Spatiotemporal filtering of regional GNSS network’s position time series with missing data using principle component analysis

The existing spatiotemporal analysis methods suppose that the involved time series are complete and have the same data interval. However missing data inevitably occur in the position time series of Global Navigation Satellite Systems networks for many reasons. In this paper, we develop a modified principal component analysis to extract the Common Mode Error (CME) from the incomplete position time series. The principle of the proposed method is that a time series can be reproduced from its principle components. The method is equivalent to the method of Dong et al. (J Geophys Res 111:3405–3421, 2006) in case of no missing data in the time series and to the extended ‘stacking’ approach under the assumption of a uniformly spatial response. The new method is first applied to extract the CME from the position time series of the Crustal Movement Observation Network of China (CMONOC) over the period of 1999–2009 where the missing data occur in all stations with the different gaps. The results show that the CMEs are significant in CMONOC. The size of the first principle components for the North, East and Up coordinates are as large as 40, 41 and 37 % of total principle components and their spatial responses are not uniform. The minimum amplitudes of the first eigenvectors are only 41, 15 and 29 % for the North, East and Up coordinate components, respectively. The extracted CMEs of our method are close to the data filling method, and the Root Mean Squared error (RMS) values computed from the differences of maximum CMEs between two methods are only 0.31, 0.52 and 1.55 mm for North, East and Up coordinates, respectively. The RMS of the position time series is greatly reduced after filtering out the CMEs. The accuracies of the reconstructed missing data using the two methods are also comparable. To further comprehensively test the efficiency of our method, the repeated experiments are then carried out by randomly deleting different percentages of data at some stations. The results show that the CMEs can be extracted with high accuracy at the non missing-data epochs. And at the missing-data epochs, the accuracy of extracted CMEs has a strong dependence on the number of stations with missing data.     

Comparative analysis of different environmental loading methods and their impacts on the GPS height time series

Three different environmental loading methods are used to estimate surface displacements and correct non-linear variations in a set of GPS weekly height time series. Loading data are provided by (1) Global Geophysical Fluid Center (GGFC), (2) Loading Model of Quasi-Observation Combination Analysis software (QLM) and (3) our own daily loading time series (we call it OMD for optimum model data). We find that OMD has the smallest scatter in height across the selected 233 globally distributed GPS reference stations, GGFC has the next smallest variability, and QLM has the largest scatter. By removing the load-induced height changes from the GPS height time series, we are able to reduce the scatter on 74, 64 and 41 % of the stations using the OMD models, the GGFC model and QLM model respectively. We demonstrate that the discrepancy between the center of earth (CE) and the center of figure (CF) reference frames can be ignored. The most important differences between the predicted models are caused by (1) differences in the hydrology data from the National Center for Atmospheric Research (NCEP) vs. those from the Global Land Data Assimilation System (GLDAS), (2) grid interpolation, and (3) whether the topographic effect is removed or not. Both QLM and GGFC are extremely convenient tools for non-specialists to use to calculate loading effects. Due to the limitation of NCEP reanalysis hydrology data compared with the GLDAS model, the GGFC dataset is much more suitable than QLM for applying environmental loading corrections to GPS height time series. However, loading results for Greenland from GGFC should be discarded since hydrology data from GLDAS in this region are not accurate. The QLM model is equivalent to OMD in Greenland and, hence, could be used as a complement to the GGFC product to model the load in this region. We find that the predicted loading from all three models cannot reduce the scatter of the height coordinate for some stations in Europe.     

