Subsetter

The subsetter is a suite of tools for clipping ParFlow inputs and outputs. Included are command line scripts to:

1. subset the inputs from CONUS1 or CONUS2 domains subset_conus
2. generate a rasterized mask from a shapefile rasterize_shape
3. clip data from any number of supported input files bulk_clip

Prerequisites

Packages

• parflowio
• gdal
• numpy
• pyyaml
• pandas
• pftools

Building Solid Files

To build solid files (.pfsol), the following ParFlow tool is required:

• pfmask-to-pfsol (included with ParFlow)

Environment Variables

For the SolidFileBuilder to work, it must be able to locate pfmask-to-pfsol

1. PARFLOW_DIR environment variable must be set
2. bin/pfmask-to-pfsol must be located directly under PARFLOW_DIR

Setup

It is strongly recommended that you create a clean environment using anaconda or miniconda as it will make installing the required GDAL package easier.

The easiest way to install the subsetter:

$ pip install parflow-subsetter

Otherwise, you can use the contents of this repository to install the package.

git clone https://github.com/hydroframe/subsetting
cd subsetting
conda env create -f=environment.yml
conda activate pf_subsetter

Input Files

For CONUS1 and CONUS2 domains, local copies of model input files are required.

CONUS1 Files

required_files:
  DOMAIN_MASK: Domain_Blank_Mask.tif
  SUBSURFACE_DATA: grid3d.v3.pfb
  PME: PmE.flux.pfb
  SLOPE_X: slopex.pfb
  SLOPE_Y: slopey.pfb
optional_files:
  LAND_COVER: conus1_landcover.sa
  LAT_LON: conus1_Grid_Centers_Short_Deg.format.sa
  DEM: CONUS2.0_RawDEM_CONUS1clip.tif

CONUS2 Files

required_files:
  DOMAIN_MASK: conus_1km_PFmask2.tif
  SUBSURFACE_DATA: 3d-grid.v3.tif
  PME: PME.tif
  SLOPE_X: Str3Ep0_smth.rvth_1500.mx0.5.mn5.sec0.up_slopex.tif
  SLOPE_Y: Str3Ep0_smth.rvth_1500.mx0.5.mn5.sec0.up_slopey.tif
  SINKS: conus_1km_PFmask_manualsinks.tif
  RESERVOIRS: conus_1km_PFmask_reservoirs.tif
  LAKE_BORDER: conus_1km_PFmask_selectLakesborder.tif
  LAKE_MASK: conus_1km_PFmask_selectLakesmask.tif
  CHANNELS: 1km_upscaledNWM_ChannelOrder5_mod2.tif
  CELL_TYPES: 1km_PF_BorderCellType.tif
optional_files:
  LAND_COVER: 1km_CONUS2_landcover_IGBP.tif
  LAT_LON: latlonCONUS2.sa
  DEM: CONUS2.0_RawDEM.tif

Local Filenames

If your local filenames differ from this list, update the local filenames in:

parflow/subset/data/conus_manifest.yaml

Or copy the format of the file list above into a new .yaml file and pass it as an argument to subset_conus script using the --manifest parameter

Testing

Run these tests to do some basic checks to make sure the tools are working correctly after install or update.

chmod +x run_tests.sh
./run_tests.sh

Usage

Create subset from CONUS models from a shapefile

python -m parflow.subset.tools.subset_conus --input_path -i INPUT_PATH --shapefile -s
                                     SHAPEFILE --conus_files -f CONUS_FILES
                                     [--manifest -m MANIFEST_FILE]
                                     [--version -v {1,2}] [--out_dir -o OUT_DIR]
                                     [--out_name -n OUT_NAME] [--clip_clm -c]
                                     [--write_tcl -w]
                                     [--padding -p Top Right Bottom Left]
                                     [--attribute_ids -a ATTRIBUTE_IDS [ATTRIBUTE_IDS ...]]
                                     [--attribute_name -e ATTRIBUTE_NAME]
                                     [--tif_outs -t]

Example usage:

