RoseLab is a comprehensive orientation analysis plugin for QGIS. It produces publication-quality rose diagrams and performs full circular statistics, length distribution analysis, fracture intensity metrics, sampling bias correction, and two-sample comparison for any line vector dataset.
Designed for structural geology (lineaments, faults, joints, veins, dykes), but applicable to any azimuthal line data — drainage networks, roads, coastal features, landslide scarps, or remote sensing lineaments.
- QGIS 3.16 or later (tested on 3.28 LTS and 3.36)
- Python 3 (bundled with QGIS)
- scipy — for KDE, fracture set detection, and distribution fitting
Install via the OSGeo4W Shell (Windows) or terminal (Linux/macOS):
pip install scipy
- Plugins → Manage and Install Plugins
- Search RoseLab
- Enable Show experimental plugins if needed
- Click Install Plugin
- Download the latest ZIP from Releases
- Plugins → Manage and Install Plugins → Install from ZIP
- Select the ZIP and click Install Plugin
Copy the roselab/ folder to your QGIS plugins directory:
| Platform | Path |
|---|---|
| Windows | C:\Users\<username>\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\ |
| Linux | ~/.local/share/QGIS/QGIS3/profiles/default/python/plugins/ |
| macOS | ~/Library/Application Support/QGIS/QGIS3/profiles/default/python/plugins/ |
Then restart QGIS and enable the plugin under Plugins → Manage and Install Plugins → Installed.
- Load a line vector layer (shapefile, GeoPackage, etc.)
- Open Processing Toolbox → RoseLab → Rose Diagram — Orientation Analysis
- Select your layer
- Set output file paths for PNG and CSV
- Click Run
- Click the RoseLab toolbar button (or Plugins → RoseLab → RoseLab Studio)
- Select a line layer from the Layer dropdown
- Adjust parameters — the preview updates automatically
- Click Export Outputs... to save all figures and CSVs to a folder
- Bidirectional (0–180°, for undirected features like joints and lineaments) or Full circle (0–360°, for directed features like palaeocurrent indicators)
- Half-rose option (clips display to northern hemisphere)
- Weight types — frequency count | frequency % | cumulative length | length %
- Petal scaling — linear | square-root | logarithmic | area-proportion (recommended by Sanderson & Peacock 2020)
- Attribute field weight — weight petals by any numeric field (aperture, displacement, confidence score, intensity rating)
- Auto bin-size recommendation (Sanderson & Peacock 2020)
- Von Mises KDE smooth density curve (avoids bin-boundary sensitivity)
- Automatic fracture set detection — up to 4 dominant sets (S1–S4)
- Set isolation — click a detected set to filter bearings to that window and re-run all analysis
- Mean vector ray with Watson (1956) confidence arc at 90/95/99%
- Concentric reference circles
- User-defined tectonic reference directions (e.g.
Rift:045; Shear Zone:130)
- Overlay a second dataset as outlined petals on the same rose (visual two-sample comparison)
- Watson U2 non-parametric two-sample test (Watson 1962) — recommended over Watson-Williams F-test for multimodal structural data (Landler et al. 2021)
| Statistic | Method |
|---|---|
| Circular mean azimuth (0–180° and 0–360°) | Mardia & Jupp (2000) |
| Circular standard deviation | Mardia & Jupp (2000) |
| Mean resultant length R | Mardia & Jupp (2000) |
| Anisotropy index | Woodcock (1977) eigenvalue method |
| Shannon entropy H | — |
| Rayleigh test (p-value) | Tests H₀: uniform distribution |
| Kuiper test | More powerful for multimodal data |
| Cardinal quadrant summary | N / NE / E / SE % |
- Min/max length filter (exclude short or long segments)
- Up to two azimuth filter windows (isolate a known fault set)
- Rotation offset (correct for magnetic declination or regional reference frame)
- Terzaghi (1965) correction for systematic under-sampling of features parallel to the scanline or DEM illumination direction
| Metric | Definition |
|---|---|
| P20 | Fracture traces per unit area |
| P21 | Total trace length per unit area |
| P22 | Dimensionless intensity (P21 × mean length) |
After Sanderson & Nixon (2015), J. Struct. Geol. 72:55–66.
Four distributions fitted by Maximum Likelihood Estimation:
- Power law (Pareto) — exponent α, lower bound x_min
- Lognormal — μ and σ
- Exponential — rate λ
- Gamma — shape and scale
Model selection by AIC (lower = better) and KS test p-value. Output is a 6-panel figure: PDF (linear), log-binned PDF, model comparison bar chart, log-log PDF, CDF, CCDF.
Bias warnings included for: right-censoring, truncation (detection limit), and scale dependence.
| Output | Format | Description |
|---|---|---|
| Rose diagram | PNG, PDF, SVG | Publication-quality at selectable DPI |
| Distribution analysis | PNG | 6-panel MLE figure |
| Statistics | CSV | All computed statistics |
| Bin data | CSV | Per-bin bearing edges, frequency, length — compatible with RockWorks, Stereonet, GEOrient, Dips, OpenStereo, Excel |
Fracture traces terminating at the study area boundary are recorded shorter than their true length. Treat length statistics as minimum estimates when many lineaments reach the map edge. Ref: Mauring & Tveranger (2025), Solid Earth.
Lineaments shorter than your detection threshold (~3–5 pixel lengths for DEM/satellite data) are absent. Do not interpret the distribution shape below this limit. Ref: Bonnet et al. (2001), Rev. Geophysics.
Results from regional satellite lineaments (km-scale) and outcrop fracture traces (m-scale) cannot be directly compared. Always report the scale range of your dataset.
If your layer CRS uses geographic degrees (longitude/latitude), length values will be in degrees — not meaningful for length distribution or intensity analysis. Reproject to a projected CRS (e.g. UTM) before running RoseLab.
| Method | Reference |
|---|---|
| Bin size recommendation | Sanderson & Peacock (2020), Earth-Sci. Rev. 201, 103055 |
| Power-law fitting | Clauset et al. (2009), SIAM Review 51(4) |
| Length distributions | Bonnet et al. (2001), Rev. Geophysics 39(3) |
| Distribution model selection | Rizzo et al. (2024), J. Struct. Geol. |
| Censoring | Mauring & Tveranger (2025), Solid Earth 16 |
| Rayleigh test | Mardia & Jupp (2000), Directional Statistics, Wiley |
| Kuiper test | Fisher (1993), Statistical Analysis of Circular Data |
| Watson U2 test | Watson (1962), Biometrika 49(1/2) |
| Watson U2 recommendation | Landler et al. (2021), Scientific Reports 11 |
| Confidence arc | Watson (1956) |
| Anisotropy index | Woodcock (1977), GSA Bulletin 88(9) |
| Terzaghi correction | Terzaghi (1965), Geotechnique 15(3) |
| P20/P21 intensity | Sanderson & Nixon (2015), J. Struct. Geol. 72 |
If you use RoseLab in published work, please cite the key methodological references relevant to your analysis and describe the tool as:
"Orientation analysis was performed using RoseLab v1.0.0, a QGIS plugin (Moyo 2024), implementing methods following Sanderson & Peacock (2020), Clauset et al. (2009), and Sanderson & Nixon (2015)."
GNU General Public License v2 or later. See LICENSE.
Daniel Moyo moyodanj@gmail.com github.com/moyodanj/roselab
Bug reports and feature requests via GitHub Issues.