Pat Scarbrough-Luther, Bryan Coffey
Use of ArcInfo, GRASS, ERDAS and SAS to Create ACSII Files and ArcInfo
Point Coverages For the Integrated Midwest Biomass Resource Analysis Project
ABSTRACT
The objectives of the Integrated Midwest Biomass Resource Analysis project are to characterize
and quantify the potential supply of biomass from energy crops in an 11 state area. Energy
crops are crops such as switch grass or poplar trees grown to produce fuels. The purpose of this
paper is to describe how large amounts of spatial data from various sources and in various
formats were prepared, input, transferred between and manipulated in several software packages
including ArcInfo, GRASS, ERDAS and SAS to create files describing the concurrence and
spatial location of land use and soils. The products were both ASCII files containing attributes
and statistics and ArcInfo point coverages with soil and land use attributes for each 1km
pixel in the 11 states. These files will used for transportation and economic analysis and
modeling of energy crops.
INTRODUCTION
The purpose of this project was to create ASCII files and ArcInfo point coverages to be
used for transportation and economic analysis and modeling for the Midwest Biomass Resource
Analysis project. The Midwest Biomass Resource Analysis project is an effort to assess the
feasibility of growing energy crops for fuel production. Energy crops are short rotation woody
crops such as poplar trees and switch grass which can be grown to produce fuels. In this project,
spatial data from various sources and in various formats were imported into, transferred between
and manipulated in different software packages in order to obtain the desired output files. This
paper addresses the challenges of formatting digital data, the need for and methods of
transferring it between software packages and the various data manipulations necessary to obtain
the desired output files.
The software packages used were ArcInfo, developed by the Environmental Systems
Research Institute, Geographic Resources Analysis System Software (GRASS), developed by the
U.S. Army Corps of Engineers Construction Engineering Research Laboratories, Earth
Resources Data Analysis Software (ERDAS), developed by ERDAS, Inc. and Statistical
Analysis System (SAS), developed by the SAS Institute.
SOURCE DATA
The attributes needed for the analysis were soil map unit ID (MUID), Major Land Resource
Area (MLRA), hydrologic unit codes, county FIPS code, and land cover class. The MUID,
MLRA and FIPS were obtained from State Soil Geographic Data Base (STATSGO) files
produced by the U.S. Department of Agriculture Soil Conservation Service (SCS). The land
cover classes were derived from the Conterminous U.S. Land Characteristics Data Set 1990
Prototype cdrom obtained from EROS Data Center. Data from the U.S. Geological Survey
USGSWAIS internet server provided the hydrologic unit codes.
DATA PREPARATION
The data was imported into GRASS software using Albers Equal Area projection and at a
resolution of 1km. The STATSGO data was obtained in DLG optional format, was input into
GRASS using the v.import command and converted to raster format using the v.to.rast
command. Attributes were attached to the data by extracting the desired attributes from the
text files and formatting them in a table usuing AWK script language and the VI editor. AWK
allows for printing only the desired collums from data files and inserting field separators. VI
facilitates global search and replace functions. The data files are so large and have so many
collums that editing them in a text editor is not desirable. Once the files were formatted
correctly the r.reclass command was used in GRASS with input redirected from the text file
using the following syntax; r.reclass input=rasterfile name output=new file name <table
name. The correct format for a reclass table is record number = attribute number (for
example, 1=159).
The land cover data was acquired in Lambert Azimuthal projection which GRASS does not have
the capability to reproject into Albers Equal Area. Therefore, it was necessary to import the file
into ArcInfo to reproject it. The file was in ASCII format. In order to import it into
ArcInfo, a header had to be added to the top of the file, then the data was imported using
the ArcInfo command ASCIIGRID. The ASCII file was too large to be edited in a text
editor so the header was created with a text editor and then attached to the file using the UNIX
command cat. The format of the header file can be found in the ArcInfo manual under the
ASCIIGRID command reference.
Clipping the individual states from the land cover file was done in ERDAS using the cutter
subroutine. The file was exported from ArcInfo using GRIDERDAS. The states were
clipped using the state boundaries and returned to ArcInfo using ERDASGRID. Later it
was realized that the clipping could have been done in ArcInfo by setting a mask and using
SETNULL.
After reprojection and clipping, the files were exported from ArcInfo using GRIDASCII and
imported into GRASS using r.in.ascii. This required editing the file header to get it into the
correct format for GRASS (which is given in the GRASS manual under the r.in.ascii command
reference). The UNIX tail command was used to cut off the GRID header, and the new header
was attached using the UNIX cat command.
The hydrologic unit files were obtained in ArcInfo polygon format. They were converted
to GRID format in ArcInfo and imported into GRASS by the above method.
