The tenth IMSC, Beijing, China, 2007 - International Meetings on ...

K. Mohankumar
Department of Atmospheric Sciences, Cochin University of Science and Technology
This study aims to connect different meteorological and oceanographic parameters over
the Arabian Sea and adjacent seas with the summer monsoon rainfall over Peninsular India.
Correlation between Rainfall and several global parameters were found and those areas
having 99% significance were selected for making indices. To check the consistency of the
relationship between these candidate predictors and the predictant, a 15-year sliding window
correlation (significant at 5% level) was done. Nine indices having high correlation were
selected. Indices are relative humidity at 925hPa (Mar and May), sea surface temperature
(Dec), zonal wind at 925 hPa and 850 hPa (Mar), air temperature at 925 hPa (Mar and May),
surface sensible heat flux (Mar) and meridional wind at 700 hPa (Jan) over different spatial
locations. PCA is done using these nine indices, for the period 1975-2003. Seven components
that explain the maximum variance were retained for further analysis. A multiple linear
regression model was developed using these components for the period 1975-1997. <strong>The</strong>
model has a multiple correlation of 0.84 and coefficient of determination of 71.2 %. <strong>The</strong> VIF
analysis of all the parameters that is retained in the model have values less than 2 indicating
an insignificant level of multicollinearity and has a Durbin Watson value of 1.856. <strong>The</strong> model
performance was assessed for a period 1998-2003. <strong>The</strong> model has a RMSE of 8.6 %, BIAS of
0% and ABSE of 7.2% of LPA. For the test period this becomes 23.68%, 18.13% and 4.86%
respectively. Climatological predictions were also made and found that RMSE, BIAS and
ABSE are 8.7%, 7.07% and 7.57% of LPA respectively.
Statistical downscaling of a specific weather pattern for rainfall forecast using techniques of
artificial intelligence
María Cleof Valverde Ramírez
Centro de Previsão de Tempo e Estudos Climáticos (CPTEC), Instituto Nacional de Pesquisas
Espaciais (INPE)
valverde@cptec.inpe.br
A statistical downscaling procedure derived from artificial neural network (ANN) and fuzzy
logic (FL) were development to generate quantitative forecasts of daily rainfall to specific
weather patterns. Weather pattern methods involve linking observational station data to given
weather classification schemes. In this case the classification is subjective and a specific
weather pattern namely SACZ-ULCV pattern was selected. This weather pattern is the
interaction between the upper level cyclonic vortices (ULCV) in the vicinity of Northeast Brazil
and South Atlantic Convergence Zone (SACZ). This interaction is associated with severe
rainfall and thunderstorms over Southeastern of Brazil. <strong>The</strong> SACZ-ULCV pattern is well
simulated by regional Eta model of Center for Weather Forecasts and Climate Studies
(CPTEC) in Brazil, for this reason the outputs of this model to period from December, January
and February of 2000-2003 were used to generate statistical downscaling. <strong>The</strong> predictors
were the dynamical meteorological variables of Eta model related to SACZ-ULCV and surface
142