The study assessed 13 aquifer models within the AquiFR platform, covering Nord Pas-de-Calais, Basse-Normandie, the Paris Basin, the Loire and Poitou-Charentes basins, Alsace, and Tarn-et-Garonne. AquiFR combines physically based hydrogeological models, including EauDyssée and MARTHE, with the SURFEX land-surface model, which calculates runoff and drainage used as groundwater-recharge terms. Its climate forcing came from the EXPLORE2 ensemble, a downscaled and bias-corrected subset of Euro-CORDEX projections derived from the CMIP5 climate experiment. The simulations covered 1975–2099, with a 1975–2004 reference period and future horizons of 2021–2050, 2041–2070, and 2071–2099.

Groundwater conditions were evaluated with the Standardized Piezometric Level Index, or SPLI, a normalized measure that expresses each monthly water level relative to the historical distribution for the same calendar month. Values above zero represent levels above the historical median, while negative values represent lower, drier conditions. The index also expresses conditions as historical return periods, allowing wet and dry events to be compared across heterogeneous aquifers. Kolmogorov–Smirnov tests assessed whether future and historical distributions differed, while Mann–Kendall tests evaluated whether conditions changed monotonically through time. The significance threshold was p ≤ 0.05. In the full ensemble analysis, areas associated with more moderate 2.5-year events declined, areas associated with 5-year events changed little, and the footprint of 10-year extremes expanded.

Disagreement among the climate projections was substantial, particularly over the short- and intermediate-term horizons, preventing a single trend from emerging for every scenario. Among 17 available RCP8.5 global and regional climate-model combinations, 11, or about 65%, indicated generally increasing groundwater levels, four produced decreases, and two indicated relatively stable conditions. The study linked much of this spread to the climate component of the modelling chain, including internal climate variability and differences among global and regional climate models. Historical groundwater cumulative-distribution functions derived from the climate projections generally followed those produced using the SAFRAN reanalysis, suggesting that much of the future divergence originated in projected rainfall, drainage, and runoff rather than in the construction of the groundwater index.

Four RCP8.5 storylines were selected to represent contrasting but internally consistent climate futures without combining their trajectories into an averaged projection. The orange storyline paired strong warming with very dry summers and produced the most adverse groundwater outcome. By the end of the century, nearly 60% of the modelled domain was affected by 10-year dry events in the area-based analysis, while more than 75% of the domain experienced extreme dry conditions in the separate time-proportion analysis. The green storyline, characterized by serious warming and increased rainfall, instead generated widespread wetter conditions. Under it, the area affected by 10-year wet events increased from 3.7% in the historical period to almost 20%.

The purple storyline, defined by a strong seasonal contrast in precipitation with wet winters and dry summers, increased both wet and dry extremes. Under this projection, extreme dry conditions occupied more than half of the time across more than half of the modelled domain late in the century. The yellow storyline, designed around relatively modest future changes, produced the smallest overall shifts and a more spatially varied pattern without a dominant tendency. Statistical tests identified significant differences between historical and future distributions and significant temporal changes for almost all event categories in the four storylines. Those results apply separately to each physically consistent projection, however, and do not establish one deterministic future.

Across the full RCP8.5 ensemble, the area affected by extreme wet groundwater conditions increased by a factor of 4.5 over the long-term horizon, a result supported by the green, purple, and yellow storylines. Extreme dry events also increased overall, although generally by less than wet extremes; the driest orange storyline produced a much larger increase, reaching a factor of 25. The ensemble analysis found that the surface assigned to 2.5-year dry and wet conditions decreased, while 5-year dry and wet areas changed only slightly and 10-year events expanded. This pattern points toward greater separation between ordinary conditions and the most extreme states, rather than a broad increase in the frequency of all wet or dry categories.

Maps of the proportion of time spent in extreme events revealed clear differences among the storylines. The orange projection reduced the time associated with extreme wet events to less than 5% over almost the entire AquiFR domain while extending extreme dry conditions across more than 75% of it. The green projection had the opposite tendency, strongly increasing extreme wet events while reducing extreme dry ones. The purple projection increased both, whereas the yellow projection remained more heterogeneous. In the combined RCP8.5 analysis, the time spent in extreme wet events was close to historical conditions over part of the domain in the short term but increased toward 2100, especially in northern France. Extreme dry conditions showed a contrasting long-term pattern, with lower values in the north and higher values in the south.

The simulations were designed to isolate climate effects by holding land use and groundwater abstraction constant at historical climatological levels. This separation avoids confounding climatic change with human water use, but it also means the projections do not represent how abstraction, demand, or policy might change as groundwater availability evolves. The assumption is particularly consequential in southern areas where some simulations produce lower levels. AquiFR also does not apply multiple interchangeable groundwater models throughout its full domain, so the contribution of hydrogeological-model uncertainty cannot be fully quantified. Southern coverage is sparse: the Alsace and Tarn-et-Garonne models each cover about 1,100 square kilometres, compared with roughly 7,000–66,000 square kilometres for most other modelled aquifer systems. These limitations prevent the findings from establishing a nationwide north–south divide with the same confidence across all of metropolitan France.

The study’s broad conclusion is that climate projections imply contrasting regional futures and a growing separation between exceptionally wet and exceptionally dry groundwater conditions, rather than a uniform rise or fall across France. Under RCP2.6 and RCP4.5, the 2021–2050 and 2041–2070 results were too divergent for reliable trend statements. Over 2071–2099, they showed greater convergence toward wetter conditions in the north, while the remote Tarn-et-Garonne and Alsace models continued to disagree. The authors identify expansion of southern model coverage, fuller quantification of uncertainty across groundwater software, and integration of socio-economic water demand as priorities for future assessment. Until then, the four storylines demonstrate how a single national projection can conceal materially different outcomes among aquifers, including futures in which extreme groundwater drought expands across nearly the entire modelled region.

