This paper concerns the description of the formation of a very large, quality-controlled dataset ... more This paper concerns the description of the formation of a very large, quality-controlled dataset of raw wave measurements. As part of the CresT (Cooperative Research on Extreme Seas and their impacT) Joint Industry Project (JIP), the participants provided raw field measurements of water surface elevation from various installations across the globe. This paper will describe the data collection from the different installations, the strict quality control procedure employed to ensure a reliable dataset, and an overview of the occurrences of records containing freak wave events. Part 2 of this paper will then go on to describe the analysis performed on this dataset as well as the findings from the study.
Journal of Marine Science and Engineering, Sep 6, 2022
This article is an open access article distributed under the terms and conditions of the Creative... more This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY
Journal of Atmospheric and Oceanic Technology, Aug 1, 2023
Accurate measurements of ocean waves underpin efficient offshore operations and optimal offshore ... more Accurate measurements of ocean waves underpin efficient offshore operations and optimal offshore structure design, helping to ensure the offshore industry can operate both safely and economically. Popular instruments used by the offshore industry are the Rosemount WaveRadar (Radar) and the Waverider Buoy. The Optech Laser has been used at some locations for specific studies. Recent reports indicate systematic differences of order 10% among the wave measurements made by these instruments. This paper examines the relative performance of these instruments based upon various time-domain comparisons, including results from a quality control (QC) procedure, capabilities of measuring the wave surface profile (skewness), and crest heights for varying wind sea and swell conditions. The QC check provides good-quality data that can be further investigated with an assurance of error-free data, suggesting that the Waverider produces the best-quality data with the lowest failure rate compared to the Laser and Radar. A significant number of the Waverider surface elevation records have negative skewness, particularly at higher sea states, affecting its crest height measurements, which are lower than those from the Laser and Radar. Additionally, the significant wave height (H1/3) estimates of the Radar are lower than the Laser and Waverider, but its zero-crossing wave periods (TZ), on average, are longer than the Laser and the Waverider. The significant heights (H1/3) of Laser and Waverider are in good agreement for all three datasets, but the Waverider’s zero-crossing wave period (TZ) estimates are significantly longer than the Laser.
Journal of Atmospheric and Oceanic Technology, May 1, 2023
Accurate ocean wave measurements are needed for the safe design and operation of offshore facilit... more Accurate ocean wave measurements are needed for the safe design and operation of offshore facilities, but despite many ocean wave measurements, the accuracy of wave measurement systems remains an ongoing issue. Of paramount importance are measurements during extreme sea states. This paper examines wave measurements made with an Optech Laser (Laser), a Rosemount WaveRadar (Radar), and a Datawell Waverider buoy at North Rankin A (NRA) platform, Australia; Ekofisk, North Sea; and several South China Sea locations. We evaluate the relative performance of these instruments based upon various frequency domain comparisons, including comparisons of their 1D frequency spectra using spectrograms, spectral moments, high-frequency tail slopes, and significant wave heights derived from their wave spectra. A spectral relationship (transfer function) in terms of mean spectral ratio of the instruments is developed, which can be used for spectral calibration. On average, Laser and Waverider spectral estimates agree well at all sea states. However, at low wind speeds, the higher-frequency spectral levels of the Laser are relatively high and noisy compared with the other two instruments. Radar higher-frequency spectral estimates are relatively low compared to the other two instruments, particularly at lower sea states. In addition, the higher-frequency tail slopes of all three instruments vary between f−4 and f−5. However, at higher sea states, the Waverider tail slopes become steeper than f−5. The Radar produces the lowest significant wave heights (Hm0) compared to the Laser and Waverider, but its second-moment period (Tm02) estimates are longer than the Laser and Waverider.
The proposed work will contribute to the improvement of existing third-generation (3G) wave model... more The proposed work will contribute to the improvement of existing third-generation (3G) wave models as well as to the development of the next generation of numerical wave modeling capability. The results will be applicable in the coastal zone from deep water up to and including the surf zone. Our efforts will focus on analyzing high quality datasets to support further development of the source terms for triad interactions (Snl3), depth induced wave breaking (Sbrk) and bottom friction (Sbot) in the near-shore zone. Another point of interest is the generation of bounded long waves in the surf zone. DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited.
