Multidecadal Scale Detection Time for Potentially Increasing Atlantic Storm Surges in a Warming Climate

Lee, Benjamin Seiyon; Haran, Murali; Keller, Klaus

doi:10.1002/2017gl074606

Cited by 28 publications

(32 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Knighton et al 2017, Arns et al 2013. Additionally, previous studies have demonstrated the difficulties in making robust modeling choices using a GEV/block maxima approach (Ceres et al, 2017;Lee et al, 2017). These relative strengths/weaknesses of the GEV versus PP/GPD approaches motivate the present study to focus on constraining uncertainties within the PP/GPD model.…”

Section: Introductionmentioning

confidence: 95%

An integration and assessment of multiple covariates of nonstationary storm surge statistical behavior by Bayesian model averaging

Wong

2018

Adv. Stat. Clim. Meteorol. Oceanogr.

View full text Add to dashboard Cite

Key points• Bayesian model averaging is used to generate a statistical model that accounts for multiple possible storm surge covariates and nonstationary behavior • For a site in the southeastern United States, the model marginal likelihoods are stable across covariate time series' used, but flood hazard projections are affected by model choice • A nonstationary model that includes four covariates raises the 100-year storm surge return level by up to 23 centimeters relative to a stationary model AbstractProjections of coastal storm surge return levels are a basic requirement for effective management of coastal risks. A common approach to estimate hazards posed by extreme sea levels is to use a statistical model, which may use a time series of a climate variable as a covariate to modulate the statistical model and account for potentially nonstationary storm surge behavior (e.g., North Atlantic Oscillation index). Previous work using nonstationary statistical approaches, however, has demonstrated the importance of accounting for the many inherent modeling uncertainties. Additionally, previous assessments of coastal flood hazard using statistical modeling have typically relied on a single climate covariate, which likely leaves out important processes and leads to potential biases in the projected flood hazards. Here, I employ upon a recently developed approach to integrate stationary and nonstationary statistical models, and examine the effects of choice of covariate time series on projected flood hazard. Furthermore, I expand upon this approach by developing a nonstationary storm surge statistical model that makes use of multiple covariate time series, namely, global mean temperature, sea level, North Atlantic Oscillation index and time. I show that a storm surge model that accounts for additional processes raises the projected 100-year storm surge return level by up to 23 centimeters relative to a stationary model or one that employs a single covariate time series. I find that the total marginal model likelihood associated with each set of nonstationary models given by the candidate covariates, as well as a stationary model, is about 20%. These results shed light on how best to account for potential nonstationary coastal surge behavior, and incorporate more processes into surge projections. By including a wider range of physical process information and considering nonstationary behavior, these methods will better enable modeling efforts to inform coastal risk management.

show abstract

Section: Introductionmentioning

confidence: 95%

An integration and assessment of multiple covariates of nonstationary storm surge statistical behavior by Bayesian model averaging

Wong

2018

Adv. Stat. Clim. Meteorol. Oceanogr.

View full text Add to dashboard Cite

show abstract

“…Previous studies have provided important new insights by examining the potentially sizable impacts of non-stationarity in the treatment of storm frequency, distribution and intensity (e.g. Ceres et al 2017, Lee et al 2017, Cid et al 2016, Grinsted et al 2013, Haigh et al 2010b, Menéndez and Woodworth 2010. For example, Grinsted et al (2013) use a generalized extreme value (GEV) distribution to model extreme sea levels, and incorporate non-stationarity in the model parameters by allowing them to covary with global mean surface temperature.…”

Section: Introductionmentioning

confidence: 99%

“…Another option is to process data to achieve independence, then use shorter time lengths of blocks (Grinsted et al 2013), but the choice of processing procedure is nontrivial and the fidelity with which non-stationary behavior may be detected is uncertain (e.g. Ceres et al 2017, Lee et al 2017. The PP/GPD modeling approach is an attractive option because all events above a specified threshold are considered in fitting the model, leading to a richer set of data (e.g.…”

Section: Introductionmentioning

confidence: 99%

“…Traditional approaches often favor parsimonious use of the limited data (e.g. Karamouz et al 2017, Lee et al 2017, Buchanan et al 2015, Tebaldi et al 2012. Bayesian model averaging (BMA), however, offers an avenue to combine a range of candidate model structures by allowing the data to inform the degree to which each model is to be trusted (Hoeting et al 1999).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Neglecting model structural uncertainty underestimates upper tails of flood hazard

Wong

Klufas

Srikrishnan

et al. 2018

Environ. Res. Lett.

