-
Notifications
You must be signed in to change notification settings - Fork 7
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
HYCOM-related likelihoods #80
Comments
Indeed this issue is coming from dim issues in HYCOM itself. In |
forget any hycom-based calculations for a minute. we have to solve the resolution issue at the source. in your call to
|
Yeah, that solved it. New raster has the correct dimensions. But this raises a big question for me. If we do a project only with data from the later HYCOM version, are our movement kernels and grid unit based calculations accounting for that different resolution? Everything was designed when HYCOM was 0.8 x 0.8 but now its 0.8 x 0.4. So if we use only new HYCOM in an upcoming study, shouldn't we just force it to always give us the data in 0.8 x 0.8? Or do we have internal catches to compensate for that grid unit size change? |
i think resampling to a regular grid is best. We're taking the centroid of cells as positions which would skew things latitudinally. Resampling should allow the movement kernel to have the proper influence on likelihood; without doing that it seems precision would be sacrificed. does anyone know why this change was made?? Perhaps the gurus found that HYCOM performs better with lower latitudinal precision.. |
maybe the way forward is to include a hycom template grid with the package data. anytime someone asks for a modern HYCOM grid with the weird resolution, it gets auto resampled and a warning is pushed notifying of the change. that way all downstream stuff remains unchanged. |
Running into a problem on 'dev' branch with calculating HYCOM-related likelihoods for data from 2019 and 2020 (presumably after the 2018-11-20 end of GOFS 3.0).
In calc.ohc.par (as an example), I've isolated the issue as a mismatch in the size of the latitude dimension among L.ohc (where it is correct) and lik.ohc (where it is incorrect). Notably, on line 138:
This yields the correct longitude dimension size, dim(dat)[1], but the incorrect latitude dimension size, dim(dat)[2], relative to line 119:
This mismatch ultimately results in an error on lines 242-244 where the incorrectly sized lik.ohc is assigned into the correctly sized L.ohc:
The text was updated successfully, but these errors were encountered: