Developing an AI-based method for river discharge forecasting using satellite signals
- 9 Downloads
Researchers have calibrated satellite signals successfully using novel artificial intelligence (AI) methods to estimate discharge at ungauged river sites accurately. However, common AI methods including neural networks have a recognized defect in time series forecasting known as input imitation error. The present study addresses this significant source of error by combining evolutionary polynomial regression (EPR) with the Nondominated Sorting Genetic Algorithm (NSGA-II) for multiobjective optimization. This new method of forecasting signal time series is called the evolutionary polynomial regression-time series predictor (EPR-T). EPR-T can simultaneously minimize the model prediction error based on traditional performance indices as well as a new index, peak similarity (PS), to prevent the model from imitating its input variables when forecasting. The prediction accuracy of the new EPR-T and traditional AI methods is compared for six case studies, namely the Connecticut, Missouri, Pee Dee, Red, White, and Willamette rivers. The results demonstrate the considerably superior accuracy of EPR-T over the regular EPR method.
The authors would like to thank the US Geographical Survey (USGS) Water Resources department for providing in situ data online (https://waterdata.usgs.gov/nwis/inventory?search_criteria=search_site_no&submitted_form=introduction) and the Global Disaster Alert and Coordination System (GDACS) for providing satellite information online (http://www.gdacs.org/flooddetection/searchareas.aspx).
- Cohen S, Brakenridge GR, Ke_ner AJ, Syvitski JP, Fekete BM, De Groeve T (2012) Calibration of orbital microwave measurements of river discharge using a global hydrology model. In: Proceedings AGU Chapman Conference on Remote Sensing of the Terrestrial Water Cycle, Kona, Hawaii, USA, pp 19–22Google Scholar
- Darras T, Kong-A-Siou L, Vayssade B, Johannet A, Pistre S (2017) Karst flash flood forecasting using recurrent and non-recurrent artificial neural network models: the case of the Lez Basin (southern France). In: Renard P, Bertrand C (eds) EuroKarst 2016, Neuchâtel: advances in the hydrogeology of karst and carbonate reservoirs. Springer International Publishing, Cham, pp 169–177CrossRefGoogle Scholar
- Khan SI, Hong Y, Wang J, Yilmaz KK, Gourley JJ, Adler RF, Brakenridge GR, Policelli F, Habib S, Irwin D (2011) Satellite remote sensing and hydrologic modeling for flood inundation mapping in Lake Victoria basin: implications for hydrologic prediction in ungauged basins. IEEE Trans Geosci Remote Sens 49:85–95CrossRefGoogle Scholar
- Kugler Z, De Groeve T (2007) The global flood detection system, vol 45. Office for Official Publications of the European Communities, LuxembourgGoogle Scholar
- Schmugge TJ (1980) Microwave approaches in hydrology. Photogramm Eng Remote Sens 46:495–507Google Scholar
- Sharifipour M, Bonakdari H, Zaji AH, Shamshirband S (2015) Numerical investigation of flow field and flowmeter accuracy in open-channel junctions. Eng Appl Comput Fluid Mech 9:280–290Google Scholar
- Wang J, ONeill P, Engman E (1982) Remote measurements of soil moisture by microwave radiometers at BARC test site II 83954. NASA, USAGoogle Scholar
- Zaji AH, Bonakdari H (2015b) Efficient methods for prediction of velocity fields in open channel junctions based on the artificial neural network. Eng Appl Comput Fluid Mech 9:220–232Google Scholar