Modelling of fine resolution precipitation data using GAME data product and validating against the spatial patterns of HUBEX-IOP-EEWB data
Roshan K. Shrestha (1), Yasuto Tachikawa (2), Kaoru Takara (2)
(1) Graduate School of Urban and Environmental Engineering, Kyoto University, Japan
(2) Disaster Prevention Research Institute, Kyoto University, Japan
Multiplicative Random Cascade (RC) method is a stochastic tool to model a fine resolution precipitation data, which is based on assumption that the precipitation structure is Multifractal. The random process involved within it, which often causes the results either to differ largely in between repeated trials or to differ from a measured one, dominates existing RC method. This weakness is necessary to remove before adopting the method as a reliable tool of modeling the fine resolution precipitation field. Noticing that the spatial rainfall field contains spatial correlation, we attempted to include this information in a RC method. First, we evaluate a reference matrix, which accounts for the spatial correlation effect of a coarse precipitation field and the distance from the nearest point of coarse grid to the center of a grid at the fine resolution. Then, this reference matrix assists to find the location of a cascade generator by comparing the hierarchical order. It also assists to re-allocate the statistically filtered peak values to a proper location. This method, named as Multiplicative Random Cascade HSA (RCHSA) method, was used to model the fine resolution precipitation data (10-arc minute resolution) using GAME Reanalysis 1.25 degree data as the input. Comparing the spatial patterns of over 40,000 test outputs from the RCHSA method against the spatial patterns of the HUBEX-IOP-EEWB data, it revealed that the RCHSA method was largely successful to remove the weakness of the RC method. The overall performance was improved to 0.6 from 0.34 after including the HSA method. This has helped us to use the GAME 1.25 degree data product in hydrological simulations of catchments as small as of 2000 sq. km. scale successfully.
Submittal Information
Name :
Date :
Roshan K. Shrestha
03-Aug-04-20:36:25
Organization :
Theme :
Disaster Prevention Research Institute, Kyoto University