PREDICTED UPGROWTH. MORTALITY. AND CONVERSION; ACTUAL INGROWTH 



Conversion from blackjack pine to yellow pine is applied at the end of each fourth growth 

 period. Therefore, only simulation runs longer than four growth periods will show the effect 

 of adding predicted conversion upon the resulting predicted diameter distribution. For simu- 

 lation runs (such as the even-aged runs) under five growth periods in duration, the results 

 are identical to those runs using actual conversion and ingrowth. As a result, the even-aged 

 data could not be used to check the predicted conversion models. 



The fifth and sixth columns of table 18 present the mean and variance of differences for 

 the simulator using the final predicted upgrowth, mortality, and conversion models, and actual 

 ingrowth. Results indicate that using predicted conversion does not reduce accuracy and 

 precision by much. 



ALL COMPONENTS PREDICTED 



The final component, ingrowth, was developed using the uneven-aged data alone and, for 

 this reason, is not applicable to the even-aged validation plots. The addition of the ingrowth 

 component reduces both accuracy and precision (see the last two columns of table 18), which 

 is particularly noticeable on the longer runs and on the managed, uneven-aged plots (plots 

 71 and 72) . 



Because an objective was to produce a simulator useful in predicting even-aged stands as 

 well, a substitute for the ingrowth model was devised and then tested on the even-aged valida- 

 tion plots. This consisted of extending the use of the upgrowth and mortality models into the 

 1-, 2-, and 3-inch diameter classes as well. If the number of trees existing below the 1-inch 

 diameter class and the possibility of additional regeneration are both insignificant for the 

 stand, then ingrowth should consist of those trees from the 1-, 2-, and 3-inch classes surviving 

 and upgrowing into the 4-inch or larger diameter classes. Therefore, if the available upgrowth 

 and mortality equations can be extended into these classes, reasonable ingrowth predictions 

 should result. 



The values listed in the last two columns of table 18 for the Taylor Woods plots are the 

 results of applying the proposed technique for two growth periods. On all plots, the technique 

 both increased the mean difference per diameter class and reduced the variance of the differences. 

 The resulting mean differences, however, are still small when compared to values for average 

 actual number of trees per diameter class that ranged from 9.38 (for TW-7) to 77.90 (for TW- 

 10) . I concluded from these findings that the technique is an adequate substitute for an 

 ingrowth model if it is applied to a stand with an average diameter of at least 4 inches. 

 This is necessary because all of the Taylor Woods plots fell in this category. 



Predicting Subplots and Averaging vs. Predicting Plot Averages 



Of those uneven-aged whole stand simulators developed and published, all have used 

 relatively small plots as their data base. Moser (1972) used 1/5-acre plots. Later Moser 

 (1974) used plots ranging in size from 1 to 4 acres. Finally, Ek (1974) used 1/7-acre plots. 

 The apparent assumption is that these small plots have the same structure and dynamics as the 

 total stand. This assumption is probably met in truly all-aged stands; however, if clumping 

 exists (as it does in ponderosa pine), the assumption probably cannot be met. 



To test this assumption, the average plot diameter distributions were predicted and 

 compared to the values previously obtained by averaging the predicted subplots. Table 19 

 presents mean differences and the variance of the differences for each plot and growth period 

 using the method of predicting plot averages. Comparing these values to those found in the 

 last two columns of table 18, I concluded that predicting plot averages generally reduces 

 both accuracy and precision. The loss, however, appears not too severe for most potential 

 uses of a whole stand simulator. The conclusion is fortunate since using the simulator for 

 answering managerial questions v;ould be greatly restricted if it were necessary to first 

 simulate subplots and then average to get stand values. 



36 



