Model Simulation Studies of Hurricane Isabel in Chesapeake Bay

1
Model Simulation Studies of Hurricane Isabel in
Chesapeake Bay
Jian Shen Virginia Institute of Marine
Sciences College of William and Mary
2
Isabel The 100-Year Storm !
3
Background of Storm Surge Modeling

• Numerical models have been successfully applied
  to simulate and predict tide and storm surge in
  coastal seas
• SLOSH (Sea, Lake, and Overland Surges for
  Hurricanes
• ADCIRC (Advanced Circulation Model)
• Impact of the storm surge at any particular
  location is sensitive to meteorological and
  topographic parameters
• Inundation is crucial for disaster planning
• Prediction of flooding areas depends on model
  grid resolution

4
New Challenges for Numerical Modeling

• More high resolution terrain data are available
• LIDAR (LIght Detection And Ranging)
• More real-time observation data are available
• Surface elevation
• Vertical velocity profile
• Wave
• Real-time simulation vs. prediction
• Rescue
• Inundation
• How to integrate high resolution terrain and
  real-time observation data into models ?

5
VIMS Real-Time Observation System
6
Current Observation at Gloucester Point
7
Airborne LIDAR Data

• LIDAR (LIght Detection And Ranging)
• Infrared laser ranger provides distance to ground
• Differential GPS provides aircraft location
• Inertial reference system provides aircraft
  orientation
• Post-processing provides x,y,z coordinates of
  ground surface
• Construction of Digital Elevation Models (DEMs)

8
DTMS
3 Sec (80-90 m) DTM Vertical Res. 1 m
10 DTM Vertical Res. 0.3 m
9
Example of LIDAR Data (Miami)
10
Model Domain Representation

• Small domain is inadequate for storm simulation
• Coarse grid is inadequate to resolve irregular
  shoreline and small topographic features in
  estuary
• Structured grid is difficult to represent complex
  bathymetric in estuary
• Unstructured grid has advantage of storm surge
  modeling
• Use nested grids
• Place fine grid in the areas of interest and
  coarse grid in the remaining large areas

11
Model Requirements

• Must resolve complex shoreline
• Must resolve land features
• coastal ridge, dam, inlet, and river
• small scale on the order of meters
• Must cover large modeling domain
• Must be computationally efficient

12
Example of Unstructured Grid (Miami)
13
Example of Nested Model Grids
14
Unstructured 3D Model (UnTRIM)

• UnTRIM incorporates an Unstructured grid into
  TRIM model (Tidal, Residual, Intertidal Mudflat),
  originally developed by Vincenzo Casulli
• It simulates three-dimensional hydrodynamic and
  transport processes
• It uses an orthogonal unstructured grid
• It conserves mass locally as well as globally
• It uses Eulerian-Langangian transport scheme
• It employs semi-implicit finite difference and
  finite volume method- very efficient
  computationally
• It is capable of simulating wet-dry processes

15
Grid Structure

• Use polygons to represent a prototype estuary
  (3-, 4-, 5-sides)
• Better fitting complicated geometry in estuarine
  and coastal environment
• Using orthogonal grid simplifies the numerical
  algorithm

16
Model Simulation Studies

• Study the accuracy of model prediction of Isabel
  forced by a stationary, circular wind model
• Compare model prediction with and without
  simulating inundation
• Study influence of open boundary condition
  specification on surge simulation
• still boundary condition vs. inverse pressure
  adjust boundary condition
• Study influence of model domain size on surge
  prediction
• Study influence of wind field on model prediction
• Study influence of hurricane on transport

17
Model Grids
Surface elements 121,338
Surface elements 239,541
18
Grid layouts at York and James Rivers
19
Model Calibration

• Calibrate model for tide
• forced by 9 tidal constituents M2, S2, K1,
  O1, Q1, K2, N2, M4, and M6
• Model was run for 3 months and the results of the
  last 29 days were used for computing tidal
  harmonics
• Time step 5 min.

20
Tidal Simulation (M2 tide)
21
Tidal Simulation (K1)
22
Observation Stations
23
Tidal Constituents Comparison (Amplitude)
Amplitude is in m Observations are based on 1992
data
24
Tidal Constituents Comparison (phase)
Phase is in degree
25
Wind Model (Myers and Malkin, 1961)

• r the distance from the storm center
• p(r) pressure, pa central pressure,
• the inflow angle across circular isobar toward
  the storm center
• V is the wind speed, f Coriolis parameter,
• ks and kn are friction coefficients.

26
Example of Wind Field
27
Isabel Simulation ResultsWith and Without
Simulating inundation
28
Comparison of Model ResultsWith and Without
Simulating Inundation
29
Comparison of Model ResultsWith and Without
Simulating Inundation
30
Test Influence of Open Boundary Condition

• Apply inverse pressure adjustment at BC
• Run large domain model (east coast) and apply
  time series output from large domain model to
  force small domain model

31
Large Domain Model Simulation
32
Influence of Open Boundary Condition
Inverse Pressure Adjustment
Nested Grids
33
Influence of Wind Field
34
Wind Field Analysis
35
Comparison of Using Different Wind Fields
36
Current Simulation at Gloucester Point
Model Wind Field
Modified Wind Field
37
Conclusions

• Unstructured model is very efficient in
  simulation tide and storm surge
• Open boundary condition specification influences
  the surge prediction
• Wind field is critical in the accurate simulation
  of storm surge

38
Thanks !