Journal of Advances in Modeling Earth Systems, 2021
Global wave hindcasts are developed using the third generation spectral wave model WAVEWATCH III ... more Global wave hindcasts are developed using the third generation spectral wave model WAVEWATCH III with the observation‐based source terms (ST6) and a hybrid rectilinear‐curvilinear, irregular‐regular‐irregular grid system (approximately at ). Three distinct global hindcasts are produced: (a) a long‐term hindcast (1979–2019) forced by the ERA5 conventional winds and (b) two short‐term hindcasts (2011–2019) driven by the NCEP climate forecast system (CFS)v2 and the ERA5 neutral winds , respectively. The input field for ice is sourced from the Ocean and Sea Ice Satellite Application Facility (OSI SAF) sea‐ice concentration climate data records. These wave simulations, together with the driving wind forcing, are validated against extensive in‐situ observations and satellite altimeter records. The performance of the ST6 wave hindcasts shows promising results across multiple wave parameters, including the conventional wave characteristics (e.g., wave height and wave period) and high‐order ...
The Journal of the Acoustical Society of America, 1983
Underwater ambient noise is known to be dependent on the wind. Several mechanisms have been propo... more Underwater ambient noise is known to be dependent on the wind. Several mechanisms have been proposed to explain the nature of the transfer of energy from the wind field to the acoustic noise field. Examples include wind and wave turbulence and nonlinear interactions between surface waves, which given rise to many phonemena not present in linear theory. This study examines these wind-related mechanisms from the low end of the acoustic spectrum. Data from a long-term study of ocean waves and the associated microseismic field recorded ashore has produced evidence helpful to the identification of processes active in the acoustic field at somewhat higher frequencies. The unique environment under which these recordings were made in New Zealand has contributed to the success of the study. Correlations with ambient noise recorded by a shallow water hydrophone were also made.
Journal of offshore mechanics and Arctic engineering, Apr 1, 2015
Specification of realistic environmental design conditions for marine structures is of fundamenta... more Specification of realistic environmental design conditions for marine structures is of fundamental importance to their reliability over time. Design conditions for extreme waves and storm severities are typically estimated by extreme value analysis of time series of measured or hindcast significant wave height, H S. This analysis is complicated by two effects. Firstly, H S exhibits temporal dependence. Secondly, the characteristics of H sp S are non-stationary with respect to multiple covariates, particularly wave direction and season. We develop directional-seasonal design values for storm peak significant wave height (H sp S) by estimation of, and simulation under a non-stationary extreme value model for H sp S. Design values for significant wave height (H S) are estimated by simulating storm trajectories of H S consistent with the simulated storm peak events. Design distributions for individual maximum wave height (H max) are estimated by marginalisation using the known conditional distribution for H max given H S. Particular attention is paid to the assessment of model bias and quantification of model parameter and design value uncertainty using bootstrap resampling. We also outline existing work on extension to estimation of maximum crest elevation and total extreme water level.
Inherent uncertainties in estimation of extreme wave heights in hurricane-dominated regions are e... more Inherent uncertainties in estimation of extreme wave heights in hurricane-dominated regions are explored using data from the GOMOS Gulf of Mexico hindcast for 1900-2005. In particular, the effect of combining correlated values from a neighborhood of 72 grid locations on extreme wave height estimation is quantified. We show that, based on small data samples, extreme wave heights are underestimated and site averaging usually improves estimates. We present a bootstrapping approach to evaluate uncertainty in extreme wave height estimates. We also argue in favor of modeling supplementary indicators for extreme wave characteristics, such as a high percentile (95%) of the distribution of 100-year significant wave height, in addition to its most probable value, especially for environments where the distribution of 100-year significant wave height is strongly skewed.
Understanding extreme ocean environments and their interaction with fixed and floating structures... more Understanding extreme ocean environments and their interaction with fixed and floating structures is critical for offshore and coastal design. Design contours are useful to describe the joint behaviour of environmental, structural loading and response variables. We compare different forms of design contours, using theory and simulation, and present a new method for joint estimation of contours of constant exceedance probability for a general set of variables. The method is based on a conditional extremes model from the statistics literature, motivated by asymptotic considerations. We simulate under the conditional extremes model to estimate contours of constant exceedance probability. We also use the estimated conditional extremes model to estimate other forms of design contours, including those based on the First Order Reliability Method, without needing to specify the functional forms of conditional dependence between variables. We demonstrate the application of new method in estimation of contours of constant exceedance probability using measured and hindcast data from the Northern North Sea, the Gulf of Mexico and the North West Shelf of Australia, and quantify their uncertainties using a bootstrap analysis.
