Introduction

IsoQC is an interactive quality‑control dashboard designed for analysing stable isotope records (δ¹⁸O and δ²H) from precipitation samples.
The app helps users:

• upload and inspect their own datasets,
• visualise temporal availability,
• compare a station’s measurements with nearby stations,
• identify potential outliers based on thresholds,
• export time‑series diagnostics.

This guide walks you through the complete usage of the application.

How the app is organised

The tabs follow the order of the work:

1. Dataset – import your data or load the Slovenian test dataset.
2. Network – every station screened at once: where are the problems? This tab opens after loading.
3. Station – one station in detail: time series, map and δ²H–δ¹⁸O plot.
4. Flags – every flagged sample in one list, to keep or exclude.

Three things are in the same place on every tab:

• Screening settings (search radius, thresholds, elevation gradients) in the left panel of the Network, Station and Flags tabs. It is one set of settings: changing it on any tab changes it everywhere. On tablets and phones the panel opens with Controls. On a desktop the round arrow on the panel's edge folds it away, which leaves more room for the plots; a Controls item then appears at the top of the left rail to open it again. The choice is remembered.
• Station search in the header (shortcut Ctrl K / ⌘ K or /): pick a station from any tab and the Station tab opens. On tablets and phones the magnifier button in the header opens the search.
• Export at the top right of the page holds every download of that tab.
• ? in the header explains the labels (sample flags, station status, your decisions).

1. Importing your data

Click Import your data on the Dataset tab. A four-step window guides you through the import; for most files you only confirm what IsoQC has recognised.

Supported files

• Text files: .csv, .tsv, .txt (comma, semicolon, tab, pipe or space separated)
• Spreadsheets: .xlsx, .xls, .ods (any sheet)
• Pasted tables: copy cells in Excel or another program and paste them into the window
• Maximum file size: 500 MB

The delimiter, the decimal mark (. or ,), the text encoding (UTF-8, Windows-1250, Latin-1) and the header row are detected automatically. Title rows above the header and unit rows below it (for example ‰ VSMOW) are skipped.

What the data must contain

• Station – station name or code
• Date – any common format (2005-03-15, 15/03/2005, 15.03.2005, Mar 2005, Excel dates), year and month columns, or the start and end of the sampling period (IsoQC uses the midpoint). Monthly dates without a day are placed on the 15th.
• Latitude / Longitude – decimal degrees or degrees–minutes–seconds (46°03'N). They can also come from a separate station table (another sheet or file with station, latitude, longitude, altitude), joined by station name.
• δ¹⁸O and/or δ²H (‰) – at least one of the two.
• Altitude (m) – optional. Stations without altitude are not elevation-corrected.

Both wide tables (one row per sample) and long tables (one row per isotope, such as GNIP/WISER exports with Measurand Symbol and Measurand Amount) are accepted.

The four steps

1. Source – choose a file or paste a table. On the start screen a file can also be dropped onto the page.
2. Layout – check the sheet, the header row, the table layout and the optional station table. If everything needed was recognised, you can go straight to the check.
3. Columns – each column has a role selector above a preview of its values. Green: recognised, amber: please check. IsoQC uses both the column name (English, Hungarian, German and GNIP names) and the values (ranges of coordinates, altitudes and isotope values). δ¹⁸O and δ²H are told apart by the slope of the meteoric water line, which also reveals swapped columns.
4. Check – a short report before importing: rows read, samples imported, rows skipped and why, placeholders such as −999 treated as missing, stations without altitude or coordinates, duplicate samples, station names that differ only in case or accents (merged when their coordinates agree), and the δ²H–δ¹⁸O slope. Skipped rows can be downloaded as CSV.

Template

The start screen and the Dataset tab offer a template (.xlsx, a few example rows and a sheet describing the columns).

A built-in test dataset (Slovenian stations and nearby stations, Hatvani et al. 2026) is available via Try the Slovenian test dataset. It can be screened in IsoQC; the dataset itself cannot be downloaded.

Screening settings after loading

After a dataset is loaded, IsoQC asks for the screening settings (search radius, thresholds, elevation gradients) before the first network run. The defaults are filled in; a line below the radius tells how many stations have at least one neighbour within it. Not now opens the Station tab without running the screening.

2. Dataset overview

After importing, the Dataset tab shows the number of stations, records, δ¹⁸O and δ²H values and the covered period. Use Import another file to replace the dataset, or Screen the network to continue.

