BISICLES file tools
BISICLES includes some tools used to perform miscellaneous actions on BISICLES/Chombo hdf5 files (amr files). They are located in the code/filetools directory. The most commonly used are described below.
When built (e.g by typing 'make all' in the code/filetools directory) a number of executables, with names like e.g extract2d.Linux.64.g++.gfortran.DEBUG.ex are created. Here, we will use the base name (e.g) 'extract' in examples; in practice, you would need e.g extract2d.Linux.64.g++.gfortran.DEBUG.ex. All of the filetools will provide a usage string when executed with no (or the incorrect number) of command line arguments.
extract
Extract one or more fields from one amr file and write to a new amr file.
usage: extract <input_file> <output_file> <var 1> [<var 2> [...]]
merge
combine fields from two amr files but on the same mesh into one amr file.usage: merge <input_file a> <input_file b> <output_file>
pythonf
Evaluate a python cell by cell over an amr file and save the results to a new amr file.usage: pythonf <input_file> <output_file> <python script> <python function> <input tuple> <output tuple>
pythonf can be used to carry out a wide range of post- and pre-processing tasks. For example, imagine a file 'linear_traction.2d.hdf5' that contains fields C (a drag coefficient), u (x-velocity), and v (y-velocity) that allowed the basal traction to be computed for a linear viscous law (T_x = C * u, T_y = C * v). This would be a typical result from the inverse problem. It is possible to create an hdf5 designed to work with a third-power law ( T_i = D * |u|^{-2/3} * u_i) that should produce the same initial velocity by setting D = C * |u|^2/3. Create a python file 'ctransform.py'
#ctransform.py
def c_third(c_one,u,v,*etc):
uu = (u*u + v*v)**(0.5)
c_third = c_one * uu**(2.0/3.0)
return c_third
then run
$BISICLES_HOME/BISICLES/code/filetools/pythonf2d.Linux.64.g++.gfortran.DEBUG.ex linear_traction.2d.hdf5 third_traction.2d.hdf5 ctransform c_third C,u,v D,u,v
stats
Compute summary statistics (e.g volume) from a plot file.usage: stats <plot file> <ice_density> <water_density> <gravity> [mask_file] [mask_no_start = 0] [mask_no_end = mask_no_start]
The stats tool is primarily used to compute quantities like ice volume and volume above flotation. At minimum, supply a file name and <ice_density> <water_density> <gravity> arguments. e.g, for the Pine Island Glacier example:
../../code/filetools/stats2d.Linux.64.g++.gfortran.DEBUG.ex plot.pigv5.1km.l1l2.2lev.000000.2d.hdf5 918 1028 9.81 | grep time time = 0.000000000000e+00 iceVolumeAll = 1.327650177413e+14 iceVolumeAbove = 6.305522125119e+13 groundedArea = 8.783100000000e+10 floatingArea = 5.154000000000e+09 totalArea = 9.298500000000e+10 groundedPlusOpenLandArea = 8.783100000000e+10 Total Melt = -1.995254865428e+11.
It is also possible to calculate the same statistics for a number of sub-regions. For that, you need an hdf5 (e.g mask.2d.hdf5) file (created with nctoamr, for example) including a single field (named e.g 'mask') that identifies each area by a unique double precision representation of an integer. If the Pine Island Glacier domain was split into sub-regions 0-3 inclusive, stats for regions 1,2 could be computed by running
../../code/filetools/stats2d.Linux.64.g++.gfortran.DEBUG.ex plot.pigv5.1km.l1l2.2lev.000000.2d.hdf5 918 1028 9.81 mask.2d.hdf5 1 2
Output would include one line per sub-region.
flatten
Produce a uniform mesh hdf5 or netcdf file by conservative averaging/interpolation data of from an amr file.
usage: flatten <input_file> <output_file> level [x0 [y0 [z0]]]
The main benefit of BISICLES AMR is in the solution of PDEs to compute the ice sheet state. The state itself is usually amenable to analysis at more modest resolution than is required around the grounding line, and on top of that the hdf5 amr files are a specialized format that many find awkward. The flatten tool can be used to make a uniform mesh version of the amr file, either as an hdf5 file that BISICLES can read for other purposes, or a netcdf file for wider dissemination. All of the interpolation is locally conservative.
When writing a NetCDF file, the location of the problem domain origin can be specified as a command line argument. For example, the Pine Island Glacier example is set on a domain whose origin corresponds to the point (-1707 km, -384 km ) on the usual Antarctic Polar Stereographic Projection (EPSG 3031). To make a 4km resolution file that can be directly compared with other data on that projection, run (e.g)
$BISICLES_HOME/BISICLES/code/filetools/flatten2d.Linux.64.g++.gfortran.DEBUG.ex plot.pigv5.1km.l1l2.2lev.000000.2d.hdf5 pig-epsg3031.nc 0 -1707000 -384000
For data at 1 km resolution, run
$BISICLES_HOME/BISICLES/code/filetools/flatten2d.Linux.64.g++.gfortran.DEBUG.ex plot.pigv5.1km.l1l2.2lev.000000.2d.hdf5 pig-epsg3031.nc 2 -1707000 -384000
nctoamr
Convert a uniform mesh netcdf file into a uniform mesh hdf5 file.usage: nctoamr <input_file> <output_file> <var 1> [<var 2>, ...]
If you have a NetCDF file called data.nc, which contains 2D fields named thk and topg arranged on a regular grid (with equal spacing in x and y), and must also contain a 1D field called x, you can run
$BISICLES_HOME/BISICLES/code/filetools/nctoamr2d.Linux.64.g++.gfortran.DEBUG.ex data.nc data.2d.hdf5 thk topg
to obtain a file called data.2d.hdf5 that BISICLES can read as (say) thickness and topography data.