David Ewen Robertson

Predicting streamflow in ungauged basins (PUB) remains a central challenge in hydrology. Long short-term memory (LSTM) networks trained on large samples of catchments ("global" LSTMs) have emerged as a state-of-the-art approach for PUB, outperforming conceptual rainfall–runoff models with traditional regionalisation approaches. However, global LSTMs are spatially agnostic, relying solely on static catchment attributes to differentiate regional hydrological behaviour. This study introduces Regionalised Fine-Tuning (ReFT), a strategy that adapts a pretrained global LSTM to the region surrounding each ungauged target catchment by fine-tuning on a spatially weighted set of donor catchments using an inverse-distance weighting scheme. ReFT is evaluated on 218 catchments from the CAMELS-AUS dataset under a spatial out-of-sample cross-validation framework, comparing two fine-tuning configurations: updating all model parameters versus updating only the prediction head while keeping the recurrent backbone frozen. ReFT improves Nash–Sutcliffe Efficiency relative to the base global LSTM in more than 66 % of catchments, with the largest gains occurring for catchments of moderate baseline performance. The ReFT framework combines the broad process generalisation of large-sample deep learning with the local specificity of regional adaptation, providing an efficient route to improved streamflow predictions in data-sparse regions.

Predicting streamflow in ungauged basins (PUB) remains a central challenge in hydrology. Long short-term memory (LSTM) networks trained on large samples of catchments ("global" LSTMs) have emerged as a state-of-the-art approach for PUB, outperforming conceptual rainfall–runoff models with traditional regionalisation approaches. However, global LSTMs are spatially agnostic, relying solely on static catchment attributes to differentiate regional hydrological behaviour. This study introduces Regionalised Fine-Tuning (ReFT), a strategy that adapts a pretrained global LSTM to the region surrounding each ungauged target catchment by fine-tuning on a spatially weighted set of donor catchments using an inverse-distance weighting scheme. ReFT is evaluated on 218 catchments from the CAMELS-AUS dataset under a spatial out-of-sample cross-validation framework, comparing two fine-tuning configurations: updating all model parameters versus updating only the prediction head while keeping the recurrent backbone frozen. ReFT improves Nash–Sutcliffe Efficiency relative to the base global LSTM in more than 66 % of catchments, with the largest gains occurring for catchments of moderate baseline performance. The ReFT framework combines the broad process generalisation of large-sample deep learning with the local specificity of regional adaptation, providing an efficient route to improved streamflow predictions in data-sparse regions.

Predicting streamflow in ungauged basins (PUB) remains a central challenge in hydrology. Long short-term memory (LSTM) networks trained on large samples of catchments ("global" LSTMs) have emerged as a state-of-the-art approach for PUB, outperforming conceptual rainfall–runoff models with traditional regionalisation approaches. However, global LSTMs are spatially agnostic, relying solely on static catchment attributes to differentiate regional hydrological behaviour. This study introduces Regionalised Fine-Tuning (ReFT), a strategy that adapts a pretrained global LSTM to the region surrounding each ungauged target catchment by fine-tuning on a spatially weighted set of donor catchments using an inverse-distance weighting scheme. ReFT is evaluated on 218 catchments from the CAMELS-AUS dataset under a spatial out-of-sample cross-validation framework, comparing two fine-tuning configurations: updating all model parameters versus updating only the prediction head while keeping the recurrent backbone frozen. ReFT improves Nash–Sutcliffe Efficiency relative to the base global LSTM in more than 66 % of catchments, with the largest gains occurring for catchments of moderate baseline performance. The ReFT framework combines the broad process generalisation of large-sample deep learning with the local specificity of regional adaptation, providing an efficient route to improved streamflow predictions in data-sparse regions.

