A Framework For Streamflow Prediction In The World'S Most Severely Data-Limited Regions: Test Of Applicability And Performance In A Poorly-Gauged Region Of China

Keywords

A priori parameter estimates; Hydrological modeling; Multi-objective calibration; Soft data; Streamflow prediction; Ungauged basins

Abstract

A framework methodology is proposed for streamflow prediction in poorly-gauged rivers located within large-scale regions of sparse hydrometeorologic observation. A multi-criteria model evaluation is developed to select models that balance runoff efficiency with selection of accurate parameter values. Sparse observed data are supplemented by uncertain or low-resolution information, incorporated as 'soft’ data, to estimate parameter values a priori. Model performance is tested in two catchments within a data-poor region of southwestern China, and results are compared to models selected using alternative calibration methods. While all models perform consistently with respect to runoff efficiency (NSE range of 0.67–0.78), models selected using the proposed multi-objective method may incorporate more representative parameter values than those selected by traditional calibration. Notably, parameter values estimated by the proposed method resonate with direct estimates of catchment subsurface storage capacity (parameter residuals of 20 and 61 mm for maximum soil moisture capacity (Cmax), and 0.91 and 0.48 for soil moisture distribution shape factor (B); where a parameter residual is equal to the centroid of a soft parameter value minus the calibrated parameter value). A model more traditionally calibrated to observed data only (single-objective model) estimates a much lower soil moisture capacity (residuals of Cmax = 475 and 518 mm and B = 1.24 and 0.7). A constrained single-objective model also underestimates maximum soil moisture capacity relative to a priori estimates (residuals of Cmax = 246 and 289 mm). The proposed method may allow managers to more confidently transfer calibrated models to ungauged catchments for streamflow predictions, even in the world's most data-limited regions.

Publication Date

2-1-2018

Publication Title

Journal of Hydrology

Volume

557

Number of Pages

41-54

Document Type

Article

Personal Identifier

scopus

DOI Link

https://doi.org/10.1016/j.jhydrol.2017.12.019

Socpus ID

85037677467 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/85037677467

This document is currently not available here.

Share

COinS