The study, authored by Hongye Cao of Xianyang Normal University and the China Jikan Research Institute of Engineering Investigation and Design, presents a model named VMD-DSTCN-GA, which stands for a variational mode decomposition-based dual-stream temporal convolutional network with gated attention. Evaluated on thirty years of daily hydro-meteorological observations from the Jingcun hydrological station on China’s Jing River, the framework achieved a coefficient of determination of 0.9872 and a Nash–Sutcliffe efficiency of 0.9631 during the independent evaluation period from 2010 to 2019, outperforming a suite of established benchmark models including deep learning hybrids and the widely used physically based SWAT model.

The central innovation of the work lies in how it disentangles two fundamentally different kinds of signal. Rivers possess a kind of internal dynamic memory: water stored in the subsurface, in snowpack and in channel banks is released slowly, producing smooth, slowly varying components of flow. Superimposed on this are abrupt responses to external forcing, when a rainstorm delivers a pulse of energy and water to the catchment and the hydrograph spikes within hours. Conventional single-stream neural networks must learn both behaviours simultaneously from raw inputs, which often blurs the distinction between the two regimes. The new architecture instead decomposes the historical runoff record using variational mode decomposition, a signal-processing technique that adaptively splits a time series into a set of frequency-specific sub-series, or modes, each capturing oscillations at a characteristic scale.

Variational mode decomposition, first formalised by Dragomiretskiy and Zosso in 2014, differs from classical empirical decomposition methods by framing the decomposition as a variational optimisation problem, seeking the set of modes whose sum reproduces the input signal while each mode remains narrow-banded around its own centre frequency. This makes it considerably more robust to noise and mode mixing than older approaches, a property that matters greatly in hydrology, where observed flows carry measurement error and the underlying signal is anything but stationary. In the new framework, the decomposed runoff components are fed into what the author calls a runoff-state stream, a temporal convolutional network built from causal dilated convolutions and residual blocks.

Temporal convolutional networks have been gaining ground on the long short-term memory (LSTM) architectures that dominated hydrological machine learning for much of the past decade. Where LSTMs process sequences step by step through gated recurrent units, temporal convolutional networks apply one-dimensional convolutions across the time axis, using dilated kernels to expand their receptive field exponentially with network depth. The causal design ensures that predictions at any time step depend only on past information, avoiding future leakage, while residual connections stabilise training in deep stacks. The practical advantages are considerable: convolutions can be computed in parallel across the entire input window, making training dramatically faster, and the hierarchical receptive field allows the network to capture dependencies operating at multiple timescales, from the daily rhythm of rainfall events to the seasonal pulse of snowmelt.

The second stream of the network is dedicated entirely to meteorological forcing. Precipitation, maximum and minimum temperature, solar radiation, relative humidity and wind speed are processed through an independent encoder, so that the atmospheric drivers of runoff are represented in their own feature space rather than being forced to share a representation with the river’s internal state. Only after both streams have produced their encoded features are they combined, and the combination is far from a simple concatenation. A gated fusion mechanism learns, for each time step and each feature channel, how much weight to assign to the runoff-state representation versus the meteorological representation, effectively letting the model decide dynamically whether the river’s own memory or the prevailing weather matters more at any given moment.

On top of this gated fusion sits a temporal attention module, which reweights the contributions of different time steps in the input history, allowing the network to focus on the days that matter most for the forecast, such as the immediate aftermath of a storm. The authors also introduce a peak-sensitive loss function, deliberately penalising errors on high-flow events more heavily than errors during low-flow periods. This addresses a chronic weakness of machine learning hydrology models, which, trained on ordinary mean-squared error, tend to fit the abundant mid-range flows well and systematically underestimate the extreme peaks that matter most for flood risk.

The evaluation protocol was deliberately stringent. The model was trained on daily data from 1990 to 2009 at the Jingcun station and tested on the entirely withheld decade from 2010 to 2019. Against observed flows, VMD-DSTCN-GA recorded an R² of 0.9872, a root mean square error of 7.3734 cubic metres per second, a percent bias of 13.0314 percent, and a Nash–Sutcliffe efficiency of 0.9631, the latter being a standard measure in hydrology where values above roughly 0.75 are generally considered very good and values above 0.9 exceptional. Among all models evaluated, the new framework achieved the highest R², indicating the strongest ability to reproduce the full range of observed runoff variability.

