Artificial neural network for river modelling: History
Subjects: Remote Sensing
Contributors:
Juan C. Mejuto

                 

 

  • Artificial Neural Networks
  • Prediction
  • River modelling

G. Astray[1], J.C. Mejuto[1]

Climate change has impacts on water resources which depends on rainfall and on climatic variables[2]. Precipitation is one of the most important weather variables that influence in the hydrological process[3]. In recent years, there were peaks in floods and droughts processes[4]. The reduction of droughts impacts is one of the challenges for agriculture due to their economic and environmental impacts[5]. In the same way, the mitigation of floods can be interesting due to the possible personal and material damages that this phenomenon can generate. To reduce the adverse effects of drought, and floods,arenecessary a suitable and effective monitoring system[5]. Nevertheless, the hydrologic response of watersheds presents a complex correlation withclimatic, edaphic and geomorphologic characteristics[2]. The typical rainfall-runoff model used to monitor the hydrologic response of watersheds can be classified in three different categories[6][7][8]: i) deterministic models which use physical laws to describe the rainfall-runoffprocess, ii) conceptual models that use simplified representations of the hydrological process and iii) parametric models which use mathematical transfer functions[7].

In the past decades,Artificial neural networks (ANNs) have proven to be a good tool to characterize, model and predict many processes related to the hydrological processes[2]. According to Dawson and Wilby[7], ANNs can be classified as parametric models where the rainfall-runoff process can be considered as a “black box” which does not need a previous detailed knowledge of the hydrological processes characteristics[7]. Due to this great advantage, ANNs have been successfully applied in a huge kind of fields such asi) in Engineering to model, simulate and control of wind and photovoltaic energy systems[9], ii) in Food Science and Technology to model interesting biological properties of food such as the antioxidant capacity of cruciferous sprouts[10] or to characterization and authentication of foods[11][12][13] ,  iii) in Viticulture and Enology to classify Spanish red wines using enological and mineral content parameters[14], iv) in Aerobiology predicting the pollen content in the atmosphere[15][16][17] or v) in Chemistry estimating physicochemical properties [18][19][20][21], inter alia.

 

Figure 1: Conceptual scheme to develop an artificial neural network model.

Artificial neural networks are widely used, and in Hydrology, ANNs can be used in various Hydrological inner fields. A good example of neural network application in Hydrology is, as stated earlier, to predict the River´s flow. This is an important topic for the flow regulation and water resources management and is related to different aspects such as flood or drought forecasting/prevention[22]. A good example of this is the research developed by Islam (2010) to study and predict the water level of Dhaka city using five different stations and obtaining a high accuracy with R2 values between 0.97 and 0.70[23]. In the same direction in our laboratory, we have developed different models to predict the average and maximum daily flow, one or two days ahead, of a river in a small forest headwaters with R2 upper to 0.85 that confirming the capacity of developed neural model[2]. In addition to this, our laboratory developed different neural models to predict the discharge of a mountain river, one, two and three days ahead[24]. In this study, time series analysis coupled to ANNs provided useful information to analyze the hydrologic behaviourof the river and allowed to extrapolate the model to a period morethan 12 years later[24].  Many flood events are related toheavy precipitation so that it’s also important to predict these events. In this sense, Junaida et al.[25] develop neural models coupled with an input variable selection method to find the most significant input variables (skip the trial and error procedure) to determine the best input selection and predict heavy precipitation events[25].

Another example of neural networks application is the study developed by Coppola et al.[26] about the aquifer water level elevations in a semi-confined glacial sand and gravel aquifer 30 d into the future[26]. The neural models developed by Coppola et al.[26] can predict with accuracy the potentiometric surface elevations variations under variable pumping and climate conditions. Besidesthis, the application of the developed neural model does not require knowledge of the physical properties and conditions of the aquifer system which it is helpful when these data is not available[26]. Another interesting topic in Hydrology is the suspended sediment concentration modelling, specifically in snow and glacier melt due to their impact in the environment and water resources, in this case, in upstream and downstream regions of Himalaya[27]. According to Joshi et al.[27], ANN models have a superior behaviourcompared tothe conventional sediment rating curve method (SRC) and can conclude that neural models are suitability to simulate and estimate the daily suspended sediment concentration in glacier melt runoff[27]. Besidesthese topics, neural models can be used to predict mean daily river water temperature due to its importance to the river biological communities and ecosystems process[28]. In this sense, DeWeber and Wagner[28] developed a final model which included different input variables (air temperature, landform attributes, etc.) to predict the mean daily water temperature with a good accuracy around 1.9 ºC (in terms of root mean squared error) in a large area of the East of United State of America[28].

