This repository hosts source code of LISFLOOD utilities. Go to Lisflood OS page for more information.
Other useful resources
Project | Documentation | Source code |
---|---|---|
Lisflood | Model docs | https://github.com/ec-jrc/lisflood-code |
User guide | ||
Lisvap | Docs | https://github.com/ec-jrc/lisflood-lisvap |
Calibration tool | Docs | https://github.com/ec-jrc/lisflood-calibration |
Lisflood Utilities | https://github.com/ec-jrc/lisflood-utilities (this repository) | |
Lisflood Usecases | https://github.com/ec-jrc/lisflood-usecases |
Lisflood Utilities is a set of tools to help LISFLOOD users (or any users of PCRaster/netCDF files) to execute some mundane tasks that are necessary to operate lisflood. Here's a list of utilities you can find in lisflood-utilities package.
-
pcr2nc is a tool to convert PCRaster maps to netCDF4 files.
- convert a single map into a NetCDF4 file
- convert a time series of maps into a NetCDF4 mapstack
- support for WGS84 and ETRS89 (LAEA) reference systems
- fine tuning of output files (compression, significant digits, etc.)
-
nc2pcr is a tool to convert a netCDF file into PCRaster maps.
- convert 2D variables in single PCRaster maps
- netCDF4 mapstacks are not supported yet
-
cutmaps is a tool to cut netcdf files in order to reduce size, using either
- a bounding box of coordinates
- a bounding box of matrix indices
- an existing boolean area mask
- a list of stations and a LDD (in netCDF or PCRaster format) Note: PCRaster must be installed in the conda env
-
compare is a package containing a set of simple Python classes that helps to compare netCDF, PCRaster and TSS files.
The package contains convenient classes for reading/writing:
- PCRasterMap
- PCRasterReader
- NetCDFMap
- NetCDFWriter
The easy way is to use conda environment as they incapsulate C dependencies as well, so you wouldn't need to install libraries.
Otherwise, ensure you have properly installed the following software:
- Python 3.5+
- GDAL C library and software
- netCDF4 C library
If you use conda, create a new env (or use an existing one) and install gdal and lisflood-utilities:
conda create --name myenv python=3.7 -c conda-forge
conda activate myenv
conda install -c conda-forge pcraster gdal
pip install lisflood-utilities
If you don't use conda but a straight python3 virtualenv:
source /path/myenv/bin/activate
pip install lisflood-utilities
If GDAL library fails to install, ensure to install the same package version of the C library you have on your system. You may also need to setup paths to gdal headers.
To check which version of GDAL libraries you have installed on your computer, use gdal-config
sudo apt-get install libgdal-dev libgdal
export CPLUS_INCLUDE_PATH=/usr/include/gdal
export C_INCLUDE_PATH=/usr/include/gdal
gdal-config --version # 3.0.1
pip install GDAL==3.0.1
Note: This guide assumes you have installed the program with pip tool. If you cloned the source code instead, just substitute the executable
pcr2nc
withpython pcr2nc_script.py
that is in the root folder of the cloned project.
The tool takes three command line input arguments:
- -i, --input: It can be a path to a single file, a folder or a unix-like widlcard expression like /path/to/files/dis00*
- -o, --output_file: Path to the output nc file
- -m, --metadata: Path to a yaml or json file containing configuration for the NetCDF4 output file.
Unless the input is a single file, the resulting NetCDF4 file will be a mapstack according to a time dimension.
Input as a folder containing PCRaster maps. In this case, the folder must contain ONLY PCraster files and the output will be a mapstack.
pcr2nc -i /path/to/input/ -o /path/to/output/out.nc -m ./nc_metadata.yaml
Input as a path to a single map. In this case, the output won't be a mapstack.
pcr2nc -i /path/to/input/pcr.map -o /path/to/output/out.nc -m ./nc_metadata.yaml
Input as a Unix style pathname pattern expansion. The output will be a mapstack. Note that in this case the input argument must be contained in double quotes!
pcr2nc -i "/path/to/input/pcr00*" -o /path/to/output/out.nc -m ./nc_metadata.json
Format of resulting NetCDF4 file is configured into a metadata configuration file. This file can be written in YAML or JSON format.
