The study, published in Hydrology and Earth System Sciences, relies on daily streamflow records from 1982 to 2020 at USGS gauges compiled in the GAGES-II dataset. After filtering for data completeness—requiring at least eight years of observations per decade with 95 percent of days recorded—the team settled on 3,198 gauges. To define drought objectively, they converted each day’s flow into a percentile using the Weibull plotting position, computed within a 30-day window centered on that calendar day across all years. That trick removes the seasonal cycle, so drought becomes any departure from what a river normally does at that time of year. Anything below the 20th percentile was flagged as drought, a threshold chosen to balance a usable definition against enough events to train a statistical model. Zero-flow days, common in arid and intermittent streams, were handled with a combined threshold-level and continuous-dry-period method that ranks consecutive dry days rather than treating them as ties.

For each gauge, the researchers built a random forest classifier fed with 130 candidate predictors: precipitation, soil moisture at multiple depths, potential evapotranspiration, snow water equivalent, temperature, the Standardized Precipitation Evapotranspiration Index, large-scale climate teleconnections such as El Niño–Southern Oscillation, the Pacific Decadal Oscillation, the Atlantic Multidecadal Oscillation, and the Pacific–North American pattern, plus even sunspot counts. Crucially, each variable entered the model in several forms—raw values, percentile-transformed values, and rolling averages over 30, 90, and 365 days—so the forests could learn which timescales of antecedent conditions matter most. Random forests were chosen for their resistance to overfitting, their tolerance of correlated predictors, and their built-in variable importance scores, which formed the analytical heart of the study. Models were trained on 1985–2015 and tested on the bookend periods 1982–1984 and 2016–2020, and scored with Cohen’s Kappa, a metric robust to the severe class imbalance created when drought occupies only 20 percent of the record.

The single most influential predictor nationwide turned out to be soil moisture at 40–100 centimeters depth, followed by soil moisture at 10–40 centimeters, then a trio of precipitation variables smoothed over 90 and 365 days and the 90-day SPEI. Nothing else came close. But the real story lay in the geography. Potential evapotranspiration dominated in the South, Southwest, and Southeast; snow water equivalent mattered most in the Southwest, Northern Rocky Mountains, Northwest, and West; and the teleconnection indices, while weak at individual gauges, carried outsized weight across the western half of the country. Deep soil moisture was especially important in humid regions, which the authors attribute to hydrologic memory: deficits stored deep in the soil profile sustain low flows long after a passing rainstorm has improved surface conditions, resisting rapid drought recovery.

To transform thousands of gauge-level importance scores into a coherent national picture, the team applied principal component analysis, compressing 130 variables into 27 components that together explained 95 percent of the variance. The first three components told most of the story. The first aligned teleconnections, temperature, and evaporative demand—a cluster centered on the West, Southwest, and Northern Rockies, and notably the same regions where the at-site models performed worst, largely because water management alters flows in ways the data cannot capture. The second component was a moisture axis pitting long-term soil moisture against short-term precipitation, cleanly dividing the humid East from the arid West. The third distinguished snow-driven basins in the mountains and northern latitudes from baseflow-driven, soil-moisture-dominated systems in the South and Southeast, drawing the map along the line between seasonal snowpack storage and subsurface storage.

The regressions linking these components to static basin characteristics revealed why certain regions respond to certain drivers. Basins where teleconnections and temperature loomed large tended to be warmer, rainier, forested or cultivated, and equipped with permeable soils but limited buffering capacity—meaning they lack the soil or snow storage to insulate streamflow from large-scale climate swings. Basins scoring high on the snow-and-soil-moisture component were characterized by high elevation, a large fraction of precipitation falling as snow, and cold freeze timing, consistent with evidence that earlier, slower snowmelt reduces streamflow production. In effect, the study shows that a basin’s physiography—its soils, topography, land cover, and snow regime—determines which meteorological levers can push it into drought, and how hard those levers must be pulled.

