Figure 2 - uploaded by Jian Chen
Content may be subject to copyright.
Outliers detected between August 25 and 30, 2005 from NDBC weather buoy data in the Gulf of Mexico: A) Strong outliers, B) Weak outliers
Source publication
Data points that exhibit abnormal behavior, either spatially, temporally, or both, are considered spatio-temporal outliers. Spatio-Temporal outlier detection is important for the discovery of exceptional events due to the rapidly increasing amount of spatio-temporal data available, and the need to understand such data. A tropical cyclone system or...
Contexts in source publication
Context 1
... outlier detection results for Hurricane Katrina can be seen in Figure 2. Strong outliers were detected at two buoys at multiple times (see Table 4 and Figure 2A for details) and each data object Figure 2A). ...
Context 2
... outlier detection results for Hurricane Katrina can be seen in Figure 2. Strong outliers were detected at two buoys at multiple times (see Table 4 and Figure 2A for details) and each data object Figure 2A). The authors were able to detect four consecutive readings from 10:00 a.m. to 13:00 p.m. on August 29, 2015 at buoy 42040 as shown in Table 4. ...
Context 3
... outlier detection results for Hurricane Katrina can be seen in Figure 2. Strong outliers were detected at two buoys at multiple times (see Table 4 and Figure 2A for details) and each data object Figure 2A). The authors were able to detect four consecutive readings from 10:00 a.m. to 13:00 p.m. on August 29, 2015 at buoy 42040 as shown in Table 4. ...
Context 4
... authors were able to detect four consecutive readings from 10:00 a.m. to 13:00 p.m. on August 29, 2015 at buoy 42040 as shown in Table 4. In Figure 2A, a 100-mile buffer along Hurricane Katrina's path was overlaid on top of the buoys to visualize the proximity of the hurricane system. If we take a close look at those buoys in the buffer zone after Katrina intensified to a Category 2 hurricane early on August 27, 2005, in Figure 2A (seven buoys in total in this case), we can see four out of the seven buoys were damaged (Buoy 42023, 42041, GISL1, and 42007). ...
Context 5
... Figure 2A, a 100-mile buffer along Hurricane Katrina's path was overlaid on top of the buoys to visualize the proximity of the hurricane system. If we take a close look at those buoys in the buffer zone after Katrina intensified to a Category 2 hurricane early on August 27, 2005, in Figure 2A (seven buoys in total in this case), we can see four out of the seven buoys were damaged (Buoy 42023, 42041, GISL1, and 42007). For those damaged buoys, we do not have any readings for this period of investiga- tion; therefore, they were excluded from this research. ...
Context 6
... those damaged sensors were able to survive and provide useful readings, most likely some outliers would be captured there as well. Among the remaining three live buoys out of the seven, two captured consecutive strong outliers as discussed earlier and one captured a series of weak outliers (Buoy 42001; see details in Table 5 and Figure 2B for weak outliers detected). All three live buoys out of the seven have some weak outliers being detected (hollow stars on Figure 2B) at moments when the hurricane intensities were not as high as those being detected as strong outliers at the same buoy. ...
Context 7
... the remaining three live buoys out of the seven, two captured consecutive strong outliers as discussed earlier and one captured a series of weak outliers (Buoy 42001; see details in Table 5 and Figure 2B for weak outliers detected). All three live buoys out of the seven have some weak outliers being detected (hollow stars on Figure 2B) at moments when the hurricane intensities were not as high as those being detected as strong outliers at the same buoy. It is not surprising that some of the other readings for Buoys 42003 and 42040 were also cap- tured as weak outliers based on the earlier discussion about their proximity to the hurricane sys- tem. ...
Context 8
... is not surprising that some of the other readings for Buoys 42003 and 42040 were also cap- tured as weak outliers based on the earlier discussion about their proximity to the hurricane sys- tem. The open ocean buoy 42001 also captured six sequential readings as outliers late on August 28 when Katrina reached its highest power as a Category 5 hurricane and the neighboring buoy 42041 was eventually damaged (see Figure 2B). From Table 5, one can observe two weak outliers out of the seventeen that might not have been relevant to Hurricane Katrina. ...
Context 9
... this time, there was a significant change in wind direction from 66 degrees in the previous hour to 263 degrees in this reading, and there was also an air temperature drop from 29.6 C to 27.5 C. It indi- cates that this reading is a legitimate outlier, although it might not relate to Hurricane Katrina because the air pressure was not low (1015 mb); however, it might be caused by local events, such as thunderstorms. The next weak outlier was detected on August 26 at station NPSF1, located on the west shore of the Florida peninsula (see Figure 2B). During this time, the eye of Hurricane Katrina was passing the area where this buoy is located. ...
Context 10
... the number of outliers in Rogers, Barbara, and Domeiconi (2009) as the benchmark, a cutoff LOF value at 2.39 produced 429 outliers in the 2005 whole year data set in this research. Among those 429 outliers in 2005, there are 11 strong outliers and 17 weak outliers detected (as discussed earlier and shown in Tables 4 and 5 and Figure 2) between August 25 and August 30,2005, when Hurricane Katrina wreaked havoc in the Gulf of Mexico (Knabb, Rhome, and Brown 2005). Rogers, Barbara, and Domeiconi (2009) also reported a strong outlier at buoy 40240 at 7:00 a.m. ...
Citations
... This paper provides a new method for anomaly detection in the specific case of spatially correlated functional data, where an anomaly is an observation that differs in both time and space from the rest of the data. Considerable literature exists on outlier detection methods for spatiotemporal data (see, for example, Rogers et al. 2009;Zhu et al. 2017;Chen et al. 2016;Wu and Xie 2016;Seidenari et al. 2010;Kut and Birant 2006;Cheng and Li 2006;Wu et al. 2010;Munawar et al. 2017;Duggimpudi et al. 2019). In particular, the authors in Wu et al. (2020) explore several categories of methods, highlighting their strengths and limitations. ...
This paper proposes a new methodology for detecting outliers in spatially correlated functional data. We use a Local Correlation Integral (LOCI) algorithm substituting the Euclidean distance calculation by the Hilbert space L2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L^2$$\end{document} distance weighted by the semivariogram, obtaining a weighted dissimilarity metric among the geo-referenced curves, which takes into account the spatial correlation structure. In addition, we also consider the distance proposed in Romano et al. (2020), which optimizes the distance calculation for spatially dependent functional data. A simulation study is conducted to evaluate the performance of the proposed methodology. We analyze the role of a threshold value appearing as an hyperparameter in our approach, and show that our distance weighted by the semivariogram is overall superior to the other types of distances considered in the study. We analyze time series of Land Surface Temperature (LST) data in the region of Andalusia (Spain), detecting significant outliers that would have not been detected using other procedures.
... The detection results can help identify suspicious activities of moving objects and therefore be used in many applications, such as severe weather prediction, security surveillance, and intelligent transportation and scientific studies [14][15][16]. As Chen et al. proposed in Literature [17], a tropical cyclone system or a hurricane can be considered an abnormal activity of the atmosphere system. Moreover, the detection and removal of trajectory outliers are significant for improving the efficiency of trajectory similarity clustering algorithm. ...
Trajectory outlier detection is one of the most popular trajectory data mining topics. It helps researchers obtain a lot of valuable information that can be used as important guidance in monitoring and forecasting. Existing methods have difficulty in detecting the outlying trajectories with continuous multi-segment exception. To address the problem, in this paper, we propose a novel trajectory outlier detection algorithm based on common slices sub-sequence (TODCSS). For each trajectory, the direction-code sequence is firstly calculated based on the direction of each trajectory segment. Secondly, the corresponding sequence consisting of trajectory slices is obtained by inflection point segmentation. And then, the common slices sub-sequences between two trajectories are found to measure their distance. Finally, the slice outliers and trajectory outliers are detected based on the new CSS distance calculation. Both the intuitive visualization presentation and the experimental results on real Atlantic hurricane dataset, real-life mobility trajectory dataset of taxis in San Francisco and synthetic labeled dataset show that the proposed TODCSS algorithm effectively detects slice and trajectory outliers, and improves accuracy and stability in trajectory outlier detection.
This study provides an extensive review of over 200 journal papers focusing on Machine Learning (ML) algorithms’ use for promoting a sustainable management of the marine and coastal environments. The research covers various facets of ML algorithms, including data preprocessing and handling, modeling algorithms for distinct phenomena, model evaluation, and use of dynamic and integrated models. Given that machine learning modeling relies on experience or trial-and-error, examining previous applications in marine and coastal modeling is proven to be beneficial. The performance of different ML methods used to predict wave heights was analyzed to ascertain which method was superior with various datasets. The analysis of these papers revealed that properly developed ML methods could successfully be applied to multiple aspects. Areas of application include data collection and analysis, pollutant and sediment transport, image processing and deep learning, and identification of potential regions for aquaculture and wave energy activities. Additionally, ML methods aid in structural design and optimization and in the prediction and classification of oceanographic parameters. However, despite their potential advantages, dynamic and integrated ML models remain underutilized in marine projects. This research provides insights into ML’s application and invites future investigations to exploit ML’s untapped potential in marine and coastal sustainability.
Cet ouvrage traite des avancées en analyse des séries chronologiques d’images de télédétection par apprentissages statistique, automatique et/ou profond. Il présente un éventail de modèles mathématiques, de méthodes d’extraction d’informations spatio-temporelles et d’applications en observation de la Terre.Détection de changements et analyse des séries temporelles d’images 1 est structuré selon les stratégies d’apprentissage supervisé, semi-supervisé et non supervisé sur les séries temporelles d’images optiques et radar à synthèse d’ouverture. L’ouvrage met en lumières des contributions sur la détection de changements multidates dans des données mono- et multi-modales, sur les classes de transitions et d’entités dynamiques, sur l’analyse des tendances et des fluctuations spatio-temporelles, ou encore sur le suivi et l’analyse prédictive d’évolution.De nombreuses applications illustrent les études sur l’évolution de la surface terrestre (cartographie d’exploitation des terres, analyse de l’état des glaciers, dynamique urbaine, évaluation de la neige, etc.) et sur les risques naturels (évaluation des inondations, activités volcaniques, surveillance des avalanches, etc.). Les orientations futures sont également exposées.
Cyclonic fields are violent atmospheric disturbances associated with swirling winds. These quasi-stochastic fields are the cause of many natural disasters, which motivate early detection methods as well as time monitoring of changing conditions. This chapter discusses some methods for time monitoring and prediction of cyclonic fields. The analysis of cyclonic fields requires the detection and localization of the cyclone eye and cyclone eyewall, as well as estimating cyclone intensity variations. The chapter addresses the intensity dynamics of the fractal cyclone parameter over time and presents the Autoregressive Fractionally Integrated Moving Average modeling and prediction of time evolution of the cyclone intensities. A cyclone field can be characterized by color and texture. The chapter proposes that statistical fractional random field modeling for a cyclone eyewall texture analysis and the retrieval of the cyclone eye position. The model derived is associated with cyclonic field strength due to a fractal intensity parameter of an isotropic observation model.
