Study on prediction of snowmelt using energy balance equations and comparing with regression method in the Eastern part of Turkey

Yerdelen C. , ACAR R.

JOURNAL OF SCIENTIFIC & INDUSTRIAL RESEARCH, cilt.64, ss.520-528, 2005 (SCI İndekslerine Giren Dergi) identifier identifier

  • Cilt numarası: 64 Konu: 7
  • Basım Tarihi: 2005
  • Sayfa Sayıları: ss.520-528


This paper presents two snowmelt models: i) Energy balance model (EBM); and ii) Linear regression model (LRM). To decide suitable model for the basin, daily mean flow data of Kirkgoze discharge gauging. station of State Hydraulic Works (DSi) was applied to Karasu-Kirkgoze mountainous basin that has 233.2 km(2) watershed drainage basin and elevation range of 1830-2854 m in the eastern pan of Turkey. EBM was applied during snow melting period, in March-May for 1987-1995. Hourly temperature (T), wind velocity (V), shortwave radiation (Rd), relative humidity (RH) and intensity of rainfall (Y) were used as input parameters. Intervals of constants in EBM that are snow surface conductance snow surface saturated conductance (K-Sat), liquid holding capacity of snow (L-k), fresh snow visible band reflectance (a(vo)) and fresh snow near infrared band reflectance (a(iro)) were determined for each period. The coefficients of correlation (R) between snow melting data calculated by EBM and gauging data were in the range of 0.88-0.98 for each year (R-2=0.77-0.96). Moreover, LRM is established for the only period of 1987 using observed discharges of basin and meteorological variables. The computed coefficient of correlation (R) between regression model including five predictor variables (T, Rd, RH, V & Y) and gauging data was obtained as 0.87 (R-2=0.757). Two models are then compared in terms of coefficient of correlations. EBM was found more representative than LRM to predict snowmelt in eastern part of Turkey due to high coefficient of correlation.