Skip to contents

Re-opening a TRIA dataset

Source: vignettes/reopen-dataset.Rmd
reopen-dataset.Rmd

Overview

The main purpose of rxs2tria is to turn raw ROXAS / ROXAS AI output into a submission-ready TRIA dataset (see vignette("workflow")). The same read_* functions that write a dataset can also read one back, so a dataset downloaded from the TRIA database — or one you prepared earlier — can be re-opened and worked with using the package.

A TRIA dataset consists of at minimum three files:

  • the QWAmetadata .json,
  • the QWAdata cells .csv,
  • the QWAdata rings .csv.

These three are all you need to re-open a dataset. The optional supplementary files (original images, raw ROXAS output, etc.) are not required for anything shown here — download them only if you specifically need the raw material behind the measurements.

Reading the individual files

Each file is read by its corresponding read_* function. read_QWAmetadata() takes the path to the .json file directly:

QWA_meta <- read_QWAmetadata("path/to/my_dataset_QWAmetadata.json")
QWA_meta$images

read_QWAdata() reads the cells and rings CSVs. Point it at the directory holding the downloaded files and it locates both by name; if the folder holds files from more than one dataset, pass dataset_name to disambiguate:

QWA_data <- read_QWAdata(dir = "path/to/downloaded_dataset")
# or with explicit paths:
QWA_data <- read_QWAdata(file_cells = "path/to/my_dataset_QWAcells.csv.gz",
                         file_rings = "path/to/my_dataset_QWArings.csv.gz")

If a profile file was included in the download, read_QWAprofile() reads it back into a QWAprofile object:

prf <- read_QWAprofile("path/to/my_dataset_QWAprofile_sector.csv.gz")

Recomputing a QWAprofile

Because the cells and rings tables carry the columns the profile calculation needs (rraddistr in cells; mrw and eww in rings, plus the cell parameters you select), you can compute a QWAprofile directly from QWA_data — even if no profile was included in the download:

prf_sector <- calculate_sector_profiles(
  QWA_data,
  n_sectors       = 5,
  sel_cell_params = c("la", "cwttan"),
  quant_probs     = c(0.1, 0.5, 0.9)
)

See vignette("QWAprofile") for the difference between sector and band profiles and the full set of options.

Reviewing ring flags in the flags app

The ring flags shipped in the downloaded rings file (incomplete_ring, missing_ring, exclude_issues, …) can be inspected and edited in the flags app. Launch it with:

In the app, choose “Load data from csv files” and provide:

  • QWA profile data — a profile CSV, e.g. one you wrote with write_QWAprofile() after computing it as above.
  • QWA rings data — the rings CSV of the dataset.
  • Images metadata — here you can upload the QWAmetadata .json directly: the app extracts its $images component automatically. (A standalone QWAimages .csv is also accepted.)

Alternatively, choose “Load data from R environment” and refer to objects you have already read in, for example prf_sector, QWA_data$rings, and QWA_meta$images.

What to keep in mind

  • Reading re-validates. read_QWAdata() checks the data against the package schema, re-derives complete year sequences, and verifies dating. If a file is missing columns the schema expects, you will see warnings — these are informative and do not stop you from working with the data that is present.
  • Supplementary files are a separate path. Re-opening (this vignette) needs only the three core files. Re-running the raw preprocessing pipeline (get_roxas_files(), collect_raw_data(), …) is a different workflow that operates on the raw ROXAS files, which are only available if they were submitted as supplementary resources (see vignette("resources")).