Artificial neural network approach for quantifying climate change and human activities impacts on shallow groundwater level: A case study of Wuqiao in North China Plain

This research attempted to present BP artificial neural network (BP-ANN) approach for studying the effect of climate change (CC) and human activities (HAs) on shallow groundwater level (SGL), taking Wuqiao in North China Plain (NCP) for example. Precipitation (P), irrigation (I) and pumping from both unconfined (UCP) and confined (CP) aquifers were found to be dominant influencing factors for annual variation of SGL in this district, which was drawn from correlation analysis and groundwater budget results in recent five years calculated by Modflow model. A BP-ANN regress model was then trained and tested using historical data from 1990 through 2008 to depict the nonlinear relationship between annual variation of SGL and the four influencing factors. Scenario analysis results indicate (1) SGL will decline by 20cm annually with no CC and HAs changes. (2)Under emission scenarios of A1B, A2 and B1(IPCC, 2001), SGL will decline by 12cm–14cm annually on average with the only consideration of direct influence of rainfall on SGL, and 15cm–18cm while both direct and indirect influences are taken into account. (3)The most effective way of alleviating SGL fall is to reduce irrigation, followed by total pumping water reduction and the increase of water supply from rivers. It will aggravate SGL decline to substitute CP for UCP and to increase non-irrigated water use.