An example of a metadata configuration file is the following
variable:
shortname: dis
description: Discharge
longname: discharge
units: m3/s
compression: 9
least_significant_digit: 2
source: JRC Space, Security, Migration
reference: JRC Space, Security, Migration
geographical:
datum: WGS84
time:
calendar: proleptic_gregorian
units: days since 1996-01-01
In variable
section you can configure metadata for the main variable:
shortname
: A short name for the variablelongname
: The long name versiondescription
: A description for humansunits
: The units of the variablecompression
: Optional, integer number between 1 and 9, default 0 (no compression). If present the output nc file will be compressed at this level.least_significant_digit
: Optional, integer number, default 2. From NetCDF4 documentation:
If your data only has a certain number of digits of precision (say for example, it is temperature data that was measured with a precision of 0.1 degrees), you can dramatically improve zlib compression by quantizing (or truncating) the data using the least_significant_digit keyword argument to createVariable. The least significant digit is the power of ten of the smallest decimal place in the data that is a reliable value. For example if the data has a precision of 0.1, then setting least_significant_digit=1 will cause data the data to be quantized using
numpy.around(scale*data)/scale
, wherescale = 2**bits
, and bits is determined so that a precision of 0.1 is retained (in this case bits=4). Effectively, this makes the compression 'lossy' instead of 'lossless', that is some precision in the data is sacrificed for the sake of disk space.
source
and reference
add information for the institution that is providing the NetCDF4 file.
In geographical
section the only setting to configure is datum
.
Currently, pcr2nc supports the following list:
WGS84
ETRS89
GISCO
This section is optional and is only required if the output file is a mapstack (a timeseries of georeferenced 2D arrays)
In this section you have to configure units
and calendar
.
units
: Can be one of the following strings (replacing placeholders with the actual date):hours since YYYY-MM-DD HH:MM:SS
days since YYYY-MM-DD
calendar
: A recognized calendar identifier, likeproleptic_gregorian
,gregorian
etc.
This tool converts single maps netCDF (time dimension is not supported yet) to PCRaster format.
nc2pcr -i /path/to/input/file.nc -o /path/to/output/out.map -c /path/to/clone.map
If input file is a LDD map, you must add the -l
flag:
nc2pcr -i /path/to/input/ldd.nc -o /path/to/output/ldd.map -c /path/to/clone.map -l
This tool cut netcdf files, using a mask, a bounding box or a list of stations along with a LDD map.
The tool accepts as input:
- a mask map (either PCRaster or netCDF format) or
- alternatively, matrix indices in the form xmini_xmaxi:ymini_ymaxi or
- alternatively, coordinates bounding box in the form xmin_xmax:ymin_ymax
- alternatively, list of stations with coordinates and a LDD map.
- a path to a folder containing netCDF files to cut or a static dataset path like LISFLOOD static files.
- a path to a folder where to write cut files.
The following command will cut all netcdf files inside /workarea/Madeira/lai/ folder and produced files will be writte in current folder. The cookie-cutter that will be used is /workarea/Madeira/maps/MaskMap/Bacia_madeira.nc. This file is a mask (boolean map with 1 only in the area of interest) where cutmaps takes the bounding box from. The mask can also be in PCRaster format.
cutmaps -m /workarea/Madeira/maps/MaskMap/Bacia_madeira.nc -f /workarea/Madeira/lai/ -o ./
Indices can also be passed as an argument (using -c argument instead of -m). Knowing your area of interest from your netCDF files,
you can determine indices of the array and you can pass in the form imin_i_max:jmin_jmax
.
cutmaps -c 150_350:80_180 -f /workarea/Madeira/lai/ -o ./
Example with coordinates and path to EFAS/GloFAS static data (-S option), with -W to allow overwriting existing files in output directory:
cutmaps -S /home/projects/lisflood-eu -c 4078546.12_4463723.85:811206.57_1587655.50 -o /Work/Tunisia/cutmaps -W
Example with stations.txt and LDD
Given a LDD map and a list of stations in a text file, each row having coordinates X/Y or lat/lon and an index, separated by tabs:
4297500 1572500 1
4292500 1557500 2
4237500 1537500 3
4312500 1482500 4
4187500 1492500 5
cutmaps -S /home/projects/lisflood-eu -l ldd.map -N stations.txt -o /Work/Tunisia/cutmaps
If ldd is in netCDF format, LDD will be converted to PCRaster format, first.
cutmaps -S /home/projects/lisflood-eu -l ldd.nc -N stations.txt -o /Work/Tunisia/cutmaps
If you experience problems, you can try to pass a path to a PCRaster clone map.
cutmaps -S /home/projects/lisflood-eu -l ldd.nc -C area.map -N stations.txt -o /Work/Tunisia/cutmaps
You will find the produced mask.map and mask.nc for your area in the same folder of ldd map; you will need it for lisflood/lisvap executions. You will also have outlets.map/outlets.nc based on stations.txt, which let you produce gauges TSS if configured in LISFLOOD.
