Easy interpolation of fesom2 output data. Your data should be created with refactoring brunch of the model.
Make something that is more light weight to install compared to pyfesom2 that has similar capabilities, but sometimes could be pain to install.
On most HPC machines you should be able to select python module, that have xarray and preferably cartopy installed, clone repository, and just use it simply as a script (e.g. python /path/to/fint/fint/fint.py [some options]). But if you want to install it in a bit more cleaner way, read instructions below.
- xarray
- netcdf4
- numpy
- scipy
- matplotlib
- pandas
- cartopy (should work without it as well, but ony interpolate to lan lat boxes)
- shapely (also should work without it)
Please follow those instructions to install Miniconda.
Change to fint directory and create separate fint environment with:
conda env create -f environment.ymlactivate it after installation with
conda activate fintand you should be good to go :)
The easiest way to test fint is to execute tests.sh script, that will download sample data and perform some tests. Let's assume you executed the script once, and everything worked. We now can repeat some of the commands from there with explanations.
We set some variables, that will be frequently used as command line arguments:
export FILE="./test/data/temp.fesom.1948.nc"
export MESH="./test/mesh/pi/"
export INFL="--influence 500000"
As bare minimum you should provide path to FESOM2 output file and path to the mesh. We also use --influence, so that output result looks a bit nicer:
python fint.py ${FILE} ${MESH} ${INFL}Currently it's just a list of help from fint. Later will expand each of the options with explination.
usage: pfinterp [-h] [--depths DEPTHS] [--timesteps TIMESTEPS] [--box BOX]
[--res N_POINTS_LON N_POINTS_LAT] [--influence INFLUENCE]
[--map_projection MAP_PROJECTION] [--interp {nn,mtri_linear,linear_scipy}]
[--mask MASK] [--ofile OFILE] [--odir ODIR] [--target TARGET] [--no_shape_mask]
[--rotate] [--no_mask_zero] [--timedelta TIMEDELTA]
data meshpath
Interpolates FESOM2 data to regular grid.
positional arguments:
data Path to the FESOM2 output data file.
meshpath Path to the FESOM2 mesh folder.
optional arguments:
-h, --help show this help message and exit
--depths DEPTHS, -d DEPTHS
Depths in meters. Closest values from model levels will be taken. Several
options available: number - e.g. '100', coma separated list - e.g.
'0,10,100,200', slice 0:100 -1 - all levels will be selected.
--timesteps TIMESTEPS, -t TIMESTEPS
Explicitly define timesteps of the input fields. There are several options:
'-1' - all time steps, number - one time step (e.g. '5'), numbers - coma
separated (e.g. '0, 3, 8, 10'), slice - e.g. '5:10', slice with steps - e.g.
'8:120:12'. slice untill the end of the time series - e.g. '8:end:12'.
--box BOX, -b BOX Several options are available: - Map boundaries in -180 180 -90 90 format that
will be used for interpolation. - Use one of the predefined regions. Available:
gs (Golf Stream), trop (Atlantic Tropics), arctic, gulf (also Golf Stream, but
based on Mercator projection.)\))
--res N_POINTS_LON N_POINTS_LAT, -r N_POINTS_LON N_POINTS_LAT
Number of points along each axis that will be used for interpolation (for lon
and lat).
--influence INFLUENCE, -i INFLUENCE
Radius of influence for interpolation, in meters. Used for nearest neighbor
interpolation.
--map_projection MAP_PROJECTION, -m MAP_PROJECTION
Map projection. Available: mer, np
--interp {nn,mtri_linear,linear_scipy}
Interpolation method. Options are nn - nearest neighbor (KDTree implementation,
fast), mtri_linear - linear, based on triangulation information (slow, but more
precise)
--mask MASK File with mask for interpolation. Mask should have the same coordinates as
interpolated data. Usuall usecase is to use mtri_linear slow interpolation to
create the mask, and then use this mask for faster (nn) interpolation.
--ofile OFILE, -o OFILE
Path to the output file. Default is out.nc.
--odir ODIR Path to the output directory. Default is ./
--target TARGET Path to the target file, that contains coordinates of the target grid (as lon
lat variables)
--no_shape_mask Do not apply shapely mask for coastlines. Useful for paleo applications.
--rotate Rotate vector variables to geographic coordinates. Use standard FESOM2 angles
(this should be fine in 99.99 percent of cases:)).
--no_mask_zero FESOM2 use 0 as mask value, which is terrible for some variables. Solution is
to create a mask using temperature or salinity, and then use this mask for
interpolation, applying this option.
--timedelta TIMEDELTA
Add timedelta to the time axis. The format is number followed by unit. E.g.
