Simulation And Analysis Of Flood In Xixi Basin In The Coastal Areas Of Southeastern China

The evolution law of water cycle under changing environment is the hot and hard issue of the hydrology. According to the literature, there are two types changing environment:climate changes and human activities. On the one hand, human activities has great impact on result in the land use/cover changes, analyzing the effect of spatial pattern of LUCC on the peak flow and flood frequency, which has important value for land use management and ecological environment construction. On the other hand, climate change makes spatial-temporal distribution of rainstorm more uniform, it is very important to understand the influence of temporal and spatial characteristics of rainstorm on peak flow and flood frequency, which not only helps to deeply understand of the mechanism of water cycle, but also is particularly important for basin flood forecasting and hydraulic engineering construction.Jinjiang Xixi basin was selected as the study area; this paper included five aspects as follows: The land use/land cover change from 1985 to 2010 was analyzed. Based on the maximum annual rainstorm flood events on the hourly scale occurred from 1955 to 2011, accumulative anomaly method and Mann-Kendall method was used to analyze the evolution trend of flood runoff. Parameter calibration method of initial loss was improved, and calculated CN used vegetation index NDVI, the distributed flood model HEC-HMS with high precision was established. The spatial variation scenarios of rainstorm and land use scenarios were established and the influences of rainstorm space change and LUCC on peak flow and flood frequency were analyzed.The results indicated that:1) The area of orchard and building increased dramatically, while forest and agriculture decreased during 1985 to 2010.2) The maximum flood peak and flood volume from 1955 to 2011 showed a decreasing trend, but the trend was not significant.3) The initial discharge could be used to show the change of initial loss or initial loss rate, the relation between initial loss (or initial loss rate) and initial discharge could use power function to fit. While initial discharge was smaller, indicating that soil moisture content was smaller, initial loss (or initial loss rate) was much larger, and with initial discharge increased, initial loss (or initial loss rate) decreased rapidly, and (?)-adually tended to a constant. Two methods comparison of initial loss calibration showed that the method calibrating initial loss through initial loss rate was better, as it can get higher simulation accuracy and consider the underlying surface characteristics, which was consistent with the theory.4) Inter-annual variation of NDVI showed segment changed, the high value of NDVI was normally stayed same from June to October. Based on the three reconstruction algorithms in TIMESAT, NDVI of Xixi basin was reconstructed from 2001 to 2010, the result showed that fidelity performance of SG method was best, AG algorithm and DLF algorithm was nearly the same; Inter-annual variation of NDVI showed segment changed, the low value of NDVI was in January to March, the high value of NDVI was in June to October.5) The method which adopted vegetation index NDVI to calculate CN was feasible. In general, the spatial distribution of NDVI_CN and Interpertaiton_CN had little difference. The simulation in HEC-HMS model also showed that the simulation accuracy by DNVI_CN was similar to traditional calculation method, but the extraction of NDVI was more convenient, so the method by NDVI_CN could greatly improve the efficiency of model and also reduce the cost.6) Devoted to LUCC from 1985 to 2010, the peak flow and flood volume increased significantly, and the average peak flow increased correspondingly, while with flood magnitude increasing, LUCC had major influence on peak flow; then the flood response of LUCC in subbasins were significantly different.7) If we did not considered LUCC that had already occurred, the calculation result of flood frequency by the actual occurrence flood turned to be small, and it was not conducive for the engineering safety. With the same conditions, only adjusted LUCC, compared actual flood frequency curve calculation in 1955 to 2011, to land use simulation in 2010, an actual occurrence of 100-year flood was only equivalent to a 73-year flood calculated by land use condition in 2010. Give not consider actual LUCC changing, only calculated the actual flood frequency, the result will also turn to be small, it is not beneficial for engineering safety.8) Rainfall spatial variation coefficient (Cy) and the center of rainstorm had significant effect on peak flow, runoff volume and the appearance moment of peak flow. The larger of the Cy the greater of the peak flow and runoff volume. Under the condition with the same Cv, the peak flow and total runoff" were the largest when rainstorm center was located in the middle basin; with the Cv, increased, the location of the center of rainstorm had great influence on peak flow and runoff volume. Under the same DSV level, different location of center rainstorm made the time of peak flow appeared earlier when the Cv was larger. Additionally, in the case of the same Cv, the time of peak flow appeared earlier when the rainstorm center position was more close to the downstream. The time to peak flow would put forward 1 hour while the rainstorm center position was in the upstream compared to the downstream.9) The calculation and prediction of flood was necessary to consider the location of rainstorm center. With the same condition, but only adjusted the location of rainstorm center to the adverse scenario, the actual 100-year flood was only equivalent to a 65-year flood. Therefore, in order to reflect flood regularity more accurately, the calculation and prediction of flood must consider the location of rainstorm center.