COMBINING DATA LAYERS
The output files needed for the economic and transportation analysis and modeling were ASCII
tables with values for each unique occurrence of all the attributes and the area for each
individual state, and ArcInfo point coverages with attributes for each 1km pixel. The data
was entered into GRASS as separate data layers which were combined using r.cross. This
facilitated the transfer of all attributes for each 1km pixel into ArcInfo as a GRID file.
The crossed data layer was output from GRASS as an ASCII file using r.out.ascii, the header
was edited into the format needed by ArcInfo, and the file was imported into ArcInfo
using ASCIIGRID.
Additional attributes, which were a breakdown of the land cover classes by percent of Anderson
code in each land cover class, were added in ArcInfo using JOINITEM.
CREATING POINT COVERAGES
ArcInfo version 6.1.1 does not have the capability of converting a GRID directly to a point
file. The GRID must first be converted to a polygon coverage. In order to preserve each 1km
pixel over such a large area covered by each of the 11 states, the FISHNET option using the
LABEL switch within the GENERATE routine of ArcInfo was used. The &DESCRIBE
command was issued on each of the 11 GRIDS (states) in order to set up the specific variables
for the FISHNET option. This would allow for the output polygon coverage to maintain the
same geographic information as the input GRID. For each FISHNET polygon coverage, the
"Fishnet Origin Coordinate (X,Y)" was set as %grd$xmax%, %grd$ymax%; the "Y-Axis
Coordinate" was set as %grd$xmax%, 0; the "Cell size" was set as 1000, 1000 (one square
kilometer) and the "Number of Rows, Columns" was set as %grd$nrows%, %grd$ncols%. The
results from each iteration created a coverage in which each polygon of the coverage was a
square kilometer and had a label point in the center of those polygons. Each newly created
polygon coverage was brought into ARCEDIT and all LABELS were SELECTed and PUT into
a new coverage. This created a new point coverage in which the spacing of each of the points
is one kilometer. The GRIDPOLY command was issued for each GRID (state) to create a
polygon coverage of the information contained in the GRID. Attributes from these polygon
coverages were transferred to their respective point coverages created from the FISHNET
routine by using the IDENTITY command.
CREATING ASCII TABLES
The original GRID files (before attaching the fishnet) were converted to polygon coverages and
used for generating the ASCII tables. Using these files allowed for smaller size having area as
an item for each unique combination of attributes in each state coverage. The data was output
from tables in ArcInfo using the UNLOAD command in a columnar format and input into
SAS, where statistics were calculated for accuracy assessment and the tables were formatted for
use in economic analysis and modeling.
DATA VALIDATION
LESSONS LEARNED
This project presented many challenges from which many useful techniques were learned,
especially methods of data transfer between software packages. There were also a lot of valuable
lessons learned after the fact which would increase the efficiency of this project. A decision was
made early on in the project planning to use GRASS software. However, all of the data
manipulations could have been done in ArcInfo which would have eliminated transferring
files between software packages. ERDAS would not have had to been used to clip out the state
files as this capability exists in ArcInfo. A more indepth knowledge of the capabilities of GIS
software packages would have made this project considerably faster and more efficient. Version
7.0.2 of ArcInfo software has the capability to convert a GRID file directly to a point
coverage but was not available at the time of this project work.
ACKNOWLEDGEMENTS
We wish to thank William Hargrove, Richard Winterfield and Mary Alice Wood for their
expertise and assistance with the various software packages used in this project, and Xiomara
Carrero-Martinez and Jessica M. Taylor for their extensive data processing work.
REFERENCES
Environmental Systems Research Institute. 1992. ArcInfo Command References, 6.0.
Environmental Systems Research Institute. Redlands, CA.
Environmental Systems Research Institute. 1992. GRID Command References, 6.0.
Environmental Systems Research Institute. Redlands, CA.
ERDAS. 1991. ERDAS Field Guide, Second Edition. ERDAS, Inc. Atlanta, GA.
U.S. Army Corps of Engineers Construction Engineering Research Laboratories. 1993.
GRASS Version 4.1 Users Reference Manual. Open GRASS Foundation. Boston,
MA,
Pat Scarbrough-Luther, GIS Facility Manager
Oak Ridge National Laboratory
P.O. Box 2008
Oak Ridge, TN 37831-6407
Telephone: (615) 241-3960
Fax: (615) 574-4665
E-mail: pz3@ornl.gov
Bryan Coffey, Computer Services Specialist IV
Statistical and Computer Services
Georgia Experiment Station
1109 Experiment Street
Griffin, Georgia 30223-1797
Telephone: (404) 228-7257
Fax: (404) 228-7270
E-mail: BCOFFEY@GAES.GRIFFIN.PEACHNET.EDU