In the early stages of our planning, we considered optimizing the selection of 

 subcompartments for estimating the acreages of particular categories of stands. In 

 effect, this procedure would have resulted in as many independent, optimized inven- 

 tories as there were categories of stands. Such a procedure would be desirable if the 

 stands not of the category for which the subcompartment had been selected could be 

 ignored. 



We rejected this approach because the stand boundaries between categories are not 

 always clearly defined nor would they be drawn in advance of the selection of sub- 

 compartments for examination. Thus, any bias generated by special attention to a 

 category in drawing the boundaries would result in bias in the final estimates of the 

 population totals. 



Consequently, we decided to sample alt stands in each selected subcompartment. 

 By so doing, any shift in stand boundaries would not affect the estimates of the popu- 

 lation totals, although the acreages in particular categories would reflect any net 

 bias in the drawing of their boundaries. 



To illustrate the distinction between these two alternatives, consider a hypo- 

 thetical subcompartment consisting of two stands classified into different strata. If 

 the expansion factor for a unit of land area sampled in one stand is the same as that 

 for a sample in the other stand, then shifting the boundary between stands can only 

 increase the average volume per acre (or of any other character) of one stratum at the 

 expense of the other stratum. Consequently, the net effect on the estimate of the 

 population total is zero. However, if the expansion factors of the two stand categories 

 differed as they would if the sampling intensity were varied so as to optimize the 

 independent inventories, then there would be no guarantee that the shift in boundaries 

 would be compensating in its effect on the population total. 



Moreover, including all stands in the management unit enhances our capability to 

 consider the limitations or opportunities imposed by adjacent stand conditions. 



We would like to emphasize that our reservations about sampling stands using 

 unequal probabilities would disappear if we had the capability to map all stands in all 

 compartments in advance of the sample selection. Forthcoming modifications of this 

 design will, we hope, utilize complete stand mapping, 



PHOTO SAMPLING PROCEDURE 



The aerial photo phase of the inventory is accomplished using resource photography 

 of 1:15,840 scale (4 inches per mile). Photo interpretation assigned stands (or non- 

 stocked areas) within the selected subcompartments and grid points on the remaining 

 area into one of 11 strata. These strata are defined on page 7. Stand boundaries are 

 defined by discontinuities in stand height, stand texture, or crown closure. The stand 

 is subordinate to the stratum, so that stand boundaries do not cross stratum boundaries. 



Within each subcompartment selected for field sampling, all stands are delineated 

 on the photos to a 10-acre minimum size. Their acreages are determined either by 

 planimeter, or by counting the ground plots that fall within the stand boundary. 



On the remaining portions of the forest, an overlay grid is superimposed on alter- 

 nate photos to establish photosample points. The spacing of the grid points is such 

 that each point represents about 100 acres. The interpreter then classifies the 

 stand in which the grid point is located (not just the condition at the point) into 

 one of the 11 strata. Along with the known acreage of the compartment, the proportion 

 of points in a stratum establishes the stratum area. 



6 