3. Data availability

Station availability plot

You can visualise how many stations have data in a given period at:

• Daily resolution
• Monthly resolution
• Yearly resolution

This helps identify data gaps and temporal coverage issues.

4. The Station view

The Station tab shows one station in detail. Open a station from the Network map, or choose one here. The time series are on the left; the map and the δ²H–δ¹⁸O plot stay in view beside them while you scroll (on narrow screens they are above the time series).

4.1 Station selection

Choose a station in the station search in the header: type part of its name or open the list (shortcut Ctrl K / ⌘ K or / from any tab), click it on the map, or step through the stations with the ‹ › buttons. Next flagged jumps to the next station that has flagged samples over its full record with the current parameters.
The app will automatically:

• focus the map on the selected station,
• refresh the date range bounds,
• update all plots,
• update the station summary (coordinates, altitude, period, number of nearby stations) and the number of flagged values next to each time series.

The status label next to the station name summarises the selected period:

• Clean – no value above the thresholds,
• Review – at least one value above a threshold (the flagged values are listed on the Flags tab),
• No comparison – no neighbour data on the station's dates.

Sharing

• Shareable view: the address in the browser always holds the current view: station, period, radius, thresholds, elevation gradients, tab, Compact or Detailed view, basemap, Flags filters and the open sample. The link button copies it. For the Slovenian test dataset the data load automatically. For uploaded data the link also stores the file name, the import settings you changed in the wizard and a fingerprint of the data: send the file with the link, the recipient is asked for it, the import is repeated the same way, and a warning appears if the data differ.
• Export on the Station tab downloads the nearby mean time series as CSV. Each plot can be saved as PNG with the camera icon in its toolbar.

4.2 Nearby stations and thresholds

The Screening settings (the same on the Network, Station and Flags tabs) let you customise:

• Search radius (km) for selecting nearby stations
• Thresholds for:
  • δ¹⁸O
  • δ²H
  • D‑excess
• Elevation‑correction parameters (default: Kern et al. 2020)

Nearby stations appear on the map:

• Blue → selected station
• Dark grey with a solid line → nearby station with values in the selected period
• Hollow marker with a dashed line → nearby station without any value in the selected period
• Light grey → other stations
• Every nearby station is connected to the selected one; hover a line to read the distance.
• The map is centred on the selected station and shows the stations around it: click any of them to select it. The globe button shows all stations (press it again to return), and the full screen button opens the map in a large view.

5. Date Range Handling

IsoQC includes an advanced date‑range controller:

• A slider above the time series, as wide as the plots, for fast interval selection
• From / To fields over the two ends of the slider: click one to choose the year, month and day in three lists (dots mark years, months and days with a measurement), then Apply; a date can also be typed
• While you drag the slider, a bubble shows the date and the length of the period, a year scale runs under the track and the From / To field follows the handle
• A reset button (↺) next to them
• Station‑specific date limits
• Smooth interaction: recalculation happens only after final slider release

If data are unavailable in the chosen interval, the app warns you (e.g., no δ¹⁸O or no δ²H).

6. Time-Series Plots

IsoQC provides multiple diagnostic plots, including:

δ¹⁸O corrected

δ²H corrected

D‑excess

δ²H vs. δ¹⁸O (XY plot)

For each variable:

• The selected station is shown in colour.
• Nearby stations appear in grey.
• The nearby average appears as a black dashed line, with a light band showing the nearby mean ± 2×SD.
• Orange rings mark dates where fewer than 3 neighbours have a value (weak comparison).
• Samples excluded during review are marked with ×.
• The lower panel shows |station − nearby mean| as bars with the threshold as a dashed line. Bars use an interpretative colour scheme:
  • Light grey → below threshold
  • Dark grey → below threshold but above 2×SD
  • Purple → above threshold
  • Purple with black outline → strongly suspicious (above threshold & above 2×SD)

Compact and Detailed. The switch next to the heading chooses the layout of the time series. Compact (the default) shows δ¹⁸O, δ²H and D‑excess as three bands of one figure, with the period slider directly above them and the dates under the last band, so all three can be compared at a glance. Detailed shows each variable in a tall panel with the list of nearby stations; on wide screens the settings panel and the map column become narrower to give the panels more room. The number of flagged values is shown next to each variable name. The choice is remembered in your browser. In the Detailed view the Legend switch shows or hides the list of stations beside each panel.

