The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
1340
Counting the frequency: Counting and marking the number of
sampling units of each group at PGT, which is called frequency
listed in the third column on the left side of table 2, signed with
f(y)-
Accumulating the frequency: A new column omitted in table
2 was correspondingly produced, which was used to store the
data of accumulating frequency when the frequency of each
group was accumulated and then marked at each PGT from top
to bottom of this data array, the data of accumulating frequency
was signed with ^ .
Accumulating^^:
A new data array was calculated and
marked by accumulating the square root of the data of
accumulating frequency at each PGT, which is listed in the
fourth column on the left side of table 2.
Thresholds of layers: The distance datum between two layers
can be calculated using the last accumulating data of the square
root of the data of accumulating frequency which was shown at
the bottom of the third column on the right side of table 2
divided by layer amount such as six. This distance datum which
is 9288.83 in table 2 was used to partition the data array of the
third column on the right side of table 2 and mark at six PGT
listed in the second column on the right side of table 2.
Marking layers: Every sampling unit can be marked with the
sign of layers when the thresholds of layers are confirmed. In
the above table, total sampling units were partitioned to six
regions according to the thresholds of layers, and every region
was marked with numerals as codes of layers. These signs of
layers were listed in the first column on the right side of table 2.
SYSTEMATIC CLUSTERING MEANS (SCM)
where y h = the average value of the collectivity of the
h layer
N = the amount of the collectivity
N h = the amount of the collectivity of the h layer
Y h . = the total crop area of the / unit of the h layer
S 2 - the summation of variable error of the h layer
±y u
4= rZ(T W -
«„“It?
(3)
(4)
where y - the average value of the samples of the h layer
n = the total amount of the samples
n h = the amount of samples of the h layer
y hj = the crop area of the i unit of the h layer
s 2 h = the variable error of the samples of the h layer
So, the least samples amount of stratified sampling is
determined by the follow formula:
(5)
It is reasonable to stratify sampling units using systematic
clustering for reference. The result of clustering is similar to
that processed by FAM although there is only one parameter
used to cluster, that is crop area such as paddy rice of sampling
units. The main steps will be introduced corresponding to the
process of stratified sampling of early rice of 2007 in China.
Where n = the total amount of samples
L = the amount of layers
V = the variable error of estimating value
The others are as same as the above
Preparing data: The data used to cluster need not to be sorted
ascending or descending. It only needs to be transformed into
appropriate format for statistical software such as SPSS.
Clustering: It is easy to cluster when the background data is
imported using statistical software. RSAC also selects the
number of six as the number of clustering in stratified sampling
of early rice of 2007 in China. The clusters of six were used as
layers of six.
3.1.4 Calculating the least samples: The following
equations should be used to calculate the least samples when
the job of stratified sampling has been done (Chen, etc., 2000).
Distribution according to proportion:
Where
The others are as same as the above
(6)
The surveying accuracy signed 5 is set to a certain value such
as 0.95, and the reliability signed (1-a) is also set to 0.95 or
other value, the value of V should be calculated according to
(2) the follow formula:
V = {d! t) 1
(7)