ABSTRACT
This chapter deals with modeling of hydrological variables of diverse complexities by employing the potential of hybrid modeling schemes involving MEMD. MEMD is used for decomposition of variables, while tools like stepwise linear regression (SLR) or data driven methods are used for modeling. MEMD allows accounting for different causal input variables in the hydrologic predictions or simulations. The significant inputs at each process scales are identified and estimates of the variables are performed for respective scales. Final aggregation of estimates from different scales helps to get the values of predictor variable. The prediction of seasonal rainfall of Kerala subdivision, Standardized Precipitation Index (SPI) of three subdivisions, monthly inflows into Hirakud reservoir, daily suspended sediment flux of Pattazhy hydrologic station, etc. are modeled using different MEMD-based hybrid schemes. The statistical performance evaluations of results showed that MEMD-based hybrid schemes are superior in performance when compared with different standalone models.
