Skip to main content
ARS Home » Pacific West Area » Kimberly, Idaho » Northwest Irrigation and Soils Research » Research » Publications at this Location » Publication #314074
Title: Estimation of furrow irrigation sediment loss using an artificial neural network

Author
item King, Bradley - Brad
item Bjorneberg, David - Dave
item Trout, Thomas
item MATEOS, LUCIANO - Ministry Of Science And Innovation, Csic
item ARAUJO, DANIELLE - Universidade Federal Do Ceara (UFC)
item COSTA, RAIMUNDO - Universidade Federal Do Ceara (UFC)

Submitted to: Journal of Irrigation and Drainage Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/20/2015
Publication Date: 7/8/2015
Publication URL: http://handle.nal.usda.gov/10113/61060
Citation: King, B.A., Bjorneberg, D.L., Trout, T.J., Mateos, L., Araujo, D.F., Costa, R.N. 2015. Estimation of furrow irrigation sediment loss using an artificial neural network. Journal of Irrigation and Drainage Engineering. 142(1). doi: 10.1061/(ASCE)IR.1943-4774.0000932.

Interpretive Summary: The area irrigated by furrow irrigation in the U.S. has been steadily decreasing but still represents about 20% of the total irrigated area in the U.S. Furrow irrigation sediment loss is a major water quality issue and a method for estimating sediment loss is needed to quantify the environmental impacts and estimate effectiveness and economic value of conservation practices. Artificial neural network (NN) modeling was applied to furrow irrigation to predict sediment loss as a function of hydraulic and soil conditions. The resulting model predicted furrow sediment loss with a correlation coefficient of 0.71 and absolute prediction error of less than ±5, ±10, ±20, and ±30 kg per furrow for 35%, 53%, 72% and 85% of the data set values, respectively. The NN model is applicable to predicting sediment loss rates between 1 and 300 kg per furrow for furrow lengths between 30 m and 400 m, slopes between 0.1% and 4%, flow rates between 5 L/min-1 and 75 L/min-1, and silt or sand particle sized fractions between 0.1 and 0.75.

Technical Abstract: The area irrigated by furrow irrigation in the U.S. has been steadily decreasing but still represents about 20% of the total irrigated area in the U.S. Furrow irrigation sediment loss is a major water quality issue and a method for estimating sediment loss is needed to quantify the environmental impacts and estimate effectiveness and economic value of conservation practices. Artificial neural network (NN) modeling was applied to furrow irrigation to predict sediment loss as a function of hydraulic and soil conditions. A data set consisting of 1926 furrow evaluations spanning three continents and a wide range of hydraulic and soil conditions was used to train and test a multilayer perceptron feed forward NN model. The final NN model consisted of 16 inputs, 19 hidden nodes in a single hidden layer and 1 output node. Prediction performance of the NN model was model efficiency (ME) = 0.66 for the training data set and ME = 0.80 for the testing data set. The prediction performance for the complete data set of 1926 furrow evaluations was ME= 0.70 with an absolute sediment loss prediction error of less than ±5, ±10, ±20, and ±30 kg per furrow for 35%, 53%, 72% and 85% of the data set values, respectively. The NN model is applicable to predicting sediment loss rates between 1 and 300 kg per furrow for furrow lengths between 30 m and 400 m, slopes between 0.1% and 4%, flow rates between 5 L/min-1 and 75 L/min-1, and silt or sand particle sized fractions between 0.1 and 0.75.