Grassland habitat mapping by intra-annual time series analysis – Comparison of RapidEye and TerraSAR-X satellite data

Remote sensing concepts are needed to monitor open landscape habitats for environmental change and biodiversity loss. However, existing operational approaches are limited to the monitoring of European dry heaths only. They need to be extended to further habitats. Thus far, reported studies lack the exploitation of intra-annual time series of high spatial resolution data to take advantage of the vegetations’ phenological differences. In this study, we investigated the usefulness of such data to classify grassland habitats in a nature reserve area in northeastern Germany. Intra-annual time series of 21 observations were used, acquired by a multi-spectral (RapidEye) and a synthetic aperture radar (TerraSAR-X) satellite system, to differentiate seven grassland classes using a Support Vector Machine classifier. The classification accuracy was evaluated and compared with respect to the sensor type – multi-spectral or radar – and the number of acquisitions needed. Our results showed that very dense time series allowed for very high accuracy classifications (>90%) of small scale vegetation types. The classification for TerraSAR-X obtained similar accuracy as compared to RapidEye although distinctly more acquisitions were needed. This study introduces a new approach to enable the monitoring of small-scale grassland habitats and gives an estimate of the amount of data required for operational surveys.     

Updating Landsat time series of surface-reflectance composites and forest change products with new observations

The use of time series satellite data allows for the temporally dense, systematic, transparent, and synoptic capture of land dynamics over time. Subsequent to the opening of the Landsat archive, several time series approaches for characterizing landscape change have been developed, often representing a particular analytical time window. The information richness and widespread utility of these time series data have created a need to maintain the currency of time series information via the addition of new data, as it becomes available. When an existing time series is temporally extended, it is critical that previously generated change information remains consistent, thereby not altering reported change statistics or science outcomes based on that change information. In this research, we investigate the impacts and implications of adding additional years to an existing 29-year annual Landsat time series for forest change. To do so, we undertook a spatially explicit comparison of the 29 overlapping years of a time series representing 1984–2012, with a time series representing 1984–2016. Surface reflectance values, and presence, year, and type of change were compared. We found that the addition of years to extend the time series had minimal effect on the annual surface reflectance composites, with slight band-specific differences (r ≥ 0.1) in the final years of the original time series being updated. The area of stand replacing disturbances and determination of change year are virtually unchanged for the overlapping period between the two time-series products. Over the overlapping temporal period (1984–2012), the total area of change differs by 0.53%, equating to an annual difference in change area of 0.019%. Overall, the spatial and temporal agreement of the changes detected by both time series was 96%. Further, our findings suggest that the entire pre-existing historic time series does not need to be re-processed during the update process. Critically, given the time series change detection and update approach followed here, science outcomes or reports representing one temporal epoch can be considered stable and will not be altered when a time series is updated with newly available data.     

地球引力场与地球自转变化

本根据现代空间技术测定地球引力场变化的进展，提出了用实测的地球自转参数和实测的低阶重力场变化结合的方法以研究地球角动量变化。其突出的优点是可以使引起地球角动量变化的质量项和速度场项解耦，使原来较复杂的问题简化。作为本提出的方法的实例，我们用Lageos-1和Lageos-2两颗卫星的SLR资料求解ΔC20，计算出ΔLOD序列，与（ΔLOD－Wind)残差序列相比，有较好的一致性，显示出本方法的有效性。     

混合GM(1,1)模型预报季节性时间序列精度的方法探讨

灰色系统已被成功用于工程、经济、物理控制等许多领域。然而在预报具有季节性的时间序列时，直接应用GM(1，1)灰色模型往往精度不高。因为GM(1，1)灰色模型只能反映时间序列的总体变化趋势，不能很好反映其季节性波动变化的具体特征。因此，作者提出运用“滑动平均去季节性波动”与GM(1，1)混合建模的方法预报具有季节特征的时间序列。并以水文地质系统中地下水位预报和安装在混凝土中的测缝计测得的建筑物形变量为一组时间序列，基于均方差、平均绝对误差和平均绝对百分误差三个精度准则，比较了此方法与其它灰色建模法的结果。结果表明，此方法不仅能反映时间序列的总体变化趋势，而且能客观反映其波动变化的具体特征，有效提高了预报精度，减少了建模的复杂度。     

EM算法的时序模型在沉降数据处理中的应用

针对时间序列分析对监测数据中出现的不完全数据(部分数据缺失)无法进行精确建模的问题,该文引入期望极大算法(EM算法),提出EM算法与时间序列分析的组合算法模型。运用EM算法的时间序列分析组合算法模型可以对沉降过程中遇到的不完全沉降数据进行建模分析,该组合算法模型可以对不完全沉降数据进行较为精确的建模,并对后期沉降数据进行较为精确的预测。将某地铁基坑点沉降数据作为实验数据,EM算法的时间序列分析的建模结果表明:所提出的组合算法模型可以对不完全沉降数据进行建模分析,绝对误差为0.23mm,建模精度较高。     

