Author(s): Chi Yang
Linked Author(s):
Keywords: Huaihe River Basin; Accumulated precipitation; Pearson type III distribution; Generalized additive models for location; Scale and shape; Downscaling
Abstract: As the sixth largest river basin in China, the Huaihe River Basin (HRB) is located in East China with a total area of 270, 000 km². Dominated by the East Asian monsoon climate, the HRB suffers from severe flood disasters. The catchment scale catastrophic floods are usually caused by large-scale persistent rainfall rather than short-term local rainstorms. In this paper, the spatial-temporal variation of annual maximal 30-day accumulated precipitation in the HRB is modeled as response to geographical effects and climate factors. Daily precipitation recorded at a network of meteorological stations covering the HRB and its contiguous areas spanning 1951 to 2012 is used to generate the annual maximum as response variable. The Pearson Type III (P3) distribution, a 3-parameter gamma distribution widely used in hydrology, is fitted within the framework of Generalized Additive Models for Location, Scale and Shape (GAMLSS). By using GAMLSS, spatial pattern and inter-annual variation of the annual maximum can be represented simultaneously through semi-parametric regression. 3-dimensional geographical coordinates are used as covariates of the model accounting for the spatial variation of the annual extremes. To investigate the relationship of the response variable to climate variability, extensive climate indices representing large-scale circulation patterns and sea surface temperature patterns are examined; statistically significant ones are selected as covariates of the model. Return periods of the annual maximum changing with location and time can be derived accordingly from the fitted GAMLSS. The results show that the fitted model can well describe both the non-stationarity-in-time and non-homogeneity-in-space of the response variable in the study area simultaneously. This method provides a useful tool for flood control planning under climate impact and for the projection of climate change scenarios onto the annual maximal cumulative precipitation in the HRB.
Year: 2013