84 
1978 land use information encoded in MAGI (Brooner and Wolfe, 1974; 
Maryland Department of State Planning, 1978), thereby determining the 
feasibility of Landsat data for future Maryland land cover/land use 
inventories. A preliminary study was completed and documented 
(Roguski, 1980). 
MAGI, implemented in 1974, was developed by DSP to organize and coor 
dinate data for comprehensive planning in accordance with state and 
federal legislation (Maryland Department of State Planning, 1974). It 
now consists of a central statewide computer data base storage, 
retrieval, and processing system for the analysis and display of geo 
graphic grid-referenced data. It is composed of 40 data planes such as 
soils, land use, highways, geology, critical areas, cultural and 
archeological features, wildlife, and housing patterns for the entire 
state in grid cells of 91.8 acres (total > 88,000 cells), and for a few 
select areas in 4.6 acre grid cells. The study area was selected to 
correspond to one of these special MAGI computer data bases created 
from aircraft and supporting land use data at a scale of 1:24,000, or 
4.6 acres per grid cell. The 74,354 acre study area is located in the 
corridor between the cities of Baltimore, Maryland and Washington, DC 
and includes parts of Howard, Prince George's and Anne Arundel Coun 
ties, extending from the city of Columbia and the Rocky Gorge Reservoir 
on the west to the cities of Glen Burnie and Odenton on the east. 
APPROACH 
MAGI System Data Base 
The MAGI System data base for the study area was acquired from DSP and 
installed on ERRSAC's image processing system, the Hewlett Packard 3000 
minicomputer using the Interactive Digital Image Manipulation System 
(IDIMS) and the Geographic Entry System (GES) software packages. The 
MAGI land cover estimates for the eight categories were derived from 
two variables in the MAGI data base. The first variable, primary land 
use, was encoded as a three digit code, where the first two digits 
represent the numerical code for the predominant land use and the last 
digit represents the percent of that land use in the cell, recorded as 
proportions in increments of 20 percent between 0 and 100 percent, but 
typically as either 60, 80 or 100 percent. 
Secondary land use, the other variable, was encoded in the same way in 
20 percent increments between 0 and 100 percent, except that the more 
typical recorded values were 0, 20 or 40 percent. Since it is common 
practice by users of the MAGI System to assume that the primary cover 
class represents 100 percent of a cell, acreage comparisons between 
Landsat and MAGI included acreages computed this way, as well as com 
putations which incorporated recorded proportions. Therefore, total 
acreages for each cover type from the DSP data base were estimated for 
primary land use (P) assumed to be 100 percent, adjusted primary (P') 
for recorded proportions of primary land use, and for the adjusted 
primary and secondary (P' + S') for recorded proportions of both the 
primary and secondary land use. (P' + S') was computed for each land 
cover category by adding the product of the total primary acreage times 
the average percent representation, to the product of the total secon 
dary acreage times the average percent representation per cover type. 
(Total Acreage = Acreage (Average Primary Percent) + Acreage (Average 
Secondary Percent), for each cover type.)