ecological modelling 195 ( 2 0 0 6 ) 307–317 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/ecolmodel The temporal transferability of calibrated parameters of a hydrological model Halit Apaydin * , Alper S. Anli, Ahmet Ozturk University of Ankara, Faculty of Agriculture, Department of Farm Structures and Irrigation, 06110 Diskapi, Ankara, Turkey article info Article history: Received 14 January 2005 Received in revised form 5 October 2005 Accepted 15 November 2005 Published on line 18 January 2006 Keywords: Hydrological model Model calibration Transferability SLURP Aksu basin Model performance Runoff Turkey abstract The SLURP model was tested for the transferability of calibrated model parameters for different time periods in the Aksu Basin in the Black Sea region in Turkey. Two different scenarios were tested. The first scenario used three data sets: 1980–1982 data for calibra- tion, 1983–1984 data for validation, and 1995–2000 data for evaluation. The second scenario used two data sets: 1995–2000 data for calibration and 1980–1984 data for evaluating trans- ferability. Because a decrease in model performance occurred whenever using the calibrated parameters before and after calibration periods, the authors recommend updating calibrated model parameters whenever marked changes occur in the watershed. © 2005 Elsevier B.V. All rights reserved. 1. Introduction Hydrological modelling is an effective technique to assist managers and decision-makers in the assessment and pro- tection of water quantity and quality. Hydrological models realistically represent the watershed’s complex system in which hydrological characteristics result from several factors (e.g., physical, vegetative, climatic, and anthropomorphic). Using modelling techniques to understand runoff genera- tion can lead to increasing water efficiency in the watershed. Models simulate water management strategies that satisfy the demands of different users, while mitigating deleterious effects on riparian ecosystems (Heathcote, 1998; Kite, 1995; Viessman et al., 1989; Romero et al., 2002; Black, 1991). ∗ Corresponding author. Tel.: +90 312 596 1562; fax: +90 312 316 7467. E-mail addresses: apaydin@agri.ankara.edu.tr (H. Apaydin), asanli@agri.ankara.edu.tr (A.S. Anli), ozturk@agri.ankara.edu.tr (A. Ozturk). Studies using hydrologic models usually consist of four phases: (1) collecting raw data of climate, topography, soil and land covers; (2) processing data; (3) calibrating and validating the model using selected periods of data; (4) using the model for evaluating past periods, predicting future periods, or pro- viding decision support for recent events. The professional literature has usually described hydro- logical models of runoff simulation that have been used or evaluated for limited periods. At the end of a particular period, a revised version or a different hydrological model has typ- ically been used. Although some models have been used for longer periods, the need for changing parameters has not been clearly established. Unrealistic results may occur whenever nonupdated parameters are applied to a basin model, one with 0304-3800/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolmodel.2005.11.032