This tool let you compare two netcdf datasets. You can configure it with tolerances (atol, rtol, thresholds for percentage of tolerated different values). You can also set the option to write diff files, so that you can inspect maps and differences with a tool like Panoply
usage: compare [-h] -a DATASET_A -b DATASET_B -m MASKAREA [-M SUBMASKAREA]
[-e] [-s] [-D] [-r RTOL] [-t ATOL] [-p MAX_DIFF_PERCENTAGE]
[-l MAX_LARGEDIFF_PERCENTAGE]
Compare netCDF outputs: 0.12.12
optional arguments:
-h, --help show this help message and exit
-a DATASET_A, --dataset_a DATASET_A
path to dataset version A
-b DATASET_B, --dataset_b DATASET_B
path to dataset version B
-m MASKAREA, --maskarea MASKAREA
path to mask
-e, --array-equal flag to compare files to be identical
-s, --skip-missing flag to skip missing files in comparison
-D, --save-diffs flag to save diffs in netcdf files for visual
comparisons. Files are saved in ./diffs folder of
current directory.For each file presenting
differences, you will find files diffs, original A and
B (only for timesteps where differences are found).
-r RTOL, --rtol RTOL rtol
-t ATOL, --atol ATOL atol
-p MAX_DIFF_PERCENTAGE, --max-diff-percentage MAX_DIFF_PERCENTAGE
threshold for diffs percentage
-l MAX_LARGEDIFF_PERCENTAGE, --max-largediff-percentage MAX_LARGEDIFF_PERCENTAGE
threshold for large diffs percentage
The tool Gfit2 was created to extract flood warning thresholds from a map stack of daily discharge for several years, given in NetCDF format. It is designed to work with any other variable as well as with different sampling (sub- or super- daily)
The tool is made of three files:
- Gfit2.sh: the main file, used to run the tool (e.g., ./Gfit2.sh ./settingFile_test.sh)
- settingFile_test.sh: a setting file, where paths and values of the input variables are defined. It can be renamed, as long as the correct name is called in the script (e.g.: set1.sh --> ./Gfit2.sh ./set1.sh)
- gfit2.r: the r script that performs the extreme value analysis
You need to have installed the following:
- CDO
- R
The tool operates in a Linux environment (also as a job with qsub). To run the tool you'll need:
- R (the script was tested with R version 3.5.0)
- the following R packages: ncdf4, lmomco, ismev
- CDO (the script was tested with CDO version 1.6.5.1)
The Gfit2 tool performs the following steps:
- It takes the input file in NetCDF and extract the series of annual maxima. an optional number of warm-up years ($warmup_yrs) are excluded at the beginning of the data, to remove potential spin-up effect. Also the map of average value is computed
- An extreme value fitting is performed on each pixel of the map, using L-moments and a 2-parameter Gumbel distribution. As option, the user can limit the analysis to a specific number of years. Also, one can use the option to remove from the fitting all values smaller than the long-term average. The fitting is performed only where at least 5 annual maxima are available, otherwise a NA is returned on the output map
- Output return level maps (corresponding to user-selected years of recurrence interval) are saved in a netcdf file (return_levels.nc) and as ascii files. If a clone map in PCRaster format is provided, maps are also saved in PCRaster (return level maps and parameters of the Gumbel distribution)
- After the EV fitting, the tool estimates some further statistics from the long-term input file, including minimum, maximum, and different percentile maps. This part can be commented out if not of interest
L-Moments: Analysis and Estimation of Distributions Using Linear Combinations of Order Statistics
Hosking, J.R.M., 1990. L-Moments: Analysis and Estimation of Distributions Using Linear Combinations of Order Statistics. J. R. Stat. Soc. Ser. B Methodol. 52, 105124.
You can use lisflood utilities in your python programs.
from lisfloodutilities.cutmaps.cutlib import mask_from_ldd
from lisfloodutilities.nc2pcr import convert
from lisfloodutilities.readers import PCRasterMap
ldd = 'tests/data/cutmaps/ldd_eu.nc'
clonemap = 'tests/data/cutmaps/area_eu.map'
stations = 'tests/data/cutmaps/stations.txt'
ldd_pcr = convert(ldd, clonemap, 'tests/data/cutmaps/ldd_eu_test.map', is_ldd=True)[0]
mask = mask_from_ldd(ldd_pcr, stations)
mask_map = PCRasterMap(mask)
print(mask_map.data)