⇔ max-stability ⇔ multivariate regular variation Assumes all components extreme ⇒ Perfect indepen... more ⇔ max-stability ⇔ multivariate regular variation Assumes all components extreme ⇒ Perfect independence or asymptotic dependence only Composite likelihood for spatial extremes (Davison et al. 2012) Point process / multivariate GP process Extremal dependence: (Ledford and Tawn 1997) Assumes regular variation of joint survivor function Yields more general forms of extremal dependence ⇒ Asymptotic dependence, asymptotic independence (with +ve,-ve association), "hidden regular variation" "Ray" extensions Hybrid spatial dependence model (Wadsworth and Tawn 2012) Conditional extremes: (Heffernan and Tawn 2004) Assumes, given one variable being extreme, convergence of distribution of remaining variables Allows some variables not to be extreme Extensions
The process of estimating the effect of a changing climate on the severity of future ocean storms... more The process of estimating the effect of a changing climate on the severity of future ocean storms is plagued by large uncertainties; for safe design and operation of offshore structures, it is nevertheless important that best possible estimates of climate effects is made given the available data. We explore the variability in estimates of 100-year return value of significant wave height (H S), and changes in estimates over a period of time, for output of WAVEWATCH-III models from 7 representative CMIP5 GCMs, and the FIO-ESM v2.0 CMIP6 GCM. Non-stationary extreme value analysis of peaks-over-threshold (POT) and block maxima (BM) using Bayesian inference provide posterior estimates of return values as a function of time; MATLAB software for the extreme value analysis is provided. Best overall estimates for return values, and changes in return value over the period 1979-2100, are calculated by averaging estimates for individual GCMs. We focus attention on neighbourhoods of locations east of Madagascar (EoM) and south of Australia (SoA) where a previous study of CMIP5 output by Meucci et al. (2020) reported significant decrease and increase in H S respectively, under RCP4.5 and RCP8.5 climate scenarios. There is large variation between return value estimates from different GCMs, and with longitude and latitude within each neighbourhood for estimates based on samples corresponding to ≤ 165 years of model output; these sources of uncertainty tend to be larger than that due to typical modelling choices (such as choice of threshold for POT, or block length for BM). However, we also find that careful threshold choice and block length are critical EoM, because of the presence of a mixed population of storms there. Nevertheless, there is general evidence supporting the trends reported by Meucci et al. (2020), but these findings are conditional on the choice of 8 GCMs being representative of climate evolution. We use simple randomisation testing to identify "significant" departures from steady climate. The long 700-year pre-industrial control (piControl) output of the CMIP6 GCM offers an excellent opportunity to quantify the apparent inherent variability in return value as a function of time, estimated using a subsample of output corresponding to a continuous time interval of between 20 and 160 years in length, where no climate forcing is present. We find large variation in return value estimates of approximately ±15% made from samples corresponding to periods of time of around 50 years drawn from piControl data.
Existing metocean design criteria for offshore facilities in the South China Sea have been estima... more Existing metocean design criteria for offshore facilities in the South China Sea have been estimated using different data and procedures, some of which are at least partly ad hoc. As a result, it is probable that existing criteria are inconsistent, in the sense that assets designed to the same design codes have different realised levels of integrity. To address this concern in this paper, we apply a large-scale extreme value model adapted to parallel computing environment, applied to the recent high-resolution SEAFINE hindcast database. This not only ensures design criteria that are statistically and spatially consistent but is also faster by avoiding the need for repetitive site-specific analysis. We have to overcome several challenges before we can apply a large-scale extreme model on the SEAFINE database. These include identifying a spatially consistent set of storm peaks and validating the hindcast data with real measurements. We then estimate marginal return values for significant wave height for locations within a large spatial neighbourhood, accounting for spatial and storm directional variability of peaks over threshold. A quantile regression identifies the extreme value threshold, the rate of exceedance of which is described using a Poisson process. The size of threshold exceedances is described by a generalised Pareto model. The characteristics of the threshold, rate and size models are all non-stationary with respect to directional and spatial covariates, parameterised in terms of (multidimensional) penalised B-splines. Parameter estimation is computationally challenging, but a combination of efficient generalised linear array algorithms executed within a parallel computing environment enable maximum likelihood estimation of all models. Bootstrap resampling is used to estimate uncertainties of model parameters and return values. We thus estimate consistent marginal return values for significant wave height and their uncertainties, at all locations in the spatial neighbourhood. In addition, we quantify directional variability of return values across different return periods. We rigorously validate the proposed spatiodirectional model with that of a direction-only model by deriving model diagnostics for the same site and demonstrating equivalent goodness of fits. To our knowledge this is the first of a kind of application of large-scale estimation for the South China Sea. With this approach, design criteria for large spatial domains, non-stationary with respect to the appropriate environmental covariates, can be estimated efficiently, consistently and with quantified uncertainty.
