Modeling rainfall-runoff process using artificial neural network with emphasis on parameter sensitivity
- PDF / 2,560,004 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 49 Downloads / 169 Views
ORIGINAL ARTICLE
Modeling rainfall‑runoff process using artificial neural network with emphasis on parameter sensitivity Vikas Kumar Vidyarthi1 · Ashu Jain2 · Shikha Chourasiya3 Received: 8 March 2020 / Accepted: 26 May 2020 © Springer Nature Switzerland AG 2020
Abstract The gradient descent (GD) and Levenberg–Marquardt (LM) algorithms are commonly adopted methods for training artificial neural network (ANN) models for modeling various earth system and environmental processes. The performance of these algorithms is sensitive to the initialization of their parameters. The initialization of the algorithm’s parameters for modeling different physical processes also varies process to process. However, there is a minority that tried to verify the sensitivity of the parameters of the algorithm than the sensitivity of the input data to the model. This work investigates the sensitivity of the popular ANN training algorithms to initial weights while modeling one of the earth system processes, i.e., the rainfallrunoff (RR) process. A novel methodology consisting of basic statistics for assessment of sensitivity of ANN parameters is proposed. The rainfall and flow data derived from three contrasting catchments are employed to establish the conclusions of this study. The results indicate that the RR model trained by LM algorithm is more robust in achieving performance with less variance irrespective of the existence of randomness in initialization of parameters than that of the GD trained models. Keywords Rainfall-runoff · Artificial neural network · Gradient descent · Levenberg–Marquardt · Hydrology
Introduction The runoff is one of the most desired hydrologic variables for design, operation, and maintenance of various water resources projects. Hydrologists and researchers proposed several methods to forecast runoff accurately in the past by developing various rainfall-runoff (RR) models. These methods are broadly classified into two categories: (i) conceptual models that consider the physics of the underlying process, and (ii) data-driven models that learn and behave according to the information present in the data without considering the physics of the system (Chen and Adams 2006). The rainfall-runoff process is highly nonlinear and extremely complex as they are interrelated to various subprocesses (involved in hydrologic cycle) which are still not
* Vikas Kumar Vidyarthi [email protected] 1
SRM IST NCR Campus, Modinagar, UP 201204, India
2
Department of Civil Engineering, Indian Institute of Technology Kanpur, Kanpur, UP 208016, India
3
Water Engineering and Management, Central University of Jharkhand, Ranchi, India
understood clearly (Zhang and Govindaraju 2000), and hence, conceptual models are not always suitable for modeling the rainfall-runoff process. In addition to this, many conceptual rainfall-runoff models need a large number of data for calibration and validation which make them computationally extensive and so, not become very popular (Lu et al. 2012; Grayson et al. 1992). The artificia
Data Loading...
