Sampling Design 



The revised inventory procedure is similar to its predecessor in that it uses 

 photo interpretation of a large number of photo points to stratify the forest acreage 

 into classes.^ However, whereas former inventories stressed volume strata, the strata 

 used here emphasize conditions that indicate certain opportunities for timber manage- 

 ment. IVhen the acreages within the strata are compiled compartment by compartment, 

 the distribution of timber management opportunities over the entire forest, or for 

 variously defined smaller planning units, can be compiled. 



Our objectives for this inventory system emphasize the importance of examining a 

 recognizable stand in relation to the surrounding conditions of the forest lands and 

 their uses. This emphasis is realized by defining the ground sampling unit to be a 

 subcompartment--the smallest permanently defined record unit suitable for long-range 

 land management planning, including a comprehensive road system. For the Coeur d'Alene 

 National Forest, a subcompartment encompasses an area of about 1,200 acres defined by 

 topographic, drainage, and political boundaries. The list of 604 subcompartments and 

 their acreages that make up the Coeur d'Alene National Forest provides a logical 

 sampling framework for not only timber management planning inventories, but for many 

 other kinds of inventories as well. Watershed conditions, recreational use and 

 developmental opportunities, range condition, and many other data needs that would 

 benefit from coordination with knowledge of timber conditions could use this same 

 overall sampling framework. 



Within the selected sample subcompartments, stands are mapped from the aerial 

 photographs and examined on the ground using a systematic grid of plots. Each of these 

 plots is established and data recorded exactly as they have been under the former 

 design. The location which was sampled by a 10-point cluster under the former design, 

 is directly analogous to the stand under the revised procedure; hence, this revised 

 procedure should retain all the capabilities of the former design. 



SELECTION OF SAMPLE SUBCOMPARTMENTS 



The subcompartments to be included in the field sample were selected with proba- 

 bility proportional to their total National Forest land area. In other words, sub- 

 compartments were arrayed in order of size and systematically selected in such a 

 manner that every acre on the forest had an equal chance to be included in the sample. 

 The specific details of this procedure for drawing a randomized sample have been 

 described by Stage (1971). 



Because we had no past experience with sampling variation between subcompartments, 

 we were "in the dark" as to the intensity of field sampling that would be needed to 

 provide sufficient accuracy for the many kinds of analysis to which the data would be 

 subjected. So four possible sample sizes were selected. The first sample included 

 21 subcompartments, distributed throughout the forest. The succeedingly larger sample 

 sizes included the next smaller size as a subset. Thus, the sample could include 21, 

 28, 34, or 40 subcompartments. By selecting four possible sample sizes, we could termi- 

 nate the inventory at four possible levels depending on the rate of production or the 

 variability encountered. Even with the smallest sample, the volume estimates should 

 meet the standards that have been specified in the past for forest -wide totals. 

 Whether the accuracy is appropriate for the decisions to which these data relate is a 

 question that has yet to he satisfactorily analyzed. 



^If facilities for easily storing^ modifying , and compiling mapped information were 

 available, this stratification could be based on stand mapping on the photos with 

 several ensuing advantages. 



5 



