Research And Application Of The Mid-Long Term Water Resource Dispatching Model Based On Multiband Autoregression And Wavelet Analysis

As a kind of fundamental natural resource and strategic social resource, water is of great significance for the survival of mankind, economic development and social progress. While the spatial and temporal distribution of water resources is unbalanced in China, where the total amount of water resources has been increasingly decreased in recent fifty years, which is unfavorable to the utilization of water resources. Meanwhile along with the social and economic development, water consumption structure has changed significantly, the competition of water utilizing has appeared frequently, and the balance of supply and demand still faces challenges. How to meet the demand of water udder the condition of limited water supply is a subject to research on water resources dispatching.Water resources dispatching is a way to tackle the problems of unbalanced spatial and temporal distribution of water and competition of water utilizing. In this paper, dispatching belongs to mid-long term operation, including water supply and irrigation. As a basis for water resources dispatching, hydrological forecasting is crucial. As forecasting period extend, forecast precision often drops. To deal with it, a new mid-long term runoff forecasting method based on multiband autoregression and wavelet analysis was proposed. Based on period analysis, trend analysis and multiband autoregression model, forecasted runoff could be calculated and entered into dispatching model. Finally, according to available dispatching demand, proposed a method to map the dispatching chart of important reservoirs.The new method was applied in Qujing, Yunnan Province. Mid-Long term water resource dispatching in Qulu region involves 4 areas,15 units,43 reservoirs and other water supply source, including ground water, recycling water and rainwater. In runoff forecasting stage, compared with traditional autoregression model, the correlation coefficients between measured runoff and forecasting runoff based on multiband autoregression model were above 0.8, and Nash coefficients were above 0.7. Then according to dispatching results, map the dispatching charts of important reservoirs. The mid-long term water resource dispatching model based on multiband autoregression and wavelet analysis has superiority in simulation accuracy compared with traditional regression model.