Groundwater recharge estimates using the Cummulative Rainfal Deparure (CRD) method have been widely used in literature (Bredenkamp, 1995), in areas where groundwater levels and rainfall present a strong correlation. This study further explores the utilisation of this method for hindcasting and estimation of baseline groundwater levels in areas where monitoring data is scarce.
The proposed methodology has been applied in the Lake Muir-Unicup Natural Diversity Recovery Catchment (MUNDRC). The catchment consists of a complex system of lakes, swamps and flood plains, located in southwestern Australia. Long term rainfall time-series in the area have shown a systematic decline over the past decades, with an abrupt change since the 70’s.
Conceptual and numerical groundwater-surface water models have been developed to assess the effects of rainfall decline in the lakes and surrounding environment. The lack of groundwater and lake level monitoring data prior to the rainfall decline makes the establishment of baseline levels and initial conditions difficult.
The CRD method has been utilised in a backward form to estimate groundwater levels prior to the decline of rainfall rates. The simplified nature this method has inherent non-uniqueness between rainfall ratios and evapotranspiration. To address that, multiple CRD models were carried at different locations within the catchment and calibrated simultaneously using regularized parameter inversion.
The calibration was undertaken with PEST-HP using 91 models, following by uncertainty analysis through the generation and sampling of the posterior covariance matrix, with a total of 500 realisations per model.
The results from the optimisation indicated a good agreement between CRD results and respective observations. Furthermore, the distribution of the calibrated parameters values is reasonable and in agreement with the conceptual model.