Multivariate Multisite Model MV.MS. Reg for water Demand Forecasting

Abstract

A new multivariate multi site MV.MS.Reg model is developed in thisresearch depended on regression analysis mixed with Auto regressive multisiteMatalas model (AMMM)and used for water demand forecasting .This developedmodel was applied to Kerkuk city as a case study for long term forecasting ofwater demand for different types such as domestic demand,industrial,commercialand public demand.This was done by dividing the city into four sites anddividing the total water demand in each site into three types ofdemand(domestic,industrial with commercial and public demand) .Each type ofwater demand in each site was analyzed by multivariate regression base then thecross correlation between this type of demand for the four sites were included inthe model using multi site Matalas model.Many explanatory variables wereconcluded to be most effective factors affecting different types of demands suchas monthly temperature,monthly evaporation ,number of residential units,number of industrial and commercial units and number of public units whichwere forecasted successfully using Stochastic weather generation (SWG)method.