Analyzing the low flow trends in northwest Indiana using neural network

This research focuses on the low flow regime of Lake Michigan watershed in the northwest Indiana, located in the southern tip of Lake Michigan. Low flow is mainly contributed through groundwater flow. Ecosystem survival depends on the low flow regime (Sala et al., 2000). Changes in landuse pattern, climate change, river training works and many other factors influences the low flow. This region had encountered lot of changes in the landuse pattern during the last 50 years. Runoff from five watersheds was used in this work. Analysis such as single station trend, regional trend, 7Q10 flow, flow duration curve and seasonal trends were conducted systematically to understand the behavior of low flow regime. Historic data such as daily flow, monthly mean temperature, monthly rainfall and land use changes were considered. To study the influence of different factors, artificial neural network based approach was used. Using relative strength effect (RSE) associated with each input neuron, the influencing factors were examined. This study indicates an increasing trend in the low flow regime for this region.