search

sandialabs / SNL-ESSC

Sandia National Laboratories - Extreme Sea State Contour
Apache License 2.0
3 stars 1 forks source link

Bimodally Distributed Data #3

Open sugarl14 opened 3 years ago

sugarl14 commented 3 years ago

Hi,

I am trying to predict the extreme sea states at a specific reference site on the east coast using data from NDBC 44056. I've have read online that the IFORM technique does not work well for buoy datasets that are bimodally distributed, and I believe this buoy has a dataset that is bimodally distributed.

Would the ESSC code be suitable for bimodally distributed data?

aubreyeckert commented 3 years ago

Thank you for your interest in the ESSC @sugarl14. Unfortunately most contour methods, including those in the ESSC, have challenges with bimodally distributed data because the underlying conditional distribution structure does not work well with bimodality. The best option might be to try the Kernel Density Estimation (KDE) contour method, though this has not been extensively tested on bimodal data.

sugarl14 commented 3 years ago

Thank you!