Subject of Research: Modelled climate-change impacts on groundwater levels and extremes across French aquifers

Article Title: Climate change impacts on groundwater simulated using the AquiFR modelling platform

Article References: Jeantet, A., Vergnes, J.-P., Munier, S., & Habets, F. (2026). Climate change impacts on groundwater simulated using the AquiFR modelling platform. Hydrology and Earth System Sciences, 30(20), 6381-6420. https://doi.org/10.5194/hess-30-6381-2026

Image Credits: AI Generated

DOI: 10.5194/hess-30-6381-2026

Keywords: groundwater, climate change, France, AquiFR, hydrogeological modelling, climate projections, extreme events, aquifer recharge, RCP8.5, water resources, Climate, change

The determination at the heart of the dispute is the US Environmental Protection Agency’s 2009 Endangerment Finding, the formal legal conclusion that greenhouse gas emissions endanger public health and welfare in the United States. That finding underpins federal regulation of carbon dioxide and other heat-trapping gases, and moves to revoke it have prompted concern among climate scientists, public health experts and environmental lawyers alike. The new review argues that the scientific basis for the Endangerment Finding was already compelling when it was issued, and that the evidence has grown steadily stronger with every subsequent year of observations. Rescinding it, the authors warn, could weaken efforts to reduce emissions and potentially open the door to dismantling other environmental regulations in the United States.

Climate fingerprinting, formally known as detection and attribution research, is a discipline built on a simple but powerful idea. Natural forces such as solar variability and volcanic eruptions, and internal climate fluctuations such as those associated with El Niño, each leave characteristic patterns of change in the climate system. Human-caused greenhouse warming leaves a different pattern, one that is distinctive in its geography, its vertical structure and its timing. By comparing observed climate changes against simulations from climate models driven by different combinations of these influences, scientists can identify which factors are responsible for the changes we actually measure. The technique is, in essence, a forensic method for the atmosphere, and it has been refined continuously since the early 1990s.

Professor Santer has been central to that refinement for more than thirty years, collaborating with research groups around the world to identify human fingerprints in many different components of the climate system. His early work contributed to a historic turning point: the 1995 conclusion of the Intergovernmental Panel on Climate Change that the balance of evidence suggested a discernible human influence on global climate. That cautious phrase marked the first time the IPCC formally attributed observed warming to human activity. The new review traces the evolution of the field from that milestone to the present, focusing on fingerprint studies of annual, decadal and seasonal changes in climate, and touching on the extensive body of evidence addressing how human-caused climate change alters the properties of extreme events.

What has changed since 2009 is not merely the accumulation of more data, but a qualitative shift in the reach of attribution science. Fingerprinting techniques have matured to the point where human influence can now be identified throughout Earth’s climate system, from atmospheric temperature structure to ocean heat content, and the review notes that this influence is increasingly affecting the ability of societies and ecosystems to thrive. Methods have also been developed to quantify how greenhouse gases have changed the frequency and intensity of extreme events, and to assess how any individual weather event observed today has been altered by emissions. Extreme events, including heat stress, flooding, wildfires and crop damage, affect human and ecosystem health and welfare directly, and fingerprint studies now build a documented link between human-caused increases in greenhouse gases and these threats.

One of the most striking statistical points made in the review concerns the interplay between signal and noise. Natural climate variability acts as background noise that can obscure or exaggerate the human-caused warming signal in any given year. The post-2009 period, however, included the ten warmest years in the instrumental surface temperature record, which has enhanced the signal of human-caused warming relative to that natural noise. In practical terms, the planet itself has run an experiment over the past seventeen years, and the results have made the human fingerprint easier, not harder, to detect. This is precisely the kind of accumulating, independently verifiable evidence that the review says the Department of Energy assessment failed to account for.

Professor Santer’s summary of the field’s verdict is unambiguous. “The climate science that’s been published since the 2009 Endangerment Finding is now stronger than in 2009, the evidence is incontrovertible,” he said. He argued that the growing body of evidence linking human activity to observed climate changes reinforces the need for policies aimed at limiting greenhouse gas emissions and addressing climate-related risks. He also issued a warning about the state of the science itself, coining a pointed phrase: “A different kind of ‘Endangerment Finding’ is that the continuation of rigorous attribution science is endangered. It’s inconvenient science for those who seek a fossil fuel intensive future, because it connects the dots between fossil fuel burning and harms to our personal health, to the health of our planet, and to the health of our societies and economies.”

Co-authors of the review bring substantial authority to that assessment. Professor Susan Solomon, the Lee and Geraldine Martin Professor of Environmental Studies at the Massachusetts Institute of Technology, noted that many rigorous fingerprint studies by multiple research groups worldwide have enhanced confidence in the reality and seriousness of human-caused climate change. “These studies reveal that many independently monitored aspects of Earth’s climate have all been altered by global warming in physically consistent ways, just as we expect — from the top of the stratosphere to the depths of the world’s oceans,” she said. That phrase, physically consistent, captures the technical core of the argument. Attribution science does not rest on a single temperature curve; it rests on the coherence of many separate measurements, each responding to warming in the way that physics predicts, from stratospheric cooling above to deep-ocean warming below.

Professor Gabi Hegerl of the University of Edinburgh, another co-author, emphasised the extension of these methods to extreme weather. “Methods have also been developed to understand how greenhouse gases have changed the frequency and intensity of extreme events and how a weather event we observe today has been changed due to greenhouse gas emissions,” she said. The review’s author team, which also includes David WJ Thompson of Colorado State University and Qiang Fu, spans institutions on both sides of the Atlantic and represents several decades of accumulated expertise in detection and attribution research. The paper, titled ‘Fingerprint Evidence for Human Effects on Climate: Advances after the 2009 Endangerment Finding’, was published on October 12.

