Artificial neural network simulation of ground water levels in uplands of a coastal tropical riparian

Abstract

Wetlands play an important role in the ecological balance of the coastal region. Understanding groundwater level behaviour in uplands is important for the management and the development of coastal tropical riparian wetland. Artificial Neural Networks has proved to be robust techniques in modeling and prediction of hydrological processes. This paper presents the application of ANNs to model groundwater levels in uplands around a wetland environment. Weekly hydro meteorological observations have been used as an input to model groundwater fluctuation observed in sevel open wells in the region. A comparison of different training algorithms has also been carried out. The results obtained show that the use of Artificial Neural Networks in modeling the groundwater levels was successful. With Root Mean Square Error values in the range of 0.09 to 0.16, the study also reasserts that the same training algorithm need not provide the best results for different conditions. Key words: Artificial neural networks, wetlands, stepwise linear regression.