search

ropensci / rnoaa

R interface to many NOAA data APIs
https://docs.ropensci.org/rnoaa
Other
330 stars 84 forks source link

Imputation of missing weather parameters #158

Closed lukas-rokka closed 6 years ago

lukas-rokka commented 8 years ago

enchantment/new feature request:

It would be very useful with a missing values imputation feature. Personally I'm mostly interested in the ISD dataset, but this should be useful for other NOAA datasets as well. For many stations and weather parameters it should be possible to use data from nearby stations and do the imputation with existing R-packages (e.g. MICE, hmisc::aregImpute()). For shorter gaps (<3 hours?) interpolation should work fine.

But there are some domain specific issues. E.g. temperature is probably strongly correlated to a nearby stations, while precipitation can vary quite much spatially. Sometimes a new stations replaces an old stations, in these situations the datasets could be simple merged. In other cases two relatively nearby stations can have very different climate (e.g. in Switzerland where two nearby stations could have a mountain peak between them).

sckott commented 8 years ago

@lukas-rokka thanks for this.

enchantment I like it 😏

Want to send a pull request to get started? and we can discuss from there

lukas-rokka commented 8 years ago

haha, enchant code would definitely be an enhancement.

I'll not work on it before the next month or so, I let you know when I get something useful done.

sckott commented 6 years ago

@lukas-rokka any thoughts on this?

lukas-rokka commented 6 years ago

I don't have time work on this issue in any near future. You can close it, @sckott.

I've started using reanalysis data instead of station observation data, and then missing data is already handled.

sckott commented 6 years ago

thansk @lukas-rokka - okay, will close