The broader stakes of the dispute extend well beyond an academic disagreement. The Endangerment Finding is the legal hinge on which US federal greenhouse gas regulation turns, and the review’s authors make clear that the scientific record does not support the case for rescinding it. Their argument is that the evidence available in 2009 was sufficient, that the intervening years have strengthened it across every independently monitored component of the climate system, and that the techniques used to detect human influence have become more sophisticated, more numerous and more consistent in their conclusions. For a field that has spent three decades distinguishing a faint human signal from the noise of natural variability, the irony is stark: as the signal has grown louder, the political willingness to hear it has, in at least one government assessment, grown quieter. The review stands as a reminder that the forensic evidence linking fossil fuel burning to a changing climate has never been clearer, and that the scientific community, having connected those dots, shows no sign of unconnecting them.

Subject of Research: Detection and attribution of human-caused climate change through climate fingerprint studies

Article Title: Scientists reaffirm ‘incontrovertible’ evidence of human influence on climate

Article References: Scientists reaffirm ‘incontrovertible’ evidence of human influence on climate. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: climate change, attribution science, climate fingerprinting, greenhouse gases, Endangerment Finding, EPA, IPCC, global warming, extreme events, US Department of Energy, Earth's Future, Benjamin Santer

The research team, led by Jiayu Xu and Biying Qiu of Henan University and Liaoning University together with colleagues, set up rainout shelters across an alpine meadow site on the plateau between 2021 and 2024. Using a two-factor randomized complete block design, they simulated extreme drought either early or late in the growing season, or in both windows sequentially, and then measured aboveground and belowground biomass, community characteristics, and soil physicochemical properties every year. The statistical toolkit was correspondingly rigorous: two-way repeated-measures analysis of variance to isolate treatment effects, structural equation modeling to trace causal pathways among biotic and abiotic variables, and linear regression to quantify correlations across the four years of data.

The headline result is deceptively simple but ecologically profound. Late season drought increased the ratio of root biomass to shoot biomass by 56 percent, compared with a 24 percent increase under early season drought. The root-to-shoot ratio is one of the most fundamental currencies in plant ecology, a measure of how a plant splits its carbon budget between the organs it needs to forage for water and nutrients below ground and the organs it needs to capture light and carbon dioxide above ground. A larger ratio typically signals a plant hedging its bets against water scarcity, building more root infrastructure to chase receding soil moisture at the expense of leafy growth.

Why would late season drought push plants to reallocate so much more aggressively than early season drought? The researchers traced the answer to a combination of legacy effects and constrained plasticity. Early season drought, they found, created a legacy by depleting shallow soil moisture and altering the physiological status of the plants, leaving them in a compromised state heading into the rest of the season. But it was the late season drought that limited the plants’ capacity to respond plastically, because by that point in the season soil moisture, plant cover, and soil nitrogen availability had all been reduced. In other words, plants hit by drought late in the season face the stress with fewer resources and less time to reorganize their growth, forcing a stronger shift of biomass below ground.

Perhaps the most striking discovery, however, lies in what happens when both droughts occur in the same year. Early season and late season drought did not simply add their effects together; they showed a positive synergistic interaction. Early season drought alone increased the root-to-shoot ratio by 1.0 unit, but when it was followed by late season drought, that increase swelled to 6.4 units. A mirrored pattern appeared for late season drought, whose effect was likewise amplified when an early season drought had preceded it. The sequential double punch, in which an early drought primes the system and a late drought delivers the blow, produces allocation shifts far larger than either stress in isolation, a signature of compound extreme events that most climate models have yet to capture.

This synergy matters because climate change is not only making droughts more intense; it is shifting when they occur. Warming is altering the seasonal distribution of precipitation across the globe, and on the Qinghai-Tibet Plateau specifically, studies have documented changes in precipitation concentration, multi-type drought propagation, and amplified precipitation seasonality. Alpine meadows cover enormous areas of the plateau and underpin the livelihoods of pastoral communities, the water resources of major Asian rivers, and significant carbon stocks. If the timing of drought determines how these ecosystems restructure themselves, then projections that treat drought as a seasonally anonymous stressor may systematically misjudge future vegetation change, forage production, and carbon sequestration.

The structural equation modeling in the study helps explain the mechanism in more detail. By linking soil moisture, plant cover, and nitrogen availability to biomass allocation within a single causal framework, the analysis showed that late season drought constrains plant plasticity through multiple simultaneous pathways. Reduced soil moisture directly limits water uptake; reduced plant cover diminishes the photosynthetic machinery available to generate new biomass; and reduced soil nitrogen availability constrains the construction of new tissue, since nitrogen is a core building block of proteins and chlorophyll. A plant facing this triple constraint late in the season has little option but to favor roots, the organs that determine whether it can survive into the next year.

The findings also connect to a broader literature on drought legacy effects in grasslands. Previous work has shown that drought can alter soil respiration through root exudates, that high belowground bud abundance can speed ecosystem recovery, and that drought timing differentially affects aboveground and belowground productivity in mesic grasslands. What the new study adds is a quantified, multi-year demonstration from an alpine system that the sequence of droughts within a single season, not just their individual occurrence, governs the magnitude of allocation shifts. The amplification factor, from a 1.0-unit increase to a 6.4-unit increase in the root-to-shoot ratio, gives modelers a concrete parameter to test against.

For the carbon cycle, the implications run deep. Roots are the primary conduit through which photosynthesized carbon enters the soil, and shifts in allocation toward belowground biomass can change rhizodeposition, soil organic matter formation, and ultimately whether an ecosystem acts as a carbon sink or source. If late season droughts become more frequent under future precipitation regimes, alpine meadows could channel progressively more of their carbon underground, altering soil carbon storage in ways that depend critically on drought timing. The authors argue that their results provide empirical evidence for improving predictions of alpine meadow ecosystem responses to future climate scenarios, and the four-year duration of the experiment lends that claim unusual weight for a field manipulation at this altitude.

