Echopype: A quick exercise

Echopype: A quick exercise#

In this exercise we will do 4 things:

convert a .raw file collected by an EK80 echosounder
compute volume backscattering strength (Sv) from the parsed raw data
regrid the Sv data to mean volume backscattering strength (MVBS)
visualize using interactive plot

The details verbally discussed during the workshop can be found in the Getting started with Echopype notebook with some additional information, in the “echopype-examples” repository.

import echopype as ep  # we recommend using "ep"
import xarray as xr
import hvplot.xarray  # for interactive plots

import matplotlib.pyplot as plt

Convert a `.raw` file#

Echopype supports accessing data directly from multiple sources, such as a local filesystem (e.g., hard disk), a cloud object store (e.g., AWS S3), an http server, etc.

# Convert from an S3 bucket
raw_path = "s3://noaa-wcsd-pds/data/raw/Bell_M._Shimada/SH2306/EK80/Hake-D20230811-T165727.raw"
ed = ep.open_raw(
    raw_path,
    sonar_model="EK80",
    storage_options={"anon": True},  # open bucket, otherwise need credential
)

Save the `EchoData` object#

# Save to zarr
ed.to_zarr("./resources", overwrite="w")  # save with the same filename, different extension

# Save to netCDF
ed.to_netcdf("./resources/Hake-D20230811-T165727.nc", overwrite="w")  # fully specify the filename also works

Im-memory vs lazy-loaded data#

# Lazy-load EchoData using xarray kwargs (keyword arguments)
ed_lazy = ep.open_converted("./resources/Hake-D20230811-T165727.zarr", chunks={})

# Lazy-loaded
ed_lazy["Sonar/Beam_group1"]["backscatter_r"]

<xarray.DataArray 'backscatter_r' (channel: 5, ping_time: 213,
                                   range_sample: 36198)> Size: 154MB
dask.array<open_dataset-backscatter_r, shape=(5, 213, 36198), dtype=float32, chunksize=(5, 213, 23474), chunktype=numpy.ndarray>
Coordinates:
  * channel       (channel) <U25 500B 'WBT 400140-15 ES120-7C_ES' ... 'WBT 40...
  * ping_time     (ping_time) datetime64[ns] 2kB 2023-08-11T16:57:27.277163 ....
  * range_sample  (range_sample) int64 290kB 0 1 2 3 ... 36194 36195 36196 36197
Attributes:
    long_name:  Raw backscatter measurements (real part)
    units:      dB

Compute Sv#

# Use the correct waveform_mode and encode_mode combination
ds_Sv = ep.calibrate.compute_Sv(ed, waveform_mode="CW", encode_mode="power")
ds_Sv = ep.consolidate.add_depth(ds_Sv, ed)

Check how the Sv data look like:

ds_Sv["Sv"].plot(
    x="ping_time", 
    row="channel", col_wrap=3,
    vmin=-80, vmax=-30,
    cmap="RdYlBu_r", yincrease=False
)

<xarray.plot.facetgrid.FacetGrid at 0x284e78470>

_images/e9e235932863c350d515610daa7699a70e40a5d96706e48514b695fe00319f80.png

Regrid to MVBS#

# Compute MVBS
ds_MVBS = ep.commongrid.compute_MVBS(
    ds_Sv,
    range_var="depth",
    range_bin="1m",
    ping_time_bin="5s",
    range_var_max="500m",  # max range to bin
)

See how MVBS data are now all aligned with one another (hence the name commingrid):

ds_MVBS = ds_MVBS.swap_dims({"channel": "frequency_nominal"})
ds_MVBS["Sv"].plot(
    x="ping_time",
    row="frequency_nominal", col_wrap=3,
    cmap='RdYlBu_r', yincrease=False,
    vmin=-80, vmax=-30
)

<xarray.plot.facetgrid.FacetGrid at 0x28aeac4d0>

_images/5f3042b0340c30442ea664d72a702348c400b7edf8d8927ff14c062dfd8d3093.png

Try interactive plotting#

ds_MVBS["Sv"].hvplot.quadmesh(
    x="ping_time", clim=(-80, -30), figsize=(6, 6), cmap="RdYlBu_r", rasterize=True,
    groupby="frequency_nominal"
).opts(invert_yaxis=True, height=400)

Save the MVBS dataset for future use#

ds_MVBS.to_zarr("./resources/Hake-D20230811-T165727_MVBS.zarr", mode="w")

<xarray.backends.zarr.ZarrStore at 0x28b148280>

Echopype: A quick exercise

Contents

Echopype: A quick exercise#

Convert a `.raw` file#

Check what’s in the raw file#

Save the `EchoData` object#

Im-memory vs lazy-loaded data#

Compute Sv#

Regrid to MVBS#

Try interactive plotting#

Save the MVBS dataset for future use#

Echopype: A quick exercise

Contents

Echopype: A quick exercise#

Convert a .raw file#

Check what’s in the raw file#

Save the EchoData object#

Im-memory vs lazy-loaded data#

Compute Sv#

Regrid to MVBS#

Try interactive plotting#

Save the MVBS dataset for future use#

Convert a `.raw` file#

Save the `EchoData` object#