结合邻域聚类分割的高光谱图像异常检测支持向量数据描述方法

支持向量数据描述方法在高光谱图像小异常目标检测中具有较好的检测性能,但是待检异常的几何形状受到约束和背景的选择具有盲目性影响检测效果,且检测需要对整幅图像进行遍历导致计算量大。提出邻域聚类分割和支持向量数据描述相结合的异常检测方法,首先利用邻域聚类方法分割图像,将几何尺寸小的分割块作为潜在异常目标;其次选择与潜在异常的形状和大小相适应的背景窗进行背景像元收集;最后采用SVDD方法从潜在异常中快速且准确地检测出异常目标。对HYMAP图像的实验结果表明,该算法提高了复杂地物背景下异常的检测性能,降低了SVDD用于高光谱图像异常检测的计算量。     

A nonlinear background removal method for seismo-ionospheric anomaly analysis under a complex solar activity scenario: A case study of the M9.0 Tohoku earthquake

A precise determination of ionospheric total electron content (TEC) anomaly variations that are likely associated with large earthquakes as observed by global positioning system (GPS) requires the elimination of the ionospheric effect from irregular solar electromagnetic radiation. In particular, revealing the seismo-ionospheric anomalies when earthquakes occurred during periods of high solar activity is of utmost importance. To overcome this constraint, a multiresolution time series processing technique based on wavelet transform applicable to global ionosphere map (GIM) TEC data was used to remove the nonlinear effect from solar radiation for the earthquake that struck Tohoku, Japan, on 11 March, 2011. As a result, it was found that the extracted TEC have a good correlation with the measured solar extreme ultraviolet flux in 26–34 nm (EUV_26–34) and the 10.7 cm solar radio flux (F_10.7). After removing the influence of solar radiation origin in GIM TEC, the analysis results show that the TEC around the forthcoming epicenter and its conjugate were significantly enhanced in the afternoon period of 8 March 2011, 3 days before the earthquake. The spatial distributions of the TEC anomalous and extreme enhancements indicate that the earthquake preparation process had brought with a TEC anomaly area of size approximately 1650 and 5700 km in the latitudinal and longitudinal directions, respectively.     

First results from the LUCID-Timepix spacecraft payload onboard the TechDemoSat-1 satellite in Low Earth Orbit

The Langton Ultimate Cosmic ray Intensity Detector (LUCID) is a payload onboard the satellite TechDemoSat-1, used to study the radiation environment in Low Earth Orbit ($～$635?km). LUCID operated from 2014 to 2017, collecting over 2.1 million frames of radiation data from its five Timepix detectors on board. LUCID is one of the first uses of the Timepix detector technology in open space, with the data providing useful insight into the performance of this technology in new environments. It provides high-sensitivity imaging measurements of the mixed radiation field, with a wide dynamic range in terms of spectral response, particle type and direction. The data has been analysed using computing resources provided by GridPP, with a new machine learning algorithm that uses the Tensorflow framework. This algorithm provides a new approach to processing Medipix data, using a training set of human labelled tracks, providing greater particle classification accuracy than other algorithms. For managing the LUCID data, we have developed an online platform called Timepix Analysis Platform at School (TAPAS). This provides a swift and simple way for users to analyse data that they collect using Timepix detectors from both LUCID and other experiments. We also present some possible future uses of the LUCID data and Medipix detectors in space.     

Investigation of GPS-TEC measurements using ANN method indicating seismo-ionospheric anomalies around the time of the Chile (Mw = 8.2) earthquake of 01 April 2014

In 94 km NW of Iquique in Chile (19.610°S, 70.776°W) a powerful earthquake of M_w = 8.2 took place at 23:46:47 UTC (20:46:47 LT) on April 01, 2014. Using GPS-TEC (Total Electron Content) measurements, potential unusual variations around the time and location of the Chile earthquake have been detected based on the median and Artificial Neural Network (ANN) methods. The indices D_st, K_p, A_p and F10.7 were used to distinguish pre-earthquake anomalies from the other anomalies related to the solar-geomagnetic activities. Using the median method, striking anomalies in time series of TEC data are observed 4 days before the earthquake at 14:00 and 16:00 UTC. The ANN method detected a number of anomalies, 4 (02:00 and 16:00 UTC) and 13 (24:00 UTC) days preceding the earthquake. The results indicate that the ANN method due to its capability of non linear learning is quite promising and deserves serious attention as a robust predictor tool for seismo-ionospheric anomalies detection.     

地磁暴期间中国中低纬电离层的CIT研究

利用电离层层析成像技术(Computerized Ionospheric Tomography, CIT)处理115°E子午圈附近6个台站的GPS观测数据, 分析了2004年11月地磁暴期间中国中低纬电离层的响应情况. 结果表明, 电离层呈正相扰动, 且不同高度上的响应不同, 800 km以下电子密度有不同程度的增加, 且在峰值高度附近增幅最大, 800 km以上地磁暴的影响并不显著; 伴随地磁能量的注入, 赤道异常峰极向扩展; 随磁扰强度的降低, 电子密度也逐渐恢复至平静水平. 这些结果与以往的理论和观测结果一致, 初步估计扰动是由热层暴环流引起的, 并受到赤道异常峰移动的影响.      

Variations of ionospheric total electron content before three strong earthquakes in the Qinghai-Tibet region

We investigate the ionospheric total electron content (TEC) anomalies occurred in the Qinghai-Tibet region before three large earthquakes (M > 7.0). The temporal and spatial TEC variations were used to detect the ionospheric possible precursors of these earthquakes. We identified two TEC enhancements in the afternoon local time 9 days and 2–3 days before each earthquake, between which a TEC decrement occurred 3–6 days before earthquakes. These anomalies happened in the area of about 30° in latitude and the maximum is localized equatorward from the epicenters. These TEC anomalies can be found in all three earthquakes regardless the geomagnetic conditions. The features of these anomalies have something in common and may have differences from those caused by geomagnetic storms. Our results suggest that these ionospheric TEC perturbations may be precursors of the large earthquakes.     