Predicting streamflow in ungauged basins (PUB) remains a central challenge in hydrology. Long short-term memory (LSTM) networks trained on large samples of catchments ("global" LSTMs) have emerged as a state-of-the-art approach for PUB, outperforming conceptual rainfall–runoff models with traditional regionalisation approaches. However, global LSTMs are spatially agnostic, relying solely on static catchment attributes to differentiate regional hydrological behaviour. This study introduces Regionalised Fine-Tuning (ReFT), a strategy that adapts a pretrained global LSTM to the region surrounding each ungauged target catchment by fine-tuning on a spatially weighted set of donor catchments using an inverse-distance weighting scheme. ReFT is evaluated on 218 catchments from the CAMELS-AUS dataset under a spatial out-of-sample cross-validation framework, comparing two fine-tuning configurations: updating all model parameters versus updating only the prediction head while keeping the recurrent backbone frozen. ReFT improves Nash–Sutcliffe Efficiency relative to the base global LSTM in more than 66 % of catchments, with the largest gains occurring for catchments of moderate baseline performance. The ReFT framework combines the broad process generalisation of large-sample deep learning with the local specificity of regional adaptation, providing an efficient route to improved streamflow predictions in data-sparse regions.

Predicting streamflow in ungauged basins (PUB) remains a central challenge in hydrology. Long short-term memory (LSTM) networks trained on large samples of catchments ("global" LSTMs) have emerged as a state-of-the-art approach for PUB, outperforming conceptual rainfall–runoff models with traditional regionalisation approaches. However, global LSTMs are spatially agnostic, relying solely on static catchment attributes to differentiate regional hydrological behaviour. This study introduces Regionalised Fine-Tuning (ReFT), a strategy that adapts a pretrained global LSTM to the region surrounding each ungauged target catchment by fine-tuning on a spatially weighted set of donor catchments using an inverse-distance weighting scheme. ReFT is evaluated on 218 catchments from the CAMELS-AUS dataset under a spatial out-of-sample cross-validation framework, comparing two fine-tuning configurations: updating all model parameters versus updating only the prediction head while keeping the recurrent backbone frozen. ReFT improves Nash–Sutcliffe Efficiency relative to the base global LSTM in more than 66 % of catchments, with the largest gains occurring for catchments of moderate baseline performance. The ReFT framework combines the broad process generalisation of large-sample deep learning with the local specificity of regional adaptation, providing an efficient route to improved streamflow predictions in data-sparse regions.

Climate change is expected to alter both the mean and variability of precipitation. Hydrological consequences of higher precipitation variability remain less understood than that induced by changes in precipitation mean. This study evaluates the combined effects of changes in annual precipitation mean and variability on runoff and water supply system across south‐east Australia (SEA) through sensitivity experiments and hydrological projections informed by climate change signals from global climate models. Runoff is modeled using three rainfall‐runoff models and water supply system performance is modeled using a concatenated behavior analysis of a virtual reservoir, using climate and streamflow data from 392 catchments across the region. The results show that annual runoff variability will increase but by less than the increase in precipitation variability due to the effects of catchment storage. The higher annual runoff variability will reduce water supply reliability. The impact on regulated systems is smaller as there are storages to buffer the variability in inflow. This is particularly so in this region, which will be much more influenced by the significant projected decline in rainfall and runoff. Median projections, for ∼2060 relative to ∼1990, from hydrological modeling informed by change signals from climate models indicate that mean annual runoff will decline by 18% in the southern part of the region and by 6% in the northern part, and the reliability of water supply systems will fall by 13% and 7% in the south and north respectively.

Accurate streamflow prediction in ungauged basins remains one of the central challenges in hydrology. While deep learning models, particularly long short-term memory (LSTM) networks, have demonstrated strong performance in large-sample rainfall-runoff modelling, current continental-scale implementations are spatially agnostic. They rely on static catchment attributes to encode regional differences, but do not explicitly exploit spatially structured hydrological similarity, limiting their ability to represent local hydro-climatic and geological controls in ungauged settings. In contrast, traditional regionalisation approaches explicitly transfer information from spatially or hydrologically similar donor catchments.We introduce Regionalised Fine-Tuning (ReFT), a method that integrates donor-based regionalisation LSTM by spatially weighting the fine-tuning of a pretrained continental model. Starting from a global model trained across all gauged basins, ReFT adapts the model for each ungauged target catchment by minimising a distance-weighted loss over neighbouring donor catchments. This transforms spatially agnostic LSTMs into spatially aware regional specialists while preserving the broad process representations learned from large-sample training.The method is evaluated using 222 Australian catchments from CAMELS-AUS under a spatial out-of-sample cross-validation framework. Two LSTM variants are tested: a climate-only model and a hybrid model that also uses runoff from the Australian Water Resources Assessment - Landscape (AWRA-L) land surface model as a dynamic predictor. ReFT-LSTMs are compared against a continental LSTM without fine-tuning, a regionalised GR4J conceptual model, and AWRA-L.Results show that ReFT produces systematic improvements across most of the Nash–Sutcliffe Efficiency distribution, with median and upper-quantile skill exceeding both the continental LSTM and traditional benchmarks. In most catchments, ReFT-LSTMs outperform both AWRA-L and regionalised GR4J, demonstrating that spatially weighted fine-tuning can effectively substitute for local calibration data in ungauged basins. Remaining weaknesses are confined to specific challenging regions, where conceptual model constraints remain advantageous. Overall, ReFT provides a principled pathway to reconcile large-sample deep learning with regional hydrological realism for PUB applications.

