ients. If these correlations had not been significant, the 
spectral stratification band could have been combined with 
one or more of the other bands to produce a hybrid band for 
the stratification. 
2. The estimated cost of constructing the digital data 
bank was $.56 per acre. Initially, this is a high cost, 
but many of these costs can be amortized over the life of 
the data bank because the only data in the bank, as it is 
now configured, that will degrade with time is the Landsat 
spectral and classified data and the Park boundary data. 
In addition, the per acre cost would be considerably lower 
for large park units. 
3. The biggest difficulty encountered in implementing 
this inventory system was establishing the ground plots 
accurately. In this project, the UTM coordinates of the 
cells that were randomly selected from each spectral strat 
um were plotted on an orthophoto map sheet, this location 
was transferred to 1:20,000 color photography, and the an 
notated photos were used to navigate to the point in the 
field. This procedure was very time consuming and subject 
to considerable error in areas of dense vegetative cover. 
Other sampling approaches that may prove useful are the use 
of transect samples or of large-scale sample photography. 
4. Finally, we think that the approach used in this 
Park can be applied to fuel inventories in other parks with 
minor modification due to the existing vegetation complex 
and to the availability of various data types. By con 
structing a digital data bank with Landsat spectral and 
classified data and with digital terrain data, a stratifi 
cation can be designed which accounts for most of the range 
of conditions present in the gradients of interest to a 
wildland fuel manager. 
62