Infrastructure Engineering - Theses
Permanent URI for this collection
Search Results
1 - 1 of 1
-
ItemEvaluation and use of reanalysis rainfall data for catchment hydrology applications( 2022)Accurate information on rainfall is required for developing modelling tools and for the design and investigation of existing and planned infrastructure. More specifically, a long time series of spatially and temporally consistent high-resolution rainfall datasets are necessary for many hydro-meteorological applications, such as understanding the variation of extreme rainfall across space and time, modelling impactful weather events such as floods, assessing hydrological risks under climate variability and change, and forecasting agricultural water needs in the short and medium terms. However, such datasets are generally not available due to the sparse and variable nature of gauging networks. Global and regional reanalyses are alternative sources of precipitation data with consistent spatial and temporal resolution and coverage. A regional reanalysis, nested in a global reanalysis, typically assimilates more local observations using a higher resolution model to improve the representation of local climate features and extreme events in the global reanalysis. BARRA-R (The Bureau of Meteorology Atmospheric high-resolution Regional Reanalysis for Australia) is one such example for Australia. Rainfall observations are not assimilated in this reanalysis, and precipitation is estimated by model physics and parameterisation. As such, a comprehensive examination of the dataset is essential to ascertain its utility across various hydrological applications. The objective of this thesis is to assess the reanalysis rainfall data for its use in catchment hydrology applications. This research has been investigated in two stages. First, it examines whether estimates of rainfalls produced via an atmospheric reanalysis is representative of historical rainfall behaviour at daily and sub-daily scales. Second, it explores ways in which the dataset can be used in hydrological modelling and engineering design. A range of reference datasets at different temporal and spatial scales and assessment metrics are selected to evaluate the BARRA-R precipitation. The BARRA-R rainfall is evaluated at daily and sub-daily temporal scales across point and areal scales using a range of existing and bespoke metrics. The reference datasets used include daily gauge and pluviograph measurements, daily Australian Water Availability project (AWAP) rainfalls, sub-daily and daily ERA-Interim rainfalls, and a blended radar product. It is found that daily rainfalls from BARRA-R provide an improved representation of spatial and temporal variability compared to estimates from a coarse-scale global reanalysis (ERA-Interim). In addition, general precipitation statistics and the frequency of large rainfall events are closely reproduced by BARRA-R. Further assessments of sub-daily rainfall at both point- and spatial-scales suggest that there is some potential for spatial displacement of precipitation fields during large rainfall events. The analysis also indicates that BARRA-R is more skilful in the temperate zone than in the tropical and arid zones. The potential use of sub-daily rainfall information from BARRA-R for hydrological modelling and engineering design is explored via hydrological modelling experiments. First, a simple disaggregation of daily AWAP rainfall is evaluated using the day-to-day distribution of hourly BARRA-R rainfall. The disaggregated sub-daily rainfall showed a mixed performance in characterising some of fine-scale characteristics while improving rainfall detection. Such disaggregated rainfall, with appropriate checks focused on target application, is deemed useful for analysis of rainfall statistics and parameterisation of stochastic weather generators. Similarly, hydrological modelling at sub-daily time steps is improved by applying the disaggregated areal rainfalls. Second, areal temporal patterns are derived from BARRA-R and then compared to corresponding design data from the Australian Rainfall and Runoff data hub, and these data sets are used to derive flood frequency curves using an event-based model. Comparisons of the derived flood frequency estimates are undertaken for catchments in Victoria and Queensland and are found to be in general agreement with flood frequency curves fitted using gauged maxima. This research underscores the significance of regional reanalysis-based rainfall datasets through evaluation and diverse hydrological applications. The analysis presented in this thesis demonstrates that the reanalysis precipitation data have good potential to complement existing datasets for hydrological applications, particularly in sparsely gauged areas, or with applications that could benefit from spatial and temporal consistency in the representation of rainfall fields.