DOI: 10.5593/SGEM2016/B31/S12.110


T.Kozel, M. Stary
Wednesday 7 September 2016 by Libadmin2016

References: 16th International Multidisciplinary Scientific GeoConference SGEM 2016, www.sgem.org, SGEM2016 Conference Proceedings, ISBN 978-619-7105-61-2 / ISSN 1314-2704, June 28 - July 6, 2016, Book3 Vol. 1, 843-850 pp

The main advantage of stochastic forecasting is fan of possible value, which deterministic method of forecasting could not give us. Future development of random process is described much better by stochastic then deterministic forecasting. We can categorize discharge in measurement profile as random process. Contents of article are development of forecasting model for managed large open water reservoir with supply function. Model is based on zone linear autoregressive model, which forecasting values of average monthly flow from linear combination previous values of average monthly flow, autoregressive coefficients and random numbers. All data were sorted to zone with same size (last zone has different size due to residue of data). Computing zone was chosen by last measurement average monthly flow. Matrix of correlation was assembled only from data belonging to matching zone. Autoregressive coefficient was calculated from Yule-Walker equations (Yule, Walker, 1927, 1931). The model was compiled for forecast of 1 to 12 month with backward correlation from 2 to 11 months. Data was got rid of asymmetry with help of Box-Cox rule (Box, Cox, 1964), value r was found by optimization. In next step were data transform to standard normal distribution. Our data were with monthly step and forecasting was recurrent. We used 90 years long real flow series for compile of the model. First 75 years were used for calibration of model (autoregressive coefficient), last 15 years were used only for validation. Outputs of model were compared with real flow series. For comparison between real flow series (100% successfully of forecast) and forecasts, we used histogram and average error between each forecasted flow and measurement flow. Results were statistically evaluated on monthly level. Results show that the longest backward correlation did not give the best results. Flows forecasted by the model give very fine results in drought period. Higher errors were reached in months with higher average monthly flows. This higher flow is caused by floods. The floods are very complexly predictable. If we evaluate all months together, we will decreased precision of outputs, but in months with higher average monthly flows is enough water. This is reason, why we could not give this time period same importance as drought periods. Due to good results in drought periods was model tested for managed large open water reservoir with supply function. The Result was much better than results from non-zone forecasting stochastic models.

Keywords: Stochastic, forecasting, average monthly flow, zone