Introduction

Linux CLI Utility to convert .dss file formats into .csv/.asc format and back.

It has 2 tools:

dss2csv

Converts dss timeseries into CSV files
Converts dss grid into ASCII files

csv2dss

Converts CSV files into dss timeseries
Converts ASCII files into dss grid

Development Notes

It’s really hacky, and only does basic things, but since DSS files are binary files that we have no other way to read except for HEC built softwares like HEC-DSSVue, and the export from HEC-DSSVue doesn’t even have CSV format, and except export fails if tried on too many series at once. I have been trying to make it for more than a year, and have finally succeeded in a prototype.

I made it as all libraries I can find depends on the C library heclib, and don’t compile at all in linux machine I have. And all libraries seem to only include the compiled binaries from the C library that are either not compatible, or have some problems in them. Even replacing the c files from the file downloaded from HEC-DSS Programmers Guide for C didn’t help.

And while trying everything I found https://github.com/HydrologicEngineeringCenter/hec-dss/, trying to use the heclib.a compiled form that also didn’t work. But the libheclib.so worked, but some functions were not available in the .so. Turns out there is a separate file to generate .so that looked like it was made for java applications they have (like HEC-DSSVue) in mind, so I remade a shared library from the actual function definitions in all the source files. So I’ve made this library, that can be compiled with the shared library and will need the shared libraries to run.

I’ve included a PKGBUILD file to install it in archlinux (just do makepkg -si to install), it’ll compile and copy the binary, shared library and the bash completion script to corresponding directories. There is also Makefile if you want to compile it and install/run it some other way. You only need gcc (c compiler).

You can also compile your own shared libraries from hec-dss linked above and use that if the current one doesn’t work. It might work with windows/mac as well that way but I haven’t tested it.

Install

use the make command to compile the c code using the Makefile. It’ll generate dss2csv and csv2dss binaries in src/.

Or for Arch Linux, use makepkg -si command to build and install the binaries, shared library and bash completion.

Usage

Help menu dss2csv:

Usage: dss2csv command dss_file.dss [rng]

Commands:
    help[h]       : print this help menu.
    list[l]       : list the available paths.
    timeseries[t] : extract the timeseries for paths.
    grid[g]       : extract the grid for paths.

    using grid command on timeseries file or vice versa will fail.
Arguments:
    dss_file.dss  : dss file to operate on.
    rng           : Range of the chosen timeseries use in format M-N
                    where M is start and N is the end number (inclusive)
                    Omitting the start or end will default in available
                    start or the end. (e.g. 1-5 or -5 or 5-)
                    [Optional: Defaults to all available]

Help menu csv2dss:

Usage: csv2dss command dss_file.dss input_files...

Commands:
    help[h]       : print this help menu.
    timeseries[t] : extract the timeseries for paths.
    grid[g]       : extract the grid for paths.

    using grid command on timeseries file or vice versa will fail.
Arguments:
    dss_file.dss  : dss file to operate on.
    input_files...: Input files to convert to dss format
                  : filename should have fields A-F separated by _
                    e.g. A_B_C_D_E_F.csv will become /A/B/C/D/E/F/

General

Filenames for input/output .csv/.asc files will/should be PATH in the DSS but separated by _ instead of /. e.g. A_B_C_D_E_F.csv will become /A/B/C/D/E/F/ and vice versa.
CSV files are simple comma delimited text files.
ASCII grid files are Esri ASCII raster format, and can be read with GIS softwares.

Only the first character of the commands are actually checked, so you only need it. Long name are useful for readability only.

Use timeseries[t] command for timeseries file and grid[g] for grid file, using the incompatible command will just make it exit without doing anything. the functions return error code, but I couldn’t figure out how to print what error codes mean, instead of just knowing it’s success or failure. I don’t want to use the enums to compare all different errors and print them myself.

I’ve tried to turn off all the log prints from the libraries with zsetMessageLevel(0, 0);. So if you want more error messages being printed, you can change that and recompile. If I can figure out how to save those logs into a log file I’ll do that, feel free to make any pull requests.

Grid Data

ASCII grid files, along with grids (.asc file) can have .prj projection files and .dssinfo file with extra informations for the dss grid metadata. For now .prj files are ignored, but .dssinfo files are created when .dss files are converted to .asc, and read when doing the reverse. The .dssinfo file is fragile, use the same order and nothing more as the format made when converting DSS to ASC file. path and number of range field are ignored.

Example .dssinfo file is included in example/ directory.

Batch processing

There is an example script examples/netcdf2dss.sh that shows how you can use gdal tool to reproject and do raster calculations before converting any grid data type (here netCDF is used) to ASCII file type, and then use them in csv2dss tool to generate the DSS file.

Bugs and Pecularities

Timeseries CSV to DSS

The field E should be valid time frequency for the CSV to be compiled into a DSS file. The error message is not useful so keep it in mind. So far I’ve only tried 1Day that works.

DSS Grid to CSV

I also have it ignore projection (just assumes it’s SHG grid systems, I should add options on that). Writing reprojections is something I don’t want to do, so convert it using the definition in GIS software. Make sure the ASCII file is valid and is in the correct location in a GIS software.

The WKT definition of the projection used in DSS format is:

PROJCS["USA_Contiguous_Albers_Equal_Area_Conic_USGS_version",GEOGCS["GCS_North_American_1983",DATUM["D_North_American_1983",SPHEROID["GRS_1980",6378137.0,298.257222101]],PRIMEM["Greenwich",0.0],UNIT["Degree",0.0174532925199433]],PROJECTION["Albers"],PARAMETER["False_Easting",0.0],PARAMETER["False_Northing",0.0],PARAMETER["Central_Meridian",-96.0],PARAMETER["Standard_Parallel_1",29.5],PARAMETER["Standard_Parallel_2",45.5],PARAMETER["Latitude_Of_Origin",23.0],UNIT["Meter",1.0]]

It also ignores all other details for the grid in DSS for now. And while writing the grid, it only calculates the minimum, mean and maximum. Range table and other Attributes are just garbage values, (it crashes if I don’t put something there).

Atreyagaurav/dss2csv