Create a subset of the CONUS1 domain with CLM inputs based on the shapefile at ~/downloads/shapfiles/WBDHU8.shp and write the .tcl file to run the model

python -m parflow.subset.tools.subset_conus -i ~/downloads/shapefiles -s WBDHU8 -f ~/downloads/conus1 -v 1 -w -n watershedA_conus1_clip

Rasterize a shapefile for use as a mask

python -m parflow.subset.tools.rasterize_shape -i <path_to_shapefile_parts> -s <shapefile_name> -r <reference_dataset> 
                              -o [path_to_write_outputs=.] 
                              -n [output_filename=shapfile_name] 
                              -p [padding clockwise from top (top,right,bottom,left)=(0,0,0,0)]                              
                              -e [shapefile_attribute_name='OBJECTID'] 
                              -a [shapefile_attribute_ids=[1]]

Example usage:

Re-project the shapefile at ~/downloads/shapfiles/WBDHU8.shp to the CONUS1 projection and extent

python -m parflow.subset.tools.rasterize_shape -i ~/downloads/shapefiles -s WBDHU8

Use a mask to clip multiple files to PFB or TIF

assumes all files are identically gridded and same as the mask file, if write_tifs=1 then you must supply at least one tif with correct projection and transform information as either the mask file, as a reference dataset with the -r option, or in the list of datafiles to clip

python -m parflow.subset.tools.bulk_clipper (--maskfile -m MASK_FILE | --bboxfile -b BBOX_FILE | --inline-bbox -l X1 Y1 NX NY)
       (--datafiles -d DATA_FILES [DATA_FILES ...] | --glob -g GLOB_PATTERN)
       [--input_path -i INPUT_PATH] [--ref_file -r REF_FILE] [--out_dir -o OUT_DIR]
       [--pfb_outs -p] [--tif_outs -t]

Example usage with mask file:

Clip the domain outputs to the bounds of a mask generated from rasterize_shape or subset_conus

python -m parflow.subset.tools.bulk_clipper -m ~/outputs/WBDHU8.tif -d ~/outputs/runname.out.press.00001.pfb ~/outputs/runname.out.press.00002.pfb

Example usage with bounding box file:

Clip the domain outputs, starting at x, y, and extending for nx, ny

python -m parflow.subset.tools.bulk_clipper -b ~/outputs/bbox.txt -d ~/outputs/runname.out.press.00001.pfb ~/outputs/runname.out.press.00002.pfb

where bbox.txt is a tab-separated text file in the format:

Example bbox.txt clipping only the very first (lower left) cell in a domain:

Example usage with bounding box declared in-line:

Clip the domain outputs, starting at x, y, and extending for nx, ny

python -m parflow.subset.tools.bulk_clipper -l 1 1 10 20 -d ~/outputs/runname.out.press.00001.pfb ~/outputs/runname.out.press.00002.pfb

clips a 10 x 20 grid anchored in the lower left from each of the inputs

Optional Arguments Explanation

Many optional arguments are available for subset_conus and rasterize_shape. Below is an explanation of the options.

-n [name for output files=shapfile_name] The name to give the output raster, defaults to shapefile name
-v [conus verson=1] The version of the ParFlow CONUS model to subset from (1 or 2), defaults to version 1
-o [path_to_write_outputs=.] The path to write the output files, defaults to current directory
-c [clip_clim] Whether or not to clip the CLM lat/lon and vegm data. Defaults to False.
-w [write_tcl] Whether or not to write the .tcl file to run the ParFlow model. Defaults to False
-p [padding for mask areas] specified clockwise from top (top, right, bottom, left). Default (0,0,0,0)
-e [shapefile_attribute_name='OBJECTID'] The name of the attribute table column to uniquely ID objects. 
-a [shapefile_attribute_ids=[1]] The list of objects in the shapefile to rasterize. Default [1]
-t [tif_outs] Whether or not to write outputs as .tif files. Defaults to False.

Supported File Types

1. .pfb
2. .tif
3. .sa