'1D' or '10h'. Valid units are 'D' (days), 'h' (hours), 'm' (minutes), 's'
(seconds). To substract timedelta, put argument in quotes, and prepend ' -', so
SPACE and then -, e.g. ' -10D'.You can also use this from the provided Docker container with both docker
and singularity. Here's a usage example for Docker:
$ docker run --rm -v "$(pwd):/app" ghcr.io/fesom/fint:latest ./test/data/temp.fesom.1948.nc ./test/mesh/pi --ofile testing_new.nc
/opt/conda/lib/python3.9/site-packages/cartopy/io/__init__.py:241: DownloadWarning: Downloading: https://naturalearth.s3.amazonaws.com/50m_physical/ne_50m_ocean.zip
timesteps -1
time: 0, depth:2.5
time: 1, depth:2.5
time: 2, depth:2.5
time: 3, depth:2.5
time: 4, depth:2.5
time: 5, depth:2.5
time: 6, depth:2.5
time: 7, depth:2.5
time: 8, depth:2.5
time: 9, depth:2.5
time: 10, depth:2.5
time: 11, depth:2.5
<xarray.Dataset>
Dimensions: (time: 12, depth: 1, lat: 170, lon: 360)
Coordinates:
* time (time) datetime64[ns] 1948-01-31T23:45:00 ... 1948-12-30T23:45:00
* depth (depth) float64 2.5
* lon (lon) float64 -180.0 -179.0 -178.0 -177.0 ... 178.0 179.0 180.0
* lat (lat) float64 -80.0 -78.99 -77.99 -76.98 ... 87.99 88.99 90.0
longitude (lat, lon) float64 -180.0 -179.0 -178.0 ... 178.0 179.0 180.0
latitude (lat, lon) float64 -80.0 -80.0 -80.0 -80.0 ... 90.0 90.0 90.0
Data variables:
temp (time, depth, lat, lon) float64 nan nan nan ... nan nan -1.527
Attributes: (12/22)
FESOM_model: FESOM2
FESOM_website: fesom.de
FESOM_git_SHA: 8d28c8f
FESOM_MeshPath: /home/ollie/nkolduno/pi/fesom2/test/...
FESOM_mesh_representative_checksum:
FESOM_ClimateDataPath: /home/ollie/nkolduno/pi/fesom2/test/...
... ...
FESOM_autorotate_back_to_geo: -1
box: -180.0, 180.0, -80.0, 90.0
influence: 80000
interp: nn
data: /app/test/data/temp.fesom.1948.nc
meshpath: /app/test/mesh/piFor singularity, the command is a little bit more complicated:
$ singularity run --pwd /app -B $(pwd):/app -w docker://ghcr.io/fesom/fint:main ./test/data/temp.fesom.1948.nc ./test/mesh/pi -o out.nc
$ ncdump -h out.nc
netcdf out {
dimensions:
time = 12 ;
depth = 1 ;
lat = 170 ;
lon = 360 ;
variables:
float temp(time, depth, lat, lon) ;
temp:_FillValue = NaNf ;
temp:coordinates = "latitude longitude" ;
double time(time) ;
time:_FillValue = NaN ;
time:units = "days since 1948-01-31 23:45:00" ;
time:calendar = "proleptic_gregorian" ;
double depth(depth) ;
depth:_FillValue = NaN ;
double lon(lon) ;
lon:_FillValue = NaN ;
double lat(lat) ;
lat:_FillValue = NaN ;
double longitude(lat, lon) ;
longitude:_FillValue = NaN ;
double latitude(lat, lon) ;
latitude:_FillValue = NaN ;
// global attributes:
:FESOM_model = "FESOM2" ;
:FESOM_website = "fesom.de" ;
:FESOM_git_SHA = "8d28c8f" ;
:FESOM_MeshPath = "/home/ollie/nkolduno/pi/fesom2/test/meshes/pi/" ;
:FESOM_mesh_representative_checksum = "" ;
:FESOM_ClimateDataPath = "/home/ollie/nkolduno/pi/fesom2/test/input/global/" ;
:FESOM_which_ALE = "zstar" ;
:FESOM_mix_scheme = "KPP" ;
:FESOM_use_partial_cell = -1 ;
:FESOM_force_rotation = -1 ;
:FESOM_include_fleapyear = 0 ;
:FESOM_use_floatice = 0 ;
:FESOM_whichEVP = 1 ;
:FESOM_evp_rheol_steps = 120 ;
:FESOM_opt_visc = 5 ;
:FESOM_use_wsplit = -1 ;
:FESOM_autorotate_back_to_geo = -1 ;
:box = "-180.0, 180.0, -80.0, 90.0" ;
:influence = 80000LL ;
:interp = "nn" ;
:data = "/app/test/data/temp.fesom.1948.nc" ;
:meshpath = "/app/test/mesh/pi" ;
}