3. Data analysis and computing procedure
Monsoon rainfall over India during the months of
June to September (122 days) exhibits interesting
oscillations over the country. In the present analysis,
the rainfall over the country has been
expressed as a linear combination of orthogonal
functions. This technique was suggested by
Lorenz (1956) and later used by Kutzbach
(1967), and Weare (1977) to evaluate the principal
component of sea-level pressure over the
Northern Hemisphere and sea-surface temperature
over the Atlantic Ocean. Also, the daily
OLR data sets (1979–1988) for the monsoon period
(June to September) have been used at 2.5
longitude=latitude intervals with 68 stations in
India. The location of 68 stations has been chosen
on the basis that they have represented the
true network of rainfall distributions over India.
Interestingly, the set of 68 stations selected for
the rainfall exhibited equally promising results
alongwith a higher correlation with all India seasonal
rainfall (Singh, 1994).
Let P be a (n m) matrix of monsoon rainfall
over m stations and a series of n years. Here,
n ¼ 10 years and m ¼ 68 stations.
The element (Prs) of P represents departure of
rainfall from their mean value for the s-th station
and r-th year.
Let P represents the time and space variability
of rainfall. It can be defined that
P ¼ Q F; ð1Þ
where the matrix Q represents the time variation
and F represents the space variation of monsoon
rainfall. The element of the P matrix is
given by
Prsðx; y; tÞ ¼ Xm
k¼1
qrkðtÞ fksðx; yÞ; ð2Þ
where the element qrk (t) represents the time and
fks (x, y) represents the space respectively. The
matrix F is an orthonormal matrix, hence the
transpose and product of this matrix should be
represented as a unique identity matrix. It is
defined that
F F0 ¼ I; ð3Þ
where F0 is the transpose of F and I is the identity
matrix. It is further stated that F and Q matrix
derive from the matrix P after defining the matrix
S, where
P0
P ¼ S:
This matrix S is a square matrix and P0 is the
transpose of the matrix P, hence from the above
equation, it is concluded that
F0
S F ¼ Q Q0 ¼ D; ð4Þ
where D is a diagonal matrix. The columns of F
are the eigenvectors of S, while the element of D
is the eigenvalue of S. Every element of D is
a measure of the percentage variance explained
by the corresponding eigenvector. F and D
were calculated from S by Jacobi’s method
(Greenstadt, 1960). This is an iterative process,
which utilizes successive rotations of each element
of s.