Trajectory models and reference frames for crustal motion geodesy

We sketch the evolution of station trajectory models used in crustal motion geodesy over the last several decades, and describe some recent generalizations of these models that allow geodesists and geophysicists to parameterize accelerating patterns of displacement in general, and postseismic transient deformation in particular. Modern trajectory models are composed of three sub-models that represent secular trends, annual oscillations, and instantaneous jumps in coordinate time series. Traditionally the trend model invoked constant station velocity. This can be generalized by assuming that position is a polynomial function of time. The trajectory model can also be augmented as needed, by including one or more logarithmic transients in order to account for typical multi-year patterns of postseismic transient motion. Many geodetic and geophysical research groups are using general classes of trajectory model to characterize their crustal displacement time series, but few if any of them are using these trajectory models to define and realize the terrestrial reference frames (RFs) in which their time series are expressed. We describe a global GPS reanalysis program in which we use two general classes of trajectory model, tuned on a station by station basis. We define the network trajectory model as the set of station trajectory models encompassing every station in the network. We use the network trajectory model from the each global analysis to assign prior position estimates for the next round of GPS data processing. We allow our daily orbital solutions to relax so as to maintain their consistency with the network polyhedron. After several iterations we produce GPS time series expressed in a RF similar to, but not identical with   ITRF2008. We find that each iteration produces an improvement in the daily repeatability of our global time series and in the predictive power of our trajectory models.     

Combination of multi-mission altimetry data along the Mekong River with spatio-temporal kriging

River water-level time series at fixed geographical locations, so-called virtual stations, have been computed from single altimeter crossings for many years. Their temporal resolution is limited by the repeat cycle of the individual altimetry missions. The combination of all altimetry measurements along a river enables computing a water-level time series with improved temporal and spatial resolutions. This study uses the geostatistical method of spatio-temporal ordinary kriging to link multi-mission altimetry data along the Mekong River. The required covariance models reflecting the water flow are estimated based on empirical covariance values between altimetry observations at various locations. In this study, two covariance models are developed and tested in the case of the Mekong River: a stationary and a non-stationary covariance model. The proposed approach predicts water-level time series at different locations along the Mekong River with a temporal resolution of 5 days. Validation is performed against in situ data from four gauging stations, yielding RMS differences between 0.82 and 1.29 m and squared correlation coefficients between 0.89 and 0.94. Both models produce comparable results when used for combining data from Envisat, Jason-1, and SARAL for the time period between 2002 and 2015. The quality of the predicted time series turns out to be robust against a possibly decreasing availability of altimetry mission data. This demonstrates that our method is able to close the data gap between the end of the Envisat and the launch of the SARAL mission with interpolated time series.     

Vertical deformations from homogeneously processed GRACE and global GPS long-term series

Temporal variations in the geographic distribution of surface mass cause surface displacements. Surface displacements derived from GRACE gravity field coefficient time series also should be observed in GPS coordinate time series, if both time series are sufficiently free of systematic errors. A successful validation can be an important contribution to climate change research, as the biggest contributors to mass variability in the system Earth include the movement of oceanic, atmospheric, and continental water and ice. In our analysis, we find that if the signals are larger than their precision, both geodetic sensor systems see common signals for almost all the 115 stations surveyed. Almost 80% of the stations have their signal WRMS decreased, when we subtract monthly GRACE surface displacements from those observed by GPS data. Almost all other stations are on ocean islands or small peninsulas, where the physically expected loading signals are very small. For a fair comparison, the data (79 months from September 2002 to April 2009) had to be treated appropriately: the GPS data were completely reprocessed with state-of-the-art models. We used an objective cluster analysis to identify and eliminate stations, where local effects or technical artifacts dominated the signals. In addition, it was necessary for both sets of results to be expressed in equivalent reference frames, meaning that net translations between the GPS and GRACE data sets had to be treated adequately. These data sets are then compared and statistically analyzed: we determine the stability (precision) of GRACE-derived, monthly vertical deformation data to be ~1.2 mm, using the data from three GRACE processing centers. We statistically analyze the mean annual signals, computed from the GPS and GRACE series. There is a detailed discussion of the results for five overall representative stations, in order to help the reader to link the displayed criteria of similarity to real data. A series of tests were performed with the goal of explaining the remaining GPS–GRACE residuals.     