The second half of the study tackles a long-standing gap: roughly one-fifth of the nation’s land area drains through watersheds too small, too high, or too remote to appear in the gauge network, and the gaps are not randomly distributed. The team’s solution is a novel dynamic regionalization scheme built on donor gauges. For any hypothetical ungauged location, principal component scores are estimated from its physical characteristics using regressions trained on the 1,900 gauges whose models passed a Kappa threshold of 0.4. Each candidate donor is then scored by how closely its component profile matches the target, weighted by the variance each component explains. The 19 most similar donors—regardless of whether they sit across the country or across the county—are enlisted, their random forest models run on the target’s climate data, and their drought probability predictions averaged and thresholded at 0.35 to yield a final daily drought forecast.

The results were striking: donor-based predictions achieved a mean Kappa of 0.40 across all sites, only 0.02 below the at-site models trained locally on that same gauge’s own data. Even more intriguingly, the ensemble sometimes outperformed the local models, particularly where those models were weak—likely because averaging 19 similar gauges suppresses the noise that a single model overfits. Two worked examples illustrate the method’s flexibility. For Brodhead Creek in Pennsylvania, the selected donors averaged just 219 kilometers away and agreed tightly in their predictions. For the Gunnison River in Colorado, the nearest characteristic donors sat an average of 2,891 kilometers away and disagreed more, yet the ensemble still matched the local model’s accuracy almost exactly. Similarity, not proximity, is what counts.

Beyond its practical value for water managers, the framework offers a partial answer to the black-box criticism that haunts machine learning in high-stakes applications. Because the workflow is compartmentalized—drought definition, random forest training, dimensionality reduction, donor regression, ensemble prediction—a modeler can inspect and adjust each stage, swapping donor sets and observing how predictions change to gauge each donor’s influence. The authors note that tools such as partial dependence analysis or SHAP values could further disentangle the overlapping drivers, especially in the West, where multiple drought typologies stack on top of one another, and they caution that a season-specific modeling approach could better separate snowmelt deficits in spring from evaporative-demand-driven drought in late summer.

The timing of such a tool could hardly be better. Under continued warming, rising evaporative demand and shrinking snowpack are expected to amplify the importance of temperature and energy-balance drivers in the West and Northern Rockies, while shifting precipitation seasonality could redraw the driver map in the East. The drought typology established here provides a baseline against which those future shifts can be measured—and, because the donor method needs no local streamflow record, it extends early-warning capability to the small headwater basins and high-elevation catchments that supply so much of the nation’s water but that gauges have long overlooked.

Subject of Research: Machine learning prediction of streamflow drought and its meteorological drivers across the conterminous United States

Article Title: Predicting streamflow drought in the conterminous United States using machine learning and a donor-gage approach, 1982–2020

Article References: Heldmyer, A., Sando, R., Simeone, C., Wieczorek, M., Hamshaw, S., Goodling, P., McShane, R., Diaz, J., Watkins, D., Pulver, B., Shastry, A., Hafen, K., & Hammond, J. (2026). Predicting streamflow drought in the conterminous United States using machine learning and a donor-gage approach, 1982–2020. Hydrology and Earth System Sciences, 30(18), 5925-5945. https://doi.org/10.5194/hess-30-5925-2026

Image Credits: AI Generated

DOI: 10.5194/hess-30-5925-2026

Keywords: streamflow drought, machine learning, random forest, hydrology, donor gage, ungauged basins, snow water equivalent, soil moisture, teleconnections, USGS, drought prediction, principal component analysis

The meteorological setup was extraordinary. Remnant moisture from Tropical Storm Barry, which had made landfall in Mexico days earlier, was drawn northward into a persistent mid-level trough, where a mesoscale convective vortex and a strengthening low-level jet fueled successive back-building thunderstorms over the Hill Country. Between 3 and 6 July, parts of the upper Guadalupe basin accumulated more than 500 millimeters of rain, with localized totals approaching 460 millimeters on 4 July alone. Radar-based estimates and rain gauges showed hourly rates reaching 100 millimeters per hour between 02:00 and 05:00, and rainfall intensities surpassed 100-year return period thresholds for durations from 3 to 24 hours when compared against NOAA Atlas 14 frequency curves.

