Flower2d

Geodetic tool package for 2-dimensional explorations of fault geometry and slip rates, written in Python 3. It can be used to plot topography, InSAR and GPS data profiles and to model tectonic deformations. It provides various tectonic geometries such as ramps, flower structures, pop-up, and bookshelf faults.

Installation

git clone https://github.com/simondaout/Flower2d.git
cd Flower2d

To update:

git pull

All scripts are standalone Python 3 modules — no installation required. Add src/flower2d/ and src/profiles/ to your PYTHONPATH or PATH.

Requirements

Package	Version
Python	≥ 3.9
NumPy	≥ 1.22
SciPy	≥ 1.9
PyMC	≥ 4.0
pytensor	≥ 2.0
arviz	≥ 0.14
matplotlib	≥ 3.5
pyproj	≥ 3.0

Install dependencies with conda:

conda install -c conda-forge pymc pytensor arviz numpy scipy matplotlib pyproj

Branches

Branch	Description
main	Python 3 — current development
python2	Legacy Python 2.7 code (read-only archive)

Organisation

flower2d/
├── src/
│   ├── flower2d/    ← Bayesian MCMC inversion of fault geometries
│   └── profiles/    ← Profile plotting and ramp estimation tools
└── examples/
    ├── haiyuam/     ← Ramp-decollement / flower structure example
    └── socal/       ← Southern California multi-structure example

src/flower2d — MCMC inversion

Inverts InSAR and/or GPS profiles for fault geometry and slip rates using PyMC (Metropolis sampler).

Key files:

• optimize.py — main entry point; defines the PyMC model and runs the sampler
• modelopti.py — fault geometry classes (mainfault, ramp, flower, popup, bookshelf, …)
• networkopti.py — data loading classes (network: InSAR dim=1, GPS dim=2/3)
• plot2d.py — posterior plot routines
• readgmt.py — GMT segment file loader

src/profiles — Profile tools

Plots data profiles and estimates residual ramps. Can be used independently of the inversion.

Key files:

• plotPro.py — main entry point
• network2d.py — data loading (same interface as networkopti.network)
• model2d.py — fault position overlays
• readgmt.py — GMT segment file loader
• atan_fit.py, tanh_fit.py — ramp-fitting utilities

Data format

InSAR (dim=1): space-separated columns

Columns	Description
x y los	average LOS angles given via av_los / av_heading
x y los look	per-pixel incidence angle (col 4); los=True
x y los look head	per-pixel incidence + heading (cols 4–5); los=True, head=True

• x, y: coordinates in km (local UTM) or degrees (if utm_proj is set)
• los: LOS displacement (mm/yr or mm)
• Nodata: pixels equal to 0.0, >9990, or the value given in nodata= are masked automatically

GPS (dim=2 or dim=3): station list file

# name  lon  lat
STAX    120.5  35.2
...

Individual station files in wdir/reduction/<name>:

• dim=2: date east north sigma_e sigma_n
• dim=3: date east north up sigma_e sigma_n sigma_up

Data can be provided in WGS84 (lon/lat) by setting utm_proj=<EPSG> (e.g. utm_proj=32647). The code reprojects automatically using pyproj.

network class — key parameters

network(network, reduction, wdir, dim,
        weight=1., scale=1., errorfile=None, perc=100,
        los=False, head=False, av_heading=None, av_los=None, nodata=None,
        mask=None, cov=None, base=None,
        color='black', samp=1, lmin=None, lmax=None, width=0.5,
        utm_proj=None, ref=None)

network    : input text file name
reduction  : label used for output plots and GPS sub-directory
wdir       : path to input files
dim        : 1=InSAR, 2=GPS (E+N), 3=GPS (E+N+Up)
weight     : data weight (default: 1)
scale      : unit scaling factor (default: 1)
errorfile  : optional per-pixel uncertainty file (x y sigma)
perc       : percentile threshold for outlier removal (default: 100 = no removal)
los        : if True, read incidence angle from col 4 (default: False)
head       : if True, read heading angle from col 5 (default: False)
av_los     : average incidence angle, e.g. 23 for descending Envisat
av_heading : average heading angle, e.g. -76 for descending Envisat
nodata     : nodata sentinel value to mask (default: None)
mask       : [ymin, ymax] — exclude profile-perpendicular distance range
cov        : [sill, sigm, lamb] — exponential covariance parameters
base       : ramp uncertainties. InSAR: [sigma_cst, sigma_lin] (default: [10, 0.1])
             GPS: [sigma_E, sigma_N] or [sigma_E, sigma_N, sigma_Up] (default: [50, 50, 50])
samp       : spatial subsampling factor (default: 1)
utm_proj   : EPSG code for UTM reprojection (e.g. 32647)
ref        : [lon, lat] reference origin for UTM projection

Examples

Haiyuan fault (examples/haiyuam/): ramp-decollement or flower structure geometry.

cd examples/haiyuam
plotPro.py tianzhuproin.py       # plot profiles
optimize.py tianzhuwestopti.py   # run MCMC inversion

Southern California (examples/socal/): imbricated multi-structure geometry.

cd examples/socal
plotPro.py gabrielproin.py
optimize.py gabrielflower.py

References

• Daout et al. (2016), Along-strike variations of the partitioning of convergence across the Haiyuan fault system detected by InSAR, GJI 205(1): 536–547

• Daout et al. (2016), Constraining the Kinematics of Metropolitan Los Angeles Faults with a Slip-Partitioning Model, GRL