English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

This paper considers two streamflow models of the Ovens catchment in northeast Victoria and assesses the goodness-of-fit and variability of model predictions when calibrated over three different, historical, climate regimes. The aim was to understand whether the two models, calibrated using historical data, could be used to model future scenarios of climate-change where climate attributes such as rainfall, temperature and potential evaporation often lie outside the bounds of historical variability. To address this question, we looked to establish the influence of calibration data on model behaviour by assessing whether a model, calibrated over one climate regime, could adequately model streamflow over another climate regime. The Ovens catchment in northeast Victoria, Australia was chosen as the study area. As the only largely unregulated catchment of the Murray River it provided a simplified connected system of streams to assess the validity of streamflow models. Two simple models of the Ovens catchment were established. The first model was constructed in Source (eWater, 2011) and consisted of a series of connected stream-nodes each modelled by Simhyd, a commonly used daily rainfall-runoff model. An identical stream network was constructed in CATPlus (DPI Victoria, 2011) with stream-nodes modelled using CATNode, a lumped surface-groundwater model designed to model temporal, surface and groundwater contributions of water yield and salt load to stream. Both models were calibrated using PEST, a non-linear, model-independent parameter estimation program (Watermark Numerical Computing, Australia 2010). Based on annual rainfall over the Ovens catchment from 1975 to 2010, historical climate data were divided into three rainfall regimes; wet, medium and dry. For each climate regime the Simhyd and CATNode models were calibrated against streamflow data from that regime and then validated against streamflow data from the two other respective regimes. Three measures were used to assess the two models. First, the goodness-of-fit of each calibration and the subsequent validations were measured using a number of statistical measures including the Nash Sutcliffe coefficient of efficiency, percent difference in cumulative flow and the gradient of a regression line fitted to the measured versus modelled streamflow. Second, the uncertainty in streamflow attributed to the three different calibrations was calculated as the mean of the standard deviation in the monthly flow over the three calibrations as a proportion of the mean monthly flow. Third, the relative changes in streamflow between the different regimes were compared for the three calibrations. Statistical analysis found that for each model over 90% of sub-catchments met the goodness-of-fit acceptability criteria. The models both validated well over the wet and average regimes, but validated poorly over the dry regime. Whilst the omission of factors such as extractions and dam releases may have contributed to this poor validation there was evidence that a model calibrated under wet conditions was biased towards those conditions. Consequently, if a future, drier scenario was used in these models, equivalent to moving from an average or wet regime to a dry regime, these models could potentially overpredict streamflow and under-estimate the impacts of climate change. On average the uncertainty in the monthly streamflow due to the different regimes of calibration was considerable, ranging between 8 to 35% of the mean monthly flow for the 22 sub-catchments. Although this uncertainty was large, in terms of assessing the impacts of climate change on streamflow it was perhaps more important to consider the uncertainty around the relative change in streamflow between different climate regimes. Results showed that the uncertainty associated with the percent decrease in streamflow ranged between 2.5 to 4.1% for CATNode and 4.1 to 6.6% for Simhyd. If under future scenarios of climate change the relative shift in streamflow was greater than this uncertainty we would now have greater confidence that this change was climate driven and not an artefact of the model calibration data.

The ability of streamflow models to capture the impact of climate variability on streamflow