The landscape itself amplified the danger. The region’s steep slopes and clay-rich soils, which have earned it the nickname Flash Flood Alley, convert intense rainfall into runoff with brutal efficiency. USGS stream gauges recorded hydrographs that defied belief: at Hunt, downstream of the confluence of the North Fork and South Fork of the Guadalupe River, flow reached roughly 8,261 cubic meters per second at 05:05, far exceeding the 500-year return period flow of about 4,161 cubic meters per second. At Kerrville, the peak of about 8,316 cubic meters per second crossed the 500-year threshold of 7,938. Most striking was the speed: both gauges climbed from ordinary conditions to the 500-year level within just one to one and a half hours.

Crucially, the evidence points to the ungauged South Fork Guadalupe River as the dominant contributor to the extreme peak at Hunt. The North Fork gauge never exceeded its 25-year return period flow, a difference that cannot be explained without a massive inflow from the South Fork, the catchment that drains the area around Camp Mystic, where many of the fatalities occurred. Because no operational USGS stream gauge existed on the South Fork at the time, this contribution cannot be directly verified from in-situ observations, but the timing of the peak and the spatial distribution of rainfall strongly support the conclusion.

To understand why forecasts missed the event, the team examined the National Water Model short-range streamflow forecasts, which provide hourly predictions out to an 18-hour range and are driven by the High-Resolution Rapid Refresh model’s 3-kilometer quantitative precipitation forecasts. Using the FIMserv tool, they generated 306 forecasted flood inundation maps between 3 and 4 July. The comparison with observations was stark. Early forecasts showed no flooding at North Fork, and only forecasts generated between 04:00 and 07:00 on 4 July jumped to flows of 120 to 600 cubic meters per second. At Hunt, the 04:00 forecast suggested flow crossing 500 cubic meters per second, while the observed peak was more than sixteen times larger. At Kerrville, no high-flow signal appeared at all until the 06:00 forecast, and at Comfort the first meaningful indication came only after 10:00.

The authors identify two primary sources of this failure. First, the HRRR model misplaced and underestimated the convective rainfall. Successive forecast cycles placed the convective core in different locations across the upper Guadalupe basin, a spatial inconsistency the researchers call a changing signal location, so the National Water Model produced runoff predictions that shifted from one basin to another with each update. This behavior is consistent with documented limitations of the HRRR in warm-season convection, including a tendency to dissipate nocturnal mesoscale convective systems too rapidly, displacement errors of intense precipitation cells on the order of 100 to 150 kilometers, and a two-to-four-hour lag in the diurnal evolution of convective available potential energy. For small basins under 1,000 square kilometers, which are precisely the ones prone to flash flooding, such errors translate directly into underpredicted and mistimed flood peaks.

The second failure was mechanical and equally consequential. The USGS gauge near Hunt failed during the fast-rising limb of the flood, and the peak of roughly 315,000 cubic feet per second, about 8,910 cubic meters per second, is recorded as an estimated rather than observed value, reconstructed after the event from surveyed high-water marks. The National Water Model constrains its forecasts through streamflow nudging in its analysis and assimilation cycle, adjusting simulated discharge toward gauge observations and propagating the correction downstream through Muskingum-Cunge routing. With the Hunt gauge offline, the incoming flood wave entered the routing network without observational constraint, allowing rainfall-driven errors to accumulate downstream and progressively degrade both discharge forecasts and the inundation maps derived from them.

To evaluate the maps themselves, the team built a benchmark flood extent from USGS high-water marks, filtering the survey data for quality, removing global and spatial outliers using interquartile range and Local Moran’s I statistics, and interpolating a continuous water surface with a hydrologically conditioned algorithm before subtracting a 10-meter digital elevation model. Against this benchmark, the forecast maps performed poorly until very close to the event. Near Camp Mystic, where the benchmark indicated 118 affected buildings, the 04:00 forecast captured only 18 and the 05:00 forecast 72. At Hunt, the benchmark showed 133 buildings impacted against just 2 in the 04:00 forecast. At Kerrville, the benchmark counted 140 flooded buildings while the 06:00 forecast showed one. At Comfort, 188 buildings were flooded according to the benchmark, but no buildings appeared in forecasts until 13:00, when 25 were predicted. Notably, two locations with similar categorical skill scores experienced very different levels of impact, underscoring the authors’ argument that skill metrics alone cannot convey flood consequences.

