A 'shiny' App for Reproducible QA/QC of Eddy Covariance Data

An interactive 'shiny'-based tool for exploration and quality assurance and quality control (QA/QC) of eddy covariance flux tower data processing. It generates data-point removal code via user-directed selection from a scatterplot, and can export a cleaned .csv with removed points set to NA plus an R script for reproducibility. Reference: Key (2025) .

fluxtools v0.7.0

Interactive Shiny application for exploration and QA/QC of eddy covariance data.
Streamlines post-processing of eddy covariance datasets (e.g., after EddyPro) to detect and remove outliers, enforce physical ranges, and generate reproducible R code for AmeriFlux submissions.

Fluxtools is an independent project and is not affiliated with or endorsed by the AmeriFlux Network. “AmeriFlux” is a registered trademark of Lawrence Berkeley National Laboratory and is used here for identification purposes only.

If you use fluxtools in your workflow, please cite:

Key, K. (2025). fluxtools (version 0.7.0) [Computer software]. Zenodo.
https://doi.org/10.5281/zenodo.15597159

Table of Contents

• Key Features
• Installation
• Quickstart
• Data Requirements
• How It Works
  • Timestamp Parsing & Time Zones
  • Interactive QC & Selection
  • Outlier Detection
  • Time Slider (Inside / Outside)
  • Manual Range Filters
  • Physical Range Module (PRM)
  • Code Generation & Copy-All
• Download & Reset
• Vignette & Docs
• Citation
• License

Key Features

#key-features

• Interactive scatter
  Choose TIMESTAMP_START (time) or any numeric variable for X; choose any numeric variable for Y. Drag a box or lasso to flag points

• Multi-year support
  Upload up to ~1GB CSV (all years by default, or select one/more years)

  • Time subset support
    Subset data view based on year(s), month(s), day(s), or time of day

  • Compare two datasets (A/B)
    Upload Dataset B, choose labels/colors, and compare variables side by side (with optional smoothers and bottom legend)

• Overlay multiple variables
  Plot several Y variables together with color palettes or custom overrides

• Add smoothed line(s)
  LOESS smoother with optional 95% CI; toggle “Only smoothed line(s)” to hide base traces

• ±σ outlier highlighting
  Slider marks points beyond n standard deviations from a linear fit; click Select ±σ outliers to add them

• Accumulate & undo
  Add selections to an accumulated list and remove them later if needed Use Apply removals to commit NA changes, and Reload original data to revert

• Physical Range Module (PRM)
  Clamp variables to physically reasonable ranges (AmeriFlux Technical Note, Table A1) Out-of-range → NA. QC columns are ignored by PRM by default Undo PRM reverts only PRM-applied changes

• Reproducible outputs
  Export a cleaned CSV (removed values set to NA) and an R script with the exact transformations If PRM was applied, a summary table, audit CSV, and a manual replay script are included

• Dark/Light mode
  Toggle theme on the fly

Installation

#installation

# From CRAN
install.packages("fluxtools")

# Newest version from GitHub
library(devtools)
devtools::install_github("kesondrakey/fluxtools")

Quickstart

#quickstart

library(fluxtools)

# Launch the QA/QC Shiny app
run_fluxtools()

Data Requirements

#data-requirements

• Input: AmeriFlux BASE (or FLUXNET-style) CSV
• Must include TIMESTAMP_START (e.g., YYYYMMDDHHMM for 30-min or hourly data)
• Typical missing values such as -9999, -9999.0, -9999.00, -9999.000, NaN, and empty strings are treated as NA

How It Works

#how-it-works Timestamp Parsing & Time Zones #timestamp-parsing--time-zones

• TIMESTAMP_START is parsed automatically
• Viewer can display times with a fixed UTC±offset (no DST)
• Selections, code, and exports are keyed to the original timestamp string. Underlying values and exported timestamps are preserved

Interactive QC & Selection #interactive-qc--selection

• Box/lasso select points in the plot
• Click Flag Data to stage selections; Unflag Data or Clear Selection to remove
• Click Apply removals to set Y-values to NA_real_
• Use Reload original data to revert to the initial upload

