Conceptual Models for Non-stationarity

SimHYD (Australia), Xinanjiang (China) & HBV (Sweden)

Hong Li 1 , Stein Beldring 2 & Chong-Yu Xu 1

1. Department of Geosciences, University of Oslo, Norway 2. Norwegian Water Resources and Energy Directorate , Norway

hong.li@geo.uio.no http://folk.uio.no/hongli Towards Improved Projections, IAHS-IAPSO-IASPEI Assembly, Gothenburg, Sweden, July 2013

The Wimmera catchment in Australia Coordinates: Lat: -36.98; Lon: 142.79

Catchment area: 2000 km² Country: Australia

Q: 02 Jan 1965 - 31 Aug 2009 (whole)

P, T and PE: 01 Jan 1960 - 31 Aug 2009 P1: 01/01/1966-31/12/1973 P2: 01/01/1974-31/12/1981 P3: 01/01/1982-31/12/1989 P4: 01/01/1990-31/12/1997 P5: 01/01/1998-31/12/2005

1997 Q

2

H

Short summary of the three models

Model Input Data Cal. Pars.

Country SimHYD P, PE 7 Australia XAJ P, PE 15 China HBV P, T 12 Sweden

All the models are for discharge simulation at daily time step.

XAJ: Xinanjiang P: Precipitation PE: Potential evaporation T: Temperature Cal. Pars.: Calibrated paramters 3

SimHYD

Input: P, PE Cal. Pars.: 7 Chiew, 2005 4

XAJ

Input: P, PE Cal. Pars.: 15 Jiang, 2007 Zhao, 1992

Distribution of tension water capacity

=0 5

HBV

Input: P, T Cal. Pars.:12

Temperature

Solomatine, 2011 6

Results of Nash-Sutcliffe efficiency (NSE) NSE>0.6: reasonable NSE>0.8: good

Best

Blue: the calibration in each period Red: the validation in the complete period Black: the calibration in the complete period 7

Results of Bias (Sim/Obs) 0.9-1.1: reasonable 0.95-1.05: good Blue: the calibration in each period Red: the validation in the complete period Black: the calibration in the complete period 8

What will happen if in flood frequency design?

Model (NSE): max Q

9

Conclusions: … XAJ model is the best, and limited.

New measuring technology is needed Model structures need to be updated or modified

KEEP BUSY!!!

10