| Summary: | 碩士 === 國立臺灣大學 === 土木工程學研究所 === 83 === Rain gage data is typically considered to provide good point
accuracy, but offers little information on the spatial distribution of rain storms. Radar, on the other hand, is capable of accurately delineating rainfall boundaries, but its estimates of rainfall are burdened with large errors. Thus it is hoped that combining the data from these two sensors will result in rainfall estimates which will have both spatial and point accuracy. This paper compares the estimate results of using various interpolation methods in different raingages density, and discusses the influence of using radars data.
By the statistics characteristics of various rainstorms, rain fall data are synthesized to be used as "true data". Making use of these rainfall data and adding random errors to it is regarded as the radar and rain gages observation data. Estimates of the rainstorms by various interpolate methods were compared to the "true data" to evaluate their performances.
In rain gage networks of high gage density, the estimate with kriging method shows good results. The Adjusted-Weighting-Coefficient method which estimates by integrating radar and raingage data can describe rainfall variation in space adequately in various raingages density, but its accuracy is poor. The co-kriging method also integrates rainfall observation data from radars and rain gages for rainfall estimates. In a network with high gage density, the method can describe rainfall distribution in space accurately. However, the result is not so good in low density rain gage networks. The modified co-kriging method can describe rainfall distribution in space accurately in different density rain gage network.
|
|---|
