Calibrated Delft3D model for Wax Lake Delta (~10m resolution), using remote sensed channel network, 18 water level measurements, and 5 discharge measurements by the Delta-X team.


NASA Delta-X hydrodynamic modeling team

Xiaohe Zhang (1), Kyle Wright (2), Paola Passalacqua (2), Marc Simard (3), Sergio Fagherazzi (1)
(1) Department of Earth and Environment, Boston University, Boston, Massachusetts, USA
(2) Department of Civil, Architectural, and Environmental Engineering, University of Texas at Austin, Austin, TX, USA
(3) Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA


The main reason we modify the channel width and depth for a reasonable simulation is because:
  1) The most updated topography elevation we use includes high-resolution (1mx1m) Lidar data (above water), and boat sonar bathymetry data. However, numerous narrow channels cannot allow boating for depth measurements, and this is a common issue all over the world.
  2) There are always errors when assigning elevation data into the model mesh, no matter the size of the mesh. The model mesh with the projected topography data may fail to represent the channels,
(e.g. blocking effect) which are hydrodynamically important for flow and material transport.

Therefore, we reply heavily on the real world, the remote sensing and field observation of water level and discharge in a large spatial extent, and explore what modification we need to do to improve the numerical models for systems with complex channel networks. We aim at providing a general and fast method to carve small scale channels for poorly-surveyed coastal area, but can model the most accurate hydrodynamics. 

Note: only water level and discharge are calibrated, future sediment transport model can start with this flow-calibrated model.


This dataset includes:

1. Delft3D model setup files

This contains the standard Delft3D model setup files, the *.mdf file (the master file) showing all the modules and parameters. The most important files are *.dep for topography and *.rgh for bed roughness. The calibrated model assigned different bed roughness value to 5 different land types. All these data can be visualized over model mesh (*.grd).
Note: M00 is the standard model without modifying the topography, therefore we name it as the standard case.


2. Channel networks

This contains remote sensed binary channel networks, derived from NWI&S2 and Google Earth Map. Also the dilated channel network (dilated channels with width < 40 m to 40 m) using code provided in cost function python scripts (see part 5). *MeshCord.tif are binary channels projected to model mesh, the *flit.tif are the binary channels in standard format of remote sensing. 
Note: The model mesh is spatially varying 9-12 m x 9-12 m, and the remote sensing data is standard 10 m x 10 m.


3. Modified topography

Several optimum modified topography based on calibrations of water level and discharges. Researchers can use these *.dep files and associated *.rgh bed roughness files, replace files in standard case, and run the model. We recommend the U1 case as the best one, however in practice the optimum channel depth and width are related to the model mesh. Our mesh grid is ~ 10 m x 10 m in wetlands, the U1 case is dilated with 40 m width representing the 4xwidth, and the optimum depth of 2 m is based on field data and cost function results. Therefore other sites can decide the depth modification using cost function in part 5, then adjust channel width to 4x model mesh grid.


4. Hydrodynamic model results

Model results of water level and velocity at Oct.16 12:00, 2016 (UTC). *.mat files contains matrixs of latitude and longitude, and the data. It also has data in *.txt format.


5. Cost function model 

A developed python script, to:
  1) derive remote sensing channels automatically
  2) modify the channel geometry (width) at each pixel
  3) calculate the flow time at each pixel
  4) determine the optimum depth for channel modification based on competition between amount of material removal and flow wave travel time. @ Kyle Wright


Contact information:
Xiaohe Zhang (zhangbu@bu.edu), Boston University 
Kyle Wright (kyleawright@utexas.edu), University of Texas at Austin