This repo is a quick comparison of standardized indices for U.S. west coast groundfish species, comparing using the 'n_knots' argument in sdmTMB::make_mesh() to the 'cutoff' argument, representing cutoff distance. The 'n_knots' argument has been implemented in the NWFSC's indexwc package for index standardization. One of the goals of changing the argument from 'n_knots' to 'cutoff' is that the complexity of the mesh generated by make_mesh() generally has more knots than the argument passed to 'n_knots' (like VAST, this is done with kmeans clustering). As an example with the U.S. west coast data,
To compare the cutoff distance approach to the 'n_knots' conventional approach, I looked at all models using the WCGBTS survey data, and for each found the cutoff distance that generated a mesh with the closest number of vertices (knots) to the mesh created using the 'n_knots' argument. For each, I implemented the same sdmTMB model, and for the models that passed all sdmTMB::sanity() checks, I overlayed the two indices (shown below). From these results, it appears that there is very little difference in indices generated from these two approaches.