The study, published on 8 September 2026 in Plant and Soil and supported by the National Natural Science Foundation of China, arrives at a moment when ecologists are increasingly recognizing that the impacts of climate extremes depend crucially on their timing. As one of the first experiments to disentangle early and late season drought effects on biomass allocation in an alpine meadow over multiple consecutive years, it delivers a clear message: to understand how grasslands will weather the climate of the coming decades, scientists must stop asking only how much rain falls, and start asking precisely when it fails to fall. On the roof of the world, the calendar of drought may decide the fate of the meadows below.

Subject of Research: Effects of drought timing on plant biomass allocation in alpine meadows on the Qinghai-Tibet Plateau

Article Title: Late season drought has stronger effects on plant biomass allocation than early season drought in alpine meadows

Article References: Xu, J., Qiu, B., Yang, J., Sun, Y., Guo, M., Yang, Z., Miao, R., & Li, G. (2026). Late season drought has stronger effects on plant biomass allocation than early season drought in alpine meadows. Plant and Soil. https://doi.org/10.1007/s11104-026-09068-4

Image Credits: AI Generated

DOI: 10.1007/s11104-026-09068-4

Keywords: drought timing, biomass allocation, root-to-shoot ratio, alpine meadow, Qinghai-Tibet Plateau, legacy effect, precipitation manipulation, climate change, grassland ecology, plant plasticity, soil nitrogen, extreme events

The motivation comes from a practical and increasingly urgent problem: industries that operate many installations spread across a territory need to know how often those installations will be hit by the same extreme weather event at the same time. Prolonged, spatially extensive droughts are a particular worry for the French electricity system, because nuclear plants depend on river water for cooling, and a nationwide dry spell can throttle several reactors simultaneously. Parliamentary missions in France have explicitly called for quantifying such hydro-stress on nuclear generation. Agriculture faces analogous risks when entire growing regions share the same drought. Yet historical observations offer only one realization of climate variability, and climate models, though valuable, are too computationally expensive to run in the thousands of variants needed for robust statistics of rare events. Stochastic weather generators fill this gap by churning out thousands of statistically plausible weather histories at negligible cost.

Mathematically, the new model builds on hidden Markov models, a statistical framework dating back to precipitation applications in the early 1990s. The central idea is elegant: imagine an invisible variable, one per day, that represents the overall weather regime over France. This hidden state evolves as a Markov chain, meaning that tomorrow’s regime depends only on today’s. Crucially, the model is seasonal – its transition probabilities are 366-periodic functions of the calendar day, parameterized by trigonometric polynomials so that they evolve smoothly through the year rather than jumping between months. Given the hidden regime, rain occurrence at each station follows a simple Bernoulli probability, and the model enforces conditional independence between stations. That constraint is deliberate: it forces every scrap of spatial correlation to be absorbed by the hidden states, which is exactly what makes them interpretable.

The authors added two refinements that prove decisive for performance. First, they introduced a one-day local memory: today’s wet or dry status at a station depends not only on the hidden regime but also on yesterday’s status at the same station. Without this autoregressive touch, the model systematically produced dry and wet spells that were too short, overestimating brief episodes and underestimating the long droughts that matter most for risk assessment. Second, after generating rain occurrences, the model attaches rainfall amounts using mixtures of two exponential distributions whose parameters also vary seasonally and with the weather regime, linked across stations by a Gaussian copula whose correlation strength itself depends on the regime. Model selection using the integrated complete-data likelihood landed on four hidden states, one day of memory, and first-degree seasonal polynomials – a parsimonious configuration of roughly 276 parameters instead of the more than 33,000 a day-by-day approach would demand.

The interpretability results are the paper’s showpiece. The four learned states form recognizable regional patterns: one in which it rains nearly everywhere in France, one that is wet in the north and drier in the south, its mirror image, and a fourth in which rain is scarce nationwide. When the authors averaged winter sea-level pressure anomalies from the ERA5 reanalysis conditional on each hidden state, coherent large-scale structures emerged over the whole North Atlantic – from a model trained exclusively on ten French rain gauges. State four resembles a blocking pattern, and the set of four regimes lines up closely with the classic Euro-Atlantic regimes – the positive and negative phases of the North Atlantic Oscillation, the Atlantic Ridge, and blocking – and with weather types previously derived from French rainfall data, though with anomalies centered more tightly on France, as expected from the training data.

The temporal dynamics of the regimes also ring true. The dry regime is the stickiest, with the highest probability of persisting from one day to the next, which explains the long nationwide droughts it can generate. It almost never flips directly into the wettest regime; an intermediate state intervenes, mirroring the intuition that a countrywide dry day is rarely followed by a countrywide soaking. Decoding the hidden states across history with the Viterbi algorithm, the model even detects famous events: a 27-day stretch of the dry regime beginning in early June 1976, matching the notorious drought of that summer, and a 15-day dry sequence during the August 2003 heatwave.

Validation tests show the generator reproducing observed statistics remarkably well. At nearly all stations, the historical distributions of dry and wet spell lengths fall within the envelope of five thousand simulated 64-year runs. Spatial correlations between stations, which range from near zero to about 0.5, are well captured, confirming the conditional independence assumption a posteriori. Most impressively, the model nails the distribution of large-scale droughts – episodes during which at most twenty percent of stations are wet – across all four seasons, and it can even produce dry spells longer than any yet observed. A comparison against the widely used WGEN-type generator, which links stations through pairwise Gaussian copulas and fourth-order local Markov chains, was revealing: the WGEN approach overestimated short dry episodes and underestimated long ones, because pairwise correlations simply cannot encode temporally persistent, territory-wide dry states. The hidden-state model scales as the square of the number of regimes rather than the square of the number of stations, giving it a decisive computational edge.

The authors then turned the generator loose on climate change. Training it on downscaled projections from the French DRIAS service, they compared climate models against the statistical envelope of natural variability during a reference period, a more meaningful benchmark than comparing against a single historical trajectory. Under the high-emission RCP8.5 scenario with the IPSL projection, the fitted transition matrices shift: the rainy regimes become more self-persistent in summer, projecting longer stretches of heavy rain – a pattern consistent with known biases of the European regional model used, which the generator makes easy to expose. Training the generator on future scenarios allows researchers to resample thousands of synthetic climates and extract return periods of compound extremes that no ensemble of physical models could afford to simulate directly.

Limitations remain candidly acknowledged. The conditional independence assumption requires carefully chosen, well-separated stations – a pair like central Paris and Orly, thirteen kilometers apart, would break it – and the model generates weather only at training sites, not high-resolution maps. Coastal stations such as La Hague and Chassiron, with their idiosyncratic maritime climates, fit less well, and the Gaussian copula underestimates joint extremes at the closest pairs, pointing toward heavier-tailed alternatives like the Student copula. Still, the package, released as an open-source Julia library with a fully reproducible tutorial, opens a practical path toward stress-testing entire energy systems and agricultural regions against synthetic weathers – including future ones – and toward estimating, at last, the true odds of the drought that half of France dreads sharing at once.

Subject of Research: An interpretable seasonal hidden Markov model for stochastic multisite precipitation generation and climate risk assessment in France

Article Title: Interpretable seasonal multisite hidden Markov model for stochastic rain generation in France

Article References: Gobet, E., Métivier, D., & Parey, S. (2025). Interpretable seasonal multisite hidden Markov model for stochastic rain generation in France. Advances in Statistical Climatology, Meteorology and Oceanography, 11(2), 159-201. https://doi.org/10.5194/ascmo-11-159-2025

Image Credits: AI Generated

DOI: 10.5194/ascmo-11-159-2025

Keywords: stochastic weather generator, hidden Markov model, precipitation modeling, France, drought, extreme events, weather regimes, climate change scenarios, energy systems, North Atlantic Oscillation, rainfall statistics, climate risk

The research addresses a fundamental blind spot in climate science. Climate models can simulate unprecedented extremes, and techniques such as ensemble boosting can efficiently generate large samples of record-shattering events. But because such events have no historical analogue, it has been unclear whether the simulations are physically credible or merely artefacts of model quirks. The 2021 Pacific Northwest heatwave, which broke previous records by almost 5 K, demonstrated that the real world can deliver events that models and records suggested were nearly impossible. Roemer and his team set out to test whether simulated month-long Western European heatwaves of comparable severity pass the credibility test.

To do this, the researchers built their assessment on three criteria: conformity with physical principles, internal consistency, and historical precedent. Rather than relying solely on statistical tests such as extreme value theory, they took a process-based approach, examining the actual physical machinery behind each simulated heatwave. Their benchmark events came from ensemble boosting simulations with the CESM2 climate model, in which a run is re-initialised a few weeks before an extreme event and restarted hundreds of times with tiny perturbations, generating thousands of alternative versions of the same weather situation. From these simulations they identified twenty unprecedented heatwaves whose 30-day mean temperature anomalies over Western Europe exceeded the most extreme event in the reanalysis record, the legendary heatwave of June and July 1976.

A central methodological innovation was the use of standardised anomalies relative to a time-evolving climatology. Because the simulated heatwaves occur in a warmer future climate while the historical comparison events occurred decades ago, comparing raw temperatures would be meaningless. Instead, the team expressed every variable, from temperature and geopotential height to soil moisture and surface heat fluxes, as an anomaly measured against the mean and variability of the background climate in which the event unfolds, using a centred 31-year window. This allowed a like-for-like comparison of the physical drivers of unprecedented simulated heatwaves with those of historical events in the ERA5 reanalysis.

The results are striking in their overall consistency. The unprecedented heatwaves are driven by the same well-known culprits as historical ones: strong blocking anticyclones over Central Europe, weak zonal winds, abundant surface solar radiation, and parched soils. The anomalies of these drivers are more intense in magnitude, but their spatial patterns and their relationships with one another closely mirror those seen in historical heatwaves. In the authors’ framing, these events are caused by extreme anomalies of common drivers rather than by any exotic new mechanism, a finding that strongly supports their physical plausibility.

The team went further, testing whether the temperature anomalies of the unprecedented heatwaves could be inferred from a simple multilinear statistical model using five predictors: geopotential height, zonal and meridional winds, soil moisture, and surface solar radiation. Remarkably, models trained on both the climate model’s own large ensemble and on historical heatwaves in the reanalysis could explain most of the variance in temperature anomalies, and they inferred the magnitudes of the unprecedented events with accuracies similar to those achieved for training data. Even the anomalies of daily minimum temperatures, which lie far beyond anything in the observational record, were reasonably captured. This suggests that the linear combination of familiar drivers holds together even deep in the tail of the distribution, an important internal consistency check that many had feared would fail.

Yet the study is not a clean bill of health. Digging into the temporal substructure of the events, the researchers uncovered substantial discrepancies between the simulated unprecedented heatwaves and their historical counterparts. CESM2 systematically underestimates the frequency and persistence of atmospheric blocking, the stubborn high-pressure systems that park over Europe and bake it for weeks. At the same time, it overestimates the drying of soils and the persistence of soil moisture deficits, amplifying land-atmosphere feedbacks. A temperature budget analysis at the 850 hPa level confirmed the pattern: simulated unprecedented heatwaves rely more on diabatic heating and less on adiabatic warming from subsiding air than historical events do. In the authors’ blunt assessment, the simulated heatwaves occur for partly the wrong reasons.

Counterintuitively, these biases may actually make the study’s conclusion more alarming rather than less. Because CESM2 underestimates the persistence of blocking, the fact that it still produces month-long heatwaves exceeding records by more than 5 K suggests that the simulated events may be a conservative estimate of what the real world could deliver. The return periods of the simulated heatwaves range from roughly 300 years to more than 6000 years, finite but far beyond a human lifetime, placing them firmly in the category of events that are statistically possible but not probable. Plausibility, the authors emphasise, sits between possibility and probability, and it is plausibility that should guide disaster preparedness, since preparing for credible worst cases is what stress-testing infrastructure is all about.

The researchers are careful about the limits of their findings. Their assessment refers primarily to standardised anomalies rather than absolute temperatures, meaning that impact studies, which typically depend on absolute thresholds such as hospital-triggering temperature limits, still require bias correction of raw model output. Surface variables such as soil moisture and heat fluxes carry substantial uncertainties in both climate models and reanalysis, and reanalysis values of the surface energy budget should not be treated as ground truth. The analysis also rests on a single climate model, and the authors call for similar evaluations with other models to test how dependent the conclusions are on CESM2’s particular strengths and weaknesses.

Western Europe is arguably the most urgent test case for this kind of work. Mean summer temperatures and extreme heat there are rising faster than in almost any other region, driven by anthropogenic warming, declining aerosol emissions, and circulation changes, and the region has been hammered by record-breaking heat in 2003, 2006, 2015, 2018, 2019, 2022, 2025, and 2026. Long-lasting heatwaves are especially deadly, as the excess mortality of the 2003 and 2010 European events demonstrated. The new framework, which the authors argue can be adapted to other extremes such as cold spells and droughts, offers decision-makers a way to distinguish between model-generated scare stories and credible worst-case scenarios. On the evidence of this study, a month-long, record-shattering European heatwave belongs firmly in the second category, and adaptation planners would be wise to plan for it.

Subject of Research: Process-based plausibility assessment of unprecedented month-long heatwaves in Western Europe using climate model storylines

Article Title: Assessing the plausibility of unprecedented events: a process-based approach applied to month-long heatwaves in Western Europe

Article References: Roemer, F. E., Fischer, E. M., Noyelle, R., & Knutti, R. (2026). Assessing the plausibility of unprecedented events: a process-based approach applied to month-long heatwaves in Western Europe. Weather and Climate Dynamics, 7(3), 1919-1949. https://doi.org/10.5194/wcd-7-1919-2026

Image Credits: AI Generated

DOI: 10.5194/wcd-7-1919-2026

Keywords: heatwaves, Western Europe, climate models, ensemble boosting, atmospheric blocking, soil moisture, land-atmosphere feedback, plausibility, climate storylines, extreme events, CESM2, adaptation planning

A team led by Haru Kuroki and corresponding author Kazumune Hashimoto of The University of Osaka, together with colleagues at The University of Tokyo, has developed a hybrid control algorithm they call ensemble-Kalman-guided model predictive path integral control, or EKG-MPPI. The method is designed to answer a deceptively simple question: if you could only nudge the atmosphere at one tiny location, with one small push, where should that push go and how big should it be to suppress an extreme event before it forms? The work sits within a growing research program known as control simulation experiments, in which scientists test hypothetical interventions entirely inside numerical models, using an uncontrolled simulation as a stand-in for the real atmosphere. This simulation-based approach allows researchers to probe the feasibility and side effects of intervention strategies without any real-world risk.

The architecture of EKG-MPPI is a marriage of two complementary control philosophies. The first ingredient is ensemble Kalman control, or EnKC, a technique introduced by co-author Yohei Sawada that cleverly repurposes the ensemble Kalman filter, the workhorse of modern weather data assimilation. In a conventional forecast, the filter blends observations with model predictions to estimate the current atmospheric state. EnKC flips the logic: it treats the desired future outcome, such as keeping wind speeds below a threshold, as a pseudo-observation to be assimilated. The ensemble Kalman smoother then works backward in time to compute the analysis increment, which is interpreted as the optimal perturbation to apply now to steer the system toward that target. Because the computation relies on ensemble statistics, it never requires an explicit covariance matrix, making it feasible for high-dimensional systems.

EnKC, however, has a fundamental weakness. Its mathematics assumes the dynamics are approximately linear over the prediction horizon, an assumption that becomes increasingly shaky in strongly nonlinear, chaotic flows. The second ingredient of the new method, model predictive path integral control, or MPPI, addresses exactly this gap. MPPI is a sampling-based technique with roots in robotics and autonomous driving, derived from a variational free-energy bound. Instead of linearizing the model or computing gradients, MPPI draws hundreds or thousands of candidate control sequences from a Gaussian distribution, rolls each one forward through the full nonlinear model, scores the resulting trajectories with a cost function, and combines them using exponential importance weights. The weighted average becomes the refined control input. Crucially, all the rollouts and weight computations can be run in parallel, which makes the method a natural fit for modern GPU hardware.

The problem with vanilla MPPI in geophysics is sample inefficiency. When the control space has thousands of dimensions, as it does for a 128-by-128 atmospheric grid, naive random sampling almost never discovers the rare, precisely placed perturbations that actually matter. This is where the ensemble Kalman guidance earns its place in the acronym. The researchers apply an adaptive thresholding procedure to the EnKC-derived perturbation, discarding small components as noise and retaining only the largest one, which identifies the single most promising actuator location and its nominal amplitude. That information is then embedded into the mean and covariance of the Gaussian proposal distributions from which MPPI samples. In effect, EnKC provides a physically informed educated guess about where the system is most sensitive, and MPPI polishes that guess using the full nonlinear dynamics. The final control is projected back onto a single-actuator form, applied to both the nature run and every ensemble member, and the forecast-assimilation-control loop repeats at the next assimilation time.

The team tested the method on two models of very different character. The first was the Lorenz-96 model, a 40-variable chaotic system that has become a standard proving ground for data assimilation research because it mimics the multiscale structure of the atmosphere while remaining computationally cheap. In control simulation experiments aimed at keeping state variables below a threshold of 12, EKG-MPPI reduced the number of extreme-event counts to 7,622 plus or minus 146, compared with 7,820 plus or minus 142 for EnKC alone, while simultaneously cutting the mean control-input magnitude from 0.191 to 0.166. The improvement was consistent across ten simulations with different random seeds. The researchers attribute the gain to the nonlinear evaluation introduced in the MPPI refinement stage, which captures effects that linearized EnKC simply cannot see.

The comparisons went further. Against a standalone Bayesian optimization baseline, which is the natural benchmark for expensive black-box optimization problems, EKG-MPPI delivered both fewer threshold exceedances and roughly half the control effort, with 7,622 events at an input magnitude of 0.166 versus 8,299 events at 0.340 for Bayesian optimization. The team also built an ablation experiment, BO-MPPI, in which Bayesian optimization rather than EnKC supplies the prior for the MPPI stage. Across multiple values of the MPPI temperature parameter, the EnKC-informed version consistently sat below the Pareto-like trade-off curve traced by the BO-informed variant, meaning it achieved comparable suppression with smaller interventions. A sensitivity analysis showed that smaller temperature parameters concentrate the weight updates on rare, highly effective samples, generally strengthening suppression but increasing input magnitude, with the smallest mean input occurring near a temperature of one. Notably, the entire EKG-MPPI computation took about 3.87 times ten to the minus four seconds per control decision, roughly 560 times faster than the BO-MPPI pipeline and comfortably shorter than a single model time step.

The second testbed was far more ambitious: the surface quasi-geostrophic model, a fluid-dynamical framework used to describe mesoscale atmospheric and oceanic dynamics, discretized on a 128-by-128 grid with surface buoyancy acting as an active tracer from which the full velocity field is diagnosed. Here the control objective was to suppress regional wind speed within eight different target areas, each roughly ten grid squares wide, over simulations spanning about ten days of model time. Across forty simulations combining eight scenarios and five random seeds, EKG-MPPI achieved lower maximum wind speeds in the target regions than EnKC in nearly every case while requiring smaller mean control magnitudes. Visualizations of one scenario showed strong winds persisting in the uncontrolled run, partial weakening under EnKC with localized hotspots remaining, and the strongest suppression under EKG-MPPI. The authors note that the sampling-based refinement introduces larger variability in intervention magnitude, a plausible consequence of exploration in a high-dimensional system, and that scenarios in which winds intensify late in the simulation showed more complex behavior as intervention effects accumulated.

The authors are careful to frame EKG-MPPI as a building block rather than a blueprint for weather modification. The current implementation assumes a single, spatially sparse actuator applying a one-step perturbation, and it does not yet enforce hard physical constraints on the control. Future work will extend the framework to multiple actuators acting over multiple time steps, constraint-aware formulations, and evaluation metrics that explicitly quantify unintended impacts on regions outside the target zone, along with robustness to model and observation uncertainty. Still, the significance of the result is hard to overstate. For the first time, researchers have a computationally tractable, derivative-free method that can identify where a tiny intervention will matter most in a chaotic geophysical flow, verify that judgment against the full nonlinear dynamics, and do so fast enough for online control. Whether the eventual actuators are dust plumes, cloud-brightening sprays, or turbine arrays, any serious assessment of their potential will need exactly this kind of mathematics. The era of guessing at weather intervention may finally be giving way to an era of computing it.

Subject of Research: Hybrid ensemble Kalman and model predictive path integral control for suppressing extremes in chaotic geophysical flows

Article Title: Ensemble Kalman-guided model predictive path integral control for spatially localized suppression of extremes in chaotic geophysical flows

Article References: Kuroki, H., Hashimoto, K., Uehara, Y., Sawada, Y., Le, D., & Minamide, M. (2026). Ensemble Kalman-guided model predictive path integral control for spatially localized suppression of extremes in chaotic geophysical flows. Nonlinear Processes in Geophysics, 33(3), 473-487. https://doi.org/10.5194/npg-33-473-2026

Image Credits: AI Generated

DOI: 10.5194/npg-33-473-2026

Keywords: chaos theory, weather modification, ensemble Kalman filter, model predictive control, data assimilation, extreme events, geophysical fluid dynamics, Lorenz-96 model, quasi-geostrophic model, control simulation experiments, tropical cyclones, computational optimization

At the heart of the study is a measurement called aerosol optical depth, or AOD, which quantifies how much sunlight is removed from a beam passing through the atmosphere by particles suspended in the air. An AOD of 0.1 describes relatively clean skies, while values above 1.0 indicate conditions so hazy that the sun itself can appear dimmed and diffuse. The researchers worked with daily retrievals from the MODIS instrument aboard NASA’s Terra satellite, which has been scanning the Earth since early 2000. Raw satellite AOD products, however, carry well-documented biases, particularly over the bright surfaces and complex aerosol mixtures of West Africa. To correct these errors, the team built a machine learning framework known as Gradient Boosting Quantile Regression, trained against ground truth from fourteen AERONET sun photometer stations spread across nine West African countries.

The correction proved dramatic. At the Nigeria-Ilorin AERONET site, the machine learning adjustment shrank the raw mean bias of MODIS retrievals from minus 0.252 to just minus 0.015, a near-perfect agreement on average. More importantly, the fraction of satellite observations falling within the expected error envelope of ground measurements jumped from a dismal 23.0 percent to 95 percent. When the team tested the model’s ability to generalize to locations it had never seen, using a leave-one-site-out validation strategy, the expected error agreement remained a respectable 74.5 percent. This bias-corrected, boundary-clipped daily record gives Nigeria something it has never had before: a trustworthy, continuous, quarter-century chronicle of its atmospheric particle loading, suitable for everything from climate model evaluation to public health epidemiology.