Outlier Detection #outlier-detection

• Fit a simple linear model between X and Y, compute residuals, and highlight points exceeding ±σ (user-controlled)
• Select all ±σ outliers to add them to your accumulated code; clear ±σ outliers if needed

Time Slider (Inside / Outside) #time-slider-inside--outside

• Adjustable time window on TIMESTAMP_START
• Flag inside: add only points within the chosen time range
• Flag outside: add all points except those within the chosen range
• The slider snaps to your data cadence (30 min / 60 min) and supports multi-year files

Manual Range Filters #manual-range-filters

• Define custom min/max for any variable
• Flag values outside range to add out-of-bounds points to the accumulated set
• Complements PRM by allowing site- or project-specific thresholds

Compare two datasets

• Enable Compare two datasets and upload Dataset B
• Choose labels and colors per dataset
• Plot the same variables from A and B; legend appears at the bottom (horizontal)
• Works with Scatter or Line; optional transparent points on top of lines enable lasso
• Smoother can be added per (variable, dataset); “Only smoothed line(s)” hides base points/lines and shows just the LOESS curves (with optional 95% CI)

Plot multiple variables

• Enable Plot multiple variables to select several Y variables from the same dataset
• Color palette is applied automatically; you can override per-variable colors if desired
• Flags (rings) can match variable color (darker/lighter) or use classic yellow

Add smoothed line

• Toggle Add smoothed line
• Method: LOESS with configurable span; optional 95% CI band
• Only smoothed line(s) hides base markers/lines and keeps the legend so variables/datasets remain selectable
• Legend is positioned at the bottom (horizontal)

Physical Range Module (PRM) #physical-range-module-prm

• Available in the app or through fluxtools directly in R
• Applies physical bounds to variables (e.g., SWC, P, TA, CO2) using name-family matching (e.g., ^SWC($|_).)
• Example: the "SWC" family refers to all variables starting with SWC (e.g., SWC_1_1_1, SWC_2_1_3)
• Out-of-range values → NA; QC columns are ignored by default
• Apply to all families detected in your data, or select a subset in the UI
• Undo PRM reverts only PRM-applied changes (later user edits are preserved)
• In exports, PRM produces:
  • prm_summary.csv (bounds and counts)
  • prm_removed_values.csv (cell-level audit)
  • manual_prm_removed.R (replay script)

Example:

# View PRM rule table
get_prm_rules()

# Apply to all present families
res <- apply_prm(df)
res$data     # clamped data
res$summary  # per-column stats (min, max, n_replaced, pct_replaced)

# Apply to selected families
res2 <- apply_prm(df, include = c("SWC", "P"))

PRM ranges (and name) were sourced from Ameriflux

Code Generation & Copy-All #code-generation--copy-all

• The app continuously builds tidy dplyr code that sets selected Y-values to NA via case_when()
• Switch between Current and Accumulated snippets; click Copy code to paste into your own scripts and version control

Download & Reset

• Export cleaned data: downloads a ZIP containing

• raw_df.csv (original upload, unchanged)

• fluxtools_processed_df_.csv (cleaned, NA applied)

• fluxtools_removal_script.R (replays removals)

• If PRM was applied:

• prm_summary.csv (bounds + replacements)

• prm_removed_values.csv (audit of values removed)

• manual_prm_removed.R (cell-level replay)

• Reload original data: restores the initial CSV to start over

Vignette & Docs

browseVignettes("fluxtools")
# or
vignette("introduction", package = "fluxtools")

Citation

If you use fluxtools in publications, please cite:

Key, K. (2025). fluxtools (version 0.7.0) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.15597159

License

This package is free software, released under the GNU General Public License v3 (GPL-3). See the LICENSE file for details.

Fluxtools is an independent project and is not affiliated with or endorsed by the AmeriFlux Network. “AmeriFlux” is a registered trademark of Lawrence Berkeley National Laboratory and is used here for identification purposes only.