Mapping Cropping Practices Using MODIS Time Series: Harnessing the Data Explosion

The MODIS (Moderate Resolution Imaging Spectroradiometer) 250m EVI dataset provides a valuable ongoing means of characterising and monitoring changes in land use and resource condition. However the multiple factors that influence a time series of greenness data make the data difficult to analyse and interpret. Without prior knowledge, underlying models for time series in a given remote sensing image are often heterogeneous. So while conventional time series analysis methods such as wavelet transform and Fourier analysis may work well for part of the image, these models are either invalid or must be substantially re-parameterised for other parts of the image. To overcome these challenges we propose a new approach to distil information from earth observation time series data. The characteristics of a remote sensing time series are represented by a set of statistics (which we call coefficients) selected to correspond to the dynamics of a natural system. To ensure the coefficients are robust and generic, statistics are calculated independently by applying statistical models with less complexity on shorter segments within the time series. An International Standards Organization (ISO) Land Cover classification (Jansen 2000) was generated for cropping regions in the Gwydir and Namoi catchments, in Australia. Areas identified in the classification as irrigated and rain fed cropping were analysed using a tailored time series analysis tool. The crop analysis tool identifies time series features such as the number and duration of fallow periods, crop timing, presence/absence of a crop during a year for a specific growing season. This information is combined with paddock boundaries derived from Landsat imagery to provide detailed year-by-year insight into cropping practices in the Gwydir and Namoi catchments.     

A time series for monitoring vegetation activity and phenology at 10-daily time steps covering large parts of South America

Abstract

It is widely accepted that natural resources should only be sustainably exploited and utilized to effectively preserve our planet for future generations. To better manage the natural resources, and to better understand the closely linked Earth systems, the concept of Digital Earth has been strongly promoted since US Vice President Al Gore's speech in 1998. One core element of Digital Earth is the use and integration of remote sensing data. Only satellite imagery can cover the entire globe repeatedly at a sufficient high-spatial resolution to map changes in land cover and land use, but also to detect more subtle changes related for instance to climate change. To uncover global change effects on vegetation activity and phenology, it is important to establish high quality time series characterizing the past situation against which the current state can be compared. With the present study we describe a time series of vegetation activity at 10-daily time steps between 1998 and 2008 covering large parts of South America at 1 km spatial resolution. Particular emphasis was put on noise removal. Only carefully filtered time series of vegetation indices can be used as a benchmark and for studying vegetation dynamics at a continental scale. Without temporal smoothing, subtle spatio-temporal patterns in vegetation composition, density and phenology would be hidden by atmospheric noise and undetected clouds. Such noise is immanent in data that have undergone solely a maximum value compositing. Within the present study, the Whittaker smoother (WS) was applied to a SPOT VGT time series. The WS balances the fidelity to the observations with the roughness of the smoothed curve. The algorithm is extremely fast, gives continuous control over smoothness with only one parameter, and interpolates automatically. The filtering efficiently removed the negatively biased noise present in the original data, while preserving the overall shape of the curves showing vegetation growth and development. Geostatistical variogram analysis revealed a significantly increased signal-to-noise ratio compared to the raw data. Analysis of the data also revealed spatially consistent key phenological markers. Extracted seasonality parameters followed a clear meridional trend. Compared to the unfiltered data, the filtered time series increased the separability of various land cover classes. It is thus expected that the data set holds great potential for environmental and vegetation related studies within the frame of Digital Earth.     