The comparison with benchmarks was informative rather than one-sided. A CNN–LSTM–Attention hybrid achieved a slightly lower root mean square error of 6.6113 cubic metres per second and the highest NSE of 0.9703, while the physically based SWAT model produced the smallest absolute percent bias, reflecting its grounding in water-balance physics. These results suggest a nuanced picture: the proposed dual-stream architecture excels at capturing the shape and variability of the hydrograph, while purely physics-driven approaches retain an advantage in reproducing total volumes. When the same framework was applied at monthly resolution, its performance improved further, yielding an R² of 0.9929, an NSE of 0.9660 and an RMSE of 6.1879 cubic metres per second, a finding consistent with the general observation that aggregation smooths daily noise and makes underlying dynamics easier to learn.

The Jing River basin, a major tributary of the Yellow River, provides a demanding test case. The basin is subject to pronounced hydrological droughts whose propagation from meteorological drought has intensified under environmental change, and its semi-arid to semi-humid climate produces highly variable flows with episodic floods. Data for the study were drawn from the Chinese Hydrological Yearbook of the Yellow River Basin and the China National Meteorological Information Center, spanning the full suite of variables a modern forecasting system would need in operation.

The significance of the approach extends beyond one basin. Signal decomposition combined with machine learning has become one of the most active fronts in hydrological forecasting research, with recent studies pairing wavelet methods, CEEMDAN and empirical mode decomposition variants with gradient boosting, LSTMs and other learners. What distinguishes the new work is the architectural separation of internal state and external forcing, which mirrors how hydrologists conceptually understand catchment behaviour, and the explicit attention to peak flows through the loss function. In effect, the model encodes domain knowledge about the physics of runoff generation into its structure rather than hoping a sufficiently large generic network will discover it from data alone.

The author is candid about the framework’s limitations. A systematic volume bias of just over thirteen percent persists, meaning the model tends to misestimate the total water passing the gauge even as it tracks the timing and shape of flow variations closely. More importantly, the study evaluated the model at a single station; whether the architecture transfers across basins with different geology, land cover and climate remains an open question that will be essential for real-world deployment. Generalisation, or the lack of it, has long been the dividing line between models that impress in benchmarks and models that serve water managers.

Nevertheless, the results arrive at a moment when the demand for accurate streamflow prediction is intensifying. Climate change is amplifying hydrological extremes in many of the world’s major basins, stressing water allocation systems designed around the statistics of a more stable past. Hybrid frameworks that combine signal processing, deep learning and physically informed architectures offer a pragmatic path forward, and the demonstration that a dual-stream, attention-equipped temporal convolutional network can reach NSE values above 0.96 on out-of-sample daily data marks a genuine step in that direction. The work was supported by several Chinese research programmes, including projects from Sinomach Group, Xianyang City and Chang’an University’s Fundamental Research Funds, and the full technical details, along with supplementary material, are available in the journal article.

The Mesopotamian Marshes—designated a UNESCO World Heritage Site—support distinctive ecosystems and long-standing cultural landscapes. After the marshes were reflooded in 2003, researchers expected recovery to be influenced mainly by climate variability. Instead, the study points to human-controlled water dynamics upstream as the main lever controlling marsh health.

Using satellite-derived measures of vegetation and surface water from 2000 to 2023, the team integrated multiple Earth observation sources, including NASA MODIS, EU Copernicus Sentinel-2, and the JRC Global Surface Water dataset. These remotely sensed indicators were evaluated alongside streamflow and precipitation records, allowing the researchers to separate rainfall-driven effects from hydrological impacts linked to dam operations and regional management.

The analysis reveals that vegetation and water generally increased over the study period, but recovery was repeatedly interrupted by steep declines. The most severe drops occurred during 2008–2009 and again in 2022–2023, aligning with periods of regional drought and altered upstream operations.

Critically, river flow emerged as the dominant factor. Higher streamflow was strongly associated with healthier vegetation, while local precipitation did not show a measurable influence on wetland extent. This suggests that the marsh system responds primarily to whether enough water actually arrives from the Tigris and Euphrates, rather than to how much rain falls directly over the wetlands.

The study also uncovers a feedback mechanism involving temperature. Marsh vegetation correlated with cooler land surface temperatures, while vegetation loss was followed by warming. The authors interpret this as a positive feedback loop: when vegetation declines, the wetland surface warms, which can further stress the ecosystem and hinder regrowth.

Overall, the results strengthen the case for transboundary water cooperation among countries sharing the Tigris–Euphrates river system. During droughts, maintaining environmental flows may be essential to prevent cascading ecological losses and preserve the marshes’ climate-moderating role.

Lead author Akram Alqaraghuli emphasizes that sustaining river flows is vital for both ecosystem conservation and regional temperature regulation. Co-author Peter North notes that long-term satellite records make it increasingly possible to distinguish human intervention from climate effects, providing evidence directly relevant to management decisions.

Subject of Research: Not applicable
Article Title: The impact of climate factors and surface water management on the Mesopotamian Marshes for the period 2000–2023
News Publication Date: 15-Jul-2026
Web References: https://www.nature.com/articles/s41598-026-61808-9
References: 10.1038/s41598-026-61808-9
Image Credits: Akram Alqaraghuli

Keywords: Mesopotamian Marshes, Iraq, Tigris and Euphrates, satellite observations, vegetation, surface water, river flow, drought, dam operations, land surface temperature

The research centers around three major twentieth-century tropical volcanic eruptions known for their high volcanic explosivity indices (VEI ≥5) — namely, the 1963 Agung eruption in Indonesia, the 1902 Santa Maria eruption in Guatemala, and the 1991 Pinatubo eruption in the Philippines. These events were chosen not only because of their substantial injections of aerosols into the stratosphere but also due to the distinct patterns in which their aerosol plumes were distributed across the hemispheres. Agung’s aerosols predominantly settled over the Southern Hemisphere, Santa Maria’s primarily affected the Northern Hemisphere, while the Pinatubo eruption’s plume was more evenly distributed across both hemispheres. This natural experiment allowed scientists to isolate and examine the flood responses driven by asymmetrical versus symmetrical aerosol forcings.

Leveraging state-of-the-art climate models capable of simulating coupled atmosphere–land–ocean processes, researchers reconstructed seasonal precipitation and temperature patterns following these eruptions. They then statistically linked these climatic variables to observed peak discharges at 7,886 river gauges globally. This innovative approach bridged two complex domains—volcanology-driven climate perturbations and hydrology—that rarely intersect with such spatial comprehensiveness. The results are striking: the hemispheric distribution of volcanic aerosols strongly modulates flood responses, producing contrasting signals in peak discharge patterns that hinge on both latitude and regional climatic context.

For eruptions with pronounced hemispheric asymmetry in aerosol loading, notable interhemispheric contrasts in flood behavior emerged. In the hemisphere where the eruption dispersed the majority of its aerosols, flood magnitudes generally decreased, while in the opposite hemisphere, flood magnitudes tended to increase. This pattern was especially apparent in tropical regions, which responded more rapidly and intensely to volcanic forcing compared to temperate and high-latitude zones. Such findings suggest that volcanic aerosols disrupt the regional hydrological cycles differently across the hemispheres, potentially through modulations of monsoon systems, shifts in precipitation bands, and alterations in local evaporation rates.

The Agung 1963 eruption exemplifies this pattern, as its southern hemispheric aerosol burden led to a widespread decline in seasonal peak river discharges within tropical regions of the Southern Hemisphere. Conversely, the Northern Hemisphere tropics experienced a rise in peak discharges during the analogous post-eruption period. This hemispheric dichotomy indicates that the volcanic aerosol layer may impose a form of climatic “see-saw” effect, perturbing atmospheric circulation in a way that redistributes precipitation anomalies across the equator, thereby shaping flood risks in counterintuitive ways.

In contrast, the Santa Maria 1902 eruption projected most of its stratospheric aerosols into the Northern Hemisphere, triggering the inverse hydrological response. Northern tropical basins witnessed declining peak flows, while their southern counterparts exhibited increased flood magnitudes. Such a response underscores the crucial role of aerosol placement in dictating downstream flood patterns, emphasizing the need for precise aerosol dispersal characterization in eruption forecasts and climate impact assessments.

The 1991 Pinatubo eruption, which injected aerosols fairly symmetrically into both hemispheres, revealed a different but equally illuminating scenario. Here, the response was more spatially uniform: tropical regions across both hemispheres predominantly experienced reductions in peak river discharges. Meanwhile, arid or semi-arid regions tended to exhibit the opposite response, with increased peak flows following the eruption. This dichotomy suggests that volcanic aerosols’ climatic effects are modulated by local climate regimes—moist tropical environments respond almost uniformly with drying-related flood reductions, whereas water-limited arid landscapes may paradoxically face elevated flood risks, potentially due to episodic intense rainfall events or altered runoff dynamics.

Underlying these hydrological shifts are tightly coupled changes in seasonal precipitation patterns. The study’s analysis confirms that most of the flood responses stem from modifications in the timing and intensity of rainy seasons induced by volcanic aerosol forcings. Aerosol-cloud interactions, shifts in monsoon intensity, and perturbations of large-scale atmospheric circulation collectively realign precipitation distributions. These processes consequently ripple through river basins, amplifying or dampening flood peaks depending on location. Understanding these mechanistic links is vital for accurate forecasting and risk management of secondary volcanic hazards.