The study closes with concrete prescriptions. The authors call for complementary observations beyond the telemetered USGS network, including dense networks of low-cost water-level sensors, bridge-mounted radar or acoustic flood sensors, and camera-based discharge estimation, since no single alternative can match the sub-hourly resolution needed for a flood that exceeded the 500-year threshold in under 90 minutes. They see particular promise in taskable commercial satellite radar constellations, pre-configured to image flash-flood-prone basins when heavy rain outlooks are issued, and in probabilistic flood inundation mapping that replaces binary flood-or-no-flood outputs with exceedance probabilities tied to predefined emergency action triggers, such as precautionary road closures at moderate probability over campsites or evacuation preparation at high probability over residential zones. The authors also caution that forecast quality is only one component of an effective early warning system, with warning dissemination, interpretation, and evacuation timing all shaping outcomes. Their analysis stands as both a technical diagnosis and a blueprint for making the next flash flood in Flash Flood Alley less lethal.

Subject of Research: Post-event evaluation of operational flood forecasting performance during the 4 July 2025 Guadalupe River flash flood in Central Texas

Article Title: Brief communication: Rise of the Guadalupe River – a multifaceted post event analysis of 4 July 2025, flood in Central Texas

Article References: Baruah, A., Munasinghe, D., Cohen, S., Abdelkader, M., Devi, D., Chen, Y., Vergara, H., & McDermott, R. (2026). Brief communication: Rise of the Guadalupe River – a multifaceted post event analysis of 4 July 2025, flood in Central Texas. Natural Hazards and Earth System Sciences, 26(10), 4753-4762. https://doi.org/10.5194/nhess-26-4753-2026

Image Credits: AI Generated

DOI: 10.5194/nhess-26-4753-2026

Keywords: flash flood, Guadalupe River, Texas Hill Country, National Water Model, flood inundation mapping, HRRR, data assimilation, high-water marks, early warning systems, NOAA, USGS, probabilistic forecasting

The problem the researchers set out to solve is one that hydrologists and machine learning engineers have been circling for years. Deep learning models have repeatedly proven that they can forecast river levels and streamflow with impressive accuracy, often outperforming classical hydrological simulation in data-rich settings. But accuracy alone is not enough for early warning. Many state-of-the-art networks carry enormous parameter counts and require long inference times, meaning the interval between feeding in the latest gauge readings and receiving a prediction can stretch unacceptably long. At the same time, standard recurrent architectures tend to smooth over exactly the signals that matter most during a disaster: the abrupt, nonlinear jumps in water level that mark the onset of an extreme flood event. A model that averages away a flood peak is, for warning purposes, worse than useless.

BOA-LSTM attacks both weaknesses simultaneously through three coordinated design choices. First, the authors deliberately reduce the number of LSTM layers and the number of hidden units within each layer. Long short-term memory networks, introduced by Sepp Hochreiter and Jürgen Schmidhuber in 1997, use gated cells to preserve information over long sequences, but stacking many such layers multiplies parameters and computation. By trimming the architecture to the minimum needed for hydrological time series, the team cut computational complexity and slashed inference time, the critical metric for real-time deployment on modest hardware at remote monitoring stations.

Shrinking a network, however, risks throwing away its sensitivity to subtle patterns, so the second design choice compensates by adding a scaled dot-product attention mechanism. Attention layers, which rose to fame in natural language processing, allow a model to dynamically weight the most informative parts of an input sequence rather than treating all time steps equally. In the flood forecasting context, this means the network can amplify sudden rises in discharge or water level in the recent past, the abrupt hydrological changes that precede extreme events, instead of letting them dissolve into the longer record of routine variability. The attention weights effectively teach the model where to look when a river begins to misbehave.

The third pillar is Bayesian optimization, a sample-efficient strategy for tuning hyperparameters such as learning rates, hidden unit counts, and window sizes. Traditional hyperparameter search relies on grid or random searches that evaluate huge numbers of configurations, an expensive process for any deep learning pipeline. Bayesian optimization, famously formalized for machine learning by Snoek, Larochelle, and Adams in 2012, builds a probabilistic surrogate of the performance landscape and intelligently selects which configurations to test next. In BOA-LSTM this automates the search for key hyperparameters, improving configuration efficiency and removing a significant degree of manual trial and error from the modeling workflow, which matters for agencies that cannot employ teams of specialists to hand-tune every new river basin.

