In the latest years the capacity and complexity of climate and environmental modeling has increased considerably. Therefore, tools and criteria for model performance evaluation are needed to ensure that different users can benefit from model selection. Among graphical tools, Taylor’s diagram is widely used to provide evaluation and comparison of model performances, with particular emphasis on climate models. Taylor’s diagram accounts for different statistical features of model outputs and observations, including correlation, variability and centered root mean square error. Not included is model bias, which is an essential feature for climate model evaluations, and it is usually calculated separately to complement the information embedded in Taylor’s diagram. In this paper a new diagram is proposed, referred to as Aras’ diagram, which allows for visual assessments of the correspondence between model outputs and reference data in terms of total error, correlation, as well as bias and variability ratios through an easy-to-interpret two-dimensional (2D) plot, allowing for proper weighting of different model features. The strengths of the new diagram are exemplified in a case study of performance evaluation of EURO-CORDEX historical experiment over Southern Italy using E-OBS as reference dataset, for three hydrological variables (i.e. daily precipitation, daily surface minimum temperature, and daily maximum surface temperature), and four popular climate indices (i.e. total annual precipitation, annual maxima of daily precipitation, annual minima of daily minimum temperatures, and annual maxima of daily maximum temperatures). The proposed diagram shows interesting properties, in addition to those already included in Taylor’s diagram, which may help promoting climate model evaluations based on their accuracy in reproducing the climatological patterns observed in time and space.

Extreme climate-related events such as droughts, floods, and heatwaves are becoming more frequent, making climate models essential tools for understanding and predicting these phenomena despite their inherent uncertainties. Continuous performance evaluation is crucial to select the most reliable models for impact assessments and to guide ongoing model improvement. This study evaluates the performance of all available EURO-CORDEX regional climate model simulations (RCMs) providing daily variables at 0.11° resolution over Portugal for the period 1971 to 2004. Three key surface variables, daily precipitation, daily minimum temperature, and daily maximum temperature, were assessed using nine mean and extreme climate indices, benchmarked against two high-resolution observational datasets, Iberia01 and E-OBS. The evaluation employed the Aras diagram to analyze model bias, variability, correlation, and total error. Results indicate that model performance varies notably by variable and index. Maximum temperature mean indices showed the highest accuracy, with total errors generally below 30%, followed by minimum temperature mean indices with errors under 40%. In contrast, precipitation indices exhibited the greatest uncertainties, with total errors exceeding 50% for extreme precipitation indices such as annual maxima and consecutive dry days across all models. Mean climate indices were better captured than extremes, with over 85% of model simulations achieving total errors below 50% for mean temperature indices. Spatial analysis of total error was performed at the individual grid-cell level, revealing localized strengths and weaknesses across Portugal and highlighting regional variations in model skill. These findings demonstrate that while EURO-CORDEX RCMs reliably capture temperature-based indices, challenges remain in simulating precipitation extremes. The comprehensive evaluation across multiple indices provides valuable guidance for model selection and future improvements in regional climate and impact assessments for Portugal.

Daily streamflow prediction is essential for effective water resources management, particularly in small, rapidly responding Mediterranean catchments characterised by pronounced seasonal variability and flashy hydrological regimes. This study evaluates two ensemble tree-based machine learning models; LightGBM and Random Forest, for daily discharge prediction in the Fiumarella di Corleto basin (33 km²), Southern Italy, using 21 years of daily hydrometeorological observations (2002–2022). Predictor variables included lagged discharge, three-day moving averages, multi-day cumulative precipitation, and harmonic seasonal indicators derived from the day of year. Hyperparameters were optimised using an expanding walk-forward cross-validation approach to preserve the temporal structure of the time series. Model performance was primarily assessed using the Aras diagram, which quantifies total error and its decomposition into bias, variability, and correlation errors across both training and independent test periods; additional evaluation metrics included the Kling-Gupta Efficiency (KGE), Nash-Sutcliffe Efficiency (NSE), and Root Mean Squared Error (RMSE). LightGBM substantially outperformed Random Forest, achieving total errors of 5% in both periods according to the Aras diagram, compared to 21% for Random Forest. This superior performance was consistent with higher KGE (0.971 vs. 0.795), NSE (0.970 vs. 0.870), and lower RMSE (0.125 vs. 0.259 mm day⁻¹), with near-zero systematic bias for both models (PBIAS < 1%). SHAP analysis computed on the full test set identified the three-day moving average of antecedent discharge (Q_ma3) as the dominant predictor, followed by precipitation and lagged discharge variables, with cross-model agreement in feature rankings confirming the robustness of these findings. These results demonstrate that LightGBM, combined with rigorous temporal validation, Aras-based performance evaluation, and SHAP interpretability, provides a reliable and physically consistent framework for daily streamflow prediction in small Mediterranean basins.