search

USGS-R / drb-gw-hw-model-prep

Code repo to prepare groundwater and headwater-related datasets for modeling river temperature in the Delaware River Basin
Other
0 stars 3 forks source link

Add a column to SB vars file to indicate how attributes should be aggregated #47

Closed lekoenig closed 2 years ago

lekoenig commented 2 years ago

This PR addresses sub-task 2 in #40. I've added a new column to 1_fetch/in/nhdv2_attributes_from_sciencebase.csv called CAT_aggregation_operation. This column tells us how we should translate the NHDv2-scale values to aggregated values for each NHM segment.

Many of these are also used in the DRB inland salinity project and so I've copied over the aggregation operations from that project for shared attributes. One attribute that I'm not convinced of my assignment is STRM_DENS (stream density) and I plan to calculate that variable myself for a few NHM catchments before just going with the values on ScienceBase that we re-aggregate to NHM.

@janetrbarclay Do you want to take a look at these to see if you agree with these aggregation operations? If so, would you be able to do so by this coming Tuesday (9/27)? I know that the formatting on GH is ugly so in case you don't want to pull down my branch I've added the file to our "data" folder on Sharepoint.

janetrbarclay commented 2 years ago

Thanks for putting a copy of the file in the data folder. Definitely MUCH easier to look at it there. My thought on stream density is that we might want an area-weighted mean for that. My other thought is that the ones currently indicated as "mean" might best be represented as length-averaged mean, but since that might not make much of a difference in the resulting values and and we don't yet know if these attributes will help (though we think they might), we could start with aggregations as used for inland salinity and then modify them to what we'd expect to be more accurate if they turn out to be ones we're keeping.

lekoenig commented 2 years ago

My other thought is that the ones currently indicated as "mean" might best be represented as length-averaged mean

Thanks, Janet - I agree! I've edited the variables sinuosity and STREAM_SLOPE to use length-weighted mean, and STRM_DENS to use area-weighted mean. I've also added STATSGO ROCKDEP as an attribute to this table, although we may end up not using it because this attribute represents soil thickness rather than depth to bedrock.