195 
registered, defoliated image is then multiplied by the 
forest/non-forest mask to produce a new "defoliated forest 
image" in which forested areas in the scene depicting 
defoliation conditions have been isolated from other cover 
types. 
Step 3 - Application of the Ratio Vegetation Index 
The assessment of forest disturbance is carried out on the 
"defoliated forest image" created in Step 2. The Ratio 
Vegetation Index is applied to this image by computing the 
ratio of the infrared to red response (MSS Band 7/MSS Band 
5) for each pixel within the image. When the RVI is 
applied to the defoliation image, a new image, the 
"assessment image", is created. In this assessment image, 
low ratio values indicate heavy defoliation, whereas, high 
ratio values indicate healthy forest. Zeros are non 
forest . 
Step 4 - Identification of Defoliation Levels 
The "assessment image" is compared to available ground 
reference data (usually aerial surveys) to determine the 
numerical cut-off points for healthy, moderately defoliated 
and heavily defoliated forests. Figure 1 illustrates the 
defoliation assessment procedure. 
It is important to note that the key requirement associated 
with the successful utilization of this defoliation assess 
ment procedure is the ability to register several different 
images to a common reference base. 
CREATION OF A STATEWIDE DATA 
BASE FOR DEFOLIATION ASSESSMENT 
Pennsylvania's Division of Forest Pest Management is 
legislatively mandated to conduct annual assessments of 
insect-related damage to the forests throughout the State. 
Yearly statistics must be compiled and stored not only for 
planning management alternatives, but also to study trends 
in insect population dynamics. Over the years, a wealth 
of information has been acquired, but its utility is 
limited because: (a) the data exists in various hard copy 
formats (maps, airphotos) which are not conducive to 
computer storage and retrieval, and (b) the non-standardized 
format, coupled with the subjectivity of analysis 
procedures, makes meaningful trend analyses almost 
impossible. By comparison, Landsat offers a standardized 
data source (i.e., MSS) which has been collected for nearly 
a decade. The data is available in a digital format which 
can be processed in a quantitative, repeatable manner, and 
both the original data and derived results can be readily 
stored and retrieved by computer. However, the use of 
Landsat data for defoliation assessments over an area as 
extensive as Pennsylvania presented a unique challenge. 
In order to be effective, large volumes of data must be 
processed, stored, and registered to a common cartographic 
reference base. Therefore, an appropriate system which 
could accommodate digital image processing, storage, and 
retrieval needed to be devised.