Computational Statistics & Data Analysis, Mar 1, 2016
Safe and reliable design and operation of fixed and floating marine structures often located in r... more Safe and reliable design and operation of fixed and floating marine structures often located in remote and hostile environments is challenging. Rigorous extreme value analysis of meteorological and oceanographic data can greatly aid the design of such structures. Extreme value analysis is typically undertaken for single spatial locations or for small neighbourhoods; moreover, non-stationary effects of covariates on extreme values are typically accommodated in an ad-hoc manner. The objective of the work summarised here is to improve design practice by estimating environmental design conditions (such as return values for extreme waves, winds and currents) for a whole ocean basin, including additional covariate effects (such as storm direction) as necessary, in a consistent manner. Whole-basin non-stationary extreme value modelling is computationally complex, requiring inter-alia the estimation of tail functions, the parameters of which vary with respect to multi-dimensional covariates characterised by us using tensor products of penalised B-splines. We outline two technical contributions which make whole-basin non-stationary analysis feasible. Firstly, we adopt generalised linear array methods to reduce the computational burden of matrix manipulations. Secondly, using high-performance computing, we develop a parallel implementation of maximum likelihood estimation for the generalised Pareto distribution. Together, these innovations allow estimation of rigorous whole-basin extreme value models in reasonable time. We evaluate the new approach in application to marginal extreme value modelling of storm peak significant wave heights in two ocean basins, accommodating spatial and directional covariate effects.
Numerous approaches are proposed in the literature for non-stationarity marginal extreme value in... more Numerous approaches are proposed in the literature for non-stationarity marginal extreme value inference, including different model parameterisations with respect to covariate, and different inference schemes. The objective of this article is to compare some of these procedures critically. We generate sample realisations from generalised Pareto distributions, the parameters of which are smooth functions of a single smooth periodic covariate, specified to reflect the characteristics of actual samples from the tail of the distribution of significant wave height with direction, considered in the literature in the recent past. We estimate extreme values models (a) using Constant, Fourier, B-spline and Gaussian Process parameterisations for the functional forms of generalised Pareto shape and (adjusted) scale with respect to covariate and (b) maximum likelihood and Bayesian inference procedures. We evaluate the relative quality of inferences by estimating return value distributions for the response corresponding to a time period of 10× the (assumed) period of the original sample, and compare estimated return values distributions with the truth using Kullback-Leibler, Cramer-von Mises and Kolmogorov-Smirnov statistics. We find that Spline and Gaussian Process parameterisations estimated by Markov chain Monte Carlo inference using the mMALA algorithm, perform equally well in terms of quality of inference and computational efficiency, and generally perform better than alternatives in those respects.
We report a new approach to model the frequency-direction spectrum, in which the frequency-direct... more We report a new approach to model the frequency-direction spectrum, in which the frequency-direction spectra from measurements or hindcast studies are fitted simultaneously in two dimensions, frequency and direction. Depending on the amount of wind forcing on the partition, either a unimodal (swell) or bimodal (wind-sea) wave spreading function is adopted together with the spectral form which best fits the frequency spectrum. This paper describes the new method and presents the results on a measured dataset.
2016 IEEE 23rd International Conference on High Performance Computing Workshops (HiPCW), 2016
Modelling rare or extreme events is critical in many domains, including financial risk, computer ... more Modelling rare or extreme events is critical in many domains, including financial risk, computer security breach, network outage, corrosion and fouling, manufacturing quality and environmental extremes such as floods, snowfalls, heatwaves , seismic hazards and meteorological-oceanographic events like extra-tropical storms, hurricanes and typhoons. Statistical modelling enables us to understand extremes and design mechanisms to prevent their occurrence and manage their impact. Extreme events are challenging to characterise as they are, by definition, rare and unusual even in a big data world. The frequency and extent of extreme events is typically driven by both primary attributes (dependent variables) and secondary attributes (independent variables or covariates). Studies have shown that improved inference can be gained from including covariate effects in predictive models but this inclusion comes at a heavy computation cost. In this paper, we present a framework for risk estimation from extreme events that are non-stationary; i.e., they are dependent on multi-dimensional covariates. The approach is illustrated by estimation of offshore structural design criteria in a storm environment non-stationary with respect to storm direction, season and geographic location. The framework allows consistent assessment of structural reliability with thorough uncertainty quantification. The model facilitates estimation of risk for any combination of covariates, which can be exploited for improved understanding and ultimately optimal marine structural design. The computational burden incurred is large, especially since thorough uncertainty quantification is incorporated, but manageable using slick algorithms for linear algebraic manipulations and high-performance computing.
