Assessing current and global-change driven behaviour of the semi-arid Onkaparinga catchment by means of spatially-explicit simulations of flow and nutrient loads based on the modelling tool SWAT

Abstract

The semi-arid rural Onkaparinga catchment in South Australia is vulnerable to future change in climate and land use because of its extreme rainfall patterns and periods of drought along with intensive horti-and viticulture. The catchment provides up to 40% of the drinking water supply to the metropolitan area of Adelaide and hence the risks of eutrophication in the downstream reservoirs are of great concern. Effective management of such catchment requires development of a robust model that sufficiently represents the diverse land use and climate system of the catchment and thus facilitate in improved understanding of spatial and seasonal flow and nutrient dynamics. Hence, to achieve this, the semi-distributed catchment modelling tool SWAT (Soil and Water Assessment Tools) was utilised with following objectives: 1) to investigate different calibration approaches for enhancing model’s validity, 2) to determine the combined effects of future climate and land use change on flow and water quality of the Onkaparinga catchment and 3) to better understand the spatial nutrient dynamics in the Cox Creek catchment by combining site-specific monitoring and spatially-explicit modelling. The models developed by means of SWAT resulted in realistic simulations of the unique flow conditions of the semi-arid Onkaparinga catchment. Experiments with different calibration approaches have shown that multi-site calibration produced better simulation results for total nitrogen (TN) and phosphorus (TP) loads than single-site calibration, but had no significant effects on results for flow and total suspended sediments (TSS) loads. Further analysis revealed a high uncertainty in the simulation results of TSS pointing at the necessity of improving the sediment modules in SWAT. The multi-sited calibrated model has been applied for future projections of climate and land use change to assess their effects on flow and water quality in the Onkaparinga catchment. The climate models suggested high uncertainty in terms of seasonally varying flow and nutrient loads, however a decreasing trend was observed. The effects of climate change were clearly dominating compared to effects of the projected land use change. Prospective simulations of combined effects for the period from 2046 to 2070 revealed highest decrease in water yield, TN and TP loads by -23.3%, -42.5% and - 49.5%, respectively during the spring season. Results for summer months suggest the declines in flow and increase in nutrient concentration, mainly driven by land use changes, and hint at potential risks of algal blooms in downstream drinking water reservoir. An approach that combines both monitoring and modelling for better understanding of nutrient dynamics was demonstrated in the Cox Creek catchment. Spatially intensive monitoring of flow and nutrients helped to identify the nutrient hotspots and established the strong link between market garden and TN and TP concentrations. Simulated nutrient export from different sub-basins matches well with field collected data for most of the sub-basins except at one, which is highly influenced by farm dam regulations. Hence future model efforts can be identified through combined monitoring and modelling. In summary, this study has highlighted the benefit of utilising spatial data for improving the performance of catchment models and identifying model deficiencies. Resulting validated models can then serve as credible tools assessing effects of future scenarios on flow and water quality in catchments. Such approach provides scientific evidence to water resource policy-makers for making informed decisions.