Premium
Periodic‐stochastic modeling of separation of precipitation into rainfall and snowfall
Author(s) -
Yevjevich Vujica,
Harmancioğlu Nilgün Bayraktar
Publication year - 1990
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/wr026i010p02613
Subject(s) - snow , precipitation , environmental science , surface runoff , quantitative precipitation forecast , climatology , hydrology (agriculture) , meteorology , geology , geography , ecology , geotechnical engineering , biology
Modelers must separate precipitation into rainfall and snowfall because the two processes have significant differences as inputs to runoff. The purpose of the present study is to model this separation, provided that available data permit the coding of observed precipitation as rainfall, snowfall, and/or mixed rainfall‐snowfall. Modeling of precipitation is realized by a simple approach such as the periodic‐stochastic Markov chain description of occurrence and nonoccurrence of time interval precipitation. The exponential probability distribution based functions, with periodic parameters, are used for modeling the nonzero values of precipitation. The generated series are then separated into rainfall and snowfall components by means of the rainfall‐precipitation ratio. This ratio is modeled as a periodic‐stochastic process in the range zero to one. The bounded (0, 1)‐beta distribution function is selected as the appropriate distribution function to describe the ratio when it can also assume values between zero and unity. Most parameters in modeling precipitation and the ratio show periodicity, often fitted well by one‐ or two‐harmonic Fourier functions. Two sets of daily precipitation series, each ten years long, are used to demonstrate the results of modeling daily precipitation and daily rainfall‐precipitation ratio: the Besanson precipitation gauging station in the Durance River basin, France (as a point process), and the W‐3 watershed area, near Danville, Vermont (as a daily average areal precipitation process). The investigation shows the feasibility of modeling the separation when observed precipitation is coded for rainfall, snowfall, and/or mixed rainfall‐snowfall.