search

dakotamm / research_blog

1 stars 0 forks source link

2023/09/26/VFC-spatial-avg #11

Open utterances-bot opened 1 year ago

utterances-bot commented 1 year ago

Improving VFC Spatial Averaging | Dakota’s Research Blog

Goals From Last Meeting: Dive into spatial averaging in VFC method - toward constructing climatologies and understanding interannual variation. This requires a significant code rewrite. Meet with UBC on 9/20. Read Scully 2013.

https://dakotamm.github.io/2023/09/26/VFC-spatial-avg.html

ahornerdevine commented 1 year ago

I really like your approach this past week: optimize what you have and consider other alternatives. I also think Option 2 is worth considering. I'm assuming (hopeful that) implementation would be a fair amount easier than what you did with the KDTree. If that is true, then con1 isn't as much of a concern. Just have to think about what you'd be giving up with this averaging approach. It is often true that a simpler, more consistent calculation leads to more reliable and easier to interpret results, even if it gives up some accuracy for individual calculations.

parkermac commented 1 year ago

I also like Option 2 (using Implementation 2a and combining segments as needed (I'm not sure I understand excluding segments - how would they be filled?)