遥感时间序列影像变化检测研究进展

同一区域、不同时期大量历史数据的积累,以及同一区域能够方便地获取高时间分辨率遥感数据,使遥感时间序列影像变化检测成为近年来遥感技术与应用的研究热点。本文系统总结和评述了当前遥感时间序列影像变化检测的相关研究进展和应用状况,在阐明遥感时间序列分析的意义,以及时间序列影像在变化检测中的优势的基础上,从非遥感领域时间序列变化检测方法出发,针对遥感时间序列影像变化检测的需求,明确和归纳了遥感时间序列变化检测的问题与类型,并对当前最新研究进行了综述,总结了各种方法的优点与不足,重点介绍了基于经验模态分解的遥感时间序列影像异常信息检测方法和基于隐马尔可夫模型的土地利用/覆盖变化检测方法,以期能够为相关研究提供参考。最后总结了该研究领域的发展趋势和存在问题,并对今后的研究工作和未来发展方向进行了展望。     

基于时间序列相似度的城市功能区识别研究

随着城市化的快速发展,城市空间结构愈发复杂,城市功能区的快速有效识别对资源的有效配置和城市规划具有重要意义.传统的功能区识别缺乏对居民这一城市空间活动主体的动态表征,而长时间序列的出租车数据能动态表征居民出行行为,进而反映城市空间结构.动态时间扭曲(DTW)距离比传统的欧氏距离更能有效挖掘高维数据,泛化后的LB_Keo...     

基于 DTW 距离的时序 NDVI 数据植被信息提取--以秦巴山区为例

秦巴山区是我国重要的生态屏障,对该区的植被信息提取开展研究,可为区内生态服务功能及自然资源开发利用提供基础数据。通过加窗处理改进DTW距离相似性算法,结合临近度模糊分类方法对2005—2014年的MODIS NDVI时序数据进行植被信息提取。首先利用S-G滤波对MODIS NDVI时序数据进行重建;再利用2013年的采样数据构建各类植被的标准NDVI时序曲线,逐像元计算与标准NDVI时序曲线的加窗DTW距离,利用临近度模糊分类实现植被信息提取;最后验证提取精度。结果表明,算法具有较高的运行效率,可避免错误匹配,以较高的精度(总体精度83.8%,kappa系数0.77)实现长时间序列的植被信息提取。     

融合高时空分辨率数据估算植被净初级生产力

植被净初级生产力NPP(Net Primary Production)遥感估算与分析,有赖于高时空分辨率的遥感数据,但目前中高分辨率的遥感数据受卫星回访周期及天气的影响,在中国南方地区难以获取连续时间序列的数据,从而影响了高精度的区域植被净初级生产力的遥感估算。为此,提出一种基于多源遥感数据时空融合技术与CASA模型估算高时空分辨率NPP的方法。首先,利用多源遥感数据,即Landsat8 OLI数据与MODIS13Q1数据,采用遥感数据时空融合方法,获得了时间序列的Landsat8 OLI融合数据;然后,基于Landsat8 OLI时空融合数据,并采用CASA模型,以长株潭城市群核心区为例,进行区域植被NPP的遥感估算。研究结果表明,基于时间序列Landsat融合数据估算的30m分辨率的NPP具有良好的空间细节信息,且估算值与实测值的相关系数达0.825,与实测NPP数据保持了较好的一致性。     

含缺值GNSS基准站坐标序列的非插值小波分析与信号提取

受多种因素影响,GNSS基准站坐标序列通常都含有缺值,传统小波分析需要对缺值数据进行内插或补零处理。本文基于小波系数与时间序列观测数据的重构关系,提出了一种非插值的二进小波变换的最小范数解法,导出了相应的计算式,并严格证明了传统的补零处理算法与本文的最小范数解法等价。最后利用中国地壳运动观测网络一期27个基准站实测数据以及模拟数据进行了验证分析。结果表明,本文的非插值算法与插值算法提取的信号差异较小,27个基准站坐标序列的平均残差中误差仅相差2.01%(North),0.54%(East)和1.26%(Up),两种算法提取的信号之差与信号平均方差比仅相差1.16%(North),0.54%(East)和1.62%(Up)。     

