search

ajleeson / research_blog

Weekly updates of research progress
0 stars 0 forks source link

research_blog/2023/09/11/first-look-at-evaluation-results #39

Open utterances-bot opened 1 year ago

utterances-bot commented 1 year ago

First look at model evaluation results | Weekly Research Update

Early in the week I wrapped up TRAPS readmes. Afterwards, I began looking at the preliminary model results. I first compared the modeled bottom DO to ORCA buoy data. Then I practiced running the model-model-data comparison scripts in Parker’s obsmod folder. I used the model-model-data comparison scripts to compare the new model run to observational data as well as the GRC run (WWTPs as tiny river implementation). Preliminary results suggest that the new model run appears to outperform the GRC run in some cases, but underperform in other. More details below. Model evaluation This week the model evaluation run has continued chugging along on klone, and so far it has made it to the middle of August. ORCA buoy DO compar

https://ajleeson.github.io/research_blog/2023/09/11/first-look-at-evaluation-results.html

parkermac commented 1 year ago

My suspicion is that a lot of the poor performance is related to the initial condition. Perhaps we should try a new run that has the same initial condition as the current cas6 model, thereby isolating the reasons for the differences.

ajleeson commented 1 year ago

I'd be happy to set that up. Since the ocean forcing is otherwise the same (I think), is it correct to assume that I only need to re-make forcing for 2017.01.01? Ocean forcing generation is the bottleneck to starting a new run.