Article
Engineering, Civil
V Agilan, N. Umamahesh, P. P. Mujumdar
Summary: The study aims to quantify the threshold uncertainty in peaks over threshold method and finds that under nonstationary conditions, the choice of threshold leads to higher uncertainty in extreme rainfall return levels.
JOURNAL OF HYDROLOGY
(2021)
Article
Construction & Building Technology
Junyong Zhou, Cuimin Hu, Zhixing Chen, Xiaoming Wang, Tao Wang
Summary: This study proposed using the peaks-over-threshold (POT) method to model extreme coincident lane load effects (LLEs) for multi-coefficient MLF calibration, which was found to be accurate and more effective than using the block maxima (BM) method with limited data.
ADVANCES IN STRUCTURAL ENGINEERING
(2021)
Article
Engineering, Civil
Xin Fang, Qi Wang, Jingchen Wang, Yunyun Xiang, Yifan Wu, Yifei Zhang
Summary: The study established nutrient criteria for inorganic nitrogen and reactive phosphate in Xiangshan Bay, providing a rational and reliable basis for subsequent monitoring and assessment.
JOURNAL OF HYDROLOGY
(2021)
Article
Computer Science, Artificial Intelligence
Mi Zhang, Daizong Ding, Xudong Pan, Min Yang
Summary: Time series prediction is widely used in many safety-critical scenarios, but conventional square loss fails to model extreme events. In this study, we propose a unified loss form called Generalized Extreme Value Loss (GEVL) to bridge the misalignment between the estimation and the ground-truth, and introduce three heavy-tailed kernels to enhance the modeling of extreme events.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Engineering, Environmental
Meghana Nagaraj, Roshan Srivastav
Summary: Interactions and feedback between the atmosphere, oceans, and land contribute to natural climatic variations and severe weather events. This study focuses on understanding the lagged effect of teleconnections and modeling extreme precipitation. A new approach called NL-GRNN is proposed, which uses Granger causality to select lagged climate indices and a non-stationary modeling framework. The results show that NL-GRNN based NSGEV models are computationally efficient and outperform linear and artificial neural network approaches. ENSO-related indices and northern hemisphere climate variability modes are found to have dominant influences on extreme precipitation in Chennai and San Diego, respectively.
STOCHASTIC ENVIRONMENTAL RESEARCH AND RISK ASSESSMENT
(2023)
Article
Engineering, Marine
Anna Maria Barlow, Ed Mackay, Emma Eastoe, Philip Jonathan
Summary: In this work, non-stationary models are proposed to predict the size and rate of occurrence of metocean extremes with respect to covariates such as direction and season. The models use piecewise-linear functions to describe the variation of model parameters with covariates, and the parameter roughness is optimized using cross-validation. Bootstrap resampling is used to quantify parameter uncertainty. These models are applied to estimate storm-peak significant wave height extremes with respect to direction and season for a specific location in the northern North Sea. The covariate representation based on a triangulation of the direction-season domain with six nodes shows good predictive performance. The penalised piecewise-linear framework provides a flexible representation of covariate effects at a reasonable computational cost.
Article
Engineering, Mechanical
Jinhua Li, Desen Zhu, Liyuan Cao, Chunxiang Li
Summary: This study proposes a novel simplified wavelet transform-based stationary transformation method (PNGEV) for extreme value analysis. The effectiveness and practicality of the PNGEV method are verified through comparisons with the traditional method (TNGEV).
PROBABILISTIC ENGINEERING MECHANICS
(2023)
Article
Computer Science, Interdisciplinary Applications
Benjamin D. Youngman
Summary: This article introduces the R package evgam, which provides functions for fitting extreme value distributions including the generalized extreme value and generalized Pareto distributions. The package also supports quantile regression using the asymmetric Laplace distribution, which is useful for estimating high thresholds. The main addition of package evgam is the ability to model extreme value distribution parameters using generalized additive models with objectively estimated smoothness using Laplace's method.
JOURNAL OF STATISTICAL SOFTWARE
(2022)
Article
Engineering, Civil
Chi Zhang, Xuezhi Gu, Lei Ye, Qian Xin, Xiaoyang Li, Hairong Zhang
Summary: Recent years have seen an increase in extreme precipitation events, challenging the assumption of stationarity in frequency analyses. However, there is a lack of research in China on the link between extreme precipitation and climate change. This study aims to identify the dominant climate indices and time scales affecting extreme precipitation in China and assess the rainstorm risk under non-stationary conditions. Through correlation analyses, non-stationary models are constructed and the optimal predictors for extreme precipitation are determined. The results show that ignoring non-stationarity leads to misperceptions of rainstorm risks, and the spatial distribution of design rainstorms differs significantly under non-stationary conditions.
