by: Mitchell 1968; Newnham and Smith 1964; Lee 1967; and Amey 1972) that maintain the 

 map coordinates of the trees through time and the stand projection programs such as 

 TRAS (Larson and Goforth 1970) that combine trees from widely diverse stands into 

 common cells. The effect of competition from neighboring trees on the growth of an 

 individual tree is retained through the early stages of projected time by maintaining 

 the identity of the neighboring trees sampled at the same small diagnostic plot within 

 the stand. In addition, variables expressing competitive status are computed from the 

 relative position in the stand table and from the crown description (Appendix I). 



Diagnostic examinations of forest stands are routine in the practice of silvi- 

 culture. This program is designed to use sample data from these routine stand examina- 

 tions as starting values for the prognosis . Two types of stand examinations are 

 accommodated. In the first type, which is appropriate for surveying the regeneration 

 phase of tree development, the sampling unit is an area of land such as a 4-milacre 

 quadrat or a 1/300-acre circular plot. For such samples, the characteristics being 

 modeled would be the species and heights of the dominant trees, a measure of competing 

 vegetation, and little else. In the second type, the individual sample tree is the 

 record unit. The tree characteristics recorded in this type of inventory enphasize the 

 information needed to estimate its future course of development. As stands described 

 initially by the stocked quadrat survey are projected through time by the program, the 

 records are converted to the individual tree type of data for prognosis of subsequent 

 stand development. 



The functions that drive the prognosis are expressions for finite differences — that 

 is, for the periodic rates of change of the various aspects of tree growth. Coeffi- 

 cients in the tree growth functions are estimated from past records of growth. Sources 

 may include growth recorded in management inventories, and in research studies of 

 silvicultural treatments or of insect and disease impacts. However, at the start of 

 the prognosis, these coefficients based on prior analysis of growth are modified if 

 the growth records of the stand being modeled provide sufficient evidence that the 

 growth rates specified by the growth functions are not appropriate. Through this 

 process of "self-caJLibration," the model can accommodate itself to local peculiarities 

 of site quality, genetic character, and tree vigor. In fact, the calibration variable 

 can be interpreted as a measure of local site quality in healthy stands, or as a 

 measure of impact of insect or disease outbreaks on the rate of accretion. 



Growth is a process that amplifies the effects of previous departures from the 

 mean growth level. To incorporate this characteristic of the growth process, special 

 techniques of computation have been developed to retain the effects of the stochastic 

 aspect of the growth process in the prognosis program. 



The prognosis is developed by first estimating the changes to be expected in the 

 tree conformation- -diameter, height, and crown — during the next growth period. Then, 

 the trees -per- acre corresponding to each sample-tree record is reduced for the expected 

 mortality rate appropriate to a tree of its characteristics growing in such an environ- 

 ment. The tree projection process is repeated for successive growth periods and 

 appropriate displays of the stand's development are produced for each period along 

 the way. 



Harvests (i.e., partial cuts, thinnings, or cleaning) can be scheduled at the 

 start of any growth period. Selection of trees to be cut can utilize any of the char- 

 acteristics describing the sample tree and the stand to simulate a variety of silvi- 

 cultural prescriptions. If these silvicultural prescriptions are keyed to the timber 

 classes and management alternatives specified for the Timber RAM Matrix algorithm 

 (Navon 1971), then this growth prognosis model can be used to provide the yield 

 schedules required by RAM for scheduling timber harvests. 



2 



