Feature ArticleDevelopment of a Forecasting System For the Dubai Coastal Zone
By Hind Mahmoud Mahaba • Dr. Paul Anid • Kelly Knee
The coastline of Dubai represents one of the most unique coastal environments in the world, with extensive developments supporting a wide range of tourism and commercial activity. In order to support these coastal developments and provide information to the public on environmental conditions, Dubai Municipality (DM) determined that an operational forecasting system coupled with an advanced coastal monitoring system was required. The system couples a suite of meteorology and ocean models with data management infrastructure and Web-based information tools. It is an important step toward providing accurate and detailed information on metocean forecasts and extreme conditions along the entire Dubai coastline for use in delivering data and decision support information to government and public users.
Diagram of model connections for
operational forecasting system.
operational forecasting system.
The approach taken by the project team, which included HDR Inc. (Omaha, Nebraska) and ASA, now part of the RPS Group (Abingdon, England), was to implement the latest open-source modeling technology available from the research and academic communities to build an operational forecasting system (OFS) from a suite of interconnected components that allow for seamless interaction of models with data distribution mechanisms. Given the size and complexity of the project, it was conducted in two phases. Phase one included individual model setup, calibration, validation, configuration of the oil spill system and definition of the inundation scenarios. Phase two included model coupling, connection to global models for boundary conditions, integration with ASA’s Environmental Data Server (EDS) for data management and connection of model results to the coastal forecast and warning system website, oil spill system and inundation predictions.
Phase One: Model Implementation
The OFS is a coupled suite of meteorology and ocean models, which provides three-day forecasts of wind, waves and currents. The operational numerical models used in this system are meteorological, hydrodynamic and wave models. Metocean observations used for model setup, calibration and validation were collected in an ongoing monitoring project being led by Dr. Paul Anid at HDR Inc., and HDR collaborated closely on the integration of monitoring data with model forecasts for this project.
In order to resolve the Arabian Gulf atmospheric circulation and dynamics, the Nonhydrostatic Mesoscale Model (NMM) core of the Weather Research and Forecasting (WRF) system was implemented. WRF-NMM was developed by NOAA/National Center for Environmental Prediction (NCEP) and is designed to be flexible, portable and efficient in a massively parallel computing environment. An important feature of the WRF implementation for Dubai is the use of nested grids, which allowed for higher resolution in the immediate vicinity of Dubai. The first grid (G1) represents atmospheric dynamics for the entire Arabian Gulf at a resolution of 0.1 degrees (approximately 10.5 kilometers). The nested grid (G2) represents atmospheric dynamics in greater detail for the Dubai coastal zone. The nesting ratio for WRF-NMM is three; thus, the grid resolution of the nested grid is 0.033 degrees (approximately 3.5 kilometers).
Initial and boundary conditions for WRF are from NOAA’s NCEP’s Global Forecast Model (GFS), which is a global spectral data assimilation and forecast model system. The horizontal resolution of the global forecast is 0.5 degrees, with three hourly outputs. GFS variables used to force WRF include wind components, temperature, pressure, humidity, soil temperature and moisture content.
The Regional Ocean Model System (ROMS), a free-surface, terrain-following, primitive-equation ocean model, was chosen to model Arabian Gulf hydrodynamics using two model domains. The ARG domain spans the Arabian Gulf and the DUB domain covers the coastal zone of Dubai. This dual grid approach is necessary to provide a numerical solution that is both physically accurate and numerically stable.
The horizontal resolution of each domain was optimized based on a balance between resolving the ocean patterns along the Dubai coast, allowing sufficient model sophistication for simulating regional physics, machine limitations, and the stability and reliability of the numerical solution.
ROMS requires tidal forcing, as well as surface and artificial boundary conditions. To create a robust operational modeling structure, much of the necessary forcing was internally modeled, with WRF providing the wind forcing for both domains and the ARG domain providing the boundary conditions for the DUB domain.
Tidal forcing was provided by the TOPEX/Poseidon model, which showed that the most important (largest amplitude) components in the region are M2, S2, N2, K1, K2, O1, P1 and Q1.
At the surface boundary, ocean-atmosphere bulk variables are required. These variables include wind components, radiational fluxes, air pressure, temperature, specific humidity and precipitation, all of which were provided by the WRF meteorological model for both model domains.
The artificial boundary of the ARG domain is at the entrance to the Arabian Gulf, known as the Strait of Hormuz. The information required at the boundary, including temperature, salinity, barotropic and baroclinic velocities, and sea surface elevation were generated using HYCOM, a primitive-equation, ocean general circulation model. Because the tidal variations are already being represented by TOPEX, only the low-frequency, large-scale ocean patterns are required. To generate these, HYCOM was run for several years. The model was forced by NCEP reanalysis climatological data, which included atmospheric forces, such as winds, heat flux, precipitation, and hydrological data, such as salinity and temperature. The ARG model provides all the initial and boundary information to the local domain (DUB).
To represent wind-induced wave conditions, SWAN, a third-generation, spectral-phase-averaging wave model developed by the Delft University of Technology, was implemented. SWAN is a numerical model for generating estimates of wave parameters in coastal areas for given wind, bottom and current conditions.
The SWAN model application was used to simulate spatially varying wave conditions over a domain spanning the Arabian Gulf. The size of the wave model domain provides sufficient fetch to capture wave propagation in the Gulf. In spite of the large domain, the model grid must also be of sufficient resolution to accurately resolve the nearshore wave dynamics of the complex Dubai coastline. To meet these requirements, the unstructured mesh version of SWAN was employed, allowing the Gulf to be covered with a variable resolution mesh with cells ranging in size from 60 meters to 45 kilometers. The key outputs from the model used in this study included significant wave height (Hs), peak wave period (Tp), mean wave period (Tm) and peak wave direction (θ peak).
Given the Arabian Gulf geometry, SWAN only requires offshore boundary conditions at a single open boundary, the eastern entrance to the Gulf. These offshore boundary conditions are defined using deepwater wave parameters obtained from the NOAA NCEP WAVEWATCH III (WW3) global wave model. Wave parameters are extracted from the WW3 forecast dataset for the closest point to the open boundary. Other boundary conditions are supplied by the WRF and ROMS model outputs. To continue this article please click here.
Hind Mahmoud Mahaba is head of the coastal monitoring and design unit of the coastal zone and waterways management section at the Environment Department of Dubai Municipality. Her vision for a coastal zone and waterways operation decision support system led to the development of the Dubai Operational Forecasting System.
Dr. Paul Anid is vice president of water quality management services at HDR Inc. and has 28 years of experience in environmental management, including modeling the fate and transport of pollutants, assessing water quality criteria, applying conventional and emerging approaches for assessing contaminated sediments and developing environmental decision support models.
Kelly Knee is manager of the ASA Coastal Hazards Group and has 10 years of experience in coastal risk and climate change assessments, including modeling and visualization of storm surge, waves, sea level rise and tsunami impacts for coastlines around the world.