Folding the map away. On wide screens the Hide button above the map folds the map and the δ²H–δ¹⁸O plot away to the right, so the time series use the full width; the narrow tab that remains at the right edge brings them back.

The three time‑series panels share one time axis: zooming (drag a box) in one panel zooms all three; double‑click resets the view. Hovering a date shows a guide line on all panels and highlights the same sample in the δ²H vs. δ¹⁸O plot; hovering a neighbour's line highlights that station on the map, and hovering a station on the map highlights its line. Click a bar to open the sample detail.

The XY plot distinguishes:

• All points
• Points under thresholds
• Points above thresholds

7. Exporting Time-Series Data

Use Export → Nearby mean time series (CSV) on the Station tab to export a .csv containing:

• Selected station values
• Nearby mean values
• Differences
• Date range metadata
• Thresholds
• Elevation correction parameters
• Search radius

The export includes UTF‑8 encoding with BOM for Excel compatibility.

8. Network screening

The Network tab compares every station with its neighbours over its full record, using the screening settings in the left panel. It shows:

• the number of compared and flagged samples,
• a map coloured by station status (Clean or Review; size = number of compared samples), zoomed to where your stations are; the globe button shows every station and the full‑screen button opens a large view.

Click a station on the map to open it; Next flagged on the Station tab steps through the stations with flagged samples. The Export menu has the Station summary and the Flagged samples (CSV). Changing a threshold updates the results immediately; changing the radius or an elevation gradient asks you to run the screening again.

9. Reviewing flagged samples

The Flags tab lists every flagged sample of the network. Filter by isotope, flag type and decision, then open a sample (or click a bar on the Station tab) to see:

• the station value, the nearby mean, |diff|, 2×SD and the number of neighbours,
• every neighbour's raw and elevation‑corrected value on that date, with distance and altitude.

Mark the sample as Keep, Exclude or Unsure (with an optional note); the next sample of the list opens automatically. Decisions are kept for the session:

• Export → Dataset with QC columns exports the full dataset with qc_flag_*, qc_decision_* and qc_note_* columns for δ¹⁸O, δ²H and D‑excess,
• Export → Review decisions saves your decisions as CSV, and Load review restores them later.

10. Display options

• Dark theme: the moon / sun button in the header (the choice is remembered in your browser).
• Phones: navigation moves to the bottom bar; the controls open from the Controls button.
• Tablets: the controls open as a side drawer from the Controls button in the header.

11. Filling Missing Months (Optional)

If the dataset contains gaps for a specific month, you may:

1. Insert a dummy numeric value for that month in your original file.
2. Upload the updated file again.

IsoQC will compute an average for that month, which can be used as an estimated replacement.

12. Data Privacy Notice

IsoQC does not store user-uploaded datasets. Uploaded files are processed only for the active session and are released when the session ends.

To support operation, troubleshooting and basic usage statistics, IsoQC keeps anonymous technical usage logs in an internal SQLite database. No cookies are used, and no name, e-mail or IP address is stored. A visit may be recorded with:

• session start and end time, total time and active time (the time the IsoQC tab was in front and in use),
• the tabs viewed and the actions taken (loading data, running the screening, exports, shared links, review decisions), with the screening settings in use,
• app version and a random session token that is not linked to the person,
• the country of the visit,
• the browser, operating system and device family (for example Chrome · Windows · Desktop), the browser language,
• the referring site (only the site name, for example Google or doi.org, never the full address),
• whether the visit came from an automated browser or a crawler.

The country is looked up on the IsoQC server from the visitor's IP address in an offline table; no external service is contacted. The IP address is used only for this lookup and is not stored. If no address is available, the country is estimated from the browser's time zone. IP to country data: DB-IP (IP to Country Lite, CC BY 4.0).

Visits are not counted when the browser sends Do Not Track or Global Privacy Control, or when counting is switched off below. Switching it off also removes what was already recorded in the current visit. The choice is kept only in this browser.

These limited logs are used only for:

• secure and reliable operation of the app,
• troubleshooting and maintenance,
• basic aggregated usage statistics.

For general information about HUN-REN CSFK website data processing, please see the institutional Privacy Notice: [HUN-REN CSFK Website Privacy Notice]

Need help?

If you encounter any unexpected behaviour, refer to:

• the Changelog for known updates and fixes,
• contact the developer: IsoQC GitHub repository

Thank you for using IsoQC!