ParNCL 1.0.0b1 - RELEASE NOTES

Sample NCL scripts

Some sample NCL scripts that work with ParNCL 1.0.0b1 are available here.

Supported NCL Language Features

ParNCL is a parallel version of NCL that runs NCL scripts in parallel and performs data analysis using ParGAL. All the graphics functions in ParNCL are executed serially using existing NCL routines. The features available in ParNCL as of 1.0.0b1 are given below,

Variables

ParNCL supports NCL variables with ancillary data like attributes, dimension and coordinates. The multi-dimensional data stored in climate data files are read by ParNCL using a parallel netcdf reader and distributed across multiple processes that are part of a ParNCL job. ParNCL currently supports viewing attributes and dimensions of a variable. Although variable coordinates are read from the climate data file and used for data analysis operations they are not exposed currently to the user. This will be fixed in upcoming releases. ParNCL also supports missing values associated with a data variable (via the "_FillValue" attribute).

ParNCL also supports standard variable subscripting for climate data read from CAM Eulerian Spectral and Finite Volume grids. Range and vector subscripting is supported. This version does not support Coordinate or Named subscripting. It also does not support variable subscripting for the CAM HOMME grid. This support will be added in upcoming releases.

Variables can be assigned to one another and the ancillary data corresponding to the variables are copied as in NCL. File variables in climate data files in the netcdf format can be read using the addfile() and addfiles() functions.

This version does not support user data creation functions like new(), fspan() to create distributed data. Support for these functions will be added soon.

Statements

ParNCL supports blocks, if statements and loop statements like do and while. It also supports user defined and built-in (see the list below) functions and procedures.

Supported data-format

ParNCL currently only supports NetCDF files output from either the: CAM Eulerian Spectral grid, CAM Finite Volume grid or CAM Spectral Element (HOMME) grid. We intend to support other types of climate model output in future releases.

Built-in Functions

The list of built-in functions support by ParNCL is given below. Please refer to the  NCAR website for more information on these built-in functions.

Supported NCL Functions

The table below gives a summary of the built-in functions supported by ParNCL for each grid type.

Eulerian Spectral GridFinite Volume GridHOMME Grid
addfileYES YESYESOpens a climate data, *.nc, file
addfilesYES YESYESCreates a reference that spans multiple climate data files
dim_avg_nYES YESYESComputes the average of a variable's given dimension(s) at all other dimensions
dim_max_nYES YESYESFinds the maximum of a variable's given dimensions at all other dimensions
dim_min_nYES YESYESFinds the minimum of a variable's given dimensions at all other dimensions
dimsizesYES YESYESReturns the dimension sizes of the input variable
maxYES YESYESComputes the maximum value of a multi-dimensional array
minYES YESYESComputes the minimum value of a multi-dimensional array
printYES YESYESPrint a variable
printVarSummaryYES YESYESPrint summary of a variable
systemfuncYES YESYESExecutes a shell command and returns the output
uv2vrAYES YESYESComputes vorticity from the wind components
uv2dvAYES YESYESComputes divergence from the wind components
absYES YESYES
acosYES YESYES
asinYES YESYES
atanYES YESYES
atan2YES YESYES
cosYES YESYES
expYES YESYES
fabsYES YESYES
floorYES YESYES
logYES YESYES
log10YES YESYES
sinYES YESYES
sinhYES YESYES
ceilYES YESYES
VisualizationYES YESYES All NCL graphics routines should work with ParNCL.
These routines are executed serially by ParNCL

Installing ParNCL

You can download ParNCL binaries here.

Information on how to build and install ParNCL from source is available here.

Running an NCL script

Information on how to run an NCL script using ParNCL is available here.

Problems or Questions?

Send email to Parvis-users at lists.mcs.anl.gov