信息化软件研发和推广提供有效指导,可在空间信息化系统用户行为分析和研究中推广使用。     

基于GPS RTK的过江隧道测图研究与应用

过江隧道测图包括1:500陆上地形测量和隧道穿越的水下地形测量.陆上地形测量采用GPS RTK或全站仪全野外数据采集,水下地形测量采用GPS RTK配合测深仪进行全野外数据采集.本文以珠海市十字门过江隧道工程为背景,介绍了过江隧道1:500地形图成图方法,特别是水下地形测量的方法和应用,建立了完整的技术流程,得到符合精...     

Contributions to a quality description of areal objects in spatial data sets

In this paper, we present a quality evaluation of two-dimensional building acquisition. We propose methods for identification and quantification of differences between independently acquired regions, and we present a systematic classification of those differences. Differences between acquired sets R_j={r_i}_j of region r_ij depend on the context of observation, on the technique of observation, etc. We distinguish topological and geometrical differences. Topological differences refer to the interior structure of a set of regions as well as to the structure of the boundary of a single region. Geometrical differences refer to the location of the boundary of a single region or of a set of regions, independent of their representation and of the structure of the boundaries.Identification of differences requires a matching of two data sets R₁ and R₂, which is done here by weighted topological relationships. For the identification of topological differences between two sets R₁ and R₂ of regions, we use the two region adjacency graphs (RAGs). For an identification of geometrical differences, we use the zone skeleton between two matched subsets r_p1 and r_q2 of the given sets. The zone skeleton is labeled with the local distances of the corresponding boundaries of the subsets; especially, we investigate its density function. An example based on two real data sets of acquired ground plans of buildings, shows the feasibility of the approach.