The empirical results are the heart of the study’s claim. Using the USGS 02337000 benchmark dataset, a river gauging record from the United States Geological Survey network, the team showed that BOA-LSTM achieves a mean squared error of 0.000295 with an inference time of 124 seconds. Benchmarked against FAIRDNN, a comparable deep learning approach, the lightweight model delivered lower overall prediction errors, a notably better representation of flood peaks, and substantially higher inference efficiency. The numbers are striking: mean squared error fell by 4.8 percent while inference time dropped by 81.8 percent. In practical terms, the model is simultaneously more accurate and roughly five times faster than its competitor, a rare combination in machine learning where speed and precision usually trade off against each other.

The improved flood-peak representation deserves particular emphasis, because peaks are where forecasting models most often fail and where failure is most costly. Extreme events are by definition rare, so they are underrepresented in training data, and loss functions dominated by routine conditions can teach a network to systematically underestimate the highest stages. By reducing overall error while sharpening the model’s attention to abrupt changes, the architecture appears to preserve precisely the tail behavior that converts a generic water level forecast into an actionable warning. For emergency managers, an underpredicted crest of even a few centimeters can mean the difference between a precautionary evacuation and a rescue operation.

To demonstrate that the approach is not tied to a single American gauging station, the researchers conducted a case study using 2024 water level observations from Cambridgeshire in the United Kingdom. This second dataset exercised the full operational pipeline on data from a different hydrological setting, including the warning classification process in which predicted levels are translated into graduated alert categories. The case study, the authors report, demonstrates the model’s applicability to other stations and provides a reference workflow for how raw predictions become graded warnings, the step where machine learning output meets the public communication machinery of civil protection agencies.

The broader context makes the work timely. Flood risk is intensifying in many regions as climate change amplifies rainfall extremes and as development pushes populations into floodplains, while the scientific literature has seen a rapid proliferation of attention-enhanced and Bayesian-assisted forecasting models, from dual-stage attention LSTMs for multi-step flood prediction to hybrid architectures for data-scarce mountain catchments and Bayesian deep learning frameworks for uncertainty estimation. What distinguishes BOA-LSTM within this crowded field is its explicit commitment to lightweight deployment, treating inference speed and parameter economy as first-class objectives rather than afterthoughts. That orientation reflects a growing recognition that a forecasting model earns its keep only when it runs reliably on the infrastructure actually available in the field.

Caveats remain, as they do with any data-driven forecasting system. The model’s performance was established on benchmark and case study datasets rather than through live operational trials during an actual flood emergency, and the published data availability statement indicates that no new datasets were generated or analyzed during the study beyond those used in the experiments. Deep learning forecasters also inherit the limitations of their training records: under nonstationary climate conditions, tomorrow’s extremes may look statistically unlike yesterday’s, a challenge the wider hydrology community continues to grapple with. Even so, the study offers a concrete demonstration that flood warning need not choose between intelligence and immediacy. With an 81.8 percent reduction in inference time and improved error performance, BOA-LSTM suggests that the next generation of flood early warning systems may be both smarter and lighter, bringing life-saving minutes to communities that live downstream of the world’s increasingly volatile rivers.

Subject of Research: A lightweight LSTM model using Bayesian optimization and attention mechanisms for real-time extreme flood early warning

Article Title: BOA-LSTM: a lightweight LSTM model integrating bayesian optimization and attention mechanism for real-time extreme flood early warning

Article References: Zhang, Y., Zhou, M., Chen, M., & Jia, H. (2026). BOA-LSTM: a lightweight LSTM model integrating bayesian optimization and attention mechanism for real-time extreme flood early warning. International Journal of Machine Learning and Cybernetics, 17(10), Article 477. https://doi.org/10.1007/s13042-026-03313-z

Image Credits: AI Generated

DOI: 10.1007/s13042-026-03313-z

Keywords: flood early warning, BOA-LSTM, LSTM, Bayesian optimization, attention mechanism, machine learning, hydrology, time series forecasting, deep learning, USGS, flood prediction, real-time systems