FREC 682 Spatial Analysis -- Gory Details

FREC 682 Spatial Analysis

GRASS Data Structures: the Gory Details

To review, GRASS is a set of separate modules operating within a specific environment. The GRASS prompt GRASS4.1> is transparent to UNIX, and merely signifies that the necessary global environment variables have been set to run GRASS modules. All standard UNIX functions are available at the GRASS prompt.

GRASS raster and vector files are actually multiple files in separate subdirectories within your mapset. The entire GRASS database structure begins with the GRASS database /home/strauss/usrb/grass.data. This includes several subdirectory locations. When you start a GRASS session in a location, GRASS creates a mapset subdirectory for you under that location.

One of the big advantages of GRASS is that its data structures are all open, meaning that you can mess around with the data at any level you like. You can edit individual vector feature vertices or raster header data with a text editor. (Just be sure you know what you're doing!)

Your mapset has multiple subdirectories containing component files for raster, vector and site maps. These are created by GRASS as needed. Complete information on GRASS data structures is available in the GRASS Programmer's Guide. The basics are summarized here.

Raster Data Structures

Raster maps have as many as 7 component files in the following subdirectories:

The cell directory contains the raster cell value files. Grass automatically applies run-length encoding, line by line, if appropriate. GRASS version 4.1 handles integer cell values only (single- or multi-byte, by line). GRASS Version 5 (to be released early 1996) is supposed to support reals also.
Reclass maps don't have their own cell files, but merely reference cell values in another raster map.
The cellhd directory contains the header information files necessary to structure the raster data into georeferenced grids (projection, zone, bounding coordinates, cell resolution, number of rows and columns), e.g.:
```
proj:       1
zone:       13
north:      4928000
south:      4914000
east:       609000
west:       590000
cols:       633
rows:       467
e-w resol:  30
n-s resol:  30
format:     0
compressed: 1
```
Supported projections are unreferenced (0), UTM (1), state plane (2) and lat-long (3). The bounding coordinates define the actual map edges, not the centers of the grid cells at the map corners. Although the cellhd file defines the default region for the raster map, it will be displayed and analyzed according to whatever current region settings are defined in the WIND file.
A reclass map's cellhd file simply contains a list of reclassification values for the cell values in the raster map it references. The region information is taken from cellhd file of the map it references.
The cats directory assigns category labels (text or numeric) to cell values. The cats file has the format:
```
# 6 categories
map title
<automatic label format>
<automatic label parameters>
0:no data
1:description of category 1
...
6:description of category 6
```
Category files are created automatically with blank default labels, and can be edited with r.support.
Note that GRASS uses zero to denote missing data, and thus does not distinguish between missing data and true zero values. While this can cause occasional inconveniences, you can generally find simple work-arounds to handle true zeroes.
The colr directory contains the color table used to display the map. The color tables generally specifies starting and ending RGB values for cell value ranges, defining a color ramp or series of color ramps. The colr file has the format:
```
% 0 240
1:255:0:0        80:255:255:0
80:255:255:0     160:0:255:0
160:0:255:0      240:255:255
```
The first line specifies the range of values; the next line specifies a color ramp from red (255,0,0) to yellow (255,255,0) for values between 0 and 80; etc.
If a map doesn't have a color table file, GRASS uses a default yellow-green-cyan-blue-magenta-red color table.
The colr2 directory contains alternate color tables created by other users.
The hist directory contains history information for your maps (metadata) which you can edit with r.support or display with r.info. This is where you store metadata about your map: the data sources, date of acquisition, accuracy standards, etc.
The cell_misc directory contains 1-line files defining the minimum and maximum positive and negative cell values in each raster map.

Vector Data Structures

Vector maps have as many as 6 component files in the following directories:

The dig directory contains the binary arc data (vertices or sequences of vertices) comprising vector features.

The dig_ascii directory contains ASCII equivalents of dig files, created with the v.out.ascii module. Each file contains 14 lines of header information, followed by sets of vertices defining arcs. Header labels occupy columns 1-14; header information starts in column 15. The first line of each arc entry specifies that it is an area edge (A) or line (L) in the first column, and specifies the number of vertices to following beginning in the third column. For example:

ORGANIZATION: USGS-NMD  DLG DATA
DIGIT DATE:   3/6/1977    
DIGIT NAME:   grass
MAP NAME:     RAPID CITY, SD
MAP DATE:     1977     
MAP SCALE:    100000
OTHER INFO:   RC3.RRT01           
ZONE:         13
WEST EDGE:    579501.42
EAST EDGE:    620264.89
SOUTH EDGE:   4872144.15
NORTH EDGE:   4928285.02
MAP THRESH:   100.00
VERTI:
A  2
   4872144.15    580178.06
   4927677.48    579501.42
L  3
   4927677.48    579501.42
   4927950.66    599367.48
   4928063.45    606138.21
....

The dig_att directory contains ASCII files assigning vector ID's or attribute codes to marker points topologically matched to vector features. For example:
```
A     123456.89    6789012.34         5
L     123457.90    6789123.45         6
....
```
The format is fairly strict: the A or L is in column 1; the easting coordinate is right-justified to column 15; the northing coordinate is right-justified to column 29; the feature attribute number (ID) is right-justified to column 39; the (non-printing) newline character is in column 50. (The size of each dig_att file should be a multiple of 50.) Note that the points in the dig_att file are stored as easting-northing, while vertices in the dig_ascii file are stored as northing-easting.
The v.to.rast module converts vector features to raster maps. Each vector feature with an attribute value entry in the dig_att file is rasterized with that value. Features lacking an entry in the dig_att file are not converted.
The dig_cats directory contains files matching feature ID's to categories. These have the same format as raster cats files.
The dig_plus directory contains vector topology files matching feature ID's in the dig_att file with component arcs in the dig file. These are binary (unreadable).
The reg directory contains files of digitizer registration points, which are only used by the v.digit, GRASS's digitizing module.

Sites (Point Feature) Data Structure

Sites maps require only a single file in the site_lists directory. A sites file has two header lines followed by a line for each site specifying easting and northing coordinates and the site name or attribute value, e.g.:

name|archsites
desc|Potential historic and archealogical sites
593493|4914730|1 Signature Rock
591950|4923000|2 No Name
589860|4922000|3 Canyon Station
590400|4922820|4 Spearfish Creek
....

The "|" character is default field separator in a site_lists file.

Basic File Management Modules

The GRASS file management modules g.copy, g.rename, g.remove handle all component files for raster and vector maps simultaneously. It is generally more efficient to use these modules than the UNIX cp, mv or rm commands.

g.mapsets accesses (read-only) others' mapsets.

g.list displays contents of your and other accessed mapsets.

g.access changes permissions on your own mapset.

course syllabus