Human activity has significant impacts on the river systems. In-stream infrastructure, such as dams and weirs, creates artificial barriers that impede fish passage during periods of low flow; and off-stream structures, such as agricultural farm dams, intercept and extract water. These structures reduce inflow to rivers and extend the frequency and duration of low flow conditions. While the impacts of water interception by farm dams on long-term streamflow have previously been investigated, the consequences for frequency and duration of low flow periods, are poorly understood. In this study, we estimated the influence of farms dams on annual streamflow and on low flow requirements that ensure water flows over in-stream weirs of different heights for the Moorabool River catchment located in southeast Australia. Streamflow in the Moorabool River was simulated using the hydrological model Genie Rural a 4 parametres Journalier (GR4J) + CHEAT1 to assess the potential impact of farm dams on low flows, using data between 1980 to 2020. The spatial and temporal distribution of farm dams and their storage capacity was estimated using remote sensing. Low flow requirements were calculated measuring cease-to-flow conditions in three weirs in the river, and a low flow spell analysis was conducted to assess the impact of farm dams. From 1990-2020, the number of farm dam increased by 163% across the catchment. Including farm dams and water extractions in the hydrological model increased its performance as assessed by the NSE and logNSE. Streamflow reduction was estimated to be between 3% to 65%, contingent on the level of water extractions and number of dams in the catchment. The cumulative impact of farm dams resulted in fewer high flow events and low flows of longer duration. Prevalent low flows would potentially affect the ability of water to go over some weirs. This study presents a comprehensive approach to quantify water resources by including farm dams and measuring cease-to-flow conditions over weirs in the hydrological model. Ongoing hydrological assessments provides a holistic estimate of water resources and determining objective minimum flow requirements to adequately overtop small weirs in a catchment. Keywords: farm dams, hydrological model, low flows, river barriers.

Accurate large-scale hydrological predictions are essential for water resource planning. However, many land surface models encounter difficulties in capturing streamflow timing and magnitudes, particularly in large catchments and when calibrated across broad regions and multiple hydrological variables. In this study, two Long Short-Term Memory (LSTM)-based approaches are assessed to enhance streamflow predictions across Australia: (i) LSTM-C, a standalone rainfall–runoff LSTM that relies solely on precipitation and potential evapotranspiration as inputs, and (ii) LSTM-QC, a rainfall–runoff LSTM that incorporates runoff outputs from the Australian Water Resources Assessment–Landscape model (AWRA-L), which can also be interpreted as a post-processor for AWRA-L. These approaches are tested in 218 minimally impacted catchments from the CAMELS-AUS dataset under three cross-validation strategies – temporally out-of-sample, spatially out-of-sample, and spatiotemporal out-of-sample – to evaluate their robustness for historical reconstructions, predictions in ungauged basins, and a proxy for climate-projection scenarios. The results indicate that both LSTM-QC and LSTM-C consistently outperform AWRA-L runoff across nearly all catchments and exceed the predictive skill of a widely used conceptual model (GR4J) in most basins. Under a temporally out-of-sample framework, LSTM-QC demonstrates a performance advantage over LSTM-C by leveraging information embedded in AWRA-L, particularly when fine-tuned to local catchment observed data. This advantage is primarily attributed to the LSTM's ability to correct systematic biases in AWRA-L and enhance channel-routing signals. However, under spatial and spatiotemporal cross-validation LSTM-C performs comparably well, suggesting that a purely data-driven approach can generalize effectively to ungauged or future conditions without reliance on AWRA-L.