NEEShub services are beginning to be phased out. Please begin using DesignSafe ( Please submit a support ticket ( if you need any assistance.


Support Options

Submit a Support Ticket


Questions and Answers

0 Like

Kris Vanneste

Auto-discretization in Strata: does it work correctly?

Dear developers,

We have experimented with Strata to determine the transfer function at a site in Belgium using equivalent-linear modeling. We have compared the result with two similar tools: EERA and DEEPSOIL.

In Strata (version 3.69), we used a wavelength fraction of 0.2 and a maximum frequency of 25 Hz. Varying the wavelength fraction within the allowed bounds (0.1 – 0.35) did not change the result noticeably. The result was also very similar to that obtained with DEEPSOIL (which has no settings at all to control layer discretization). In EERA (an Excel implementation of SHAKE91), layer discretization has to be specified manually for each layer. If we subdivide the layers in a way that should be comparable to the parameters we used in Strata, we obtain different results. However, changing the subdivision of layers in EERA results in significant changes, with peaks shifting to lower frequencies as subdivision increases and vice versa. So far, we assumed EERA was wrong, and Strata and DEEPSOIL are right.

We noticed that in version 3.81 of Strata, there is an option to disable auto-discretization. So, we are wondering if there was a problem with this feature after all?

Thanks for your time, Kris Vanneste

Report abuse

1 Responses

  1. 0 Like 0 Dislike

    Albert Kottke

    The auto-discretization works. An option was added to disable the auto-discretization after a user requested it. However, if you are varying shear-wave velocity and/or layering then it is recommended that auto-discretization is used.

    As far as your observations, bounds on the wave-length fraction are provided such that provided values range within the appropriate range. Furthermore, EQL site response is not very sensitivity to the layer thickness because it does not influence the computed transfer function, but rather captures the changes in strain through a specific soil layer. I am not familiar with EERA, but Deepsoil is much more sensitive to the layer thickness because the layers govern the highest frequency at which can be accurately modeled.

    Reply Report abuse

    Please login to answer the question.