In mikejohnson51/opendap.catalog: R OpenDap Client

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  warning = FALSE,
  error = FALSE,
  message = FALSE
)

library(opendap.catalog)
library(terra)

As of r format(Sys.Date(), format = "%m/%Y"), there are r formatC(nrow(opendap.catalog::params),big.mark = ",",digits = 0,format = "f") web resources documented in the catalog.

This offers too many resources to search through by hand unless you know exactly what you want which voids the possibility of serendipitous discovery. So, we have added a generally fuzzy search tool to help discover datasets.

Search

Say we want to find what snow water equivalent data (SWE) is available for a research problem. We can search the catalog on that key word:

(swe = search("swe"))

In total there are r nrow(swe) resources available across r length(unique(swe$id)) data sets. Viewing even a reduced catalog can still be onerous so a search summary is provided.

Search Summary

search_summary can be applied to any catalog object, but is meant for those reduced with search. It groups resources by ID and varname in order to consolidate the number of resources that are generated from many model/ensemeble/scenario runs (e.g. count)

search("swe") |> 
  search_summary()

This shows us more clearly the data sets and variables available. Importantly, count references number of variable data sets within each id/variable pair.

Mulit term search

Say looking through the returned results that we know NLDAS produces SWE measures we can expand the search to look for 'NLDAS swe', or 'swe NLDAS'.

search("swe NLDAS")

The result is that we see NLDAS has three models (VIC, NOAH, and MOSAIC) as well as two time horizons (M and H). We can continue to narrow from there!

Puting it all together

Say we want monthly SWE for 2018 along the Rocky Mountain corridor. We found NLDAS to be a good source and want to look at how the models compare in January:

swe = dap(catolog = search("month swe NLDAS"),
          AOI = AOI::aoi_get(state = c("CO", "UT", "ID", "MT", "WY")), 
          startDate = "2018-01-01")

terra::plot(terra::rast(swe), main = c("NLDAS_MOS0125_M.2.0", 
                                       "NLDAS_NOAH0125_M.2.0", 
                                       "NLDAS_VIC0125_M.2.0"))

plot(mean(rast(swe)), main = "Ensemble Jan 2018 NLDAS SWE")

Alternatively, say you wanted one year of monthly data from just the VIC model. All you would need to do is refine your search:

swe = dap(catolog = search("month swe NLDAS VIC"),
          AOI = AOI::aoi_get(state = c("CO", "UT", "ID", "MT", "WY")), 
          startDate = "2018-01-01",
          endDate = "2018-12-31")