bruges

Bruges: reflection and rock physics

Use Bruges for interface reflectivity, elastic properties, fluid substitution, and analytic wavelets. Use a wave-equation solver for propagation modelling; use a log reader for LAS input. Companion skills are optional.

Inputs and conventions

• Convert velocity to m/s, density to kg/m³, and moduli to Pa for the examples below. Divide Pa by 1e9 only when reporting GPa.
• Reflection functions take incidence angles in degrees. Supply finite layer properties; positive density and shear modulus alone are insufficient: an isotropic solid also needs vp**2 > (4/3) * vs**2 for positive bulk modulus.
• Keep missing log intervals marked. Do not turn NaNs into zero-valued rock properties or convolve through missing intervals as if they were observed.
• A synthetic seismic trace requires regularly sampled two-way time. Convert and resample depth logs first, retaining the velocity/datum choices.

AVO at one interface

Runnable example with illustrative upper/lower properties:

import numpy as np
from bruges.reflection import zoeppritz, shuey, akirichards

vp1, vs1, rho1 = 3000.0, 1700.0, 2300.0  # m/s, m/s, kg/m³
vp2, vs2, rho2 = 3200.0, 1800.0, 2350.0
theta_deg = np.arange(0.0, 31.0)
r_exact = zoeppritz(vp1, vs1, rho1, vp2, vs2, rho2, theta_deg)
r_shuey = shuey(vp1, vs1, rho1, vp2, vs2, rho2, theta_deg)
r_aki = akirichards(vp1, vs1, rho1, vp2, vs2, rho2, theta_deg)
intercept, gradient = shuey(vp1, vs1, rho1, vp2, vs2, rho2,
                            return_gradient=True)

return_gradient belongs to shuey, not akirichards; it returns the intercept/gradient pair, not reflectivity plus gradient. Shuey's intercept is an approximation to the exact normal-incidence impedance contrast. Verify the exact result against (rho2*vp2 - rho1*vp1) / (rho2*vp2 + rho1*vp1).

At or beyond critical angles, Zoeppritz responses can be complex. Retain their phase or explicitly define the amplitude being plotted. An approximate angle limit alone does not establish accuracy for every elastic contrast. Interpret AVO classes using polarity convention, angle coverage and lithology; an intercept/gradient sign pair is not a unique fluid diagnosis.

Wavelet and time-domain synthetic

Bruges wavelets return a named tuple with amplitude then time. Use the fields to avoid swapping a wavelet for its time axis.

import numpy as np
from bruges.filters import ricker
from scipy.signal import convolve

dt_s = 0.001
time_s = np.arange(257) * dt_s  # Regular two-way-time sampling.
impedance = np.full(time_s.size, 3000.0 * 2300.0)
impedance[128:] = 3200.0 * 2350.0
reflectivity = np.zeros_like(impedance)
# Place an interface at the first sample of the lower layer.
reflectivity[1:] = np.diff(impedance) / (impedance[:-1] + impedance[1:])
wavelet = ricker(duration=0.128, dt=dt_s, f=25.0)
synthetic = convolve(reflectivity, wavelet.amplitude, mode='same')

Check sample interval, time alignment, peak polarity and amplitude on a single known interface before generating a well tie. Read wavelet details for Ormsby/Klauder and phase rotation.

Rock properties and fluid changes

For elastic moduli, Gardner/Castagna estimates or Gassmann substitution, read rock physics examples. The tested API uses moduli.mu, moduli.pr, moduli.vp, moduli.vs, and avseth_fluidsub; there is no top-level gassmann or castagna in bruges.rockphysics. Gardner is in bruges.petrophysics and returns kg/m³ by default.

Use measured/calibrated mineral and fluid properties at reservoir conditions. Gassmann assumes a connected pore system, pressure equilibration at low frequency, and an unchanged solid frame/shear modulus. Check the resulting density change and shear-modulus invariance; unchanged Vs is not the invariant.

Verification and resources

Examples were checked with Bruges 0.5.4, NumPy 1.26.4 and SciPy 1.17.1 on 2026-09-14. Bruges 0.5.4 imports pkg_resources; the tested environment uses setuptools 80.9.0. Install scientific dependencies in an isolated environment. These examples do not certify every Bruges operation.

• Reflection API
• Rock physics API
• Wavelet API
• scripts/avo_analysis.py: tested AVO helper. Its CLI takes velocity in m/s and density in g/cm³, unlike the SI examples above. Exact reflection requires Bruges; missing imports fail explicitly. Plots preserve real/imaginary components beyond critical angles. Classification is a screening heuristic; IIp is not inferred from gradient sign alone.