This paper concerns the description of the formation of a very large, quality-controlled dataset ... more This paper concerns the description of the formation of a very large, quality-controlled dataset of raw wave measurements. As part of the CresT (Cooperative Research on Extreme Seas and their impacT) Joint Industry Project (JIP), the participants provided raw field measurements of water surface elevation from various installations across the globe. This paper will describe the data collection from the different installations, the strict quality control procedure employed to ensure a reliable dataset, and an overview of the occurrences of records containing freak wave events. Part 2 of this paper will then go on to describe the analysis performed on this dataset as well as the findings from the study.
Journal of Marine Science and Engineering, Sep 6, 2022
This article is an open access article distributed under the terms and conditions of the Creative... more This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY
Journal of Atmospheric and Oceanic Technology, Aug 1, 2023
Accurate measurements of ocean waves underpin efficient offshore operations and optimal offshore ... more Accurate measurements of ocean waves underpin efficient offshore operations and optimal offshore structure design, helping to ensure the offshore industry can operate both safely and economically. Popular instruments used by the offshore industry are the Rosemount WaveRadar (Radar) and the Waverider Buoy. The Optech Laser has been used at some locations for specific studies. Recent reports indicate systematic differences of order 10% among the wave measurements made by these instruments. This paper examines the relative performance of these instruments based upon various time-domain comparisons, including results from a quality control (QC) procedure, capabilities of measuring the wave surface profile (skewness), and crest heights for varying wind sea and swell conditions. The QC check provides good-quality data that can be further investigated with an assurance of error-free data, suggesting that the Waverider produces the best-quality data with the lowest failure rate compared to the Laser and Radar. A significant number of the Waverider surface elevation records have negative skewness, particularly at higher sea states, affecting its crest height measurements, which are lower than those from the Laser and Radar. Additionally, the significant wave height (H1/3) estimates of the Radar are lower than the Laser and Waverider, but its zero-crossing wave periods (TZ), on average, are longer than the Laser and the Waverider. The significant heights (H1/3) of Laser and Waverider are in good agreement for all three datasets, but the Waverider’s zero-crossing wave period (TZ) estimates are significantly longer than the Laser.
Journal of Atmospheric and Oceanic Technology, May 1, 2023
Accurate ocean wave measurements are needed for the safe design and operation of offshore facilit... more Accurate ocean wave measurements are needed for the safe design and operation of offshore facilities, but despite many ocean wave measurements, the accuracy of wave measurement systems remains an ongoing issue. Of paramount importance are measurements during extreme sea states. This paper examines wave measurements made with an Optech Laser (Laser), a Rosemount WaveRadar (Radar), and a Datawell Waverider buoy at North Rankin A (NRA) platform, Australia; Ekofisk, North Sea; and several South China Sea locations. We evaluate the relative performance of these instruments based upon various frequency domain comparisons, including comparisons of their 1D frequency spectra using spectrograms, spectral moments, high-frequency tail slopes, and significant wave heights derived from their wave spectra. A spectral relationship (transfer function) in terms of mean spectral ratio of the instruments is developed, which can be used for spectral calibration. On average, Laser and Waverider spectral estimates agree well at all sea states. However, at low wind speeds, the higher-frequency spectral levels of the Laser are relatively high and noisy compared with the other two instruments. Radar higher-frequency spectral estimates are relatively low compared to the other two instruments, particularly at lower sea states. In addition, the higher-frequency tail slopes of all three instruments vary between f−4 and f−5. However, at higher sea states, the Waverider tail slopes become steeper than f−5. The Radar produces the lowest significant wave heights (Hm0) compared to the Laser and Waverider, but its second-moment period (Tm02) estimates are longer than the Laser and Waverider.
The proposed work will contribute to the improvement of existing third-generation (3G) wave model... more The proposed work will contribute to the improvement of existing third-generation (3G) wave models as well as to the development of the next generation of numerical wave modeling capability. The results will be applicable in the coastal zone from deep water up to and including the surf zone. Our efforts will focus on analyzing high quality datasets to support further development of the source terms for triad interactions (Snl3), depth induced wave breaking (Sbrk) and bottom friction (Sbot) in the near-shore zone. Another point of interest is the generation of bounded long waves in the surf zone. DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited.
