| In the routine practice affairs about the water resources analysis and assessment, it is necessary to calculate and analyze the hydrological data stored in the historical hydrological database. The calculating task is very huge. The paper built an integrated forecast model with data digging technology and hydrological time series analysis tools, to supply a technological support for scientific integrated water resources planning, designing and assessing.The paper set out from the fundamental principles and technologies about data mining, to summarize the concept, meaning and implementing methodology, and to analyze the various data mining methods, including the data mining technologies about classification, clustering and Association rules finding, time series analysis, and data summary. Out of which, the Association rules finding, time series analysis were the important study objects.The paper applied statistics analysis method, to descript the main characteristics of the hydrogeological variables. According to the characteristic of hydrological variables changing over time, the time series analysis technology was applied to analyze the trends (including step trend, linear trend, exponential trend), period and auto regression characteristic of the hydrogeological variables with long-term series. Based on the analysis, an integrated forecast model for time series variables was built. The model can simulate the time series variable, so supply a strong base for estimating historical and future hydrological events.Finally, integrating practice, a time series analysis tool package was developed. A case study based on some long-term observed groundwater head data was presented. With the tools software and time series analysis technology, the trends, periods, auto regression characteristic of the groundwater heads were analyzed. According the output of the analysis, an integrated forecast equation for the time series variable was built. |
PDF Full Text Request