Real-Time Forecasting of Drought Discharge with Long-and Short-Term Runoff Model

Abstract

River discharges with lead time of up to several months during drought years were made with the Long-and Short-Term Runoff Model (LST Model). Fourteen years of hydrologic data from 1975 to 1988 at Eigenji Dam (132km²) were used.
A simple method for modifying daily areal rainfall and evapotranspiration was introduced into the operation of the LST Model instead of the application of a Kalman filter, and the model states were adjusted to be consistent with measured streamflow. The climatological time series from a specific date during drought for a given return period were drawn statistically on the basis of several 10-day cumulated rainfall amounts. Monthly evapotranspiration and simplified daily rainfall patterns which consist of a sequence of rainy days with average intensity and that of no-rain days were assumed. Several 10-day cumulated runoff volumes during drought for a given return period were obtained using those climatological time series and compared with observed volumes.
The results were summarized as follows: The effects of modification of daily data on runoff forecasting is limited within a month at most. The forecasted river discharge volumes for 1-9 ten-days ahead periods in drought years for 5-, 7-and 15-years return periods agree with the observed cumulated runoff volumes.