The Multisite Rainfall Downscaling (MRD) is a tool that conditionally simulates the rainfall occurrence and amount by using the modified Markov model-kernel density estimate (MMM-KDE). The conditioning variables include previous day rainfall/state, aggregated wetness state over a period of time and atmospheric variables. By ignoring atmospheric variables, model reduces to standard multisite rainfall simulator. The package provides a stochastic modeling framework for multisite downscaling of daily rainfall incorporating low-frequency variability in the simulations. This framework simulates the daily rainfall occurrences and amounts as a two stage process at each individual location. The rainfall occurrence is modeled with a Markov chain conditional on the rainfall occurrence of previous days, wetness state over a predefined past aggregated days and atmospheric variables. The rainfall amounts on the wet days are simulated using a kernel-density estimation procedure conditional on the previous day’s rainfall and atmospheric variables.

The spatial dependence across stations is simulated by making use of spatially correlated random numbers. Spatial correlation in the random numbers is introduced on the basis of the at-site observed cross correlations in rainfall occurrences and amounts. The MMM- KDE modelling framework is presented here in the form of an interactive tool and is named as Multisite Rainfall Downscaling (MRD).


Mehrotra R., Sharma A. (2010) Development and Application of a Multisite Rainfall Stochastic Downscaling Framework for Climate Change Impact Assessment. Water Resources Research. VOL. 46, W07526, 17 PP.,  doi:10.1029/2009WR008423.

Mehrotra, R., A. Sharma, D. Nagesh Kumar and T. Reshmidevi (2013) Assessing future rainfall projections using multiple GCMs and a multi-site stochastic downscaling model, Journal of Hydrology, 488, Pages 84-100.