Research On Complexity Measure And Forecast Of Hydrology Factors In Harbin

Harbin is a national old industrial base; it is an important center city in the northeast of China and enjoys the reputation of ‘cock’ in Heilongjiang province. Harbin occupies very important strategic position in the strategy of revitalizing old Northeastern industrial bases. Rec ent years, with the rapid development of urbanization and industrialization in Harbin, obvious changes have taken place in urban hydrological system; these changes lead to a series of hydrological problems, such as over development of groundwater, surface water quality deterioration, and land subsidence and so on. It is of profound theoretical and practical significance to make an intensive study of the complexity of hydrological elements, reveal the space-time change of hydrological elements, analyze the development of hydrological elements and realize scientific and reasonable allocation of water resources.Traditional multifractal detrended fluctuation analysis(MF-DFA) use polynomial to fit data in the process of Generalized Hurst index fitting. However, polynomial fitting is quite cumbersome and can only fit polynomial data. In this paper, the Empirical Mode Decomposition(EMD) is used to optimize the MF-DFA method, for using the modified method the efficiency and performance of MF-DFA is improved. The modified Multifractal spectrum analysis(MFS-EMD-DFA) is used to measure the temporal and spatial changes of precipitation, underground water and evaporat ion capacity in 11 districts in Harbin.Hydrological sequence always contains many kinds of noise, but most of the traditional hydrological sequence complexity measurement methods do not analyze the noise resistance. In this paper, MFS-EMD-DFA, LZC and ApEn were compared in the aspects of assistant to white noise and color noise. The result shows that in some extent both LZC and Ap En show robust in the aspect of assistant to white noise, but it is easy to collapse when there is color noise, MFS-EMD-DFA method is much more robust. MFS-EMD-DFA modified the multifractal measurement method for the degree of sequence complexity, it promote the study of multifractal, and provides a new idea for the measurement of sequence complexity.Based on the pace-time laws of the complexity of hydrological elements in Harbin, from the prospective of complexity, this paper built prediction models, the BP Network model and ARIMA Time Series model, to predict hydrological elements in Harbin. Through comparative analysis, we can discover that both of the classical prediction models perform well in data fit and predict. According to the Mean Absolute Error(MAE) index and Root Mean Square Error(RMSE) index, BP model is select for both precipitation and groundwater sequences in Shuangcheng and Mulan, so as the evaporation sequences in Wuchang. And ARIMA model is select for evaporation in the downtown. And the forecast length is 60. It has a higher reference value when the sequence has lower complexity. As for the higher complex sequence, we should pay attention to uncertainties and make preparation. The complexity which changes with time could reflect the impact of previous decision on the hydrological element, and provide a solid foundation for further implement Harbin optimization allocation of water resources, hydrology environmental monitoring and management. At the same time, it also provided a scientific basis for the other related departments to make decision.