Storm hazard analysis over extended geospatial grids utilizing surrogate models

Kyprioti, Aikaterini P.; Taflanidis, Alexandros A.; Nadal-Caraballo, Norberto C.; Campbell, Madison

doi:10.1016/j.coastaleng.2021.103855

Cited by 27 publications

(17 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, Taflanidis et al (2013a)'s approach adopted herein to fill the missing peak storm surge values in this study is sensitive to the peak storm surge of the nearest point. Thus, future studies should investigate several interpolation approaches (e.g., k-nearest neighbor interpolation with an inverse distance weighting (Kyprioti et al, 2021)) that can improve the performance of the C1PKNet model.…”

Section: Future Workmentioning

confidence: 99%

“…However, these surrogate models are designed only to predict one or two locations at a time and did not consider overland locations, which are too ine cient to apply across an extensive coastal region that spans a large number of geographic locations. For these reasons, several other surrogate modeling approaches have been developed, such as a moving least squares response surface (Taflanidis et al, 2013a,b), kriging with principal component analysis (PCA) (Jia and Taflanidis, 2013;Jia et al, 2016;Bass and Bedient, 2018), kriging with PCA and k-means clustering (Kyprioti et al, 2021), and an MLP with PCA (Bass and Bedient, 2018), to predict storm surges at a large number of geographic locations at the same time. Moreover, to consider overland locations, several methods were developed to estimate storm surge at the overland location from the water level at the o shore location using a one-dimensional Boussinesq model (Taflanidis et al, 2013b), from the water level at the nearest location (Taflanidis et al, 2013a), or from the water level at the nearby locations using k-nearest neighbor interpolation (Kyprioti et al, 2021).…”

Section: Introductionmentioning

confidence: 99%

“…Secondly, several surrogate modeling studies did not consider bypassing storms that pass by the study area but do not make landfall within or near the study area (see Table 1). For example, Hashemi et al (2016) and Kyprioti et al (2021) used the North Atlantic Comprehensive Coastal Study (NACCS) synthetic tropical cyclone surge database of the US Army Corps of Engineers (Cialone et al, 2015), which contains storm surge simulations for 1050 synthetic tropical cyclones (TCs) in the US North Atlantic region (Virginia 2015) and references therein), the white circles are the NACCS save points (Cialone et al, 2015), and the red circles represent the NOAA stations (Tides and Currents, 2021) (see Table 2). In the right map, the black lines are the NACCS synthetic tropical cyclone (TC) tracks.…”

Section: Introductionmentioning

confidence: 99%

“…The NACCS defined the bypassing storm as a storm where track headings is larger than zero degree clockwise from North. Although bypassing storms make up approximately 40% of the NACCS database, Hashemi et al (2016) and Kyprioti et al (2021) excluded bypassing storms in surrogate modeling. The lack of capacity to respond to bypassing-track storms is a major problem for surrogate models because TCs, such as Hurricane Irene (2011), cannot be addressed.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Rapid prediction of peak storm surge from tropical cyclone track time series using machine learning

Lee

Irish

Bensi

et al. 2021

Coastal Engineering

View full text Add to dashboard Cite

Rapid and accurate prediction of peak storm surges across an extensive coastal region is necessary to inform assessments used to design the systems that protect coastal communities' life and property. Significant advances in high-fidelity, physics-based numerical models have been made in recent years, but use of these models for probabilistic forecasting and probabilistic hazard assessment is computationally intensive. Several surrogate modeling approaches based on existing databases of high-fidelity synthetic storm surge simulations have been recently suggested to reduce computational burden without substantial loss of accuracy. In these previous studies, however, the surrogate modeling approaches relied on a tropical cyclone condition at one moment (usually at or near landfall), which is not always most correlated with the peak storm surge. In this study, a new one-dimensional convolutional neural network model combined with principal component analysis and a k-means clustering (C1PKNet) is presented that can rapidly predict peak storm surge across an extensive coastal region from time-series of tropical cyclone conditions, namely the storm track. The C1PKNet model was trained and cross-validated for the Chesapeake Bay area of the United States using existing database of 1031 high-fidelity storm surge simulations, including both landfalling and bypassing storms. Moreover, the performance of the C1PKNet model was evaluated based on observations from three historical hurricanes (Hurricane Isabel in 2003, Hurricane Irene in 2011, and Hurricane Sandy in 2012. The results indicate that the C1PKNet model is computationally e cient and can predict peak storm surges from realistic tropical cyclone track time-series. We believe that this new surrogate model can enhance coastal resilience by providing rapid storm surge predictions.

