DATA REQUIREMENTS FOR 

 MORTALITY MODELING 



Large quantities of individual tree data must be available to fit mortality models 

 that predict the probability of an individual tree dying within a specified interval. 

 Individual tree characteristics on sample trees must be measured at the beginning of 

 the interval. The status of the tree, that is, whether it is alive or dead, must be 

 recorded at the end of the interval. 



Mortality models from such data are directly applicable only for the interval and 

 geographic area actually measured. We extend model capability by distributing plots 

 over the geographic range of interest and by measuring them over several successive 

 intervals. Variability in mortality rates that is caused by uncontrollable climatic 

 factors may be accounted for in two ways. Plots may be measured at intervals of at 

 least 5 to 10 years in order to include a range of climatic conditions within the 

 measurement interval. Alternatively, plots may be remeasured at shorter intervals. 

 For example, we could measure trees annually for 10 to 20 years and record mortality 

 when it occurs. When this is done, variability in mortality rates caused by climatic 

 variation is accounted for by the fact that a range of climatic conditions is repre- 

 sented in the data set used for modeling. 



The time span required to gather data is often longer than we can afford; thus, 

 we frequently work with available data. To do this, geographic range, distribution of 

 sample trees, and time span must be representative of the target population. When 

 this is not the case, we must recognize and state the limitations of the resulting 

 models . 



Models reported in this paper reflect processes or conditions in the forest at 

 the time the data were collected. False mortality estimates may result when the 

 equations are applied to trees growing under different conditions. For example, 

 during the data gathering period mortality from white pine blister rust (Cronartiton 

 ribicola) and from the mountain pine beetle {Dendvootonus montioolae) were at epi- 

 demic levels. Thus, the mortality model reported in this paper will produce high 

 levels of mortality in western white pine [Pinus montiaola) . Models for other species 

 reported in this paper do not appear to produce abnormal mortality rates for the 

 species involved. However, as the dynamics of the north Idaho forests change, the 

 models will require continual updating. 



Data Used in This Study 



Data were collected over two periods on Pot latch Corporation permanent variable- 

 radius plots, using a 25-factor prism. In the first measurement period (1964-1970), 

 part of the data was measured over a 5-year interval and the remainder was measured 

 over a 3-year interval. In the second measurement period (1968-1974), data were 

 recorded over a 6-year interval. 



3 



