Query via h5netcdf - Connected Data Products Demo

Query NISAR GUNW via h5netcdf (baseline)¶

For comparison, open the same NISAR granule the traditional way using earthaccess and h5netcdf, and extract the same spatial subset.

import warnings

warnings.filterwarnings("ignore", category=FutureWarning, module="earthaccess")

Open with earthaccess + h5netcdf¶

import time

import earthaccess
import xarray as xr

earthaccess.login()

results = earthaccess.search_data(
    short_name="NISAR_L2_GUNW_BETA_V1",
    point=(174.1192, -39.3379),  # lon, lat
    temporal=("2026-01-01", "2026-01-04"),
)

files = earthaccess.open(results)
t0 = time.perf_counter()
ds = xr.open_dataset(
    files[0],
    group="/science/LSAR/GUNW/grids/frequencyA/unwrappedInterferogram/HH",
    engine="h5netcdf",
    decode_timedelta=False,
    phony_dims="access",
)
print(f"Opened in {time.perf_counter() - t0:.2f}s")
ds

Extract the same spatial subset¶

import numpy as np

ny, nx = ds.dims["yCoordinates"], ds.dims["xCoordinates"]

# Same 500x500 pixel window from the center of the scene
cy, cx = ny // 2, nx // 2
subset_slice = dict(
    yCoordinates=slice(cy - 250, cy + 250), xCoordinates=slice(cx - 250, cx + 250)
)

t0 = time.perf_counter()
subset = ds.isel(**subset_slice).load()
elapsed = time.perf_counter() - t0

print(f"Loaded 500x500 subset in {elapsed:.2f}s (from {ny}x{nx} full scene)")
subset

Visualize the subset¶

import pygmt

fig = pygmt.Figure()
pygmt.makecpt(
    cmap="SCM/romaO", series=[0, 2 * np.pi, np.pi / 2], cyclic=True, continuous=True
)
fig.grdimage(grid=subset.unwrappedPhase, cmap=True, frame=True)
fig.colorbar()
fig.show()

Visualize coherence magnitude¶

fig = pygmt.Figure()
fig.grdimage(grid=subset.coherenceMagnitude, cmap="gray", frame=True)
fig.colorbar()
fig.show()

Note: earthaccess.open streams the file through HTTPS via fsspec. Loading a spatial subset may still transfer a large portion of the file depending on the HDF5 chunking layout. Compare the wall-clock time with the Icechunk notebook, which fetches only the chunks overlapping the subset.