TITLE:
Southern and Tropical Indian Ocean SST: A Possible Predictor of Winter Monsoon Rainfall over South India
AUTHORS:
Ravi P. Shukla, Shailendra Rai, Avinash C. Pandey
KEYWORDS:
Winter Monsoon Rainfall over South India; Southern/Tropical Indian Ocean; Multivariate/Linear Regression Models
JOURNAL NAME:
Atmospheric and Climate Sciences,
Vol.3 No.4,
August
26,
2013
ABSTRACT:
The complexities in the relationship
between winter monsoon rainfall (WMR) over South India and Sea Surface temperature
(SST) variability in the southern and tropical Indian Ocean (STIO) are evaluated statistically. The data
of the time period of our study (1950-2003) have been divided exactly in two halves
to identify predictors. Correlation analysis is done to see the effect of STIO
SST variability on winter monsoon rainfall index (WMRI) for South India with a
lead-lag of 8 seasons (two years). The significant positive correlation is
found between Southern Indian Ocean (SIO) SST and WMRI in July-August-September
season having a lag of one season. The SST of the SIO, Bay of Bengal and North
Equatorial Indian Ocean are negatively correlated with WMRI at five, six and seven seasons before
the onset of winter monsoon. The maximum positive correlation of 0.61 is found
from the region south of 500 S having a lag of one season and the
negative correlations of 0.60, 0.53 and 0.57 are found with the SST of the
regions SIO, Bay of Bengal and North Equatorial Ocean having lags of five, six
and seven seasons respectively and these correlation coefficients have
confidence level of 99%. Based on the correlation analysis, we defined
Antarctic Circumpolar Current Index A and B (ACCIA (A) & ACCIB (B)), Bay of
Bengal index (BOBI (C)) and North Equatorial Index (NEI (D)) by averageing SST
for the regions having maximum correlation (positive or negative) with WMRI
index. These SST indices are used to predict the WMRI using linear and
multivariate linear regression models. In addition, we also attempted to detect
a dynamic link for the predictability of WMRI using Nino 3.4 index. The
predictive skill of these indices is tested by error analysis and Willmott’s
index.