show abstract

Section: Future Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Rapid prediction of peak storm surge from tropical cyclone track time series using machine learning

Lee

Irish

Bensi

et al. 2021

Coastal Engineering

View full text Add to dashboard Cite

show abstract

“…Further, we may be able to more effectively utilize these high fidelity simulations for Event Maps through surrogate modeling techniques (Bass and Bedient, 2018;Zahura et al, 2020;Contreras et al, 2020;Kyprioti et al, 2021), similar to how the NWC-HAND and Fathom-US utilize a precomputed riverine hydraulics in those implementations (Zheng et al, 2018;Wing et al, 2019). This paper seeks to investigate if different modeling frameworks produce substantially different Event Maps during compound coastal flood events.…”

Section: Introductionmentioning

confidence: 99%

Comparison of Flood Inundation Modeling Frameworks within a Small Coastal Watershed during a Compound Flood Event

Gutenson

Tavakoly

Islam

et al. 2022

Preprint

View full text Add to dashboard Cite

Abstract. The flooding brought about by compound coastal flooding can be devastating. Before, during, and immediately following these events, flood inundation maps, or Events Maps, can provide essential information to emergency management. However, there are a number of frameworks capable of estimating Event Maps during flood events. In this article, we evaluate three such Event Map frameworks in the context of Hurricane Harvey. Our analysis reveals that each of the three frameworks provide different inundation maps that differ in their level of accuracy. Each of the three Event Maps also produce different exposure and consequence estimates because of their physical differences. This investigation highlights the need for a centralized means of vetting and adjudicating multiple Event Maps during compound flood events empowered by the ability to distribute Event Maps as geographic information system (GIS) services and coalesce Event Maps into a common operating picture. Furthermore, we provide evidence that the ability to produce multi-model estimates of Events Maps to create probabilistic Event Maps may provide a better product than the use of a lone Event Map.

show abstract

Improved metamodels for predicting high-dimensional outputs by accounting for the dependence structure of the latent variables: application to marine flooding

Rohmer

Sire

Lecacheux

et al. 2023

Stoch Environ Res Risk Assess

View full text Add to dashboard Cite

Metamodelling techniques have shown high performance to overcome the computational burden of numerical hydrodynamic models for fast prediction of key indicators of marine flooding (e.g. total flooded area). To predict flood maps (e.g. spatial distribution of maximum value of water depth during a flood event), a commonly-used approach is to rely on principal component analysis to reduce the high dimensionality of the flood map (related to the number of pixels typically of several 1,000s) by transforming the spatial output into a low number of latent variables (typically <10). One commonly-used approach is to build one metamodel per latent variable by assuming independence between the latent variables. Using two real cases of marine flooding, we show that the predictive performance of the metamodelling approach (relying on kriging metamodels) can significantly be improved when the dependence structure of the latent variables is accounted for. Our tests show that the most efficient approach relies on the clustering in the space of the latent variables (here with k-means algorithm).Complementing the approach with a kriging metamodel specifically dedicated to handle vectorvalued variables allows an additional increase of predictability for the case with the larger size of the training dataset.

show abstract

Storm hazard analysis over extended geospatial grids utilizing surrogate models

Cited by 27 publications

References 27 publications

Rapid prediction of peak storm surge from tropical cyclone track time series using machine learning

Rapid prediction of peak storm surge from tropical cyclone track time series using machine learning

Comparison of Flood Inundation Modeling Frameworks within a Small Coastal Watershed during a Compound Flood Event

Improved metamodels for predicting high-dimensional outputs by accounting for the dependence structure of the latent variables: application to marine flooding

Contact Info

Product

Resources

About