Article
Environmental Sciences
Demetris Koutsoyiannis, Theano Iliopoulou, Antonis Koukouvinos, Nikolaos Malamos, Nikos Mamassis, Panayiotis Dimitriadis, Nikos Tepetidis, David Markantonis
Summary: In Greece, a comprehensive rainfall dataset was compiled to construct rainfall intensity-timescale-return period relationships for the entire country. Ground rainfall data as well as non-conventional data from reanalyses and satellites were included. The dataset was also examined from a climatic perspective to assess its support for the climate crisis doctrine. Monte Carlo simulations and stationary Hurst-Kolmogorov (HK) stochastic dynamics were used for data comparison. The study found that rainfall extremes conform to statistical expectations under stationarity, with notable climatic events including water abundance in the 1960s and drought conditions around 1990.
Article
Environmental Sciences
Jenny Kupzig, Robert Reinecke, Francesca Pianosi, Martina Floerke, Thorsten Wagener
Summary: Global hydrological models (GHMs) provide important information for simulating water cycles and supporting decision-making. However, inaccuracies in GHM simulations can hinder valuable decision support. In this study, we introduce a transparent and efficient method to understand parameter control in GHMs and improve parameter estimation using global sensitivity analysis (GSA). Our findings show that traditionally neglected model parameters have a significant influence on GHM simulations, and basin attributes explain the spatial variability of parameter importance better than climate zones. Overall, our results demonstrate the effectiveness of GSA in guiding parameter estimation and improving the accuracy of GHM simulations.
ENVIRONMENTAL RESEARCH LETTERS
(2023)
Article
Environmental Sciences
Enda O'Connell, Greg O'Donnell, Demetris Koutsoyiannis
Summary: Precipitation deficits are the main drivers of droughts, and their level of persistence can be characterized by the Hurst coefficient H (0.5 < H < 1). Previous analyses of global precipitation datasets showed weak long-term persistence (LTP), but a new finding reveals that LTP increases with the spatial scale of averaging. This discovery has important implications for characterizing regional drought severity and understanding the annual flows of major rivers and aquifer recharge.
WATER RESOURCES RESEARCH
(2023)
Article
Engineering, Civil
Hanlin Li, Longxia Qian, Jianhong Yang, Suzhen Dang, Mei Hong
Summary: This study proposes an improved Bootstrap method and combines it with three commonly used parameter estimation methods, i.e., improved Bootstrap with method of moments (IBMOM), maximum likelihood estimation (IBMLE), and maximum entropy principle (IBMEP). A series of numerical experiments and a case study on the estimation of distribution parameters demonstrate that the proposed methods provide more accurate and less deviated results compared to conventional Bootstrap and without-Bootstrap approaches. Moreover, the improved Bootstrap method shows significant improvement in parameter estimation when smaller sample size is used. The method based on improved Bootstrap offers a new solution to the requirement of large sample size in quality hydrological frequency analysis.
WATER RESOURCES MANAGEMENT
(2023)
Editorial Material
Environmental Sciences
A. Gupta, R. S. Govindaraju, R. Morbidelli, C. Corradini
Summary: Bayes theorem provides a framework for parameter estimation by combining prior and sample information, but the availability and vagueness of prior knowledge may require the use of a reference prior for objective analysis. This study pursues an information-theoretic approach to derive reference priors and compares them to results obtained using a uniform prior.
WATER RESOURCES RESEARCH
(2022)
Article
Computer Science, Artificial Intelligence
Zaigham Tahir, Hina Khan, Faten S. Alamri, Muhammad Aslam
Summary: This study presents a generalized neutrosophic ratio-type exponential estimator (NRTEE) for estimating location parameters and achieving the lowest mean square error (MSE) possible for interval neutrosophic data (IND). Unlike typical estimators, its findings are not single-valued but rather in interval form, which reduces the possibility of over-or under-estimation caused by single crisp outcomes and also increases the likelihood of the parameter dwelling in the interval. The suggested NRTEE's efficiency is further addressed by utilizing real-life IND of temperature and simulations.