Ground and satellite-based observations of ionospheric plasma bubbles and blobs at 5.65° latitude in the Brazilian sector

This investigation uses simultaneous observations from all-sky imager system and an ionosonde collocated at Araguatins (5.65° S, 48.07° W and dip-latitude of 4.17° S), a near-equatorial region in Brazil. These simultaneous observations were used to investigate the occurrence of plasma bubbles and blobs in the field of the imaging system and their association with atypical range Spread-F signature in ionograms. Also, in-situ observation of plasma density from Swarm satellites were used to support the ground-based observations. Using a few cases, a methodology will be established to identify in the plasma blobs (atypical ESF) in the ionograms when there is the simultaneous observation of plasma bubbles and blobs in the field of view of the ionosonde. For this purpose, simultaneous sequence of OI 630.0 nm nightglow images and ionograms are presented for different case studies; 1. when there is the absence of a plasma bubble or blob, 2. when there is only the occurrence of plasma bubbles and 3. when there is the occurrence of plasma bubbles and blobs, in order to compare traces in the ionogram in all these case studies. With these we can cover all kinds of signatures in the ionograms corresponding to no irregularities, plasma bubbles only and plasma bubbles-blobs. These OI 630.0 nm nightglow and ionograms recorded simultaneously make it possible to establish a novel methodology to recognize in ionograms cases when there is the occurrence of Spread-F signature associated with bubble-blob in the FOV of the ionosonde.     

Extrasolar solar-sail trajectories and dark matter

Hyper-thin, high-speed solar-photon sail space probes exploring the Sun?s Oort comet cloud could also be used to set an upper bound to the concentration of WIMPS (weakly interacting massive particles), one of the suggested (but unconfirmed) forms of dark matter within the vicinity of the solar system. Newton?s Shell Theorem would be applied to determine variations in apparent solar mass as the probe moves further out from the Sun. Application of this technique to the trajectories of Pioneer 10/11 reveals that the upper limit to WIMP concentration within ~60 AU of the Sun is ~0.2 Earth masses, as revealed in studies of the Pioneer Anomaly. If the published accuracy of the Pioneer acceleration measurements can be increased by an order of magnitude, probe trajectory measurements out to ~10,000 AU may confirm or falsify the hypothesis that WIMP mass within the solar vicinity is ~3X star mass. It is shown that a space-manufactured ~40-nm thick beryllium hollow-body solar sail deployed from a ~0.07 AU perihelion is a candidate spacecraft for such a mission. Possible science-team organization strategy for a ~100-year mission to ~10,000 AU is discussed.     

Aircraft engine fault detection based on grouped convolutional denoising autoencoders

Many existing aircraft engine fault detection methods are highly dependent on performance deviation data that are provided by the original equipment manufacturer. To improve the independent engine fault detection ability, Aircraft Communications Addressing and Reporting System (ACARS) data can be used. However, owing to the characteristics of high dimension, complex correlations between parameters, and large noise content, it is difficult for existing methods to detect faults effectively by using ACARS data. To solve this problem, a novel engine fault detection method based on original ACARS data is proposed. First, inspired by computer vision methods, all variables were divided into separated groups according to their correlations. Then, an improved convolutional denoising autoencoder was used to extract the features of each group. Finally, all of the extracted features were fused to form feature vectors. Thereby, fault samples could be identified based on these feature vectors. Experiments were conducted to validate the effectiveness and efficiency of our method and other competing methods by considering real ACARS data as the data source. The results reveal the good performance of our method with regard to comprehensive fault detection and robustness. Additionally, the computational and time costs of our method are shown to be relatively low.     

An Adaptive Network-based Fuzzy Inference System for the detection of thermal and TEC anomalies around the time of the Varzeghan,Iran, (Mw = 6.4) earthquake of 11 August 2012

Anomaly detection is extremely important for forecasting the date, location and magnitude of an impending earthquake. In this paper, an Adaptive Network-based Fuzzy Inference System (ANFIS) has been proposed to detect the thermal and Total Electron Content (TEC) anomalies around the time of the Varzeghan, Iran, (M_w = 6.4) earthquake jolted in 11 August 2012 NW Iran. ANFIS is the famous hybrid neuro-fuzzy network for modeling the non-linear complex systems. In this study, also the detected thermal and TEC anomalies using the proposed method are compared to the results dealing with the observed anomalies by applying the classical and intelligent methods including Interquartile, Auto-Regressive Integrated Moving Average (ARIMA), Artificial Neural Network (ANN) and Support Vector Machine (SVM) methods. The duration of the dataset which is comprised from Aqua-MODIS Land Surface Temperature (LST) night-time snapshot images and also Global Ionospheric Maps (GIM), is 62 days. It can be shown that, if the difference between the predicted value using the ANFIS method and the observed value, exceeds the pre-defined threshold value, then the observed precursor value in the absence of non seismic effective parameters could be regarded as precursory anomaly. For two precursors of LST and TEC, the ANFIS method shows very good agreement with the other implemented classical and intelligent methods and this indicates that ANFIS is capable of detecting earthquake anomalies. The applied methods detected anomalous occurrences 1 and 2 days before the earthquake. This paper indicates that the detection of the thermal and TEC anomalies derive their credibility from the overall efficiencies and potentialities of the five integrated methods.