So what does 25 years of data actually show? Perhaps surprisingly, the national annual trend is essentially flat. The linear slope across the full record is minus 0.0020 per year, with a p-value of 0.226, meaning the change cannot be distinguished from random year-to-year variability. But seasonal decomposition told a subtler story. The dry season, when Saharan dust and harmattan winds dominate, shows a suggestive decline of minus 0.0043 per year, equivalent to roughly a 16 percent reduction over the study period. This trend sits just at the edge of conventional statistical significance, with a p-value of 0.059 after accounting for autocorrelation through lag-1 pre-whitening, so the authors are careful to describe it as indicative rather than definitive. The wet season shows only a marginal, non-significant decrease of minus 0.0012 per year. In other words, Nigeria’s aerosol environment has been remarkably stable, with at most a gentle cleaning of its dustiest months.

The seasonal rhythm itself, however, is unmistakable. The monthly climatology is predominantly unimodal, peaking in February at a mean AOD of 0.865, when dry northeasterly harmattan winds carry desert dust deep into the country, and bottoming out in November at 0.370, as the monsoon transition washes the atmosphere clean. That February peak is more than double the November minimum, a swing that shapes everything from solar power output to aviation visibility to respiratory disease patterns across the region. Previous research has linked these dust episodes to meningitis outbreaks in the Sahel, and the harmattan season has been implicated in altered transmission dynamics of respiratory infections, making the timing and intensity of this annual maximum a matter of direct public health consequence.

To bring order to the daily chaos of aerosol variability, the researchers applied an unsupervised K-means clustering algorithm to the full record, letting the data sort itself into distinct atmospheric states. Three regimes emerged. Clean days, with a mean AOD of 0.399, account for 57.7 percent of the record. Transitional days, averaging 0.647, make up 31.6 percent. And then there are the Extreme days, averaging a staggering AOD of 1.163, which occupy 10.7 percent of all days in the record. These extreme events are overwhelmingly a dry season phenomenon: they occur on 18.1 percent of dry season days but only 3.2 percent of wet season days. That asymmetry confirms quantitatively what residents of Kano or Maiduguri know viscerally: the harmattan is when the air becomes a hazard.

One of the most practically useful contributions of the study is its characterization of how often extreme aerosol events recur. Using empirical return period analysis, the team calculated that days with AOD exceeding 1.0 arrive on average every 12.5 days, based on 711 exceedances across the record, with a 95 percent confidence interval of 11.6 to 13.4 days. Thresholds above 1.5 are crossed roughly every 61 days, with 145 exceedances recorded. And truly exceptional events, with AOD above 2.0, occur about once every 6.06 years, though with only four such events in 25 years the confidence interval is wide, spanning 3.03 to 24.25 years. These recurrence benchmarks give atmospheric hazard planners, health officials, and solar energy operators a climatological yardstick for what counts as a normal bad day versus a once-in-a-decade crisis.

The spatial structure of Nigeria’s aerosol burden is equally striking. A global spatial autocorrelation analysis yielded a Moran’s I of 0.949, with a p-value of 0.001 and a z-score of 163.9, confirming that aerosol loading is not randomly distributed but intensely clustered in space. Local Indicators of Spatial Association, corrected for the false discovery rate, flagged 40.1 percent of grid cells as statistically significant hotspots or coldspots. The pattern reveals a stable High-High hotspot belt across Southern and Southwestern Nigeria, where dense populations, urban pollution, and biomass burning combine, and persistent Low-Low spots in the Northeast. This geographic fingerprint has remained stable across the record, suggesting that the underlying drivers, from dust transport corridors to emission sources, are entrenched features of the regional environment rather than transient anomalies.

The study also probed the atmosphere’s memory, asking whether today’s aerosol conditions predict tomorrow’s. The answer is a strong yes, at least over short horizons. The lag-1 daily autocorrelation is 0.75, decaying to 0.21 by lag 30, indicating that aerosol states persist for days to weeks before the signal fades. An ARIMA (2,0,2) time series model successfully captured these synoptic dynamics, achieving a pseudo R-squared of 0.58. This short-range predictability has real-world value: it means that once a major dust outbreak begins, forecasters can anticipate with reasonable confidence that hazy conditions will persist for several days, giving hospitals, airports, and power grid operators time to prepare.

Taken together, these findings establish a rigorous long-term atmospheric baseline for a country whose air quality has historically been monitored far less intensively than its population warrants. The stability of the national trend, punctuated by a possible dry season improvement, offers a measure of reassurance, but the sheer frequency of extreme events, one in every nine days nationally and nearly one in five during the dry season, underscores the scale of the ongoing exposure. As West Africa’s combustion emissions continue to grow and climate change reshapes dust mobilization across the Sahara, records like this one will serve as the indispensable reference point against which future change is measured. For Nigeria’s policymakers, the message is clear: the dustiest skies are predictable, clustered, and recurring on a known clock, and that predictability is itself a tool for protecting public health.

Subject of Research: Long-term satellite analysis of aerosol optical depth dynamics, seasonal regimes, and extreme dust events over Nigeria from 2000 to 2024

Article Title: A 25-year analysis of aerosol optical depth dynamics over Nigeria (2000–2024): spatial-temporal dynamics, seasonal regimes, and extreme event characterization

Article References: Ekpe, E. E., Shi, C., Yin, S., Wang, H., Zhao, Y., Letu, H., & Raihan, M. Z. U. (2026). A 25-year analysis of aerosol optical depth dynamics over Nigeria (2000–2024): spatial-temporal dynamics, seasonal regimes, and extreme event characterization. Air Quality, Atmosphere & Health, 19(10), Article 224. https://doi.org/10.1007/s11869-026-02111-4

Image Credits: AI Generated

DOI: 10.1007/s11869-026-02111-4

Keywords: aerosol optical depth, Nigeria, MODIS Terra, machine learning bias correction, AERONET, Saharan dust, harmattan, air quality, spatial autocorrelation, extreme events, seasonal climatology, West Africa