search

djhocking / Trout_GRF

Application of Gaussian Random Fields in a Dendritic Network
3 stars 0 forks source link

Add weather covariates #6

Closed djhocking closed 8 years ago

djhocking commented 8 years ago

It would be good to have the previous winter precipitation (maybe summer precipitation) and the previous year's fall and summer temperatures. This is easy for New England. The problem for the West Susquehanna in PA is that the daymet data has not been linked to our catchment delineations and added to the database. This is not be done until the new delineation is done. Basically it will be months until it's ready for me to use. I will move forward without this.

There is one option that might be possible for the first paper if we thought it was important. We could just assign the temperature to the nodes (confluences and sampling locations) based on lat and lon with a spatial join from daymet tiles. This wouldn't average over the catchment area but would give a decent estimate of the temperature around the nodes each season in each year. However, even this might be challenging to get done in a reasonable amount of time based on all the years and nodes.

Kyle or Ben have thoughts?

James-Thorson commented 8 years ago

My main vote is just to but bite off more than we can chew. given that you're low on time Dan, I'd suggest we try to construct a biological story given the covariates we have in hand.

Sent from my phone

On Oct 10, 2015, at 5:36 PM, Daniel J. Hocking notifications@github.com wrote:

It would be good to have the previous winter precipitation (maybe summer precipitation) and the previous year's fall and summer temperatures. This is easy for New England. The problem for the West Susquehanna in PA is that the daymet data has not been linked to our catchment delineations and added to the database. This is not be done until the new delineation is done. Basically it will be months until it's ready for me to use. I will move forward without this.

There is one option that might be possible for the first paper if we thought it was important. We could just assign the temperature to the nodes (confluences and sampling locations) based on lat and lon with a spatial join from daymet tiles. This wouldn't average over the catchment area but would give a decent estimate of the temperature around the nodes each season in each year. However, even this might be challenging to get done in a reasonable amount of time based on all the years and nodes.

Kyle or Ben have thoughts?

— Reply to this email directly or view it on GitHub.

Tom-Bombadil commented 8 years ago

Dan and I discussed this earlier today. It should be pretty simple to assign climate records to the nodes based on their association with catchments. Using the R package I've developed to process the Daymet records this should take a couple of hours. I hope to have the temp and precip records loaded to the database this afternoon.

Tom-Bombadil commented 8 years ago

I forgot to factor in Conte's internet speed. Hopefully it should be loaded for tomorrow.

Tom-Bombadil commented 8 years ago

The Daymet climate record (1980 - 2014) for tmin, tmax, and prcp for the West Susqehanna watershed is up on Osensei as a postgres database titled "NHDHRDV1_daymet_SusqWest". The schema.table ID is "data.daymet". Dan, please take a look at this and let me know if it works for you.

djhocking commented 8 years ago

Kyle I think you have to give me permission to use the database and maybe assign me a password. I'm not sure but I can't get in currently.