units was felt» So far an area of nearly 14.5 million ha. covering
about 2 0 % of the entire forest area in different zones has been covered
under regional inventories. The results have indicated useful potential
for paper and pulp, saw milling, plywood and other wood based
industries. One of the interesting facts revealed is that the total
growing stock has been very much underestimated in the past small
area inventories.
"Role of Aerial Photographs"
Aerial photographs were mainly used for forest type delinea
tion, preparation of base maps and inventory maps, area determination
of forest types and location of ground sample plots in the field. By
restricting the inventory field, work in forested areas alone saved the
cost and time by about 30% o Location of ground sample plots was
found to be easier and much more reliable than conventional topographic
line maps. The experience was varied in different types of forests on
different scales. On smaller scales and in dense forests with poor
identifiable features the plot centres were located by bearings and
distance measurement for short traverses. In most of the cases the
plot centre could be located by the image of the surroundings. Worst
cases were in tropical wet Evergreen forests where plot location was
mainly done by traversing.
"Sampling Designs"
In the initial stages, three different types of sampling designs
were used in three different regions, in the Southern zone consisting of
tropical wet evergreen forests, systematic strip plot sampling was used.
In the central zone consisting of tropical moist and dry deciduous forests,
a systematic cluster sampling with two plots in each cluster was used.
In the Northern coniferous forest zone simple random point sampling was
adopted. The usual method has been to systematically lay out sample
plots on grid intersections (Latitude and Longitude) which provided the
centre of the first plot of cluster. The intensity of the grids varied
in different regions from 5'x5‘ to 1 1' x 11' depending on the varia-
4
bility and precision fixed. In general a precision of _+ 10% was
prescribed'.
Investigations into the optimum sampling design for different
forest types were undertaken as exercises at IPI. Designs based on
simple random sampling, line plot sampling, stratified random sampling
with fixed plot size and variable plot sampling were tried. In the coni
ferous forest zone of the Himalayas, stratification in volume classes was
done and volume per unit area measured by variable plot sampling.
Due to topographic consideration this method was found useful. In
tropical forest areas line plot and stratified plot sampling were tried
and in areas where stratification into different forest types or volume
classes could be done this gave useful results, with fixed plot size.
TIWARI ( 9 ) conducted investigations into fixed plot and variable plot
sampling in Sal forests of Dehra Dun and no significant differences
were found for basal area and volume estimates but tree count estimates
were poor with variable plot sampling. However, in the majority of the
tropical deciduous type of forest where forests are irregular and trees