28 Aug 2023

Scaling and Sensitivity Analysis of Machine Learning Regression on Periodic Functions

Abstract: In this report we document the scalability and sensitivity of machine learning (ML) regression on a periodic, highly oscillating, and 𝐶∞ function. This work is motivated by the need to use ML regression on periodic problems such as tidal propagation. In this work, TensorFlow is used to investigate the machine scalability of a periodic function from one to three dimensions. Wall clock times for each dimension were calculated for a range of layers, neurons, and learning rates to further investigate the sensitivity of the ML regression to these parameters. Lastly, the stochastic gradient descent and Adam optimizers wall clock timings and sensitivities were compared.

8 Dec 2022

Coastal Hazards System–Puerto Rico and US Virgin Islands (CHS-PR)

Abstract: The South Atlantic Coastal Study (SACS) was completed by the US Army Corps of Engineers to quantify storm surge and wave hazards allowing for the expansion of the Coastal Hazards System (CHS) to the South Atlantic Division (SAD) domain. The goal of the CHS-SACS was to quantify coastal storm hazards for present conditions and future sea level rise (SLR) scenarios to aid in reducing flooding risk and increasing resiliency in coastal environments. CHS-SACS was completed for three regions within the SAD domain, and this report focuses on the Coastal Hazards System–Puerto Rico and US Virgin Islands (CHS-PR). This study applied the CHS Probabilistic Coastal Hazard Analysis (PCHA) framework for quantifying tropical cyclone (TC) responses, leveraging new atmospheric and hydrodynamic numerical model simulations of synthetic TCs developed explicitly for the CHS-PR region. This report focuses on documenting the PCHA conducted for CHS-PR, including the characterization of storm climate, storm sampling, storm recurrence rate estimation, marginal distributions, correlation and dependence structure of TC atmospheric-forcing parameters, development of augmented storm suites, and assignment of discrete storm weights to the synthetic TCs. As part of CHS-PR, coastal hazards were estimated for annual exceedance frequencies over the range of 10 yr⁻¹ to 10⁻⁴ yr⁻¹.

31 Aug 2022

Coastal Hazards System–Louisiana (CHS-LA)

Abstract: The US Army Engineer Research and Development Center (ERDC), Coastal and Hydraulics Laboratory (CHL) expanded the Coastal Hazards System (CHS) to quantify storm surge and wave hazards for coastal Louisiana. The CHS Louisiana (CHS-LA) coastal study was sponsored by the Louisiana Coastal Protection and Restoration Authority (CPRA) and the New Orleans District (MVN), US Army Corps of Engineers (USACE) to support Louisiana’s critical coastal infrastructure and to ensure the effectiveness of coastal storm risk management projects. The CHS-LA applied the CHS Probabilistic Coastal Hazard Analysis (PCHA) framework to quantify tropical cyclone (TC) responses, leveraging new atmospheric and hydrodynamic numerical model simulations of synthetic TCs developed explicitly for the Louisiana region. This report focuses on documenting the PCHA conducted for the CHS-LA, including details related to the characterization of storm climate, storm sampling, storm recurrence rate estimation, marginal distributions, correlation and dependence structure of TC atmospheric-forcing parameters, development of augmented storm suites, and assignment of discrete storm weights to the synthetic TCs. As part of CHS-LA, coastal hazards were estimated within the study area for annual exceedance frequencies (AEFs) over the range of 10 yr-1 to 1×10-4 yr-1.

