Cautionary Hints on using GeoClaw

As with all of Clawpack, the GeoClaw code is provided as a research and teaching tool with no guarantee of suitability for any particular purpose, and no liability on the part of the authors. See the License for more details.

The authors believe that GeoClaw can be used for some real-world modeling of geophysical hazards, but it is the responsibility of the user to fully understand the model and its limitations and validate it for the intended purpose.

Tsunami hazard modeling

GeoClaw is currently in use for tsunami hazard assessment by several research groups. Version 4.6.1 of the code was approved in 2012 by the US National Tsunami Hazard Mitigation Program (NHTMP) for use in modeling work supported by the program, after an extensive benchmarking project, the results of which can be found on the NTHMP benchmarking page.

However, users who wish to apply GeoClaw to the real world should be aware that doing so properly requires a good understanding of the capabilities and limitations of the code, the equations they model, and the suitability of using these equations to model any particular real-world scenario.

The authors of this code have invested considerable time in learning about appropriate aspects of geohazard modeling, through reading the literature and working directly with geoscientists who are domain experts. Even so we are very cautious in using any results from GeoClaw without performing sensitivity studies, grid refinement studies, etc., and if possible comparing results with those obtained by other modeling groups and confirming with experts that the results are reasonable. It is impossible to encapsulate the knowledge needed to deal with all the inaccuracies and uncertainties of geohazard modeling in any piece of software or its documentation, and there is no replacement for extensively reading the literature and working with domain experts.

It is also important to understand the various parameters in GeoClaw and if necessary experiment with different settings and perform sensitivity studies. See Specifying GeoClaw parameters in setrun.py.

Here are a few of the things that should be considered in any GeoClaw simulation:

  • The depth-averaged shallow water equations are a fairly good model for the fluid dynamics of tsunamis provided the wave length is long relative to the depth of the water. In particular, for large tsunamis generated by subduction zone earthquakes propagating over the ocean, these equations may be adequate. However, even then, they are only an approximation. More accurate depth-averaged equations such as Boussinesq equations that include dispersive terms may be more accurate.
  • For short wavelength tsunamis such as those generated by landslides, shallow water equations are less accurate since dispersive terms can be very important. Incorporating dispersive terms in GeoClaw is planned for the future but not yet available. These limitations should be clearly understood.
  • GeoClaw solves the nonlinear shallow water equations and can capture turbulent bore formation to some extent via the formation of shock waves. It does not model wave breaking directly and in the nearshore region the use of depth-averaged equations may be inaccurate since the flow becomes fully three-dimensional. Reasonable agreement with observations from historic events and wave tank experiments have been seen in validation studies, both of GeoClaw and other shallow water codes, but caution is required.
  • The empirical Manning formulation is used to model bottom friction, as described further in the section Manning friction term, where some limitations are discussed.
  • For most tsunami simulations including the Coriolis terms in the momentum equations makes little difference in the observed results and so these terms are often turned off for efficiency (coriolis_forcing = False).
  • The geoclaw parameter sea_level determines the initial fluid depth relative to the topography, as specified by the topo files. It is important to know what vertical datum the topography is relative to. Coastal bathymetry developed for tsunami modeling is often relative to Mean High Water (MHW) at some point, in which case setting sea_level = 0. corresponds to assuming the water level being initially at MHW. See sea_level for more information.
  • Tsunami modeling generally requires specifying a seafloor displacement in order to initiate the tsunami, by specifying a dtopo file. This may be a time-dependent displacement, as explained in dtopo. However, it is important to understand that any displacement of the seafloor causes the entire water column above this point to be shifted upwards by the same amount (since the depth h is held constant), and so is immediately observed in the sea surface elevation. In reality, displacement of the seafloor leads to the propagation of acoustic waves that result in a surface displacement