Table 19 . --Mean difference and variance of differences between actual and predicted average 

 plot number of trees per diameter class for a specified plot and growth period 



p^^^ Growth Mean Variance of 



period difference difference 



61 1 0.0275 0.0226 



2 .0563 .0595 



3 .0636 .0939 



4 .0474 .0125 



5 .0742 .0168 



6 .1076 .0746 



7 .1368 .1100 



8 .1745 .1451 

 10 .2724 .5277 



62 1 -.0224 .0449 



2 -.0464 .0647 



3 -.0607 .1314 



4 .0174 .0070 



5 .0263 .0057 



6 .0284 .0134 



7 .0300 .0119 



8 .0642 .0664 



9 -.0453 .2520 



71 1 -.0137 .0780 



2 -.0847 .4184 



3 .0564 .0426 



4 .0362 .0283 



5 .0189 .0292 



6 -.0486 .0807 

 8 -.4189 2.0891 



72 1 -.0264 .0991 



2 -.0773 .3569 



3 .0615 .0399 



4 .0409 .0368 



5 .0469 .0196 



6 .0378 .0223 

 8 -.3097 1.7529 



RESULTS OF STUDY 



Analysis of the validation runs demonstrates that the simulator provides reasonable results 

 for runs at least 40 to 50 years in duration (that is, 8 to 10 growth periods). This is 

 fortunate because cutting cycles can be from 5 to 30 years long, and the method for determining 

 optimal diameter distribution requires good estimates of the future stand at the end of the 

 cutting cycle. In addition, solving the optimal conversion strategy problem could require 

 simulation runs as long as two or three cutting cycle lengths. 



The final equations that make up the simulator are found in table 20. Those equations 

 incorporating predicted basal area growth as an independent variable (the mortality and 

 through-growth equations) have had their regression coefficients changed to adjust for the 

 final removal of the log bias correction. A description of the control cards for the simulator 

 can be found in appendix J. 



37 