Mapping wetland using the object-based stacked generalization method based on multi-temporal optical and SAR data

Wetland ecosystems have experienced dramatic challenges in the past few decades due to natural and human factors. Wetland maps are essential for the conservation and management of terrestrial ecosystems. This study is to obtain an accurate wetland map using an object-based stacked generalization (Stacking) method on the basis of multi-temporal Sentinel-1 and Sentinel-2 data. Firstly, the Robust Adaptive Spatial Temporal Fusion Model (RASTFM) is used to get time series Sentinel-2 NDVI, from which the vegetation phenology variables are derived by the threshold method. Subsequently, both vertical transmit-vertical receive (VV) and vertical transmit-horizontal receive (VH) polarization backscatters (σ0 VV, σ0 VH) are obtained using the time series Sentinel-1 images. Speckle noise inherent in SAR data, resulting in over-segmentation or under-segmentation, can affect image segmentation and degrade the accuracies of wetland classification. Therefore, we segment Sentinel-2 multispectral images to delineate meaningful objects in this study. Then, in order to reduce data redundancy and computation time, we analyze the optimal feature combination using the Sentinel-2 multispectral images, Sentinel-2 NDVI time series, phenological variables and other vegetation index derived from Sentinel-2 multispectral images, as well as time series Sentinel-1 backscatters at the object level. Finally, the stacked generalization algorithm is utilized to extract the wetland information based on the optimal feature combination in the Dongting Lake wetland. The overall accuracy and Kappa coefficient of the object-based stacked generalization method are 92.46% and 0.92, which are 3.88% and 0.04 higher than that using the pixel-based method. Moreover, the object-based stacked generalization algorithm is superior to single classifiers in classifying vegetation of high heterogeneity areas.     

黄土台塬滑坡变形的时序InSAR监测分析

黄土台塬由于经常性的农业灌溉容易造成边缘区域滑坡发育。因此,需要利用有效手段对这些潜在的滑坡隐患进行早期识别与监测。利用时间序列合成孔径雷达干涉测量（interferometric synthetic aperture radar,InSAR）技术对2016-01至2018-08期间获取的升降轨Sentinel-1数据集进行分析,获取了甘肃永靖黑方台典型台塬地区的滑坡隐患分布情况。将InSAR结果与GPS观测资料进行对比,验证了时序InSAR处理方法的有效性。对该地区滑坡的历史变形分析表明,持续的农业灌溉引起的地下水位抬升是台塬边缘坡体失稳的主要诱因。同时,InSAR时序分析发现,研究区域内的跨黄河大桥受季节更替和温度波动的影响,存在周期性变形现象。实验结果证明了时序InSAR方法在地表变形监测中的有效性,可在黄土滑坡识别与监测防治中发挥重要作用。     

NDVI dynamics as reflected in climatic variables: spatial and temporal trends – a case study of Hungary

Understanding climate change and revealing its future paths on a local level is a great challenge for the future. Beside the expanding sets of available climatic data, satellite images provide a valuable source of information. In our study we aimed to reveal whether satellite data are an appropriate way to identify global trends, given their shorter available time range. We used the CARPATCLIM (CC) database (1961–2010) and the MODIS NDVI images (2000–2016) and evaluated the time period covered by both (2000–2010). We performed a regression analysis between the NDVI and CC variables, and a time series analysis for the 1961–2008 and 2000–2008 periods at all data points. The results justified the belief that maximum temperature (TMAX), potential evapotranspiration and aridity all have a strong correlation with the NDVI; furthermore, the short period trend of TMAX can be described with a functional connection with its long period trend. Consequently, TMAX is an appropriate tool as an explanatory variable for NDVI spatial and temporal variance. Spatial pattern analysis revealed that with regression coefficients, macro-regions reflected topography (plains, hills and mountains), while in the case of time series regression slopes, it justified a decreasing trend from western areas (Transdanubia) to eastern ones (The Great Hungarian Plain). This is an important consideration for future agricultural and land use planning; i.e. that western areas have to allow for greater effects of climate change.