This paper presents version 2 of the Australian edition of the Catchment Attributes and Meteorology for Large-sample Studies (CAMELS) series of datasets. CAMELS-AUS (Australia) v2 comprises data for an increased number (561) of catchments, each with long-term monitoring, combining hydrometeorological time series with attributes related to geology, soil, topography, land cover, anthropogenic influence and hydroclimatology. It is freely downloadable from https://zenodo.org/doi/10.5281/zenodo.12575680.

Improving how rainfall-runoff models estimate evapotranspiration is key to better reproducing water partitioning under current conditions, and will increase model realism under future changing conditions. We tested how well different conceptual rainfall-runoff model equations simulate evapotranspiration using Australian catchment and flux tower data. We found one equation consistently worked better than the others. However, even this equation had flaws, pointing to missing vegetation processes.

This study adopts actual evapotranspiration (AET) signatures to diagnose deficiencies in simulation of AET within conceptual rainfall-runoff models. Five models are assessed using flux tower data at 14 Australian sites. Even when AET is included in the calibration, the models struggle to represent aspects of AET dynamics, including interannual variability and timing on seasonal and event scales. The approach shows promise for more insightful critique of model simulations.

Hydrologic models are needed to provide simulations of water availability, floods, and droughts. The accuracy of these simulations is often quantified with so-called performance scores. A common thought is that different models are more or less applicable to different landscapes, depending on how the model works. We show that performance scores are not helpful in distinguishing between different models and thus cannot easily be used to select an appropriate model for a specific place.

Multivariate calibration has become a widely used method to improve model realism. We found that multivariate calibration can lead to less constrained flux maps and more uncertain hydrographs relative to univariate calibration. These symptoms could be caused by non-overlapping behavioural parameter distributions for the individual calibration variables. The results emphasize that the value of non-discharge data in calibration is contingent on the suitability of the model structure.