12 Aug 2021

Rapid Tidal Reconstruction for the Coastal Hazards System and StormSim Part II: Puerto Rico and US Virgin Islands

Abstract: This Coastal and Hydraulics Engineering Technical Note (CHETN) describes the continuing efforts towards incorporating rapid tidal time-series reconstruction and prediction capabilities into the Coastal Hazards System (CHS) and the Stochastic Storm Simulation System (StormSim). The CHS (Nadal-Caraballo et al. 2020) is a national effort for the quantification of coastal storm hazards, including a database and web tool (https://chs.erdc.dren.mil) for the deployment of results from the Probabilistic Coastal Hazard Analysis (PCHA) framework. These PCHA products are developed from regional studies such as the North Atlantic Coast Comprehensive Study (NACCS) (Nadal-Caraballo et al. 2015; Cialone et al. 2015) and the ongoing South Atlantic Coast Study (SACS). The PCHA framework considers hazards due to both tropical and extratropical cyclones, depending on the storm climatology of the region of interest. The CHS supports feasibility studies, probabilistic design of coastal structures, and flood risk management for coastal communities and critical infrastructure. StormSim (https://stormsim.erdc.dren.mil) is a suite of tools used for statistical analysis and probabilistic modeling of historical and synthetic storms and for stochastic design and other engineering applications. One of these tools, the Coastal Hazards Rapid Prediction System (CHRPS) (Torres et al. 2020), can perform rapid prediction of coastal storm hazards, including real-time hurricane-induced flooding. This CHETN discusses the quantification and validation of the Advanced Circulation (ADCIRC) tidal constituent database (Szpilka et al. 2016) and the tidal reconstruction program Unified Tidal analysis (UTide) (Codiga 2011) in the Puerto Rico and US Virgin Islands (PR/USVI) coastal regions. The new methodology discussed herein will be further developed into the Rapid Tidal Reconstruction (RTR) tool within the StormSim and CHS frameworks.

12 Aug 2021

Rapid Tidal Reconstruction with UTide and the ADCIRC Tidal Database

Abstract: The quantification of storm surge is vital for flood hazard assessment in communities affected by coastal storms. The astronomical tide is an integral component of the total still water level needed for accurate storm surge estimates. Coastal hazard analysis methods, such as the Coastal Hazards System and the StormSim Coastal Hazards Rapid Prediction System, require thousands of hydrodynamic and wave simulations that are computationally expensive. In some regions, the inclusion of astronomical tides is neglected in the hydrodynamics and tides are instead incorporated within the probabilistic framework. There is a need for a rapid, reliable, and accurate tide prediction methodology to provide spatially dense reconstructed or predicted tidal time series for historical, synthetic, and forecasted hurricane scenarios. A methodology is proposed to combine the tidal harmonic information from the spatially dense Advanced Circulation hydrodynamic model tidal database with a rapid tidal reconstruction and prediction program. In this study, the Unified Tidal Analysis program was paired with results from the tidal database. This methodology will produce reconstructed (i.e., historical) and predicted tidal heights for coastal locations along the United States eastern seaboard and beyond and will contribute to the determination of accurate still water levels in coastal hazard analysis methods.

9 Jul 2020

PUBLICATION NOTICE: Rapid Tidal Reconstruction for the Coastal Hazards System and StormSim Part I: Coastal Texas and Louisiana

Abstract: incorporating a rapid tidal time series reconstruction and prediction subroutine within the Coastal Hazards System (CHS) framework. The CHS (https://chs.erdc.dren.mil) is a national database and web tool that provides probabilistic coastal hazard analysis (PCHA) products developed from regional studies such as the North Atlantic Coast Comprehensive Study (Nadal-Caraballo et al. 2015; Cialone et al. 2015). PCHA considers hazards due to both tropical and extratropical cyclones, depending on the storm climatology of the region of interest. The CHS supports feasibility studies, probabilistic design of coastal structures, flood risk management for coastal communities, and critical infrastructure. In the case of tropical cyclones (TCs) or hurricanes, both the timing of landfall and the level of the astronomical tide at the landfall location are critical in determining the magnitude of the still water level (i.e., storm surge + wave setup + astronomical tide). Therefore, a robust and accurate tide prediction methodology is needed to provide reliable reconstruction of tidal time series for historical, synthetic, and forecasted hurricane scenarios. This CHETN also discusses the quantification and validation of the Advanced Circulation (ADCIRC) tidal constituent database in the coastal Texas and Louisiana region as well as the implementation of the tidal reconstruction program Unified Tidal analysis (UTide) in the CHS framework.