Study On Annual Runoff Of Key Watersheds And Its Forecast In The Loess Plateau

The Loess Plateau is the most serious areas with soil erosion in our country, and its ecological environment is very fragile, simultaneously, it is also main runoff yield area of the middle yellow river. Because the fluvial of lower reaches of Yellow River is higher than both banks area obviously, other rivers can not go into the Yellow River, therefore, how much the runoff volume of Loess Plateau has important influence on the block of lower Yellow River or not. For many years, large-scale hydrologic engineering and water and soil conservation measures have been taken to avoid water and soil erosion, which will have important effects to the cycle of precipitation in the surface basin. So, analyzing the changing characteristic and the evolution of runoff, forecasting the runoff precisely, have significance to appraise the effects of the water and soil conservation, instruct the hydrologic engineering instruction and the water resources exploitation.This thesis, on the base of the absorbing the past results and using modern new theories (such as chaos, wavelet transformation, approximate entropy and life cycle model) and traditional method, making use of the Xinshui river basin's runoff data who located at northwest Shanxi province, study systematically variant characteristic (such as period, chaos, complexity and characteristic of sudden-change), discuss the influence of climatic variations and human activity to the annual runoff, build annual and month prediction model, and the results provide the future annual runoff for the people who choosing the water resources policy, researcher and public. Principal findings are as follows:Firstly, to the whole river basin, from seventh month to ninth month is the main flood season, and from first month to third month is the main dry season; the distribution non-uniform index Cv y is 0.63, the distribution of runoff is non-uniform; the annual runoff has close relation with the flood season runoff, and the concentration degree was high; the runoff of Xinshui river basin in 1950s and 1060s is abundant, after those times, is dry. In conclusion, the changing extent of runoff is big, the multi-year changing curve of annual runoff great.Secondly, the high-flow period of annual runoff was short(usually 1-3 years), while the low-flow period was long(the longest is 7 years), and the runoff had a 5-6 years changing periodicity; the runoff varied obviously, the volume of runoff reduced obviously too; Based on the results of R/S, the Hurst value is 0.8565, it is said that the runoff of Xinshui river will decreased in the future.Thirdly, the annual and month runoff of Xinshui river basin has chaos characteristic, from the built-in dimension and saturation correlation dimension, we know, the prediction length of month runoff is 10 month, the prediction length of annual runoff is 8.26 years. Fourthly, according to the results of the approximate entropy, the complexity of the primitive serial is higher than the de-noising serial; The complexity of the runoff appears a decrease trend, and the variety range is big, it is said that the complexity of the runoff varied big.Fifthly, The correlations between annual runoff and precipitation are positive, and the correlations between annual runoff and temperature are negative, and their correlations are all obvious, it is said that the precipitation and temperature have important influence on the runoff; and the influence of climatic variations and human activity to the annual runoff are analyzed quantificationally, it reveals that, compared with the rising temperature and the decreasing precipitation, the artificial factors are the main factor leading to reduction of annual runoff,Sixthly, The life cycle model is a new long term prediction model for river runoff, which requires less data, simple calculation. However, due to the physical mechanism, the prediction results of the life cycle model can't reflect the fluctuate and random characteristics of the river annual runoff, and it can not meet required precision for hydraulics and hydropower engineering management, which usually need one year prediction period. Aiming at these problems, the life cycle-Markov combination prediction model was put forward, in which the life cycle model was used to predict the tendency item of the river annual runoff, periodic modified method is adopted to predict the runoff periodicity item, and the Markov model is also involved to predict the random item. The annual runoff at Xinshui river basin of the Yellow river from 1956 to 1997was predicted with the new combination model, and the annual runoff from 1998 to 2000 was tested. The results showed that: The average precision of the combination model is 88.26%, 86.85% prediction results is qualified. So the new model can meet the requirement of real-time prediction for the management of hydraulics and hydropower engineering.Seventhly, the AR(P) month runoff prediction model is established, its precision is very high, and it can be used to predict the month runoff in the future; The month runoff and month precipitation show good linear correlations, and the month runoff and month temperature show good exponential correlations, so, the non-linear regress model is established, and this model is simple, its precision is also very high.To sum up the whole study adopted modern new theories, methods and various ways, variety characteristic of runoff at the Xinshui river basin of The Loess Plateau is systemically studied. Its research and the analysis methods provided intending variety foreground of runoff for decision-making, researcher and public.