Data assimilation, sensitivity analysis and uncertainty quantification in a semi-arid terminal catchment subject to rainfall decline

Abstract

<p>Quantification of long-term hydrologic change in groundwater often requires the comparison of states pre- and post- change. The assessment of these changes in ungauged catchments is particularly difficult from a conceptual point of view and due to parameter non-uniqueness and associated uncertainty of quantitative frameworks. In these contexts, the use of data assimilation, sensitivity analysis and uncertainty quantification techniques are critical to maximise the use of available data both in terms of conceptualisation and quantification. This paper summarises findings of a study undertaken in the Lake Muir-Unicup Natural Diversity Recovery Catchment (MUNDRC), where a number of techniques were applied to inform both conceptual and numerical models. The MUNDRC is and small-scale endorheic basin located in southwestern Australia listed under the Ramsar Convention as a Wetland of International Importance and have been subject to a systematic decline in rainfall rates since 1970. Conceptual and numerical frameworks have been development to understand and quantify impacts of rainfall decline on the catchment using a variety of metrics involving groundwater and lake levels, as well as fluxes between these compartments and mass balance components. Conceptualisation was facilitated with the use a novel data-driven method relating rainfall and groundwater response running backwards in time, allowing the establishment of baseline conditions prior to rainfall decline, estimation of net recharge rates and providing initial heads for the forward numerical modelling. Parameter and predictive uncertainties associated with data gaps have been minimised and quantified utilising an Iterative Ensemble Smoother (White, 2018), while further refinement of conceptual model was undertaken following results from sensitivity analysis, where major parameter controls groundwater levels and other predictions of interest were quantified.</p>