Streamflow estimation in ungauged basins using watershed classification and regionalization techniques

  • PDF / 2,471,644 Bytes
  • 18 Pages / 595.276 x 790.866 pts Page_size
  • 110 Downloads / 203 Views

DOWNLOAD

REPORT


 Indian Academy of Sciences (0123456789().,-volV)(0123456789( ).,-volV)

StreamCow estimation in ungauged basins using watershed classiBcation and regionalization techniques GANVIR KANISHKA and T I ELDHO* Department of Civil Engineering, Indian Institute of Technology Bombay, Powai, Mumbai 400 076, India. *Corresponding author. e-mail: [email protected] MS received 18 January 2020; revised 22 April 2020; accepted 2 June 2020

Classifying watersheds prior to regionalization improves streamCow predictions in ungauged basin. Present study aims to assess the ability of combining watershed classiBcation using dimensionality reduction techniques with regionalization methods for reliable streamCow prediction using soil and water assessment tool (SWAT). Isomap and principal component analysis (PCA) are applied to watershed attributes of 30 watersheds from Godavari river basin in India to classify them. The best classiBcation technique is determined by calculating similarity index (SI). The results showed that Isomap is better at classifying hydrologically similar watersheds than PCA with an average SI value of 0.448. The regionalization methods such as global mean, inverse distance weighted (IDW) and physical similarity were applied to transfer the parameters from watersheds of best watershed classiBcation group to the pseudoungauged watersheds, using SWAT model. The present study suggests that classifying watersheds with Isomap and regionalization using physical similarity improves the eDciency of streamCow estimation in ungauged basins. Keywords. Ungauged basins; regionalization; watershed classiBcation; Isomap; similarity index; SWAT model.

1. Introduction The assessment of water resources in a watershed is very important for appropriate planning and management, decision making for policy makers, water allocation for agricultural, industrial and domestic sectors, design of bridges, dams, etc., and disaster management. For this, continuous streamCow records for the watersheds are essential. Although hydrologists across the globe have achieved success to a great extent in modelling the rainfall-runoA relationship, estimation of streamCow in ungauged watersheds still remains a crucial problem. The major problem is unavailability of streamCow series data for calibration and validation of the rainfallrunoA models. Estimation of model parameters for

an ungauged watershed from a donor gauged watershed is called as regionalization (Bl€ oschl and Sivapalan 1995). Regionalization methods are widely used for the estimation of streamCow in ungauged basins. The past decade saw a paradigm shift in regionalization methods under ‘Decade on Predictions in Ungauged Basins (PUB): 2003–2012’ established by the International Association of Hydrological Sciences. The advances made in the Beld of regionalization suggests that performance of regionalization methods is different in different regions and there is no universal regionalization method that is applicable to all the regions (Oudin et al. 2008; Samuel et al. 2011). Many of the