Andrew Schepen

Drought is a recurrent and significant driver of stress on agricultural enterprises in Australia. Historically, rainfall indices have been used to identify drought and inform government responses. However, rainfall indicators may not fully reflect agricultural or economic drought conditions and are a lagging indicator. To address these shortcomings, the AADI (Australian Agricultural Drought Indicators) was recently developed to monitor and forecast drought for upcoming seasons using biophysical and agro-economic models, including crop yields, pasture growth and farm profit, at ∼ 5 km² resolution. Here, we evaluate the skill of drought indicator forecasts driven by the ACCESS-S2 dynamical global climate model over a hindcast period from 1990–2018. Analysis of the AADI hindcasts finds that antecedent landscape conditions significantly enhance predictive skill for crop yields, pasture growth and farm profit across a financial year. As lead time shortens from 12 to 3 months, forecast confidence increases: median farm profit skill rises from 43 % at 12 months to 67 % and 73 % at 6 and 3 months, respectively, whilst median farm profit biases remain below 2 % across all lead times, with high reliability indicating a well-calibrated ensemble, making the forecasts highly suitable for risk management and decision-making. Forecasts for wheat, sorghum, and pasture are also skilful and reliable in ensemble spread, although residual biases can occur (e.g. up to 20 % for sorghum) and suggests further system refinements are needed. Analysis of historical events under both dry and wet conditions demonstrated the AADI system's ability to identify drought-impacted areas with increased confidence up to 6 months earlier than rainfall deficits.

Machine learning (ML) is a powerful tool for hydrological modelling, prediction, dataset creation and the generation of insights into hydrological processes. As such, ML has become integral to the field of large-sample hydrology, where hundreds to thousands of river catchments are included within a single ML model to capture diverse hydrological behaviours and improve model generalizability. This manuscript outlines recent advances in ML for large-sample hydrology. We review new tools in explainable AI (XAI) and interpretability approaches, as well as challenges in these areas. Key research avenues for large-sample hydrology include addressing variability in interpretations resulting from different ML models and XAI techniques, enhancing hydrological predictions in data-sparse and human-impacted regions, reducing the ‘cascade of uncertainty’ inherent in hydrological modelling, developing improved methods for multivariate prediction and identifying causal relationships. This article is part of the Royal Society Science+ meeting issue ‘Hydrology in the 21st century: challenges in science, to policy and practice’.

Machine learning methods provide a promising approach for exploiting relationships between raw forecasts and observations for forecast calibration. This paper highlights the role of data transformation in rainfall forecast calibration with neural networks. We develop a distributional regression network that accounts for the positive skewness and zero bound of rainfall by incorporating a normalizing transformation (log‐sinh) in both the input and output stages. A unified loss function is formulated based on the negative log‐likelihood function for parameter optimization. To test the role of data transformation, we conduct five calibration experiments: one that does not use transformation at all (the baseline) while the others use the log‐sinh transformation in different ways. All experiments are based on 10‐day rainfall forecasts from the European Centre for Medium‐range Weather Forecasts (ECMWF) from 2011 to 2022. Overall, the calibration methods effectively correct spatiotemporally varying biases in raw forecasts and improve reliability, yielding mean skill improvements of approximately 2%–11% and in the best case reducing forecast biases to less than 2%. Without transformation, the baseline method suffers from forecast biases ranging from −30% to 50%, due to its limited ability to characterize the uncertainty of rainfall forecasts. Of the four experiments that use the log‐sinh transformation, the optimal performance is achieved by the combined use of transforming raw forecasts for the input layer and utilizing fixed transformation parameters for generating calibrated forecasts in the output layer. We show that this method marginally outperforms an advanced existing Bayesian Ensemble Model Output Statistics method in reducing forecast biases.

Drought is a recurrent and significant driver of stress on agricultural enterprises in Australia. Historically, rainfall indices have been used to identify drought and inform government responses. However, rainfall indicators overlook important factors such as drought propagation and commodity prices. To address these shortcomings, AADI (Australian Agriculture Drought Indicators) was recently developed to monitor and forecast drought for upcoming seasons using biophysical and agro-economic models, including crop yields, pasture growth, and farm profit at ~ 5 km² resolution. Here, we evaluate the skill of drought indicator forecasts driven by the ACCESS-S2 dynamical global climate model over a hindcast period from 1990–2018. Analysis of the AADI hindcasts finds that antecedent landscape conditions significantly enhance predictive skill for crop yields, pasture growth, and farm profit across a financial year. As lead time shortens from 12 to 3 months, forecast confidence increases: median farm profit indicator skill rises from 43 % at 12 months to 67 % and 73 % at 6 and 3 months, respectively, whilst median farm profit biases remain below 2 % across all lead times, with high reliability indicating a well-calibrated ensemble, making the forecasts highly suitable for risk management and decision-making. Forecasts for wheat, sorghum, and pasture are also skilful and reliable in ensemble spread, although residual biases can occur (e.g., up to 20 % for sorghum), which suggests further system refinements are needed. Analysis of historical events in both dry and wet conditions demonstrated the AADI system’s ability to identify drought-impacted areas with increased confidence up to 6 months earlier than rainfall deficits.

