Rainfall-runoff modelling in South-West Western Australia. 5. Comparison of results from LUCICAT and the 'adopted model'

[摘要] Summary and Conclusions•In this chapter the LUCICAT model and the South-West Western Australia Sustainable Yields Project ‘adopted’ models are compared in calibration, scenario projections and the models’ ability to represent non-stationary systems particularly relevant to climate change and the hydrology of south-west Western Australia. The selection of the ‘adopted model’ is described elsewhere (CSIRO, 2009; Silberstein et al., 2010b).•Rainfall and APET inputs were identical for the ‘adopted’ and LUCICAT models. However, the LUCICAT model needs additional data, namely spatial and temporal changes in land use, stream networks and topographically determined Response Units. Due to time constraints, the LUCICAT model was applied to only ten water resources catchments (mainly the water supply and Water Resources Recovery Catchments) covering an area of 15,460 km2. These catchments are located in all three project regions – Gingin to Murray, Harvey to Preston and Busselton to Denmark.•LUCICAT was calibrated by trial and error and shares a single set of parameters for the whole catchment. The daily Nash-Sutcliffe modelling efficiency (NSE) of the LUCICAT model was generally lower than that for the ‘adopted’ model.NSEs determined from monthly and annual flows were similar for LUCICAT and the adopted models. The performance of the LUCICAT model based on other statistical criteria is quite satisfactory.•Under Scenario A, the adopted model projected approximately 5% more streamflow than the LUCICAT model, mainly due to greater projections in the Denmark and Kent River catchments.•The LUCICAT projected a greater reduction in streamflow than the ‘adopted’ model under all future climate scenarios, which could be attributed to the projected declining flow associated with the growth of plantations in affected catchments and, in some catchments, a lower (and better) simulation of flow through the last ten years of record which may be related to a better representation of catchment groundwater storages.•Total calibrated flow to the major gauging stations of the 10 catchments where the LUCICAT model has been applied was 655 GL. Under the historical climate scenario (Scenario A), streamflow is projected to reduce to 641 GL. Under the median future climate scenario (Cmid) streamflow is projected to decrease to 427 GL, a 34% reduction below that of Scenario A.•On average every percent reduction in rainfall translates to approximately 3.7 percent reduction in runoff (ranging from 3.1 to 3.9), which is similar to other findings in Western Australia.•The daily flow duration distributions, particularly the low-flow periods, simulated by the LUCICAT, were better than low-flow simulation by the ‘adopted’ model. The low-flow predictions would be very useful for environmental water allocation and planning.•In most catchments, the LUCICAT model simulation error was well balanced with less temporal trend than for the ‘adopted’ model particularly during the last 10 years, indicating that LUCICAT may better represent non-stationary systems. •In catchments with significant recent plantation development the LUCICAT model better represented the impacts of plantation growth on streamflow than by the ‘adopted’ model due its lack of ability to simulate the effect of land use change.

[发布日期] 2010-12-31 [发布机构] CSIRO

[效力级别] Climate Change Processes [学科分类] 地球科学(综合)