Keywords
regional flood risk, spatial rainfall, stochastic rainfall, random cascade
Start Date
1-7-2006 12:00 AM
Abstract
Spatial rainfall is a key input into models that simulate flood behaviour at regional scale. Stochastic rainfall data provide alternative realisations that are equally likely to have occurred, and are often used to drive hydrologic models to quantify uncertainty in environmental systems associated with climatic variability. This paper describes the development and testing of a stochastic daily spatial rainfall generation approach which comprises two components. Daily temporal rainfalls in two meso-scale square regions are first generated using a bi-variate first-order transition probability matrix model. The spatial rainfall in each region is then disaggregated using a modified non-homogeneous random cascade model that utilises scaling invariance features in the historical rain field. The models are parameterised using 100 years of daily grided rainfall data across the Gippsland Lakes region in southeast Australia. The approach is used to generate 20 replicates of 100-year daily concurrent catchment average rainfall time series for the six major catchments in the region. The generated stochastic rainfalls are evaluated by comparing key spatial and temporal statistics with those in the historical data. The results indicate that the approach is suitable for regional flood risk assessment, although the simulated 1-day and 3-day rainfall AEP (annual exceedence probability) are slightly underestimated, while the simulated rainfall correlations between catchments are mostly higher than the observed spatial correlations. The main limitations of the approach are the absence of space-time correlation of rain fields on consecutive days, and problems in simulating the clustering (i.e. spatial correlation) of daily rain field during extreme storm events, both of which would require significant research to overcome.
Stochastic Generation of Daily Spatial Rainfall for Regional Flood Risk Assessment
Spatial rainfall is a key input into models that simulate flood behaviour at regional scale. Stochastic rainfall data provide alternative realisations that are equally likely to have occurred, and are often used to drive hydrologic models to quantify uncertainty in environmental systems associated with climatic variability. This paper describes the development and testing of a stochastic daily spatial rainfall generation approach which comprises two components. Daily temporal rainfalls in two meso-scale square regions are first generated using a bi-variate first-order transition probability matrix model. The spatial rainfall in each region is then disaggregated using a modified non-homogeneous random cascade model that utilises scaling invariance features in the historical rain field. The models are parameterised using 100 years of daily grided rainfall data across the Gippsland Lakes region in southeast Australia. The approach is used to generate 20 replicates of 100-year daily concurrent catchment average rainfall time series for the six major catchments in the region. The generated stochastic rainfalls are evaluated by comparing key spatial and temporal statistics with those in the historical data. The results indicate that the approach is suitable for regional flood risk assessment, although the simulated 1-day and 3-day rainfall AEP (annual exceedence probability) are slightly underestimated, while the simulated rainfall correlations between catchments are mostly higher than the observed spatial correlations. The main limitations of the approach are the absence of space-time correlation of rain fields on consecutive days, and problems in simulating the clustering (i.e. spatial correlation) of daily rain field during extreme storm events, both of which would require significant research to overcome.