Journal of Advances in Modeling Earth Systems, 2021
Global wave hindcasts are developed using the third generation spectral wave model WAVEWATCH III ... more Global wave hindcasts are developed using the third generation spectral wave model WAVEWATCH III with the observation‐based source terms (ST6) and a hybrid rectilinear‐curvilinear, irregular‐regular‐irregular grid system (approximately at ). Three distinct global hindcasts are produced: (a) a long‐term hindcast (1979–2019) forced by the ERA5 conventional winds and (b) two short‐term hindcasts (2011–2019) driven by the NCEP climate forecast system (CFS)v2 and the ERA5 neutral winds , respectively. The input field for ice is sourced from the Ocean and Sea Ice Satellite Application Facility (OSI SAF) sea‐ice concentration climate data records. These wave simulations, together with the driving wind forcing, are validated against extensive in‐situ observations and satellite altimeter records. The performance of the ST6 wave hindcasts shows promising results across multiple wave parameters, including the conventional wave characteristics (e.g., wave height and wave period) and high‐order ...
The Journal of the Acoustical Society of America, 1983
Underwater ambient noise is known to be dependent on the wind. Several mechanisms have been propo... more Underwater ambient noise is known to be dependent on the wind. Several mechanisms have been proposed to explain the nature of the transfer of energy from the wind field to the acoustic noise field. Examples include wind and wave turbulence and nonlinear interactions between surface waves, which given rise to many phonemena not present in linear theory. This study examines these wind-related mechanisms from the low end of the acoustic spectrum. Data from a long-term study of ocean waves and the associated microseismic field recorded ashore has produced evidence helpful to the identification of processes active in the acoustic field at somewhat higher frequencies. The unique environment under which these recordings were made in New Zealand has contributed to the success of the study. Correlations with ambient noise recorded by a shallow water hydrophone were also made.
Journal of offshore mechanics and Arctic engineering, Apr 1, 2015
Specification of realistic environmental design conditions for marine structures is of fundamenta... more Specification of realistic environmental design conditions for marine structures is of fundamental importance to their reliability over time. Design conditions for extreme waves and storm severities are typically estimated by extreme value analysis of time series of measured or hindcast significant wave height, H S. This analysis is complicated by two effects. Firstly, H S exhibits temporal dependence. Secondly, the characteristics of H sp S are non-stationary with respect to multiple covariates, particularly wave direction and season. We develop directional-seasonal design values for storm peak significant wave height (H sp S) by estimation of, and simulation under a non-stationary extreme value model for H sp S. Design values for significant wave height (H S) are estimated by simulating storm trajectories of H S consistent with the simulated storm peak events. Design distributions for individual maximum wave height (H max) are estimated by marginalisation using the known conditional distribution for H max given H S. Particular attention is paid to the assessment of model bias and quantification of model parameter and design value uncertainty using bootstrap resampling. We also outline existing work on extension to estimation of maximum crest elevation and total extreme water level.
Inherent uncertainties in estimation of extreme wave heights in hurricane-dominated regions are e... more Inherent uncertainties in estimation of extreme wave heights in hurricane-dominated regions are explored using data from the GOMOS Gulf of Mexico hindcast for 1900-2005. In particular, the effect of combining correlated values from a neighborhood of 72 grid locations on extreme wave height estimation is quantified. We show that, based on small data samples, extreme wave heights are underestimated and site averaging usually improves estimates. We present a bootstrapping approach to evaluate uncertainty in extreme wave height estimates. We also argue in favor of modeling supplementary indicators for extreme wave characteristics, such as a high percentile (95%) of the distribution of 100-year significant wave height, in addition to its most probable value, especially for environments where the distribution of 100-year significant wave height is strongly skewed.
Understanding extreme ocean environments and their interaction with fixed and floating structures... more Understanding extreme ocean environments and their interaction with fixed and floating structures is critical for offshore and coastal design. Design contours are useful to describe the joint behaviour of environmental, structural loading and response variables. We compare different forms of design contours, using theory and simulation, and present a new method for joint estimation of contours of constant exceedance probability for a general set of variables. The method is based on a conditional extremes model from the statistics literature, motivated by asymptotic considerations. We simulate under the conditional extremes model to estimate contours of constant exceedance probability. We also use the estimated conditional extremes model to estimate other forms of design contours, including those based on the First Order Reliability Method, without needing to specify the functional forms of conditional dependence between variables. We demonstrate the application of new method in estimation of contours of constant exceedance probability using measured and hindcast data from the Northern North Sea, the Gulf of Mexico and the North West Shelf of Australia, and quantify their uncertainties using a bootstrap analysis.