Self Cite

View full text Add to dashboard Cite

Coastal flooding drives considerable risks to many communities, but projections of future flood risks are deeply uncertain. The paucity of observations of extreme events often motivates the use of statistical approaches to model the distribution of extreme storm surge events. One key deep uncertainty that is often overlooked is model structural uncertainty. There is currently no strong consensus among experts regarding which class of statistical model to use as a 'best practice'. Robust management of coastal flooding risks requires coastal managers to consider the distinct possibility of non-stationarity in storm surges. This increases the complexity of the potential models to use, which tends to increase the data required to constrain the model. Here, we use a Bayesian model averaging approach to analyze the balance between (i) model complexity sufficient to capture decision-relevant risks and (ii) data availability to constrain complex model structures. We characterize deep model structural uncertainty through a set of calibration experiments. Specifically, we calibrate a set of models ranging in complexity using long-term tide gauge observations from the Netherlands and the United States. We find that in both considered cases, roughly half of the model weight is associated with the non-stationary models. Our approach provides a formal framework to integrate information across model structures, in light of the potentially sizable modeling uncertainties. By combining information from multiple models, our inference sharpens for the projected storm surge 100 year return levels, and estimated return levels increase by several centimeters. We assess the impacts of data availability through a set of experiments with temporal subsets and model comparison metrics. Our analysis suggests that about 70 years of data are required to stabilize estimates of the 100 year return level, for the locations and methods considered here.

show abstract

“…Unfortunately, estimated base flood levels or dike heights do not allow city planners to evaluate changes to damage associated with surge heights above the selected return level. Additionally, return periods for these estimated levels are typically long compared to the record of surge observations available (Grinsted et al, 2012(Grinsted et al, , 2013Menéndez and Woodworth, 2010;Ceres et al, 2017;Lee et al, 2017). This relative sparsity of data leads to large uncertainties surrounding estimates of long-period return levels (Coles, 2001) and the potential for bias in estimating extreme event risks using common extreme value analysis methods (Coles et al, 2003;Ceres et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

Optimization of multiple storm surge risk mitigation strategies for an island City On a Wedge

Ceres

Forest

Keller

2019

Environmental Modelling & Software

Self Cite

View full text Add to dashboard Cite

Managing coastal flood risks involves selecting a portfolio of different strategies. Analyzing this choice typically requires a model. State-of-theart coastal risk models provide detailed regional information, but they can be difficult to implement, computationally challenging, and potentially inaccessible. Simple economic damage models are more accessible but may not incorporate important features and thus fail to model risks and trade offs with enough fidelity to support decision making. Here, we develop a new framework to analyze coastal flood risk management. The framework is computationally inexpensive yet incorporates common features of many coastal cities. We apply this framework to an idealized coastal city and assess and optimize two objectives using combinations of risk mitigation strategies against a wide range of future states of the world. We find that optimization using combinations of strategies allows for identification of Pareto optimal strategy combinations that outperform individual strategy options.

show abstract

Multidecadal Scale Detection Time for Potentially Increasing Atlantic Storm Surges in a Warming Climate

Cited by 28 publications

References 19 publications

An integration and assessment of multiple covariates of nonstationary storm surge statistical behavior by Bayesian model averaging

An integration and assessment of multiple covariates of nonstationary storm surge statistical behavior by Bayesian model averaging

Neglecting model structural uncertainty underestimates upper tails of flood hazard

Optimization of multiple storm surge risk mitigation strategies for an island City On a Wedge

Contact Info

Product

Resources

About