internship/swash/processing/post.py

import argparse
import configparser
import logging
import pathlib

import matplotlib.pyplot as plt
import numpy as np
import scipy.signal as sgl

parser = argparse.ArgumentParser(description="Post-process swash output")
parser.add_argument("-v", "--verbose", action="count", default=0)
parser.add_argument("-c", "--config", default="config.ini")
args = parser.parse_args()

logging.basicConfig(level=max((10, 20 - 10 * args.verbose)))
log = logging.getLogger("post")

log.info("Starting post-processing")
config = configparser.ConfigParser()
config.read(args.config)

inp = pathlib.Path(config.get("post", "inp"))
root = pathlib.Path(config.get("swash", "out"))

log.info(f"Reading data from '{inp}'")
x = np.load(inp.joinpath("x.npy"))
t = np.load(inp.joinpath("t.npy"))

botl = np.load(inp.joinpath("botl.npy"))
watl = np.load(inp.joinpath("watl.npy"))
#vel = np.load(inp.joinpath("vel_x.npy"))
vel = np.load(inp.joinpath("vel.npy"))[0]

# Cospectral calculations
x0 = config.getint("post", "x0")
arg_x0 = np.abs(x - x0).argmin()
t0 = int(config.getfloat("post", "t0") * 1e3)
arg_t0 = np.abs(t - t0).argmin()
dt = np.diff(t).mean() * 1e-3
f = 1 / dt
nperseg = config.getint("post", "nperseg", fallback=None)
log.info(f"Computing reflection coefficient at x={x0}")

eta = watl[t > t0, arg_x0]
u = vel[t > t0, arg_x0]

phi_eta = sgl.welch(eta, f, nperseg=nperseg)
phi_u = sgl.welch(u, f, nperseg=nperseg)
phi_eta_u = sgl.csd(eta, u, f, nperseg=nperseg)

H = np.sqrt(np.abs(phi_eta[1]))
U = np.sqrt(np.abs(phi_u[1]))
G = H / U
th_eta_u = np.angle(phi_eta_u[1])

R = np.sqrt(
    (np.abs(phi_eta[1]) + np.abs(phi_u[1]) - 2 * phi_eta_u[1].real)
    / (np.abs(phi_eta[1]) + np.abs(phi_u[1]) + 2 * phi_eta_u[1].real)
)
# R = np.sqrt(
#   (1 + G**2 - 2 * G * np.cos(th_eta_u))
#   / (1 + G**2 + 2 * G * np.cos(th_eta_u))
# )

if config.has_option("post", "compare"):
    inp_comp = pathlib.Path(config.get("post", "compare"))
    x_ = np.load(inp_comp.joinpath("x.npy"))
    t_ = np.load(inp_comp.joinpath("t.npy"))

    botl_ = np.load(inp_comp.joinpath("botl.npy"))
    watl_ = np.load(inp_comp.joinpath("watl.npy"))
    vel_ = np.load(inp_comp.joinpath("vel_x.npy"))

    # Cospectral calculations
    arg_x0_ = np.abs(x_ - x0).argmin()
    arg_t0_ = np.abs(t_ - t0).argmin()
    dt_ = np.diff(t_).mean() * 1e-3
    f_ = 1 / dt_

    eta_ = watl_[t_ > t0, arg_x0_]
    u_ = vel_[t_ > t0, arg_x0_]

    phi_eta_ = sgl.welch(eta_, f_, nperseg=nperseg)
    phi_u_ = sgl.welch(u_, f_, nperseg=nperseg)
    phi_eta_u_ = sgl.csd(eta_, u_, f_, nperseg=nperseg)

    H_ = np.sqrt(np.abs(phi_eta_[1]))
    U_ = np.sqrt(np.abs(phi_u_[1]))
    G_ = H_ / U_
    th_eta_u_ = np.angle(phi_eta_u_[1])

    R_ = np.sqrt(
        (np.abs(phi_eta_[1]) + np.abs(phi_u_[1]) - 2 * phi_eta_u_[1].real)
        / (np.abs(phi_eta_[1]) + np.abs(phi_u_[1]) + 2 * phi_eta_u_[1].real)
    )


# Plotting
log.info("Plotting results")
fig, (ax_watl, ax_vel) = plt.subplots(2)

ax_watl.plot(t * 1e-3, watl[:, arg_x0], lw=1, label="watl")
ax_watl.set(xlabel="t (s)", ylabel="z (m)")
ax_watl.autoscale(axis="x", tight=True)
ax_watl.grid()
ax_watl.axvline(t0 * 1e-3, c="k", alpha=0.2)

ax_vel.plot(t * 1e-3, vel[:, arg_x0], lw=1, label="vel")
ax_vel.set(xlabel="t (s)", ylabel="U (m/s)")
ax_vel.autoscale(axis="x", tight=True)
ax_vel.grid()
ax_vel.axvline(t0 * 1e-3, c="k", alpha=0.2)

fig.tight_layout()

fig_r, ax_r = plt.subplots()
ax_fft = ax_r.twinx()

ax_fft.plot(
    *sgl.welch(eta, 1 / dt, nperseg=nperseg),
    lw=1,
    c="k",
    alpha=0.2,
    label="PSD ($\\eta$, cas 1)",
)
ax_r.plot(phi_eta[0], R, marker="+", label="R (cas 1)")
if config.has_option("post", "compare"):
    ax_fft.plot(
        *sgl.welch(eta_, 1 / dt, nperseg=nperseg),
        lw=1,
        c="k",
        alpha=0.2,
        label="PSD ($\\eta$, cas 2)",
    )
    ax_r.plot(phi_eta[0], R_, marker="+", label="R (cas 2)")
ax_r.set(xlim=(0, 0.3), ylim=(0, 1), xlabel="f (Hz)", ylabel="R")
ax_fft.set(ylim=0, ylabel="PSD (m²/Hz)")
ax_r.grid()
ax_r.legend(loc="upper left")
ax_fft.legend(loc="upper right")
fig_r.tight_layout()

fig_x, ax_x = plt.subplots()
ax_x.plot(x, -botl, color="k")
ax_x.plot(
    x,
    np.maximum(watl[arg_t0, :], -botl),
)
if config.has_option("post", "compare"):
    ax_x.plot(x, -botl_, color="k", ls="-.")
    ax_x.plot(
        x,
        np.maximum(watl_[arg_t0, :], -botl_),
        ls="-.",
    )
ax_x.axvline(x0, c="k", alpha=0.2)
ax_x.set(xlabel="x (m)", ylabel="z (m)")
ax_x.autoscale(axis="x", tight=True)
fig_x.tight_layout()

out = pathlib.Path(config.get("post", "out")).joinpath(f"t{t0}x{x0}")
log.info(f"Saving plots in '{out}'")
out.mkdir(parents=True, exist_ok=True)

fig.savefig(out.joinpath("t.pdf"))
fig_r.savefig(out.joinpath("R.pdf"))
fig_x.savefig(out.joinpath("x.pdf"))

log.info("Finished post-processing")