tdftohdf [ parameter=value ... ] [ inputfile outputfile ] tdftohdf [ parameter=value ... ] [ inputfile ... directory ] tdfnc [ parameter=value ... ] [ inputfile outputfile ] tdfnc [ parameter=value ... ] [ inputfile ... directory ]
Parameters for tdftohdf are: include_vars, include_history.
Parameters for tdfnc are: include_vars, apply_scaling, double_scale, scale_all_vars, include_history.
tdftohdf and tdfnc export TeraScan dataset variables to HDF and netCDF-formatted datasets, respectively. HDF (Hierarchical Data Format) was developed by the National Center for Supercomputing Applications at the University of Illinois at Urbana-Champaign. netCDF (network Common Data Format) was developed by Unidata Programs at the University Corporation for Atmospheric Research. Both of these dataset formats are public domain analogs to the TeraScan dataset format (see datasets). Each has independently produced a library of functions, and in some cases utility programs, to read, write, and manipulate multi-dimension, multi-type, multi-variable datasets. HDF has several interfaces built around 8- or 24-bit images, scientific data, tabulated data, etc. The translation performed by tdftohdf exports TeraScan datasets to the Scientific Data Set (SDS) format of HDF.
Input datasets can be any valid TeraScan dataset. One output dataset is created for each input dataset. The input and output dataset variables have the same data representation (e.g., short, int). Variable units, valid max, valid min, bad value, scale, and offset are all exported to the output dataset. Also exported are dimension names and global and variable attributes.
tdftohdf only:
Variables with units of std_date and std_time are converted to integer yyyymmdd values and float (or integer) hhmmss[.ss] values, respectively (see dates, timestr, formats).
tdfnc only:
tdfnc supports options that transform variable data during export. This typically results in a change in datatype. If double_scale=yes, all scaled output has datatype double (64-bit real). If double_scale=no, all scaled output has datatype float (32-bit real), except if the input datatype is double (64-bit real).
Specifies the list of variables to export. If the list is preceded by a minus sign, all variables except those listed will be exported. Wildcards * and ? are allowed. The default is to export all variables.
If apply_scaling=no, which is the default, all variables are exported exactly as input. If apply_scaling=yes, builtin variable attributes scale and offset are applied during export, e.g., output = scale factor * input value + scale offset. Scaled output is either 32-bit floating point, or 64-bit floating point, depending on the double_scale parameter.
Unless scale_all_vars=yes, output scaling is applied only to variables with non-zero scale offset, and/or non-unity scale factor.
tdfnc only. When apply_scaling=no, double_scale=yes forces all variable scale and offset attributes to be exported as 64-bit floating point. Otherwise, scale and offset attributes for 8-bit, 16-bit, and 32-bit variables are exported as 32-bit floating point. double_scale=yes is the default when apply_scaling=no.
When apply_scaling=yes, double_scale=yes forces all scaled variable data to be exported as 64-bit floating point. If double_scale=no, input 8-bit, 16-bit, and 32-bit scaled variable data is exported as 32-bit floating point. This is the default when apply_scaling=no.
tdfnc only. Relevant only if apply_scaling=yes. If scale_all_vars=yes, all exported variables are converted to 32-bit or 64-bit floating point (as specified by double_scale), even if variable scale is 1, and variable offset is 0. The default is no.
Indicates whether or not to export the TeraScan history dataset attribute. This attribute contains some non-printable characters, and may be very, very long. The default is no. Using include_history=yes is not recommended if exported data is to re-imported into TeraScan.
expbin, expasc, hdftotdf, impasc, impbin, printvar, datasets, dates, timestr, formats
HDF does not support dimensions and variables with the same name. If a dimension and a variable have the same name, tdftohdf appends an underscore to the dimension name before exporting to HDF. For example, TOVS datasets usually have a dimension and a corresponding variable named wind_levels. If these are exported to HDF, the dimension name is changed to wind_levels_ prior to export.
Both tdftohdf and tdfnc create additional global attributes for encoding TDF attribute units, and TDF dimension information. If attribute center_lat has units std_latitude, then tdftohdf or tdfnc creates an additional global attribute Xcenter_latYunitsZ = std_latitude.
The delimiter characters X, Y, and Z are controlled by hidden parameters fake_left, fake_mid, and fake_right respectively. The defaults for these parameters are fake_left=NUL, fake_mid=\, and fake_right=NUL, where NUL is the empty character. Using these defaults in the above example, the additional global attribute for encoding units for center_lat is center_lat\units.
For every TDF dimension, the following additional global attributes may be defined: XdimnameYcoordZ, XdimnameYscaleZ, XdimnameYoffsetZ, and XdimnameYvariableZ. Using the default values for delimiter characters X, Y, and Z, these additional global attributes look like dimname\coord, dimname\scale, dimname\offset, and dimname\variable.
The HDF source code and documentation, which also includes the netCDF library interface and documentation, may be obtained via public domain anonymous FTP to NCSA:
1) ftp ftp.ncsa.uiuc.edu or ftp 128.174.20.50
2) log in using "anonymous" for the name
3) enter your local login name for the password
Documentation that may be particularly helpful: About_3.3r3 Getting Started with HDF HDF Reference Manual NCSA HDF Calling Interfaces and Utilities
Addresses: National Center for Supercomputing Applications 152 Computing Applications Bldg. 605 E. Springfield Ave. Champaign, IL 61820
Unidata Program Center P.O. Box 3000 Boulder, CO 80307-3000:
Last Update: $Date: 2002/05/07 23:22:58 $