Location: Soil and Water Management ResearchTitle: Comparison of artificial neural network and empirical models to determine daily reference evapotranspiration
|CHOI, YONGHUN - Korean Rural Development Administration|
|KIM, MINYOUNG - Korean Rural Development Administration|
|JEON, JONGGIL - Korean Rural Development Administration|
|KIM, YOUNGJIN - Korean Rural Development Administration|
|SONG, WEON JUNG - Korean Rural Development Administration|
Submitted to: Journal of Korean Society of Agricultural Engineers
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/27/2018
Publication Date: 11/30/2018
Citation: Choi, Y., Kim, M., O'Shaughnessy, S.A., Jeon, J., Kim, Y., Song, W. 2018. Comparison of artificial neural network and empirical models to determine daily reference evapotranspiration. Journal of Korean Society of Agricultural Engineers. 60(6):43-54. https://doi.org/10.5389/KSAE.2018.60.6.043.
Interpretive Summary: Increasing human population increases the demand for fresh water. Agriculture for food production is the largest user of water and must be more efficient as competition for water grows. Weather data can be used to estimate seasonal crop water needs; however, the purchase, maintenance and management of a weather station is expensive. Therefore, scientists from ARS (Bushland, Texas) and South Korea used two different models of artificial intelligence to calculate crop water use. Crop water use estimates derived from the artificial intelligence and standard methods were similar. This is important because these new models can create new solutions, which may make it possible to reduce cost and increase reliability of crop water use estimates from weather stations.
Technical Abstract: The accurate estimation of reference crop evapotranspiration (ETo) is essential in irrigation water management to assess the time-dependent status of crop water use and irrigation scheduling. The importance of ETo has resulted in many direct and indirect methods to approximate its value and include pan evaporation, meteorological-based estimations, lysimetry, soil moisture depletion, and soil water balance equations. As one of the data-driven models, the artificial neural networks (ANNs) were intensively implemented for process-based hydrologic modeling due to its superior performance of nonlinear modeling, pattern recognition, and classification. This study adapted two well-known network algorithms in ANNs, Backpropagation neural network (BPNN) and Generalized regression neural network (GRNN), to evaluate their capability to accurately predict ETo using daily meteorological data. All data were obtained from two automatic weather stations, Sangju (2002-2017) and Jangsu (1988-2017) regions, respectively. The daily reference evapotranspiration was calculated using the Penman-Monteith equation as a reference, and all values of ETo and its corresponding meteorological data were separated into three datasets, training, validation and test set. The performance of ANN was evaluated against the multiple linear regression (MLR) and the overall results showed that BPNN performed best followed by MLR and GRNN in a statistical sense.