JOURNAL OF INTELLIGENT & FUZZY SYSTEMS
(2023)
Article
Engineering, Civil
Y. Sun, W. Bao, S. Qu, Q. Li, P. Jiang, Z. Zhou, P. Shi
Summary: This paper investigates the usefulness of the consider Kalman filter (CCKF) in hydrological modeling. The CCKF can improve the forecast performance of hydrological models by updating the states without updating the parameters. The results show that the CCKF is a more robust option for state estimation in the presence of parameter uncertainty.
JOURNAL OF HYDROLOGY
(2023)
Article
Computer Science, Interdisciplinary Applications
Zochil Gonzalez Arenas, Juan Carlos Jimenez, Li-Vang Lozada-Chang, Roberto Santana
Summary: This paper investigates the application of a global optimization method based on Estimation of Distribution Algorithms in calculating Innovation Estimators for diffusion processes. Through numerical simulations, it is shown that the global optimization algorithm significantly improves the effectiveness of the Innovation Estimators for diffusion processes with complex nonlinear and stochastic dynamics.
MATHEMATICS AND COMPUTERS IN SIMULATION
(2021)
Article
Mathematics, Applied
Ledys Llasmin Salazar Gomez, Soledad Torres, Jozef Kiselak, Felix Fuders, Naoyuki Ishimura, Yasukazu Yoshizawa, Milan Stehlik
Summary: The main objective of this paper is to analyze fluctuations of foreign currency exchange rates and identify the dependence structure in the associated stochastic processes. A novel methodology is introduced, which proves effective in analyzing bivariate financial time series with heavy tails. This methodology serves as a powerful tool for improving exchange rate fluctuation prediction, crucial in monetary and fiscal decision-making. It can also contribute to predicting financial crises and explaining deviations from the Purchasing Power Parity theory.
APPLIED MATHEMATICS AND COMPUTATION
(2022)
Article
Computer Science, Information Systems
H. Unozkan
Summary: This paper introduces a method for improving parameter estimation in statistical models by estimating parameters from other statistical distributions, which increases the success rate of the models.
Article
Water Resources
Panayiotis Dimitriadis, Demetris Koutsoyiannis, Theano Iliopoulou, Panos Papanicolaou
Summary: This study utilized an extensive collection of data from worldwide stations to seek stochastic analogies in key processes related to the hydrological cycle. It found stochastic similarities among the examined processes and traced similarities to turbulence-related lab recordings. These results contribute to the development of a universal stochastic view of the hydrological cycle.
Article
Engineering, Chemical
Xunyuan Yin, Song Bo, Jinfeng Liu, Biao Huang
Summary: This study proposes a consensus-based estimation mechanism for accurately estimating soil moisture in agro-hydrological systems, even when poor initial guesses of parameters and states are used.
Article
Computer Science, Interdisciplinary Applications
David Kepplinger
Summary: In high-dimensional regression problems, heavy-tailed error distributions and predictors with anomalous values are common and can undermine the validity of statistical analyses. To address this issue, a new robust regularized regression estimator called adaptive PENSE is proposed, which provides reliable variable selection and coefficient estimates even in the presence of abnormal contamination. It is demonstrated that adaptive PENSE performs better than other penalties in terms of robust and reliable variable selection, especially when there are gross outliers in the predictor space. Moreover, adaptive PENSE exhibits strong variable selection properties and possesses the oracle property even under heavy-tailed errors without the need to estimate the error scale. Numerical studies on simulated and real data sets confirm the superior finite-sample performance of adaptive PENSE compared to other robust regularized estimators in the case of contaminated samples. An R package implementing a fast algorithm for computing the proposed method and additional simulation results are provided in the supplementary materials.
COMPUTATIONAL STATISTICS & DATA ANALYSIS
(2023)
Article
Automation & Control Systems
Tushar Garg, Sayan Basu Roy
Summary: This work proposes a distributed adaptive estimation algorithm for multi-agent system architecture using online measurements. The algorithm ensures parameter convergence under relaxed condition of initial excitation, which is more realistic and suitable for practical scenarios. Simulation results validate the effectiveness of the proposed algorithm compared to traditional methods.
