Environmental Engineering Reference
In-Depth Information
Forecasting Groundwater Level Using Hybrid
Modelling Technique
Sumant Kumar and Surjeet Singh
Introduction
In India, groundwater serves about 70 % of rural population, 50 % of urban
population and about 60 % of agricultural area. There are more than 20 million
groundwater extraction structures in place which are being used to meet require-
ment for domestic, industrial and agricultural activities. There is intensive
development of groundwater in certain pockets of India, which has resulted in
over-exploitation of groundwater resources and led to steep declining trend in
levels of groundwater. As per the assessment of groundwater resources (CGWB
2007
), out of 5,723 assessment units (blocks/mandals/taluks) in the country,
839 units in various states have been categorized as “over-exploited” meaning
that annual groundwater extraction exceeds the annual replenishable resource.
In addition, 226 units are critical with stage of groundwater development hover-
ing between 90 % and 100 % of annual replenishable resource.
In recent years, the artificial neural network (ANN) technique has been found
successfully applied to solve various water resource problems including time-series
forecasting (ASCE
2000
; Sudheer et al. 2002; Yoon et al.
2007
; Maier and Dandy
2000
). Nayak et al. (
2006
) and Krishna et al. (
2008
) successfully predicted the
GWL fluctuation in coastal aquifers using ANN models with meteorological infor-
mation with GWL data as the input variables. Daliakopoulos et al. (
2005
) simulated
groundwater level using optimum ANN structure and provided acceptable predic-
tion upto 18 months ahead. Coulibaly et al. (
2001
) and Coppola et al. (
2005
) have
also applied ANN model in groundwater sector to predict water table fluctuation.
Jalalkamali and Jalalkamali (
2011
) used hybrid model of ANN and genetic algorithm
Search WWH ::
Custom Search