Calibration of Rainfall-Runoff Model by Referring to Hydrological Separation of Runoff Components using Chemical and Isotopic Characteristics of Discharge

Calibration of Rainfall-Runoff Model by Referring to Hydrological Separation of Runoff Components using Chemical and Isotopic Characteristics of Discharge Hidetaka Chikamori Graduate School of Environmental Science, Okayama University, JAPAN A rainfall-runoff model is generally calibrated by minimizing error in calculated runoff using records of hydrological components, that is, observed rainfall, discharge and observed or estimated evapotranpiration. However, calibration using only hydrological components sometimes produces a model with strange structure that does not reflect physical properties of an objective basin. It is probably due to error in referred hydrological records. In this study, the author calibrated a rainfall runoff model using not only hydrological record but also chemical and isotopic data of discharge so as to obtain a reasonably structured model from multiple viewpoints. Besides, the model structure was improved in order to simulate isotopic characteristics well. It is well known that ratio of surface flow in total flow can be estimated by change in concentration of cation or anion. Relative concentration of 18O, δ18O is well used for separating runoff of retained water in soil as "old water" from total runoff. A Long-and-Short Term Tank Model (LST2 Model) was applied to three Hinoki Cypress catchments in Mie experimental basin located in the middle of Japan. One of these catchments is of well-maintained planted forest, and two are of poor-maintained planted. A model was calibrated by Differential Evolution for each catchment using hydrological data, concentration of K+ and δ18O. In these catchments, Gomi et al (2008) showed that concentration of K+ well expresses ratio of surface runoff to total runoff, and that δ18O subsurface runoff to total runoff. The results show that an original version of LST2 Model cannot simulated delayed subsurface flow ratio estimated by δ18O, although it well simulates surface flow ratio estimated by concentration of K+. For overcoming this weakness, a LST2 Model was modified by installing a water retention layer storing soil water to each tank, and the modified LST2 model performed well on simulating runoff so that not only surface flow ratio but also subsurface flow ratio were well identified with estimated by chemical and isotopic