WATER RESOURCES MANAGEMENT
(2023)
Article
Environmental Sciences
Michael Ring, Paola Elizabeth Rodriguez-Ocampo, Rodolfo Silva, Edgar Mendoza
Summary: This study combines numerical data with empirical data to adjust the return levels of ocean currents in the Mexican Caribbean region using extreme value analysis methods. The results show that the adjusted numerical model has an underestimation error of about 22% compared to the real data, but the results are consistent within the domain. This method can be used to estimate the return levels of ocean currents provided by HYCOM.
FRONTIERS IN MARINE SCIENCE
(2022)
Article
Computer Science, Interdisciplinary Applications
Marinela Capanu, Mihai Giurcanu, Colin B. Begg, Mithat Gonen
Summary: Introduces a novel variable selection method named OPT-STABS for low-dimensional generalized linear models. OPT-STABS repeatedly subsamples the data, minimizes AIC over a sequence of nested models for each subsample, and includes predictors selected in the minimum AIC model in a large fraction of the subsamples. It outperforms other methods in most settings and exhibits competitive performance in the rest.
COMPUTATIONAL STATISTICS & DATA ANALYSIS
(2023)
Article
Environmental Sciences
Eun-Young Lee, Kyung-Ae Park
Summary: Extreme value analysis (EVA) using satellite-observed sea surface temperature (SST) data was used to understand and predict long-term return extreme values in the East/Japan Sea (EJS). The peaks-over-threshold (POT) method showed better performance in deriving SST extremes. The calculated 100-year-return SST values were higher than the average value of satellite-observed SSTs over the past decades. The distribution of the SST extremes followed the known seasonal variation, but with enhanced extreme SSTs in early summer and late autumn. Comparison with climate model simulation results showed a slightly smaller extreme SST with a negative bias. This study highlights the potential of the POT method in understanding future oceanic warming based on satellite observed SSTs.
FRONTIERS IN MARINE SCIENCE
(2022)
Article
Statistics & Probability
Laurens de Haan, Chen Zhou
Summary: This article develops a bootstrap analogue of the asymptotic expansion of the tail quantile process in extreme value theory and applies it to construct confidence intervals for estimators of the extreme value index. It shows the bootstrap consistency of the confidence intervals for the peaks-over-threshold method, but not for the block maxima method. Simulations demonstrate that the sample variance of bootstrapped estimates can be a good approximation for the asymptotic variance of the original estimator.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2022)
Article
Computer Science, Interdisciplinary Applications
Tucker S. McElroy, Agnieszka Jach
Summary: This article presents and tests a nonparametric procedure for identifying the differencing operator of a non-stationary time series. The proposed differencing operator is applied to the time series, and the spectral density is tested for zeroes corresponding to the polynomial roots of the operator. A nonparametric tapered spectral density estimator is used, and the subsampling methodology is applied to obtain critical values. Simulations explore the effectiveness of the procedure under various scenarios involving nonstationary processes.
COMPUTATIONAL STATISTICS & DATA ANALYSIS
(2023)
Article
Statistics & Probability
Raphael de Fondeville, Anthony C. Davison
Summary: This paper extends the peaks-over-threshold analysis to extremes of functional data and introduces the generalized r-Pareto process for modeling threshold exceedances. The authors provide construction rules, simulation algorithms, and inference procedures for the generalized r-Pareto processes, and apply the new methodology to extreme European windstorms and heavy spatial rainfall.
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY
(2022)
Article
Statistics & Probability
Magda Monteiro, Isabel Pereira, Manuel G. Scotto
Summary: This study evaluates the modeling performance of two bivariate models for time series of counts in forest fires analysis in two counties of Portugal. The results indicate that the bivariate dynamic factor (BDF) model is better at capturing the dependence structure.
COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
(2021)
Article
Statistics & Probability
Claudia Santos, Isabel Pereira, Manuel G. Scotto
Summary: This paper introduces and studies in detail a multivariate integer-valued autoregressive model of order one with periodic time-varying parameters, driven by a periodic innovations sequence of independent random vectors. It discusses the basic probabilistic and statistical properties of the model, adopts a composite likelihood-based approach to reduce computational burden, compares the performance with traditional competitors, and addresses forecasting and an application to a real data set.
STATISTICAL PAPERS
(2021)
Article
Economics
Andres M. Alonso, Pedro Galeano, Daniel Pena
JOURNAL OF ECONOMETRICS
(2020)
Article
Economics
P. De Zea Bermudez, J. Miguel Marin, Helena Veiga
ECONOMETRIC REVIEWS
(2020)
Article
Geosciences, Multidisciplinary
Maria Alexandra Oliveira, Manuel G. Scotto, Susana Barbosa, Cesar Freire de Andrade, Maria da Conceicao Freitas
Article
Environmental Sciences
Carlos Silveira, Ana Martins, Sonia Gouveia, Manuel Scotto, Ana I. Miranda, Alexandra Monteiro
Summary: This study highlights the importance of considering aerosol effects in weather forecasting, not just for air quality assessment. Aerosols have a significant impact on meteorological variables such as shortwave radiation and temperature, particularly over land regions.
Article
Engineering, Multidisciplinary
Naushad Mamode Khan, Hassan S. Bakouch, Ashwinee Devi Soobhug, Manuel G. Scotto
Summary: The COVID-19 pandemic has severely disrupted global economic and social activities, especially impacting vulnerable economies like Small Island Developing States. Through the development of a new integer-valued time series model for COVID-19 infections, this research has made progress in modeling and forecasting the pandemic, with the INAR-WCG model showing slightly better performance in terms of RMSEs compared to other models.
ALEXANDRIA ENGINEERING JOURNAL
(2021)
Article
Statistics & Probability
Manuel G. Scotto, Sonia Gouveia
Summary: This paper investigates the extremal properties of the max-INAR process of order one based on the binomial thinning operator, with a focus on the limiting distribution of the number of exceedances of high levels and the joint limiting law of the maximum and the minimum. The extremal behavior of the max-INAR process of order one under the assumption of random thinning parameter and the periodic case are also examined.
COMMUNICATIONS IN STATISTICS-THEORY AND METHODS
(2023)
Article
Multidisciplinary Sciences
Ana Martins, Manuel Scotto, Ricardo Deus, Alexandra Monteiro, Sonia Gouveia
Summary: This study evaluates the impact of air pollution on daily respiratory hospital admissions in 58 spatial locations of Portugal mainland from 2005 to 2017, finding temperature to be the most determinant covariate and the importance of historical data in predicting hospitalization. Despite the small variability explained by air quality, models typically include two air pollutants covariates alongside temperature on average.
Article
Computer Science, Artificial Intelligence
Andres M. Alonso, Pierpaolo D'Urso, Carolina Gamboa, Vanesa Guerrero
Summary: A new approach to cluster large sets of time series is presented in this work, taking into account the dependency among the time series. The proposed methodology involves a two-step procedure, calculating cophenetic distances and applying a non-Euclidean fuzzy relational clustering algorithm. This robust fuzzy procedure is capable of detecting groups of time series with different types of cross-dependency, showing substantial improvements over hard partitioning clustering.
INTERNATIONAL JOURNAL OF APPROXIMATE REASONING
(2021)
Article
Biology
M. de Carvalho, S. Pereira, P. Pereira, P. de Zea Bermudez
Summary: This study introduces a novel regression model for learning the effect of covariates on extreme values without the need for conditional threshold selection in an extreme value theory framework. Through simulation studies, it is shown that the proposed method can accurately recover the true conditional distribution and perform well in variable selection.
JOURNAL OF AGRICULTURAL BIOLOGICAL AND ENVIRONMENTAL STATISTICS
(2022)
Article
Engineering, Environmental
Luis Gimeno-Sotelo, P. de Zea Bermudez, Iago Algarra, Luis Gimeno
Summary: This paper investigates the extremes of transported moisture in the Great Plains Low-Level Jet system and analyzes the extremal dependence between precipitation and tropospheric stability. The study finds that the extremes of transported moisture are expected to decrease in the future and the relationship between precipitation and tropospheric instability is stronger in the case of high moisture transport.
STOCHASTIC ENVIRONMENTAL RESEARCH AND RISK ASSESSMENT
(2022)
Article
Meteorology & Atmospheric Sciences
Susana Barbosa, Manuel G. Scotto
Summary: This study analyzes extreme summer temperatures on the Iberia Peninsula using ERA5-Land reanalysis data and a mixture model. The results show significant differences in temperature between the periods from 1981 to 2000 and from 2000 to 2019, with an increase in the mean temperature in the eastern region of the peninsula.
WEATHER AND CLIMATE EXTREMES
(2022)
Article
Energy & Fuels
Daniel Foronda-Pascual, Andres M. Alonso
Summary: This paper aims to predict the final price in the Spanish electricity market by forecasting the supply curve in advance. Various machine learning approaches are utilized to obtain the day-ahead supply curves for the secondary market, with Histogram-Based Gradient Boosting performing the best. The most relevant variables for the prediction are lagged values, daily market price, gas price, and wind values in Spanish provinces.
Article
Statistics & Probability
Maria Antonia Amaral Turkman, Kamil Feridun Turkman, Patricia de Zea Bermudez, Soraia Pereira, Paula Pereira, Miguel de Carvalho
Summary: In an environmental framework, extreme values of certain spatio-temporal processes, such as wind speeds, can cause severe damage to property. This study focuses on calibrating extreme values of data using a conditional quantile matching method, aiming to make simulated data more reliable and enriching it with practical applications. The proposed method avoids reliance on threshold choices and suggests extending calibration to extreme values using methods recommended by extreme value theory.
REVSTAT-STATISTICAL JOURNAL
(2021)