⇔ max-stability ⇔ multivariate regular variation Assumes all components extreme ⇒ Perfect indepen... more ⇔ max-stability ⇔ multivariate regular variation Assumes all components extreme ⇒ Perfect independence or asymptotic dependence only Composite likelihood for spatial extremes (Davison et al. 2012) Point process / multivariate GP process Extremal dependence: (Ledford and Tawn 1997) Assumes regular variation of joint survivor function Yields more general forms of extremal dependence ⇒ Asymptotic dependence, asymptotic independence (with +ve,-ve association), "hidden regular variation" "Ray" extensions Hybrid spatial dependence model (Wadsworth and Tawn 2012) Conditional extremes: (Heffernan and Tawn 2004) Assumes, given one variable being extreme, convergence of distribution of remaining variables Allows some variables not to be extreme Extensions
The process of estimating the effect of a changing climate on the severity of future ocean storms... more The process of estimating the effect of a changing climate on the severity of future ocean storms is plagued by large uncertainties; for safe design and operation of offshore structures, it is nevertheless important that best possible estimates of climate effects is made given the available data. We explore the variability in estimates of 100-year return value of significant wave height (H S), and changes in estimates over a period of time, for output of WAVEWATCH-III models from 7 representative CMIP5 GCMs, and the FIO-ESM v2.0 CMIP6 GCM. Non-stationary extreme value analysis of peaks-over-threshold (POT) and block maxima (BM) using Bayesian inference provide posterior estimates of return values as a function of time; MATLAB software for the extreme value analysis is provided. Best overall estimates for return values, and changes in return value over the period 1979-2100, are calculated by averaging estimates for individual GCMs. We focus attention on neighbourhoods of locations east of Madagascar (EoM) and south of Australia (SoA) where a previous study of CMIP5 output by Meucci et al. (2020) reported significant decrease and increase in H S respectively, under RCP4.5 and RCP8.5 climate scenarios. There is large variation between return value estimates from different GCMs, and with longitude and latitude within each neighbourhood for estimates based on samples corresponding to ≤ 165 years of model output; these sources of uncertainty tend to be larger than that due to typical modelling choices (such as choice of threshold for POT, or block length for BM). However, we also find that careful threshold choice and block length are critical EoM, because of the presence of a mixed population of storms there. Nevertheless, there is general evidence supporting the trends reported by Meucci et al. (2020), but these findings are conditional on the choice of 8 GCMs being representative of climate evolution. We use simple randomisation testing to identify "significant" departures from steady climate. The long 700-year pre-industrial control (piControl) output of the CMIP6 GCM offers an excellent opportunity to quantify the apparent inherent variability in return value as a function of time, estimated using a subsample of output corresponding to a continuous time interval of between 20 and 160 years in length, where no climate forcing is present. We find large variation in return value estimates of approximately ±15% made from samples corresponding to periods of time of around 50 years drawn from piControl data.
Existing metocean design criteria for offshore facilities in the South China Sea have been estima... more Existing metocean design criteria for offshore facilities in the South China Sea have been estimated using different data and procedures, some of which are at least partly ad hoc. As a result, it is probable that existing criteria are inconsistent, in the sense that assets designed to the same design codes have different realised levels of integrity. To address this concern in this paper, we apply a large-scale extreme value model adapted to parallel computing environment, applied to the recent high-resolution SEAFINE hindcast database. This not only ensures design criteria that are statistically and spatially consistent but is also faster by avoiding the need for repetitive site-specific analysis. We have to overcome several challenges before we can apply a large-scale extreme model on the SEAFINE database. These include identifying a spatially consistent set of storm peaks and validating the hindcast data with real measurements. We then estimate marginal return values for significant wave height for locations within a large spatial neighbourhood, accounting for spatial and storm directional variability of peaks over threshold. A quantile regression identifies the extreme value threshold, the rate of exceedance of which is described using a Poisson process. The size of threshold exceedances is described by a generalised Pareto model. The characteristics of the threshold, rate and size models are all non-stationary with respect to directional and spatial covariates, parameterised in terms of (multidimensional) penalised B-splines. Parameter estimation is computationally challenging, but a combination of efficient generalised linear array algorithms executed within a parallel computing environment enable maximum likelihood estimation of all models. Bootstrap resampling is used to estimate uncertainties of model parameters and return values. We thus estimate consistent marginal return values for significant wave height and their uncertainties, at all locations in the spatial neighbourhood. In addition, we quantify directional variability of return values across different return periods. We rigorously validate the proposed spatiodirectional model with that of a direction-only model by deriving model diagnostics for the same site and demonstrating equivalent goodness of fits. To our knowledge this is the first of a kind of application of large-scale estimation for the South China Sea. With this approach, design criteria for large spatial domains, non-stationary with respect to the appropriate environmental covariates, can be estimated efficiently, consistently and with quantified uncertainty.
