Automated Calibration and Analysis of 'GLS' (Global Location Sensor) Data

Provides a fully automated workflow for calibrating and analyzing light-level geolocation ('GLS') data from seabirds and other wildlife. The 'glscalibrator' package auto-discovers birds from directory structures, automatically detects calibration periods from the first days of deployment, processes multiple individuals in batch mode, and generates standardized outputs including position estimates, diagnostic plots, and quality control metrics. Implements the established threshold workflow internally, following the methods described in 'SGAT' (Wotherspoon et al. (2016) < https://github.com/SWotherspoon/SGAT>), 'GeoLight' (Lisovski et al. (2012) ), and 'TwGeos' (Lisovski et al. (2019) < https://github.com/slisovski/TwGeos>).

glscalibrator: Automated GLS Calibration and Analysis

An R package for fully automated calibration and analysis of Global Location Sensor (GLS) data from seabirds and other wildlife. glscalibrator streamlines the traditionally manual and time-consuming process of GLS data analysis by auto-discovering devices, detecting calibration periods, and batch processing multiple individuals.

Key Features

• Fully Automated Workflow: Process entire datasets with a single command
• Auto-Discovery: Automatically finds all GLS devices in your directory structure
• Intelligent Calibration: Auto-detects calibration periods from the first days of deployment
• Batch Processing: Handles multiple individuals without manual intervention
• Quality Control: Automated hemisphere checks, twilight filtering, and diagnostic plots
• Standardized Outputs: Produces consistent data formats (GLSmergedata.csv) and visualizations
• Built on Proven Methods: Implements NOAA-style solar geometry to replicate the classic threshold workflow without archived dependencies

Installation

Install from GitHub (this will pull the CRAN dependencies automatically):

if (!requireNamespace("remotes", quietly = TRUE)) install.packages("remotes")
remotes::install_github("fabbiologia/glscalibrator")

Quick Start

library(glscalibrator)

# Process all GLS devices in a directory
results <- calibrate_gls_batch(
  data_dir = system.file("extdata", package = "glscalibrator"),
  output_dir = "data/processed/calibration",
  colony_lat = 27.85178,  # Colony latitude
  colony_lon = -115.17390  # Colony longitude
)

# Check summary
print(results$summary)

# Access individual bird results
bird_positions <- results$results$BW154_05Jul24_225805

What It Does

Traditional GLS analysis requires:

1. Manual identification of each bird's data file
2. Manual selection of calibration period
3. Custom scripting for twilight detection
4. Individual processing of each bird
5. Manual creation of output formats and plots

glscalibrator automates all of this:

# Traditional approach (hours of work)
# Read file → Find calibration dates → Detect twilights →
# Filter twilights → Calibrate → Calculate positions →
# Create plots → Repeat for each bird → Combine outputs

# glscalibrator approach (one command)
results <- calibrate_gls_batch(data_dir, output_dir, colony_lat, colony_lon)

Output Structure

output_dir/
├── data/
│   ├── GLSmergedata.csv              # Combined data (standard format)
│   ├── all_birds_calibrated.csv     # Combined positions
│   ├── calibration_summary.csv      # Summary statistics
│   ├── BW154_calibrated.csv         # Individual bird data
│   └── BW154_GLSmergedata.csv       # Individual bird (standard format)
└── figures/
    ├── all_tracks_combined.png      # All tracks on one map
    ├── BW154_track.png              # Individual track
    └── BW154_calibration.png        # Calibration diagnostics

Advanced Usage

Excluding Equinox Periods

# Define equinox exclusion periods
equinoxes <- list(
  c("2024-08-24", "2024-10-23"),  # Autumn equinox
  c("2024-02-19", "2024-04-19")   # Spring equinox
)

results <- calibrate_gls_batch(
  data_dir = "data/raw/birds",
  output_dir = "data/processed/calibration",
  colony_lat = 27.85,
  colony_lon = -115.17,
  exclude_equinoxes = equinoxes
)

Processing Individual Birds

# Read light data bundled with the package
light_data <- read_lux_file(gls_example("W086"))

# Auto-detect calibration period
calib <- auto_detect_calibration(
  light_data,
  colony_lat = 27.85,
  colony_lon = -115.17
)

# Detect twilights
twilights <- detect_twilights(light_data, threshold = 2)

# Filter twilights
twilights_clean <- filter_twilights(twilights, light_data, threshold = 2)

Methodology

The package implements a proven workflow:

1. Twilight Detection: Threshold-crossing method (light > 2 lux = day)
2. Auto-Calibration: Searches first 1-5 days for stable period at colony
3. Gamma Calibration: Learns an optimal sun elevation directly from calibration twilights (algorithm inspired by TwGeos)
4. Position Estimation: Applies NOAA solar geometry to derive coordinates from twilight pairs
5. Quality Filtering:
   • Removes twilights < 1 hour apart
   • Filters unusual intervals (not ~12 or ~24 hours)
   • Checks light quality around transitions
   • Validates hemisphere (Western vs Eastern)
   • Excludes equinox periods

Bundled Example Data

glscalibrator ships with three .lux files in inst/extdata/ that power the documentation, tests, and vignettes. You can explore them programmatically:

# List available example datasets and their metadata
glscalibrator_example_metadata

# Retrieve the path to a specific file
w086_path <- gls_example("W086")

# See summary information
list_gls_examples()

Use these datasets in tutorials, automated tests, or live demonstrations without needing external files. They are also summarised in inst/extdata/README.md for quick human-readable reference.

Dependencies

• dplyr / magrittr / lubridate / stringr – Data manipulation utilities
• maps – Basemap rendering for diagnostic plots
• Base R packages stats, graphics, grDevices, utils
• Historical inspiration from TwGeos, GeoLight, and SGAT algorithms (no runtime dependency)

Testing

Run the automated test suite to verify the installation:

testthat::test_local("tests")

The bundled synthetic dataset (gls_example("synthetic")) underpins most unit tests, while real bird deployments offer higher-volume scenarios for manual QA.

Citation

If you use glscalibrator in your research, please cite.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License - see LICENSE file for details

Acknowledgments

Built on the excellent work of:

• SGAT package authors
• GeoLight and TwGeos developers
• The seabird tracking community

Support

For issues and questions:

• GitHub Issues: https://github.com/fabbiologia/glscalibrator/issues
• Email: [email protected]