Time Series Analysis, Forecasting and Control
Time Series Analysis, Forecasting and Control
| Hi-index | 0.00 |
The paper deals with the development of an adaptive model that is applicable to real-time forecasting of hydrologic processes. The rainfall-runoff process is considered here. In this model the discharge was modeled as autoregressive with past discharges and a moving average representation on the precipitation. The model makes use of the Constrained Linear Systems (CLS) technique to split the precipitation into two rainfall inputs by using a threshold based on an antecedent precipitation index. This technique can be thought of as a piecewise linearization of a nonlinear process. The real-time forecasting model is a time invariant linear state model where the state variables, discharge and rainfall, are estimated by the Kalman filtering algorithm and the unknown model parameters by using the instrumental variables approach. This technique was applied in a case study using data from the Ombrone River Basin, Italy, and was implemented on a small desk-top computer.