Drought is a recurrent and significant driver of stress on agricultural enterprises in Australia. Historically, rainfall indices have been used to identify drought and inform government responses. However, rainfall indicators overlook important factors such as drought propagation and commodity prices. To address these shortcomings, AADI (Australian Agriculture Drought Indicators) was recently developed to monitor and forecast drought for upcoming seasons using biophysical and agro-economic models, including crop yields, pasture growth, and farm profit at ~ 5 km² resolution. Here, we evaluate the skill of drought indicator forecasts driven by the ACCESS-S2 dynamical global climate model over a hindcast period from 1990–2018. Analysis of the AADI hindcasts finds that antecedent landscape conditions significantly enhance predictive skill for crop yields, pasture growth, and farm profit across a financial year. As lead time shortens from 12 to 3 months, forecast confidence increases: median farm profit indicator skill rises from 43 % at 12 months to 67 % and 73 % at 6 and 3 months, respectively, whilst median farm profit biases remain below 2 % across all lead times, with high reliability indicating a well-calibrated ensemble, making the forecasts highly suitable for risk management and decision-making. Forecasts for wheat, sorghum, and pasture are also skilful and reliable in ensemble spread, although residual biases can occur (e.g., up to 20 % for sorghum), which suggests further system refinements are needed. Analysis of historical events in both dry and wet conditions demonstrated the AADI system’s ability to identify drought-impacted areas with increased confidence up to 6 months earlier than rainfall deficits.

Drought is a recurrent and significant driver of stress on agricultural enterprises in Australia. Historically, rainfall indices have been used to identify drought and inform government responses. However, rainfall indicators overlook important factors such as drought propagation and commodity prices. To address these shortcomings, AADI (Australian Agriculture Drought Indicators) was recently developed to monitor and forecast drought for upcoming seasons using biophysical and agro-economic models, including crop yields, pasture growth, and farm profit at ~ 5 km² resolution. Here, we evaluate the skill of drought indicator forecasts driven by the ACCESS-S2 dynamical global climate model over a hindcast period from 1990–2018. Analysis of the AADI hindcasts finds that antecedent landscape conditions significantly enhance predictive skill for crop yields, pasture growth, and farm profit across a financial year. As lead time shortens from 12 to 3 months, forecast confidence increases: median farm profit indicator skill rises from 43 % at 12 months to 67 % and 73 % at 6 and 3 months, respectively, whilst median farm profit biases remain below 2 % across all lead times, with high reliability indicating a well-calibrated ensemble, making the forecasts highly suitable for risk management and decision-making. Forecasts for wheat, sorghum, and pasture are also skilful and reliable in ensemble spread, although residual biases can occur (e.g., up to 20 % for sorghum), which suggests further system refinements are needed. Analysis of historical events in both dry and wet conditions demonstrated the AADI system’s ability to identify drought-impacted areas with increased confidence up to 6 months earlier than rainfall deficits.

Drought is a recurrent and significant driver of stress on agricultural enterprises in Australia. Historically, rainfall indices have been used to identify drought and inform government responses. However, rainfall indicators overlook important factors such as drought propagation and commodity prices. To address these shortcomings, AADI (Australian Agriculture Drought Indicators) was recently developed to monitor and forecast drought for upcoming seasons using biophysical and agro-economic models, including crop yields, pasture growth, and farm profit at ~ 5 km² resolution. Here, we evaluate the skill of drought indicator forecasts driven by the ACCESS-S2 dynamical global climate model over a hindcast period from 1990–2018. Analysis of the AADI hindcasts finds that antecedent landscape conditions significantly enhance predictive skill for crop yields, pasture growth, and farm profit across a financial year. As lead time shortens from 12 to 3 months, forecast confidence increases: median farm profit indicator skill rises from 43 % at 12 months to 67 % and 73 % at 6 and 3 months, respectively, whilst median farm profit biases remain below 2 % across all lead times, with high reliability indicating a well-calibrated ensemble, making the forecasts highly suitable for risk management and decision-making. Forecasts for wheat, sorghum, and pasture are also skilful and reliable in ensemble spread, although residual biases can occur (e.g., up to 20 % for sorghum), which suggests further system refinements are needed. Analysis of historical events in both dry and wet conditions demonstrated the AADI system’s ability to identify drought-impacted areas with increased confidence up to 6 months earlier than rainfall deficits.

