Using numpy arrays as input

You don’t have a xarray.Dataset ready to use as input? No worries, you can use the .from_numpy() class method to initialise the Triplets and Arrays classes from numpy.ndarray objects

import numpy as np
import xarray as xr

from scipy.io import loadmat
from matplotlib import pyplot as plt

import ewdm
from ewdm.sources import SpotterBuoysDataSource
from ewdm.plots import plot_directional_spectrum

Testing with ewdm.Triplets

Let’s see how this feature work with the ewdm.Triplets class

spotter = SpotterBuoysDataSource("../../data/displacement.csv")
dataset = spotter.read_dataset()

subset = dataset.sel(time=slice("2020-08-20 10:00", "2020-08-20 10:30"))

# extract numpy arrays from dataset
time = subset["time"].data
surface_elevation = subset["surface_elevation"].data
eastward_displacement = subset["eastward_displacement"].data
northward_displacement = subset["northward_displacement"].data
sampling_rate = subset.attrs["sampling_rate"]

Now, we can initialise the ewdm.Triplets class with the numpy arrays

spec = ewdm.Triplets.from_numpy(
    time = time,
    surface_elevation = surface_elevation,
    eastward_displacement = eastward_displacement,
    northward_displacement = northward_displacement,
    fs=sampling_rate,
    max_time_gap="30s"
)
print("Sampling rate is: ", spec.fs)
print("Maximum allowed time gap is: ", spec.max_time_gap)

Sampling rate is:  2.5
Maximum allowed time gap is:  30s

Let’s compute the spectra and print the output

output_triplets = spec.compute()
print(output_triplets)

<xarray.Dataset> Size: 95kB
Dimensions:                   (frequency: 81, direction: 72)
Coordinates:
  * frequency                 (frequency) float64 648B 0.03125 0.03263 ... 1.0
  * direction                 (direction) float64 576B -180.0 -175.0 ... 175.0
Data variables:
    directional_spectrum      (frequency, direction) float64 47kB 0.0001156 ....
    directional_distribution  (frequency, direction) float64 47kB 0.001684 .....
    frequency_spectrum        (frequency) float64 648B 0.06864 ... 0.0003756

Testing with ewdm.Arrays

It works for ewdm.Arrays class too.

# loading matlab file
mat_fname = "../../data/donelan_run62.mat"
mat_data = loadmat(mat_fname, simplify_cells=True)

sampling_rate = mat_data["fs"]
time = np.arange(len(mat_data["eta"])) / sampling_rate
elements = np.arange(len(mat_data["x"]))

print("Surface elevation shape: ", mat_data["eta"].shape)
print("Time array shape: ", time.shape)
print("Position x shape : ", mat_data["x"].shape)
print("Position y shape : ", mat_data["y"].shape)

Surface elevation shape:  (4096, 6)
Time array shape:  (4096,)
Position x shape :  (6,)
Position y shape :  (6,)

Now let’s create spec object

spec = ewdm.Arrays.from_numpy(
    time = time,
    surface_elevation = mat_data["eta"],
    position_x = mat_data["x"],
    position_y = mat_data["y"],
    fs = sampling_rate,
    max_nan_ratio = 0.05,
)

print("Sampling rate is: ", spec.fs)
print("Maximum allowed nan ratio: ", spec.max_nan_ratio)

Sampling rate is:  4
Maximum allowed nan ratio:  0.05

Let’s compute the spectra and print the output

output_array = spec.compute()
print(output_array)

<xarray.Dataset> Size: 114kB
Dimensions:                   (frequency: 97, direction: 72)
Coordinates:
  * frequency                 (frequency) float64 776B 0.03125 0.03263 ... 2.0
  * direction                 (direction) float64 576B -180.0 -175.0 ... 175.0
Data variables:
    directional_spectrum      (frequency, direction) float64 56kB 1.444e-05 ....
    directional_distribution  (frequency, direction) float64 56kB 0.003793 .....
    frequency_spectrum        (frequency) float64 776B 0.003806 ... 3.278e-05

Let’s plot the results

It’s always nice to see some specral plots, isn’t it?

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(8,4), layout="constrained")

plot_directional_spectrum(
    output_array.directional_spectrum, ax=ax1, levels=None, colorbar=True,
    axes_kw={"rmin": 0.1, "rmax": 1.2, "rstep": 0.2, "angle": 135},
    cbar_kw={"label": ""}
)

plot_directional_spectrum(
    output_triplets.directional_spectrum, ax=ax2, levels=None, colorbar=True,
    axes_kw={"rmin": 0.1, "rmax": 1.2, "rstep": 0.2, "angle": 135},
    cbar_kw={"label": "$E(f,\\theta)$"}
)