THE PROBLEM 



Although several studies have tried to determine which of the many possible photo 

 schemes result in strata most closely resembling those used in ground surveys, few seem to 

 question the value of the original ground strata in volume or management estimates. Instead, 

 they concentrate on ways of rating photo-classification schemes by photo-field differences, and 

 on methods of adjusting for these differences . 



At least three studies have broken with tradition by testing the usefulness of photo classi- 

 fications in the light of their improvement of volume estimates. Bickford (3) reported that cubic - 

 volume classes were more efficient than stand-size classes in estimating cubic volume on plot 

 sampling surveys. Kendall and Sayn- Wittgenstein (4) reported that a combination of cover type, 

 stand height, and cover density gave best results when areas were delineated (mapped) rather 

 than point sampled. Macpherson (5) reported two-way stratification by cover type and volume 

 classes about equal to three-way stratification using cover type, volume, and maturity group. 

 However, it is quite significant that the gains shown by these studies were made without the use 

 of precise photo measurements and aerial photo volume tables . 



When examining this problem of which stratifications should be used on aerial photos, it 

 is well to remember the limitations of the photo interpreter as well as his strong points . In 

 general, he studies the forest, not the trees. For example: The photo interpreter is limited in 

 the number of direct measurements he can hope to make . He can measure only what he can see 

 on his aerial photos. Direct plot measurements therefore are limited to total height, crown 

 diameter, and crown coverage of the visible trees. On medium-scale photos, he often finds it 

 difficult to measure individual trees; instead, his measurements are plot averages. His con- 

 ception of stocking must be based on crown coverage of all visible trees; he cannot directly 

 recognize dead or cull trees. He can, however, record measurements for two or more height 

 classes, and the coverage of each in many two-storied stands. Because topographic differences 

 are accentuated on aerial photos, he can often recognize broad soil and site differences and can 

 measure slope and aspect with considerable accuracy. In addition to these reasonably objective 

 measurements, the photo interpreter may be able to deduce from photographic tonal differences 

 many changes in species composition or stand condition due to past history of fire, logging, or 

 insect devastation. These varied photo measurements and interpretations have been combined 

 into numerous classification schemes. Those based strictly upon direct photo measurement are 

 quite objective. Those which rely largely on tonal differences and image shape or appearance 

 are almost wholly subjective. 



In management inventories, as in timber estimates, sampling by strata can mean a better 

 estimate at less cost. The amount of gain depends on the type of strata used. These must be 

 defined so that each stratum is fairly homogeneous and so there are real differences between 

 strata. And to afford adequate sampling in each, the total number must be kept quite low. 

 Bickford (2) states: "For timber cruising aerial photos provide the most practical basis for de- 

 fining classes needed for stratified sampling." He also emphasizes that the most effective 

 classes are those closely related to the data being estimated. When volume is being estimated, 

 the best stratification schemes will be those that are most effective in reducing volume 

 variation within classes. 



The stratification problem is further complicated when considering combined management 

 and volume inventory surveys . The photo schemes most valuable for estimating gross cubic 



Underlined numbers in parentheses refer to Literature Cited. 



2 