This research also advances the scientific narrative regarding volcanic eruptions’ role as natural experiments in earth system science. The global flood responses they incite serve as moving probes into the complex interplay between aerosols, climate dynamics, and hydrology. Unlike gradual anthropogenic climate change, volcanic eruptions induce abrupt, sharp alterations that can test the resilience and response capacity of hydrological systems worldwide on seasonal to decadal timescales.

Moreover, this work carries significant implications for disaster preparedness and infrastructure resilience globally. Flooding is among the deadliest natural hazards, and if large tropical volcanic eruptions systematically modulate flood risks regionally—as this study demonstrates—then existing flood hazard models may need adjustments to accommodate these episodic influences. This interplay becomes all the more relevant given ongoing climate variability and the potential for future eruptions as historical analogs inform contemporary risk.

In light of these findings, policymakers and climate modelers alike must consider volcanic aerosols as potent influencers beyond their direct radiative cooling or warming effects. Their cascading impacts on regional hydrology offer a critical dimension to disaster risk assessment, especially in tropical nations disproportionately vulnerable to both volcanic activity and flood hazards. Coupling volcanic eruption forecasts with hydrological early warning systems could thus form a vital piece of integrated risk management strategies.

This research also opens new avenues for cross-disciplinary collaborations blending volcanology, climatology, hydrology, and disaster science. Further exploration is needed to dissect how eruption magnitude, duration, aerosol composition, and atmospheric circulation patterns collectively govern downstream flood variability. Equally important will be assessing these dynamics under the influence of concurrent anthropogenic climate change, which may amplify or mitigate volcanic eruption impacts.

Importantly, the study underscores the heterogeneity of flood responses—a reminder that broad-brush assumptions about “volcano-induced drought” or “volcano-induced floods” are overly simplistic. Instead, the reality is nuanced and highly dependent on regional climatic context, aerosol pathways, and local hydrological conditions. This complexity elevates the need for localized impact assessments rather than generalized global predictions.

The dataset used in this research, encompassing nearly eight thousand globally distributed streamgauges, represents an unprecedented scale in hydrological observational analysis paired with global climate model outputs. This combination allows for robust statistical confidence and granular insight into the spatial and seasonal dimensions of volcanic flood impacts. Such rigor paves the way for more precise prediction models that integrate atmospheric forcing with catchment-scale hydrology.

In sum, this groundbreaking investigation transforms our understanding of how Earth’s most violent volcanic episodes imprint not just on the atmosphere, but on the planet’s surface water regimes as well. By elucidating the global-scale flood responses to eruptions with varied hemispheric aerosol dispersions, the study charts an innovative path forward in comprehending and mitigating the cascading hazards triggered by volcanic activity. These insights will be indispensable in crafting resilient strategies to confront multifaceted environmental threats in an increasingly dynamic planet.

Subject of Research: Climate impacts of tropical explosive volcanic eruptions on global flood responses

Article Title: Global response of floods to tropical explosive volcanic eruptions

Article References:
Kim, H., Villarini, G., Yang, W. et al. Global response of floods to tropical explosive volcanic eruptions. Nat. Geosci. (2025). https://doi.org/10.1038/s41561-025-01782-5

Image Credits: AI Generated

The Datong Basin represents a crucial hydrogeological zone in Shanxi Province, China, where groundwater significantly supports agricultural irrigation and domestic consumption. However, this region has faced challenges due to fluctuating water tables and data scarcity, complicating traditional interpolation methods that often fail to capture the spatial heterogeneity and temporal variability inherent in such groundwater systems. The study’s adoption of a Spatio-temporal CoKriging technique overcomes many limitations encountered by conventional geostatistical methods, delivering enhanced accuracy and resolution in groundwater level prediction.

CoKriging is a geostatistical interpolation method that leverages multiple correlated variables to improve spatial estimation. Unlike ordinary kriging, which uses known values of a single variable, CoKriging incorporates secondary variables to supplement the primary data, thereby reducing estimation errors. When extended into the spatio-temporal domain, the method accounts for variations across both space and time, crucial for dynamic systems like groundwater aquifers which change seasonally or due to human activities.

The novelty in the authors’ approach lies in coupling groundwater level data with auxiliary hydrological and meteorological information, including precipitation, temperature, and local topography. This multivariate integration within a spatio-temporal framework enables a robust model that captures not only the groundwater distribution patterns but also their evolution over time. Such comprehensive modeling provides deeper insights essential for understanding aquifer recharge, depletion rates, and potential responses to climatic variability.