Drought is a recurrent and significant driver of stress on agricultural enterprises in Australia. Historically, rainfall indices have been used to identify drought and inform government responses. However, rainfall indicators overlook important factors such as drought propagation and commodity prices. To address these shortcomings, AADI (Australian Agriculture Drought Indicators) was recently developed to monitor and forecast drought for upcoming seasons using biophysical and agro-economic models, including crop yields, pasture growth, and farm profit at ~ 5 km² resolution. Here, we evaluate the skill of drought indicator forecasts driven by the ACCESS-S2 dynamical global climate model over a hindcast period from 1990–2018. Analysis of the AADI hindcasts finds that antecedent landscape conditions significantly enhance predictive skill for crop yields, pasture growth, and farm profit across a financial year. As lead time shortens from 12 to 3 months, forecast confidence increases: median farm profit indicator skill rises from 43 % at 12 months to 67 % and 73 % at 6 and 3 months, respectively, whilst median farm profit biases remain below 2 % across all lead times, with high reliability indicating a well-calibrated ensemble, making the forecasts highly suitable for risk management and decision-making. Forecasts for wheat, sorghum, and pasture are also skilful and reliable in ensemble spread, although residual biases can occur (e.g., up to 20 % for sorghum), which suggests further system refinements are needed. Analysis of historical events in both dry and wet conditions demonstrated the AADI system’s ability to identify drought-impacted areas with increased confidence up to 6 months earlier than rainfall deficits.

Drought is a recurrent and significant driver of stress on agricultural enterprises in Australia. Historically, rainfall indices have been used to identify drought and inform government responses. However, rainfall indicators overlook important factors such as drought propagation and commodity prices. To address these shortcomings, AADI (Australian Agriculture Drought Indicators) was recently developed to monitor and forecast drought for upcoming seasons using biophysical and agro-economic models, including crop yields, pasture growth, and farm profit at ~ 5 km² resolution. Here, we evaluate the skill of drought indicator forecasts driven by the ACCESS-S2 dynamical global climate model over a hindcast period from 1990–2018. Analysis of the AADI hindcasts finds that antecedent landscape conditions significantly enhance predictive skill for crop yields, pasture growth, and farm profit across a financial year. As lead time shortens from 12 to 3 months, forecast confidence increases: median farm profit indicator skill rises from 43 % at 12 months to 67 % and 73 % at 6 and 3 months, respectively, whilst median farm profit biases remain below 2 % across all lead times, with high reliability indicating a well-calibrated ensemble, making the forecasts highly suitable for risk management and decision-making. Forecasts for wheat, sorghum, and pasture are also skilful and reliable in ensemble spread, although residual biases can occur (e.g., up to 20 % for sorghum), which suggests further system refinements are needed. Analysis of historical events in both dry and wet conditions demonstrated the AADI system’s ability to identify drought-impacted areas with increased confidence up to 6 months earlier than rainfall deficits.

Drought is a recurrent and significant driver of stress on agricultural enterprises in Australia. Historically, rainfall indices have been used to identify drought and inform government responses. However, rainfall indicators overlook important factors such as drought propagation and commodity prices. To address these shortcomings, AADI (Australian Agriculture Drought Indicators) was recently developed to monitor and forecast drought for upcoming seasons using biophysical and agro-economic models, including crop yields, pasture growth, and farm profit at ~ 5 km² resolution. Here, we evaluate the skill of drought indicator forecasts driven by the ACCESS-S2 dynamical global climate model over a hindcast period from 1990–2018. Analysis of the AADI hindcasts finds that antecedent landscape conditions significantly enhance predictive skill for crop yields, pasture growth, and farm profit across a financial year. As lead time shortens from 12 to 3 months, forecast confidence increases: median farm profit indicator skill rises from 43 % at 12 months to 67 % and 73 % at 6 and 3 months, respectively, whilst median farm profit biases remain below 2 % across all lead times, with high reliability indicating a well-calibrated ensemble, making the forecasts highly suitable for risk management and decision-making. Forecasts for wheat, sorghum, and pasture are also skilful and reliable in ensemble spread, although residual biases can occur (e.g., up to 20 % for sorghum), which suggests further system refinements are needed. Analysis of historical events in both dry and wet conditions demonstrated the AADI system’s ability to identify drought-impacted areas with increased confidence up to 6 months earlier than rainfall deficits.