Quantitative Assessment of Channel Seepage Using the Artificial Neural Network (ANN) Approach

Shahbaz Khan, Tariq Rana, Dharma Dassanayake

Research output: Book chapter/Published conference paperConference paper

12 Downloads (Pure)

Abstract

Channel seepage has been identified as a significant loss from the irrigation channels from both water quantity and environmental degradation perspectives. Recent studies have indicated that estimates of channel seepage are an essential component in the management of earthen channel systems. Seepage losses from channel or drains must be located and quantified to establish their economic and environmental importance. Because of drought scenarios and environmental concerns, there is much pressure on existing water resources in Australia. Seepage from earthen channels has therefore become an important issue in Australia for several reasons including the loss of an economically valuable resource and of channel assets and accession to groundwater. Artificial Neural Networks (ANNs) have been recently employed for the solution of many hydraulic, hydrologic, and water resources problems but ANNs do not seem to have been applied for analysis of seepage from irrigation channels. In the present study ANNs have been applied to analyse channel seepage in the Murrumbidgee region of New South Wales, Australia. It is predicted by ANN that if channel seepage is not remediated, over 42 GL of water can be lost from 500 kilometres of channel in the Murrumbidgee Irrigation Area (MIA) each year as seepage and 12.5 GL will be lost through evaporation for the measured length of the channels. The traditional seepage estimate methods such as inflow-outflow and ponding tests are only useful in providing bulk estimates of losses from the studied channel reaches. The distributed qualitative methods using EM-31 and local quantitative methods using the Idaho seepage meter were an improvement in the estimation of seepage losses. However, due to varying soil properties and underlying groundwater characteristics it was not possible to effectively determine spatial distribution of channel seepage which is necessary for cost-effective lining of channels.Due to the complexity and the nonlinearity of the seepage phenomenon and impossibility of building linear relationship between seepage and EM data an ANN method was developed to overcome this limitation. This helped spatially map the seepage extent along the supply channels of the MIA and therefore guide the most cost effective investments for reducing seepage losses. Results from this study clearly show that ANNs can be successfully applied to analyse distributed channel seepage by using key input variables since the ANN method is capable of handling nonlinearity due to quick adaptation and parallel computation power. The channel seepage study in the Murrumbidgee Irrigation Area in Australia indicates that most significant seepage (> 20 mm/day per unit area) occurs in less than 32 percent of the surveyed channel length, therefore it is important to initially target channel lining investments to the leakiest parts, 'hotspots' of the channel system. The approach using ANNs has proven its advantages in this paper, it may be the most effective way of ascertaining seepage hotspots. 2299
Original languageEnglish
Title of host publicationLand, Water & Environmental Management
Subtitle of host publicationIntegrated Systems for Sustainability
EditorsLes Oxley, Don Kulasiri
Place of PublicationChristchurch, New Zealand
PublisherModelling and Simulation Society of Australia and New Zealand
Pages2299-2305
Number of pages7
ISBN (Electronic)9780975840047
Publication statusPublished - 2007
EventInternational Congress on Modelling and Simulation (MODSIM) - Christchurch, New Zealand
Duration: 10 Dec 200713 Dec 2007

Conference

ConferenceInternational Congress on Modelling and Simulation (MODSIM)
CountryNew Zealand
CityChristchurch
Period10/12/0713/12/07

Fingerprint Dive into the research topics of 'Quantitative Assessment of Channel Seepage Using the Artificial Neural Network (ANN) Approach'. Together they form a unique fingerprint.

  • Cite this

    Khan, S., Rana, T., & Dassanayake, D. (2007). Quantitative Assessment of Channel Seepage Using the Artificial Neural Network (ANN) Approach. In L. Oxley, & D. Kulasiri (Eds.), Land, Water & Environmental Management: Integrated Systems for Sustainability (pp. 2299-2305). Modelling and Simulation Society of Australia and New Zealand.