As can be seen, the use of artificial neural networks applied to the Hydrology field is highly widespread and provides an advantage over traditional models due to does not need a previous knowledge of the processes.

 

References

  1. Physical Chemistry Department, Faculty of Science, University of Vigo at Ourense, 32004-Ourense, Spain
  2. Gonzalo Astray; J. A. Ferrerio-Lage; P. Araujo; Juan C. Mejuto; J. A. Rodriguez-Suarez; B. Soto; Multilayer perceptron neural network for flow prediction. J. Environ. Monit. 2011, 13, 35-41, 10.1039/c0em00478b.
  3. Yen-Ming Chiang; Fi-John Chang; Integrating hydrometeorological information for rainfall-runoff modelling by artificial neural networks. Hydrological Processes 2009, 23, 1650-1659, 10.1002/hyp.7299.
  4. Ashok K. Mishra; Vijay P. Singh; A review of drought concepts. Journal of Hydrology 2010, 391, 202-216, 10.1016/j.jhydrol.2010.07.012.
  5. Anteneh Belayneh; Jan Adamowski; Drought forecasting using new machine learning methods / Prognozowanie suszy z wykorzystaniem automatycznych samouczących się metod. Journal of Water and Land Development 2013, 18, 3-12, 10.2478/jwld-2013-0001.
  6. Anderson, M.G.; Burt, T.P. Hydrological forecasting; Anderson, M.G.; Burt, T.P, Eds.; Wiley: Chichester, 1985; pp. 1-616.
  7. C. W. Dawson; R. L. Wilby; Hydrological modelling using artificial neural networks. Progress in Physical Geography: Earth and Environment 2001, 25, 80-108, 10.1177/030913330102500104.
  8. Watts, G. Contemporary hydrology: Towards holistic environmental science ; Watts, G, Eds.; Wiley: Chichester, 1997; pp. 151-193.
  9. Kerim Karabacak; Numan Cetin; Artificial neural networks for controlling wind–PV power systems: A review. Renewable and Sustainable Energy Reviews 2014, 29, 804-827, 10.1016/j.rser.2013.08.070.
  10. Adam Bucinski; Henryk Zieliński; Halina Kozłowska; Artificial neural networks for prediction of antioxidant capacity of cruciferous sprouts. Trends in Food Science & Technology 2004, 15, 161-169, 10.1016/j.tifs.2003.09.015.
  11. I. Gonzalez-Fernandez; M. Esteki; J. Simal-Gandara; M.A. Iglesias-Otero; O.A. Moldes; J.C. Mejuto; A critical review on the use of artificial neural networks in olive oil production, characterization and authentication. Critical Reviews in Food Science and Nutrition 2018, 58, 1-14, 10.1080/10408398.2018.1433628.
  12. O. A. Moldes; Juan C. Mejuto; R. Rial-Otero; Jesus Simal-Gandara; A critical review on the applications of artificial neural networks in winemaking technology. Critical Reviews in Food Science and Nutrition 2015, 57, 2896-2908, 10.1080/10408398.2015.1078277.
  13. Gonzalo Astray; J. X. Castillo; J. A. Ferreiro-Lage; J. F. Galvez; Juan C. Mejuto; Artificial neural networks: a promising tool to evaluate the authenticity of wine Redes neuronales: una herramienta prometedora para evaluar la autenticidad del vino. CyTA - Journal of Food 2010, 8, 79-86, 10.1080/19476330903335277.
  14. Isabel M. Moreno; Angel J. Gutiérrez; Carmen Rubio; A. Gustavo González; Dailos González-Weller; Naouel Bencharki; Arturo Hardisson; Consuelo Revert; Classification of Spanish Red Wines Using Artificial Neural Networks with Enological Parameters and Mineral Content. American Journal of Enology and Viticulture 2018, 69, 167-175, 10.5344/ajev.2017.17021.
