This is a package written initially by Ernesto Rodriguez to estimate various river parameters starting from remote sensing data. Alex Fore, Brent Williams, Cassie Stuurman, Rui Wei, and Renato P.M. Frasson have also provided code to reflect the evolving SWOT project. The code is currently maintained by the SWOT Algorithm Definition Team.
Detailed installation instructions are in the Install.md file.
For generating data products are most similar to the SWOT project's data products, the following script is recommended (found in src/bin):
usage: swot_pixc2rivertile.py [-h] [--shpbasedir SHPBASEDIR] [-l LOG_LEVEL]
[--gdem-file GDEM_FILE]
pixc_file out_riverobs_file out_pixc_vector_file
rdf_file
where pixc_file is the SWOT high-resolution pixel cloud data product, out_riverobs_file is the filename of the output rivertile data product, out_pixc_vector_file is the filename of the output pixel cloud vector data product, rdf_file is the configuration file (see this link) for the recommended configuration). Additionally there are some optional arguments: --shpbasedir SHPBASEDIR will write out the nodes and reaches as shapefile format (written as netCDF to out_riverobs_file), -l LOG_LEVEL controls the verbosity of the logging, and --gdem-file GDEM_FILE will create a pixc_file from the GDEM file and run RiverObs on that as a type of truth processing.
RiverObs requires a prior reach and node database. The database contains fixed node locations, reach boundaries, and high-resolution reach centerlines. It is distributed as a set of netcdf files, broken by continent (first two characters in the file name) and "major basins" in the continent (3rd and 4th characters in the file name). Metadata describing the database fields and the current version of the database is available here.
RiverObs: This is the main package for associating data with river reaches, and estimating hydrology parameters base on reach averaging (or not...). In addition to the homegrown packages listed below, this package requires the following open source packages:
- scipy: Science algorithms swiss army knife.
- numpy: Numerics swiss army knife.
- netCDF4: Reading netcdf4 files, including SWOT L2 data files.
- StatsModels: Fitting and estimation tools.
- pysal: nice interface to shapefiles and shapely bridge.
- pyproj: Cartographic projections swiss army knife.
- pandas: The Python Data Analysis Library for DataFrames and HDFStore.
- pytables: easy HDF5 support, required for pandas HDFStore.
Centerline: Provides a class that can be used to project data or refine a river center line. Requires the following packages:
GeometryDataBase: Find quickly which reach intersects with a geometry of interest. The geometries are assumed to be stored in a shapefile. Requires the following packages:
- Rtree: Fast bounding box queries.
- libspatialindex: Required by Rtree.
- pysal: nice interface to shapefiles and shapely bridge.
- shapely: geometry calculations.
SWOTRiver: This package contains classes that use the RiverObs capabilities to produce hydrology outputs from SWOT (simulated) data.
- numpy: Numerics swiss army knife.
- netCDF4: Reading netcdf4 files, including SWOT L2 data files.
- pyproj: Cartographic projections swiss army knife.
- pandas: The Python Data Analysis Library for DataFrames and HDFStore.
- pytables: easy HDF5 support, required for pandas HDFStore.
GDALOGRUtilities: Provides homegrown utilities for reading and writing various GIS files. Requires the following packages:
GWDLR: This is an optional package to convert Global Width Database-Large Rivers raster data provided by Dai Yamazaki to vectors that can be used as centerlines. Requires:
- grass: for raster to vector program.
- scikit-image: for skeletonize.