Estimating sub-catchment rainfall from rain gauge observations at daily time steps

[摘要] Accurate estimation of rain gauge data is essential for many meteorological and hydrological applications. Rainfall is also one of the most important drivers for hydrological models that forecast stream flow and ash flood. In this paper, a hierarchical Bayesian method is introduced for multi-site estimation of sub-catchment rainfall at daily time steps. The methodology presented herein utilizes an approach that can efficiently parameterize a spatio-temporal dynamic model to estimate daily sub-catchment rainfall and its uncertainty. The log-sinh transformation technique is used to conform the rainfall data to an assumption of normality. The zero rainfall occurrences have been treated as censored data and incorporated into the model hierarchy. In addition, this framework allows for handling missing observations and data gaps in records over both short and long periods. An application is presented to show the effectiveness of the technique to estimate rainfall for infilling missing observations at gauged locations using a cross-validation experiment. The results suggest the potential for developing the Bayesian hierarchical modeling technique to ensure a reliable approach towards missing data infilling and spatial interpolation of daily and sub-daily sub-catchment rainfall.

[发布日期] 2013-07-01 [发布机构] CSIRO

[学科分类] 地球科学(综合)