ee Appendix 
DY 
Survey Methods 
The forest survey of Florida used aerial photo- 
graphs as a basis for preliminary estimates of forest 
area and stand-class area. A stratified sample of 
forest plots was located on the photographs and was 
then carefully cruised on the ground. The county 
was the basic work unit. Detailed statistics by survey 
units and counties have been issued in five publica- 
tions. (13; 14,-15;16;-17); 
Area Estimates 
Acreages of forest and other land were estimated 
with the use of a dot grid placed on every third 
contact print along each flight line in each county. 
(In parts of South Florida suitable photography was 
not available so the dot grid was used on Soil Con- 
servation Service maps made up from small-scale 
aerial photographs. In Taylor County lack of both 
photographs and Soil Conservation Service maps 
necessitated the use of a Florida Forest Service type 
map.) The proportion of dots falling on forest land 
when applied to the gross area of the county, as 
reported in 1940 by the Bureau of the Census, yielded 
a preliminary estimate of the acreage of forest and 
other land-use classes in each county. This estimate 
was revised after field checks were made of a sample 
of both forest and nonforest ground plots. All together 
approximately 148,000 of these dots were classified. 
Every third dot classified as forest in the preceding 
step, about 27,500, was further classified into forest 
type, stand class, and density class by careful stereo- 
scopic analysis of a one-acre plot surrounding the dot 
on the photograph. The proportion of plots falling 
in each classification when applied to the forest area 
of the county gave the total area in each classification. 
These totals were revised after a sample of plots 
was checked for proper classification on the ground. 
In estimating the areas of various categories of 
land, there were two possible sources of error: (1) 
errors in classifying the dots and plots or in compiling 
Se | 
the data, and (2) sampling errors. In this survey 
every effort. was made to maintain a high order of - 
accuracy in the collection and compilation. Frequent 
checks were made and a continuous program of 
training was carried out. The sampling intensity 
was sufficient to provide an estimate of the forest 
acreage of the State with a standard error of = 
0.4 percent. This indicates the probabilities are two 
out of three that the actual forest area is within = 
0.4 percent of the value given, exclusive of measure- 
ment and computing errors. 
Volume Estimates = 
Timber cruisers made a detailed on-the-ground 
tally of photo plots in each stand-size class to obtain 
volume, growth, cull, and mortality data and to 
check the accuracy of the photo classification. The 
number of plots tallied varied according. to the 
stand-size classe; i.e., every 3d large saw-timber 
photo plot, every 4th small saw-timber, every 6th— 
pole-timber, every 13th seedling and sapling plot, 
and every 26th poorly stocked or nonstocked plot 
was taken. The total amounted to 3,542 one-fourth 
acre plots, although these were classified by forest 
type and stand-size class on a one-acre basis. An addi- 
tional sample of 916 plots, classified as agricultural 
on the photographs or maps, was taken to provide 
a check on land-use changes since the date of photo- 
graphy. 
In estimating timber volumes, the sources of error 
include (1) errors in classifying field plots and in 
compiling the data, (2) sampling errors, (3) in- 
accurate measurements of tree diameter, height, 
form, and cull, and (4) errors resulting from im- 
proper construction or use of tree volume tables. 
As in the case of area determinations, every effort 
was made to obtain accurate classifications, measure- 
ments, and final statistics through frequent checks 
and training. The volume tables used also were 
checked and found to give reasonably accurate 
results. The standard error of estimate of the board- 
52 Forest Resource Report No. 6, U. S. Department of Agriculture’ 