INTERNATIONAL JOURNAL OF ADAPTIVE CONTROL AND SIGNAL PROCESSING
(2023)
Article
Computer Science, Interdisciplinary Applications
Xuzheng Chai, Arpad Rozsas, Arthur Slobbe, Ana Teixeira
Summary: This study investigates the impact of structural health monitoring (SHM) on the failure probability of a simulated sheet pile wall system, utilizing Bayesian statistics to estimate non-directly observable soil parameters and finding that coupling SHM with Bayesian statistics can significantly reduce failure probability estimates.
COMPUTERS AND GEOTECHNICS
(2022)
Article
Engineering, Civil
Georgia Papacharalampous, Hristos Tyralis, Yannis Markonis, Martin Hanel
Summary: In this study, a new methodological framework is proposed for exploring and comparing multi-scale analyses in a hydroclimatic context, in order to comprehensively understand the behaviors of geophysical processes and evaluate time series simulation models. By computing the feature values at different temporal resolutions and three hydroclimatic time series types, similarities and differences in the evolution patterns are identified. The computed features are also used for meaningful clustering of hydroclimatic time series, which allows for interpretation of hydroclimatic similarity at various temporal resolutions.
JOURNAL OF HYDROLOGY
(2023)
Article
Engineering, Civil
Hristos Tyralis, Georgia Papacharalampous, Sina Khatami
Summary: This paper introduces a method to estimate the uncertainty of hydrological simulations using expectiles. The method is applied to 511 basins and compares different hydrological models. The results show that the GR6J model outperforms the other two models at all expectile levels.
JOURNAL OF HYDROLOGY
(2023)
Article
Engineering, Civil
Hristos Tyralis, Georgia Papacharalampous
Summary: Hydrological post-processing using extremal quantile regression is introduced to estimate the extreme quantiles of hydrological model's responses, overcoming the limitation of conventional quantile regression. The new method demonstrates higher accuracy in estimating high quantiles compared to conventional quantile regression, while being equivalent at lower quantiles. This research provides a valuable approach for improving the uncertainty estimation of hydrological predictions.
JOURNAL OF HYDROLOGY
(2023)
Article
Environmental Sciences
Georgia Papacharalampous, Hristos Tyralis, Anastasios Doulamis, Nikolaos Doulamis
Summary: Gridded satellite precipitation datasets are widely used in hydrological applications, but they are not accurate compared to ground-based measurements. To improve their accuracy, this study compares eight state-of-the-art machine learning algorithms in correcting satellite precipitation data for the entire United States and a 15-year period. The results show that extreme gradient boosting (XGBoost) and random forests are the most accurate.
Article
Engineering, Marine
Kimon Kardakaris, Panayiotis Dimitriadis, Theano Iliopoulou, Demetris Koutsoyiannis
Summary: A combination of stochastic and deterministic models is used to study ocean wind waves. Timeseries of significant wave height and mean zero up-crossing period from floating buoys are analyzed to construct a double periodic model and select an optimal marginal distribution and dependence function. The study finds that wind waves are mostly influenced by seasonal periodicity rather than diurnal periodicity. The Pareto-Burr-Feller distribution is also found to be an appropriate choice for the marginal distribution.
Article
Environmental Sciences
Enda O'Connell, Greg O'Donnell, Demetris Koutsoyiannis
Summary: Precipitation deficits are the main drivers of droughts, and their level of persistence can be characterized by the Hurst coefficient H (0.5 < H < 1). Previous analyses of global precipitation datasets showed weak long-term persistence (LTP), but a new finding reveals that LTP increases with the spatial scale of averaging. This discovery has important implications for characterizing regional drought severity and understanding the annual flows of major rivers and aquifer recharge.
WATER RESOURCES RESEARCH
(2023)
Article
Environmental Sciences
Demetris Koutsoyiannis, Theano Iliopoulou, Antonis Koukouvinos, Nikolaos Malamos, Nikos Mamassis, Panayiotis Dimitriadis, Nikos Tepetidis, David Markantonis
Summary: In Greece, a comprehensive rainfall dataset was compiled to construct rainfall intensity-timescale-return period relationships for the entire country. Ground rainfall data as well as non-conventional data from reanalyses and satellites were included. The dataset was also examined from a climatic perspective to assess its support for the climate crisis doctrine. Monte Carlo simulations and stationary Hurst-Kolmogorov (HK) stochastic dynamics were used for data comparison. The study found that rainfall extremes conform to statistical expectations under stationarity, with notable climatic events including water abundance in the 1960s and drought conditions around 1990.