Computational Statistics & Data Analysis, Mar 1, 2016
Safe and reliable design and operation of fixed and floating marine structures often located in r... more Safe and reliable design and operation of fixed and floating marine structures often located in remote and hostile environments is challenging. Rigorous extreme value analysis of meteorological and oceanographic data can greatly aid the design of such structures. Extreme value analysis is typically undertaken for single spatial locations or for small neighbourhoods; moreover, non-stationary effects of covariates on extreme values are typically accommodated in an ad-hoc manner. The objective of the work summarised here is to improve design practice by estimating environmental design conditions (such as return values for extreme waves, winds and currents) for a whole ocean basin, including additional covariate effects (such as storm direction) as necessary, in a consistent manner. Whole-basin non-stationary extreme value modelling is computationally complex, requiring inter-alia the estimation of tail functions, the parameters of which vary with respect to multi-dimensional covariates characterised by us using tensor products of penalised B-splines. We outline two technical contributions which make whole-basin non-stationary analysis feasible. Firstly, we adopt generalised linear array methods to reduce the computational burden of matrix manipulations. Secondly, using high-performance computing, we develop a parallel implementation of maximum likelihood estimation for the generalised Pareto distribution. Together, these innovations allow estimation of rigorous whole-basin extreme value models in reasonable time. We evaluate the new approach in application to marginal extreme value modelling of storm peak significant wave heights in two ocean basins, accommodating spatial and directional covariate effects.
Numerous approaches are proposed in the literature for non-stationarity marginal extreme value in... more Numerous approaches are proposed in the literature for non-stationarity marginal extreme value inference, including different model parameterisations with respect to covariate, and different inference schemes. The objective of this article is to compare some of these procedures critically. We generate sample realisations from generalised Pareto distributions, the parameters of which are smooth functions of a single smooth periodic covariate, specified to reflect the characteristics of actual samples from the tail of the distribution of significant wave height with direction, considered in the literature in the recent past. We estimate extreme values models (a) using Constant, Fourier, B-spline and Gaussian Process parameterisations for the functional forms of generalised Pareto shape and (adjusted) scale with respect to covariate and (b) maximum likelihood and Bayesian inference procedures. We evaluate the relative quality of inferences by estimating return value distributions for the response corresponding to a time period of 10× the (assumed) period of the original sample, and compare estimated return values distributions with the truth using Kullback-Leibler, Cramer-von Mises and Kolmogorov-Smirnov statistics. We find that Spline and Gaussian Process parameterisations estimated by Markov chain Monte Carlo inference using the mMALA algorithm, perform equally well in terms of quality of inference and computational efficiency, and generally perform better than alternatives in those respects.
We report a new approach to model the frequency-direction spectrum, in which the frequency-direct... more We report a new approach to model the frequency-direction spectrum, in which the frequency-direction spectra from measurements or hindcast studies are fitted simultaneously in two dimensions, frequency and direction. Depending on the amount of wind forcing on the partition, either a unimodal (swell) or bimodal (wind-sea) wave spreading function is adopted together with the spectral form which best fits the frequency spectrum. This paper describes the new method and presents the results on a measured dataset.
2016 IEEE 23rd International Conference on High Performance Computing Workshops (HiPCW), 2016
Modelling rare or extreme events is critical in many domains, including financial risk, computer ... more Modelling rare or extreme events is critical in many domains, including financial risk, computer security breach, network outage, corrosion and fouling, manufacturing quality and environmental extremes such as floods, snowfalls, heatwaves , seismic hazards and meteorological-oceanographic events like extra-tropical storms, hurricanes and typhoons. Statistical modelling enables us to understand extremes and design mechanisms to prevent their occurrence and manage their impact. Extreme events are challenging to characterise as they are, by definition, rare and unusual even in a big data world. The frequency and extent of extreme events is typically driven by both primary attributes (dependent variables) and secondary attributes (independent variables or covariates). Studies have shown that improved inference can be gained from including covariate effects in predictive models but this inclusion comes at a heavy computation cost. In this paper, we present a framework for risk estimation from extreme events that are non-stationary; i.e., they are dependent on multi-dimensional covariates. The approach is illustrated by estimation of offshore structural design criteria in a storm environment non-stationary with respect to storm direction, season and geographic location. The framework allows consistent assessment of structural reliability with thorough uncertainty quantification. The model facilitates estimation of risk for any combination of covariates, which can be exploited for improved understanding and ultimately optimal marine structural design. The computational burden incurred is large, especially since thorough uncertainty quantification is incorporated, but manageable using slick algorithms for linear algebraic manipulations and high-performance computing.
Uploads
Papers by Kevin Ewans