| Recently, due to intensive human activities and climate change, hydrological regime in our country has changed significantly, triggering a series of problems, such as water resources shortage, water pollution, water ecological deterioration and sedimentation problems and so forth, which greatly influence the healthy development of regional social economy. Therefore, thoroughly understand hydrological regime of stream flow and grasp regional water resources development are of great importance to regional water resources planning and management, design of hydraulic engineering activities and water balancing between ecosystems and human demands in the present and future.It is well known that under the influence of climate conditions and complex underlying surface conditions, hydrological processes in natural conditions are extremely complex and variable, frequently showing complex nonlinear, non-stationary and multi-temporal scale characteristics. However, human activities greatly alter the hydrological regime of rivers and add complexities to river hydrologic processes, which brings in a huge challenge to scientific and accurate understanding of stream flow changes. Taking Shaanxi province as the study area, adopting various nonlinear analysis methods and non-stationary data processing methods to study hydrological variation regime, multi temporal scale characteristics and responses to climate factors in Han River Basin, Wei River Basin and Wuding River Basin, and analyze their spatial variation laws. Besides, we established the long term hydrological prediction model and made extrapolation analysis to hydrological regime in Shaanxi province. The main contents and conclusions are as follows.(1)Hydrological variation diagnosis. According to thought of hydrological variation diagnosis raised by professor Xie ping, we made variation diagnoses of stream flow evolution in area of Northern Shaanxi, Central Shaanxi and Southern Shaanxi respectively. All these analyses were based on three aspects, the preliminary diagnosis, elaborate diagnosis and comprehensive diagnosis. ① The preliminary diagnosis. Use the process line method, cumulative curve slope range analysis and Hurst coefficient method in preliminary analysis to judge the existence of runoff variation. ②Elaborate diagnosis. through correlation coefficient method, the Spearman Rank Correlation test and Kendall Rank correlation test to diagnosis whether a sequence exist trend variation whereas using Lee-Heghinian test, Orderly Clustering method, the Optimal Information method,Brown Forsythe Test and Mann-Kendall test to detect its jump variation situation. ③Comprehensive diagnosis. Synthesize trend variation and jump variation situation of a sequence and get the final variation conclusion. Through diagnostic analysis, we found that from the aspect of geographical distribution, the degree of hydrological variation increases normally from south to north. In detail, Han River Basin located in Southern Shaanxi has no obvious hydrological variation, Wei River Basin located in central Shaanxi has a normal hydrological variation while Wuding River Basin located in Northern Shaanxi has an extremely remarkable variation. Through further analysis, reasons leading to hydrological variation varies. Mutual influences of climate change and human activities have caused the variation in Wei River Basin whereas only human activities occupies a predominant role in driving variation in Wuding River Basin.(2)Multi-temporal scale characteristics. Hydrological processes are extremely complex and variable, with an extraordinary randomness. Combining the wavelet analysis with HHT transform method, we studied deterministic periodic components of runoff sequences within study area and compared their multi-temporal scale characteristics. The results indicate that due to different theoretical bases, conclusions using the 2 methods are also divergent. Regional cycles detected by HHT are more meticulous and complex. The detected results show that the whole study area has 2-4 year cycles, 6-8 year cycles, 9-13 year cycles and a large scale of 24-28 year cycles. However, wavelet analysis results are more clear and stable, and periods in three regions are more close, especially in 3-4 year cycles and 8-9 year cycles. In a larger time scale, their periods are slightly different. Periods in the Guanzhong area(23-25 year cycles) are shorter than that in Northern and Southern Shaanxi(27-28 year cycles). From the aspect of methodologies, we concluded that results from HHT are more careful and closer to the fact, which responses to the fact that different runoff generation will lead to different periods. Due to influences by transcendent analysis and the Fourier transform, conclusions from the wavelet analysis in three regions are close. Simultaneously, shorter periods are significant and strong in early phase while gradually weaken or even disappear with time passing by, mainly owing to intensifying human activities. Thus in later time, 23-28 year periods have become the major periods in stream flow. In a word, they are new changes in hydrological multi-temporal scale characteristics influenced by human activities. Therefore administrative departments should pay their attentions to that.(3)Responding relationship between runoff periodic evolution and climate factor. The climate change decides the deterministic components in runoff evolution. Thus in order to further reveal the change of stream flow regime, we introduced in the ENSO events, which characterize the global climate change. Utilizing the cross wavelet transform method and condensation method, we analyzed the correlation of runoff evolution law and climate change, and studied runoff periodic evolution mechanisms from the aspect of driving factors. Results indicate that the station runoff and ENSO Index(MEI index) in small scale 2- 4 year cycle existed significant negative correlations before 1970 s, but after the 1970 s, due to impacts of human activities, this correlations become complicated.(4)The long term runoff prediction. Under dual influences between climate change and human activities, stream flow process has gradually altered from a natural stable process to a complicated nonstationary linear process. Thus smoothing hydrological sequences or adopting nonlinear analysis has become indispensable. This research established RBF neural network prediction model based on the EMD, in order to solve prediction problems concerning runoff time series. Decompose measured a runoff sequence into several relatively stable IMF components and a single trend component. Then establish RBF neural network prediction model, analyze errors and check its accuracy. Finally conduct long term stream flow prediction. The prediction results indicate that stream flow in Southern Shaanxi will stay steady while stream flow in Guanzhong area and Northern Shaanxi will decrease greatly.(5)Spatial distributions of hydrological regimes. As climate, topography, geomorphology and hydrogeology conditions are different, long term hydrological evolutions in Northern Shaanxi, Guanzhong area and Southern Shaanxi are varied. Furthermore, ①Severity of hydrological variation gradually increases from south to north. ②As for multi-temporal scale characteristics, periods in central are longer than that in north and south. ③From the aspect of responding relationships to climatic factors, runoff processes in Guanzhong area are closest to MEI, then follow the Northern Shaanxi, finally the Southern Shaanxi. ④According to runoff prediction, runoff in Guanzhong area and Northern Shaanxi will continually decrease. The decreasing trend in Guanzhong area is more obvious. However, runoff in Southern Shaanxi will stay stable. |
PDF Full Text Request