Abstract—This study proposes an innovative approach for
statistical downscaling of rainfall based on scaling property of
meteorological variables. The reanalysis data of five dominant
meteorological variables mean sea level pressure, relative
humidity, surface temperature, wind velocity (zonal and
meridional components) extracted from National Centre for
Environmental Prediction (NCEP) are used as predictors to
project monthly rainfall of Kerala meteorological subdivision in
India. The multiscale decomposition of predictor dataset of the
region and the monthly rainfall of a specific grid point is
performed simultaneously by employing the Multivariate
Empirical Mode Decomposition (MEMD) technique. The
individual modes are predicted by fitting stepwise linear
regression (SLR) by considering the potential predictors based
on p-value statistics. Subsequent addition of the predicted
modes gives the monthly rainfall. The method is demonstrated
by a specific grid point of Chalakkudi river basin in Kerala,
India. The method is found to be superior over the linear
regression and M5 model tree based transfer function
approaches. Further, the MEMD-SLR hybrid model is used for
rainfall projections of the state of Kerala under three
representative concentration pathway scenarios (RCP2.6,
RCP4.5 and RCP8.5) provided by Canadian Centre for Climate
Modeling and Analysis (CCCMa).
Index Terms—Downscaling, Kerala, rainfall, MEMD.
S. Adarsh is with TKM College of Engineering Kollam 691005, India (e-mail: adarsh_lce@ yahoo.co.in).
Cite: S. Adarsh, "Finer Scale Rainfall Projections for Kerala Meteorological Subdivision, India Based on Multivariate Empirical Mode Decomposition," International Journal of Environmental Science and Development vol. 7, no. 12, pp. 896-901, 2016.