The command line, end to end¶

disp_xr runs the common steps without Python:

Command	What it does
search	find a frame's granules in NASA CMR
info	summarize local granules: frame, dates, versions, duplicates
stack	correct, mask, chain and re-reference into one store
export	write COG, GeoTIFF, Zarr, GeoZarr (for bowser), NetCDF or MintPy
geometry	download the static layers (LOS, DEM, masks) on the DISP grid
virtual	build a virtual frame in Icechunk without copying data

This page runs info, stack and export on synthetic granules. Finding and downloading real ones looks like this (it needs Earthdata credentials in ~/.netrc):

disp_xr search --frame-id 11116 --start 2024-01-01 --out granules.csv
disp_xr geometry F11116/ --out static/

In [1]:

import json
import subprocess
import tempfile
from pathlib import Path

import disp_xr as dxr
from disp_xr.tutorial import make_disp_stack

work = Path(tempfile.mkdtemp())
make_disp_stack(work / "F11116", "three_ministacks", shape=(64, 64))


def run(*args):
    """Run a disp_xr command in the work folder and print what it prints."""
    print("$ disp_xr " + " ".join(args))
    out = subprocess.run(["disp_xr", *args], capture_output=True, text=True, cwd=work)
    print(out.stdout.strip() or out.stderr.strip())
    out.check_returncode()

import json import subprocess import tempfile from pathlib import Path import disp_xr as dxr from disp_xr.tutorial import make_disp_stack work = Path(tempfile.mkdtemp()) make_disp_stack(work / "F11116", "three_ministacks", shape=(64, 64)) def run(*args): """Run a disp_xr command in the work folder and print what it prints.""" print("$ disp_xr " + " ".join(args)) out = subprocess.run(["disp_xr", *args], capture_output=True, text=True, cwd=work) print(out.stdout.strip() or out.stderr.strip()) out.check_returncode()

What is in a folder of granules¶

In [2]:

run("info", "F11116")

run("info", "F11116")

$ disp_xr info F11116

Product:          DISP-S1
Frame:            11116
Granules:         15 in 3 ministacks
Dates:            2020-01-13 to 2020-06-29
Reference dates:  2020-01-01, 2020-03-01, 2020-04-30
Versions:         1.0
Dropped duplicates: 0

One stack¶

stack applies the steps in the product's order: subtract the corrections the product has not applied, apply the recommended mask, chain the ministacks, then re-reference (auto by default). The output is a Zarr store, or NetCDF with .nc, laid out for time series (--preset timeseries) or maps (--preset maps).

In [3]:

run("stack", "F11116", "--out", "F11116.zarr", "--reference", "median")

run("stack", "F11116", "--out", "F11116.zarr", "--reference", "median")

$ disp_xr stack F11116 --out F11116.zarr --reference median

Wrote F11116.zarr (16 epochs)

Export¶

export reads a store from stack (or granules, chained without corrections, masking or re-referencing) and writes other formats. Here, a Cloud Optimized GeoTIFF of the last epoch, and a GeoZarr store that the bowser viewer opens directly (--flat: this frame is too small for the display pyramid that full frames get).

In [4]:

run("export", "F11116.zarr", "--format", "cog", "--out", "last.tif")
run("export", "F11116.zarr", "--format", "geozarr", "--out", "view.zarr", "--flat")

run("export", "F11116.zarr", "--format", "cog", "--out", "last.tif") run("export", "F11116.zarr", "--format", "geozarr", "--out", "view.zarr", "--flat")

$ disp_xr export F11116.zarr --format cog --out last.tif

Wrote 1 Cloud Optimized GeoTIFF file(s) to last.tif
$ disp_xr export F11116.zarr --format geozarr --out view.zarr --flat

Wrote view.zarr

Back in Python¶

Every output records where it came from: input granules, the steps and their parameters, and package versions.

In [5]:

ds = dxr.open_zarr(work / "F11116.zarr")
print(dict(ds.sizes))
print(ds.attrs["reference"])
print("steps:", [step["op"] for step in ds.disp.history])
# Lists are stored as JSON text, so NetCDF and Zarr outputs read the same.
print("from", len(json.loads(ds.attrs["input_granules"])), "granules")

ds = dxr.open_zarr(work / "F11116.zarr") print(dict(ds.sizes)) print(ds.attrs["reference"]) print("steps:", [step["op"] for step in ds.disp.history]) # Lists are stored as JSON text, so NetCDF and Zarr outputs read the same. print("from", len(json.loads(ds.attrs["input_granules"])), "granules")

{'time': 16, 'y': 64, 'x': 64}

{"method": "median"}
steps: ['open_disp', 'correct', 'mask', 'chain', 'rereference']
from 15 granules