search

SECOORA / skill_score

Prototypes for the SECOORA skill score
MIT License
7 stars 1 forks source link

Simple demo of HF radar comparison #16

Closed rsignell-usgs closed 10 years ago

rsignell-usgs commented 10 years ago

It should be possible to now discover the HF Radar from UCSD and at least one SECOORA model (SABGOM) via the catalog. Can we get a sample notebook going that compares the two surface currents fields? Let's use Cartopy for plotting, I'm thinking something like https://www.wakari.io/sharing/bundle/rsignell/Iris_HFR_plot but @ocefpaf , likely you have a better idea.

For calculating surface currents from the model for comparison to HFR, I imagine we eventually should do some kind of averaging of the top layers from the model, based on wave height, perhaps? Need to look at what Wilkin and Hunter did in their paper. But for now, let's just use the top layer from the model.

ocefpaf commented 10 years ago

Working in that! Will have something for you tomorrow or Wednesday at best. Just need to finish the SECOORA inundation.

jcothran commented 10 years ago

The other SECOORA model which should also be currently discoverable via the NGDC catalog is the USF one listed below covering the West Florida Shelf and the HF radars(both CODAR and WERA) for the Tampa, FL area. I still need to daily update the time extents metadata for this one, but the standard_names should be correct for discovery purposes.

The SABGOM covers the larger SECOORA domain including the Gulf, but just wanted to mention if looking to run further model comparisons. The HYCOM currents model is another thredds accessible global model, while Rich is probably aware of thredds access or lack of access to any ADCIRC models mentioned at http://commons.esipfed.org/node/1229

http://crow.marine.usf.edu:8080/thredds/catalog/WFS_ROMS_NF_model/catalog.html?dataset=WFS_ROMS_NF_model/USF_Ocean_Circulation_Group_West_Florida_Shelf_Daily_ROMS_Nowcast_Forecast_Model_Data_best.ncd

rsignell-usgs commented 10 years ago

@ocefpaf , I said "let's use Cartopy for plotting", but I only meant Cartopy as opposed to Basemap. If you have something that works better (e.g. interactive), let's use that.

ocefpaf commented 10 years ago

@rsignell-usgs Got it. I only rely on Basemap when I have to use python3, which is not the case for SECOORA. Hopefully cartopy will be py3k compatible soon and we will be able to stop using Basemap.

rsignell-usgs commented 10 years ago

@ocefpaf I just realized that for HF model/data comparison, we will need to deal with the staggered grid in ROMS. The way I've been handling this is brute force, using Rob Hetland's shrink and rot2d routines, as in this notebook: http://nbviewer.ipython.org/gist/rsignell-usgs/78a3005d0f4b4e9b8f4e

It of course would be better to handle this gracefully in Iris. If you have any ideas, I've raised the issue here: https://github.com/SciTools/iris/issues/1164

ocefpaf commented 10 years ago

The discussion on how to implement this on iris is here:

https://github.com/SciTools/iris/issues/1164

Meanwhile I will use a workaround with shrink and rot2d + netCDF4.