Article
Environmental Studies
G. -Fivos Sargentis, Demetris Koutsoyiannis
Summary: The water-energy-food (WEF) and land nexus is essential for prosperity. However, the distribution of WEF elements is uneven, and trading dynamics drive the distribution of goods. To understand the role of money in the WEF nexus, we convert all elements into stable energy units using GDP and electricity price data. We observe that land is the foundation of WEF and positively correlated with all its elements. Even wealthy countries face critical deficits in WEF. By adding money (GDP in energy units) to the WEF nexus, the balance becomes positive, highlighting the importance of trade for survival and prosperity.
Article
Mathematics, Applied
Demetris Koutsoyiannis
Summary: Knowable moments, or K-moments, are redefined as expectations of maxima or minima of a number of stochastic variables in order to make them applicable to any type of variable. They have unique features that make them useful in various applications, including the fact that they can be reliably estimated from a sample and assigned values of distribution function. K-moments also consider estimation bias in time series data, offering a strategy of model fitting not shared by other methods, which is particularly useful for modeling extremes associated with high-order moments.
Article
Water Resources
Aristoteles Tegos, Stefanos Stefanidis, John Cody, Demetris Koutsoyiannis
Summary: This paper examines the impacts of three different potential evapotranspiration (PET) models on drought severity and frequencies indicated by the standardized precipitation index (SPEI). The findings highlight the presence of uncertainty in defining the severity of drought, especially for large timescales, and suggest that the PET parametric model is a preferable model for both the standardized precipitation-evapotranspiration index and the aridity indexes. This outcome is worth further consideration for climatic studies in data scarce areas.
Article
Water Resources
Georgia Papacharalampous, Hristos Tyralis, Anastasios Doulamis, Nikolaos Doulamis
Summary: To obtain accurate precipitation datasets that cover large regions with high density, merging satellite products and ground-based measurements using machine learning regression algorithms is crucial. In this study, we compared three tree-based ensemble algorithms and found that extreme gradient boosting (XGBoost) performs the best for correcting satellite precipitation products in the contiguous United States.
Article
Geosciences, Multidisciplinary
Ning Wang, Fubao Sun, Demetris Koutsoyiannis, Theano Iliopoulou, Tingting Wang, Hong Wang, Wenbin Liu, G. -Fivos Sargentis, Panayiotis Dimitriadis
Summary: This study reveals that in 53% of countries, people tend to distance themselves from floods, especially in the Middle East. This behavior significantly reduces flood fatalities and displacements. Moreover, in regions with higher flood protection level, people are less likely to move away from floods. Over time, flood protection levels and the distance between humans and flood-prone areas have decreased in regions affected by ice jam- and hurricane-induced floods. Regions with slightly below average human-flood distance for a given flood type experience more severe flood fatalities.
GEOPHYSICAL RESEARCH LETTERS
(2023)
Article
Environmental Sciences
Georgia Papacharalampous, Hristos Tyralis, Nikolaos Doulamis, Anastasios Doulamis
Summary: This study fills the research gap of lacking sufficient ensemble learners for improving the accuracy of satellite precipitation products and their large-scale comparison. By proposing 11 new ensemble learners and extensively comparing them, the study finds that sophisticated stacking method performs significantly better than the base learners, especially when applied using the linear regression algorithm.
Article
Water Resources
Nikolaos Malamos, Dimitrios Koulouris, Ioannis L. Tsirogiannis, Demetris Koutsoyiannis
Summary: The study evaluates the accuracy of Bologna Limited-Area Model (BOLAM) predictions for precipitation, air temperature, relative humidity, and wind speed. The results show that the model performs poorly in predicting precipitation and wind speed, while it performs well in relative humidity and very well in air temperature predictions.
Article
Engineering, Electrical & Electronic
Hristos Tyralis, Georgia Papacharalampous, Nikolaos Doulamis, Anastasios Doulamis
Summary: To improve precipitation estimates, this study applies machine learning to merge rain gauge-based measurements and satellite precipitation data, and proposes issuing probabilistic spatial predictions of precipitation using LightGBM algorithm. The results show that LightGBM outperforms random forests in terms of accuracy at extreme quantiles.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
