A tool for making Digital Elevation Maps (DEMs) in binary data format (16-bit integers, little endian) for use in interferometric synthetic aperture radar (InSAR) processing (e.g. using isce2.
The sardem
command creates a cropped (and possibly upsampled) digital elevation map:
usage: sardem [-h] [--bbox left bottom right top] [--geojson GEOJSON] [--wkt-file WKT_FILE] [--xrate XRATE] [--yrate YRATE] [--output OUTPUT] [--data-source {NASA,NASA_WATER,COP}] [-isce] [--keep-egm] [--shift-rsc]
[left_lon] [top_lat] [dlon] [dlat]
Using conda:
conda install -c conda-forge sardem
# Or, if mamba is installed, mamba install -c conda-forge sardem
Using pip:
pip install sardem
This creates the command line executable sardem
Alternatively, you can clone to build/install:
git clone https://github.com/scottstanie/sardem
cd sardem
# Install requirements using either pip or conda
# conda install -c conda-forge --file environment.yml
# pip install -r requirements.txt
# the conda environment.yml is more complete, as GDAL is required for some of the functionality
pip install -e .
which will run pip install --upgrade .
and create the command line script.
The default data source, --data-source NASA
, uses the SRTM 1 arcsecond data. You can also use the newer Copernicus Digital Surface Model (DSM).
To see a comparison of the two, see the srtm_copernicus_comparison notebook.
Note: To convert the elevation values to heights about the WGS84 ellipsoid (which is the default), or to use the Copernicus data, GDAL is required.
For the Copernicus data, the minimum required GDAL version is 3.4.2; versions earlier than 3.4.0 seem to hang upon using gdalwarp
on the global VRT, and <3.4.2 have an internal bug isce-framework/isce2#556 .
sardem
uses the gdal convention ("pixel is area") where --bbox
points to the edges of the [left, bottom, right, top] pixels.
I.e. (left, bottom) refers to the lower left corner of the lower left pixel.
GDAL is required for the conversion, which is installed when using conda install -c conda-forge sardem
.
If you already are using an existing environment, make sure that the GDAL version is >=3.4.2.
conda install -c conda-forge "gdal>=3.4.2"
# or
# conda install -c conda-forge mamba
# mamba install -c conda-forge "gdal>=3.4.2"
The full options for the command line tool in sardem/cli.py
can be found using
$ sardem -h
usage: sardem [-h] [--bbox left bottom right top] [--geojson GEOJSON] [--wkt-file WKT_FILE] [--xrate XRATE] [--yrate YRATE] [--output OUTPUT] [--data-source {NASA,NASA_WATER,COP}] [-isce] [--keep-egm] [--shift-rsc]
[--cache-dir CACHE_DIR]
[left_lon] [top_lat] [dlon] [dlat]
Stiches SRTM .hgt files to make (upsampled) DEM
Pick a lat/lon bounding box for a DEM, and it will download
the necessary SRTM1 tiles, stitch together, then upsample.
The `--bbox` convention points to the *edges* of the [left, bottom, right, top]
pixels, following the "pixel is area" convention as used in gdal.
I.e. (left, bottom) refers to the lower left corner of the lower left pixel.
Usage Examples:
sardem --bbox -156 18.8 -154.7 20.3 # bounding box: [left bottom right top]
sardem -156.0 20.2 1 2 --xrate 2 --yrate 2 # Makes a box 1 degree wide, 2 deg high
sardem --bbox -156 18.8 -154.7 20.3 --data-source COP # Copernicus DEM
sardem --geojson dem_area.geojson -x 11 -y 3 # Use geojson file to define area
sardem --bbox -156 18.8 -154.7 20.3 --data-source NASA_WATER -o my_watermask.wbd # Water mask
sardem --bbox -156 18.8 -154.7 20.3 --data COP -isce # Generate .isce XML files as well
Default out is elevation.dem for the final upsampled DEM.
Also creates elevation.dem.rsc with start lat/lon, stride, and other info.
positional arguments:
left_lon Left (western) most longitude of DEM box (degrees, west=negative)
top_lat Top (northern) most latitude of DEM box (degrees)
dlon Width of DEM box (degrees)
dlat Height of DEM box (degrees)
options:
-h, --help show this help message and exit
--bbox left bottom right top
Bounding box of area of interest (e.g. --bbox -106.1 30.1 -103.1 33.1 ).
--bbox points to the *edges* of the pixels,
following the 'pixel is area' convention as used in gdal.
--geojson GEOJSON, -g GEOJSON
Alternate to corner/dlon/dlat box specification:
File containing the geojson object for DEM bounds
--wkt-file WKT_FILE Alternate to corner/dlon/dlat box specification:
File containing the WKT string for DEM bounds
--xrate XRATE, -x XRATE
Rate in x dir to upsample DEM (default=1, no upsampling)
--yrate YRATE, -y YRATE
Rate in y dir to upsample DEM (default=1, no upsampling)
--output OUTPUT, -o OUTPUT
Name of output dem file (default=elevation.dem for DEM, watermask.wbd for water mask)
--data-source {NASA,NASA_WATER,COP}, -d {NASA,NASA_WATER,COP}
Source of DEM data (default NASA). See README for more.
-isce, --make-isce-xml
Make an isce2 XML file for the DEM.
--keep-egm Keep the DEM heights as geoid heights above EGM96 or EGM2008. Default is to convert to WGS84 for InSAR processing.
--shift-rsc Shift the .rsc file by half a pixel so that X_FIRST and Y_FIRST are at the pixel center (instead of GDAL's convention of the top left edge). Default is GDAL's top-left edge convention.
--cache-dir CACHE_DIR
Location to save downloaded files (Default = /Users/staniewi/.cache/sardem)
The default data source is NASA's Shuttle Radar Topography Mission (SRTM) version 3 global 1 degree data. See https://lpdaac.usgs.gov/dataset_discovery/measures/measures_products_table/srtmgl3s_v003 . The data is valid outside of arctic regions (-60 to 60 degrees latitude), and is zeros over open ocean.
This data requires a username and password from here: https://urs.earthdata.nasa.gov/users/new
You will be prompted for a username and password when running with NASA data. It will save into your ~/.netrc file for future use, which means you will not have to enter a username and password any subsequent times. The entry will look like this:
machine urs.earthdata.nasa.gov
login USERNAME
password PASSWORD