The study meticulously addresses the underlying mathematical formulations necessary for implementing Spatio-temporal CoKriging. This involves constructing covariance functions that define relationships in both spatial and temporal dimensions, coupled with cross-covariance matrices to incorporate secondary variables. Zhang and colleagues carefully calibrate these models using historical data, optimizing parameters to reduce bias and variance in groundwater level estimates. Their method recognizes the anisotropy typical of groundwater flow and variations caused by geological heterogeneity.

One of the paper’s significant contributions is the demonstration of this methodology on the Datong Basin, a vitally important agricultural region in Northern China. Using extensive datasets collected from monitoring wells, weather stations, and satellite observations, the researchers validate their model by comparing predictions with observed groundwater levels. Results exhibit substantial improvements in accuracy over traditional kriging and inverse distance weighting methods, with enhanced ability to reproduce both spatial gradients and temporal fluctuations.

This improved groundwater level interpolation is not just an academic exercise; it bears direct implications for sustainable water resource management. Accurate mapping and forecasting enable better planning and regulation of groundwater extraction, helping to prevent overexploitation and subsequent land subsidence or ecological degradation. The authors emphasize that policymakers and stakeholders can utilize such models to enforce water quotas, design recharge projects, and evaluate the impacts of climate change scenarios.

Furthermore, the integration of remote sensing data and in situ measurements within the spatio-temporal CoKriging framework demonstrates the increasing relevance of interdisciplinary datasets in hydrological research. Satellite data provides wide-ranging spatial coverage while local sensors offer precise temporal updates, combining to form a comprehensive observational network. Zhang et al.’s work underscores the importance of harmonizing diverse data types to tackle the multifaceted challenges in groundwater science effectively.

The robustness of the model also allows it to be adapted for other regions with similar data constraints and hydrogeological complexities. By providing a rigorous statistical framework, the study sets a precedent for groundwater monitoring in arid and semi-arid environments prone to data scarcity. The model’s reliance on spatial covariance structures and temporal trends means it can integrate novel datasets as they become available, including those from emerging sensor technologies or crowdsourced water level observations.

Groundwater sustainability is an urgent global challenge amid growing populations and the intensification of agricultural demand. The approach taken by Zhang, Rui, Zhao, and collaborators exemplifies the innovative solutions required to balance human needs with environmental conservation. By improving the precision and reliability of groundwater level estimations, their research supports science-based decision-making that can lead to more resilient water resource policies, particularly in vulnerable basins experiencing rapid anthropogenic pressures and climatic shifts.

The correction issued by the authors highlights their commitment to scientific accuracy and transparency. Technical corrections ensure the validity of the results and solidify confidence in the methodology among the research community and practitioners alike. This dedication to continual refinement is crucial as new data and computational capabilities evolve, promising even greater advancements in hydrological modeling.

Future avenues stemming from this research include expanding the spatio-temporal CoKriging approach to incorporate predictive elements, transforming it into a forecasting tool capable of simulating groundwater responses under various scenarios. Integrating machine learning techniques with geostatistical models could further enhance predictive performance, while coupling with surface water dynamics may offer a unified framework for comprehensive watershed management.

In conclusion, the innovative spatio-temporal CoKriging technique developed by Zhang and colleagues addresses a critical gap in groundwater level interpolation by effectively integrating multifaceted spatial and temporal data. Their application to the Datong Basin represents a compelling case study with implications extending far beyond the region. This advancement marks a significant step forward in hydrogeological modeling, offering a powerful tool for sustainable management of groundwater resources under increasing environmental stresses.

As water scarcity escalates as a global concern, methodologies that provide accurate, timely, and dynamic groundwater assessments will be indispensable. The work presented invites a shift towards more holistic and data-integrated approaches in hydrological research. It also signals the emerging role of advanced statistical techniques in bridging observational gaps, fostering collaboration between hydrologists, statisticians, and policymakers dedicated to securing water for future generations.

Subject of Research: Groundwater level interpolation estimation using spatio-temporal CoKriging methods in the Datong Basin, Shanxi Province.

Article Title: Correction: Spatio-temporal CoKriging approach for groundwater level interpolation estimation — a case study of the Datong Basin, Shanxi Province.

Article References: Zhang, H., Rui, X., Zhao, X. et al. Correction: Spatio-temporal CoKriging approach for groundwater level interpolation estimation — a case study of the Datong Basin, Shanxi Province. Environ Earth Sci 84, 495 (2025). https://doi.org/10.1007/s12665-025-12475-y

Image Credits: AI Generated