  15. F. Javier Rodríguez-Rajo; Gonzalo Astray; J.A. Ferreiro-Lage; M.J. Aira; M.V. Jato-Rodriguez; Juan C. Mejuto; Evaluation of atmospheric Poaceae pollen concentration using a neural network applied to a coastal Atlantic climate region. Neural Networks 2010, 23, 419-425, 10.1016/j.neunet.2009.06.006.
  16. G. Astray; M. Fernández-González; F.J. Rodríguez-Rajo; D. López; Juan C. Mejuto; Airborne castanea pollen forecasting model for ecological and allergological implementation. Science of The Total Environment 2016, 548, 110-121, 10.1016/j.scitotenv.2016.01.035.
  17. M. A. Iglesias-Otero; M. Fernández-González; D. Rodríguez-Caride; G. Astray; Juan C. Mejuto; F. J. Rodriguez-Rajo; A model to forecast the risk periods of Plantago pollen allergy by using the ANN methodology. Aerobiologia 2014, 31, 201-211, 10.1007/s10453-014-9357-z.
  18. Gonzalo Astray; Manuel A. Iglesias-Otero; Oscar A. Moldes; Juan C. Mejuto; Predicting Critical Micelle Concentration Values of Non-Ionic Surfactants by Using Artificial Neural Networks. Tenside Surfactants Detergents 2013, 50, 118-124, 10.3139/113.110242.
  19. Gonzalo Astray; Juan F. Gálvez; Juan C. Mejuto; Oscar A. Moldes; Iago Montoya; Esters flash point prediction using artificial neural networks. Journal of Computational Chemistry 2012, 34, 355-359, 10.1002/jcc.23139.
  20. Gonzalo Astray; P. V. Caderno; J. A. Ferreiro-Lage; J. F. Galvez; Juan C. Mejuto; Prediction of Ethene + Oct-1-ene Copolymerization Ideal Conditions Using Artificial Neuron Networks. Journal of Chemical & Engineering Data 2010, 55, 3542-3547, 10.1021/je1001973.
  21. A. Cid; G. Astray; J. A. Manso; J. C. Mejuto; O. A. Moldes; Artificial Intelligence for Electrical Percolation of AOT-based Microemulsions Prediction. Tenside Surfactants Detergents 2011, 48, 477-483, 10.3139/113.110155.
  22. Lekkas, D.F.; Onof, C.; Lee, M.J.; Baltas, E.A.; Application of artificial neural networks for flood forecasting.. Global NEST: the international Journal 2004, 6, 205-211, 10.30955/gnj.000305.
  23. A. S. Islam; Improving flood forecasting in Bangladesh using an artificial neural network. Journal of Hydroinformatics 2010, 12, 351-364, 10.2166/hydro.2009.085.
  24. G. Astray; B. Soto; D. López; M. A. Iglesias; Juan C. Mejuto; Application of transit data analysis and artificial neural network in the prediction of discharge of Lor River, NW Spain. Water Science and Technology 2016, 73, 1756-1767, 10.2166/wst.2016.002.
  25. S. Junaida, H. Hirose; A method to predict heavy precipitation using the Artificial Neural Networks with an application. IEEE - Proccedins of 7th International Conference on Computing and Convergence Technology (ICCCT) 2012, 1, INSPEC Accession Number: 13580022, https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6530417.
  26. Emery A. Coppola; Anthony J. Rana; Mary M. Poulton; Ferenc Szidarovszky; Vincent W. Uhl; A neural network model for predicting aquifer water level elevations. Groundwater 2005, 43, 231-241, 10.1111/j.1745-6584.2005.0003.x.
  27. Rajesh Joshi; Kireet Kumar; Vijay Pal Singh Adhikari; Modelling suspended sediment concentration using artificial neural networks for Gangotri glacier. Hydrological Processes 2015, 30, 1354-1366, 10.1002/hyp.10723.
  28. Jefferson Tyrell Deweber; Tyler Wagner; A regional neural network ensemble for predicting mean daily river water temperature. Journal of Hydrology 2014, 517, 187-200, 10.1016/j.jhydrol.2014.05.035.
More
© Text is available under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/)
Feedback