The research, led by Michael Aich of the Technical University of Munich and the Potsdam Institute for Climate Impact Research, together with Sebastian Bathiany, Philipp Hess, Yu Huang and Niklas Boers, addresses one of the most persistent weaknesses in modern climate simulation: precipitation. Earth system models, or ESMs, are the workhorses of climate projection, numerically solving the equations of atmospheric motion on discretized grids. But their grids are coarse, typically on the order of one degree or more, and the processes that actually generate rain — convection, cloud microphysics, the organization of thunderstorm complexes — unfold at scales far smaller than a single grid cell. To cope, modelers rely on parameterizations: simplified, column-based schemes that approximate subgrid physics independently at each vertical column of the atmosphere. These schemes are computationally expensive, depend on subjective calibration, and ignore horizontal interactions between neighboring columns, which makes it difficult to capture phenomena such as mesoscale convective organization. The result is a suite of systematic biases, the most famous being the “double Intertropical Convergence Zone” problem, in which models paint excessive rainfall across the southern tropics.

The new framework sidesteps these column-based assumptions entirely. Instead of treating each atmospheric column as an isolated regression problem, the researchers’ models operate on global two-dimensional fields, allowing information to flow between distant locations and between different physical variables. The inputs are deliberately minimal: specific humidity at the 850 hectopascal pressure level, the eastward and northward wind components at ten meters above the surface, and mean sea-level pressure — all at a coarse one-degree resolution. Remarkably, the authors show that these four surface and near-surface fields contain enough implicit information about the three-dimensional state of the atmosphere to reconstruct detailed rainfall patterns. Mean sea-level pressure, for instance, captures the integrated mass of the entire air column, while surface winds are intrinsically coupled to the larger-scale circulation.

The pipeline works in two stages. The first is a deterministic UNet regression model, a convolutional neural network of roughly 24 million parameters with four downsampling stages built from residual blocks and a single bottleneck attention layer. Trained on the ERA5 reanalysis — a state-of-the-art dataset produced by the European Centre for Medium-Range Weather Forecasts that fuses observations with numerical weather prediction — the network learns to map the four coarse atmospheric variables to one-degree daily precipitation. A single forward pass through this network takes about nine milliseconds on a modern Nvidia H100 graphics card.

The second stage is where the generative magic happens. A conditional denoising diffusion probabilistic model, an architecture descended from the same family of techniques behind modern image generators, is trained to transform the coarse one-degree precipitation estimate into a genuine quarter-degree field — a sixteenfold increase in the number of grid cells, from a 180-by-360 global grid to 720 by 1440. Diffusion models work by learning to reverse a gradual noising process; here, the model is conditioned on a degraded, noisy version of the coarse precipitation field and learns to regenerate the fine-scale spatial texture that such coarse inputs lack. The researchers inject Gaussian noise of a carefully calibrated magnitude into the conditioning field during training, chosen by analyzing where the spatial power spectra of the coarse and fine datasets intersect, at a threshold scale of roughly 900 kilometers. Scales larger than this threshold are preserved from the conditioning input; scales below it are regenerated stochastically in a manner consistent with the ERA5 distribution. This design turns the diffusion model into both a downscaling engine and a bias-corrector: it wipes away the flawed, overly smooth small-scale features of the coarse input and replaces them with statistically realistic texture, while anchoring the result to the trustworthy large-scale structure.

Bridging the gap between reanalysis data and climate model output posed a further challenge, because the statistics of ESM fields differ from those of ERA5. The researchers solved this with quantile delta mapping, a bias-correction technique fitted on the historical overlap between the model and the reanalysis, applied to the regression output before it enters the diffusion stage. The noise-injection step then serves double duty as a domain adaptation device: by corrupting the small-scale information of the bias-corrected field to the same degree as during training, it renders the inference-time data distribution approximately identical to the training distribution, allowing the diffusion model to operate on climate model data it has never seen.

The results are striking. When applied to the GFDL-ESM4 model from the U.S. National Oceanic and Atmospheric Administration, the framework reduced the mean absolute precipitation bias from 0.561 millimeters per day in the raw climate model to 0.164 millimeters per day after the regression stage, and further to 0.111 millimeters per day after the diffusion stage. The notorious double ITCZ bias, glaring in the GFDL fields, largely disappears. Spatial power spectral analysis reveals why: the climate model’s precipitation is blurry, lacking variability at scales below about 400 kilometers, whereas the diffusion model’s output aligns closely with the observed spectrum down to the finest resolved scales. The generated fields also reproduce the distribution of rainfall intensities far more faithfully, capturing both the frequency of moderate events and the rare extremes exceeding 150 millimeters per day.

Extreme events, arguably where accurate precipitation matters most for society, received particular attention. Using the R95p index — the annual total precipitation falling on days above the 95th percentile — the team found that the AI-generated fields track ERA5 closely during the historical period, while the raw GFDL model overestimates tropical extreme rainfall. Under the SSP3-7.0 high-emissions scenario, all approaches agree that extreme precipitation will intensify in the tropics by the late twenty-first century, but the diffusion model projects a more moderate amplification than either the raw climate model or the regression stage alone, suggesting it partially corrects the wet bias baked into the driving model. Indices of consecutive wet and dry days, critical for flood and drought assessment, are likewise reproduced more faithfully than by the original ESM.

A distinctive advantage of the generative approach is uncertainty quantification. Because the diffusion model is stochastic, running it fifty times on the same coarse input yields fifty equally plausible high-resolution realizations, forming an ensemble that reflects the inherent ambiguity of inferring fine detail from coarse data. Evaluating this ensemble with the continuous ranked probability score over a validation year, the researchers found a mean score of 0.56 millimeters per day — better than both a bi-linearly interpolated ERA5 baseline at 0.73 and the interpolated regression output at 1.42 millimeters per day. The spread-skill relationship of the ensemble roughly follows the ideal one-to-one line, indicating that the model’s stated uncertainty is well calibrated rather than overconfident or underconfident.

Perhaps the most consequential test involved the future. The SSP3-7.0 scenario is, from the model’s perspective, an out-of-distribution challenge: a warming world with systematically higher specific humidity. The regression stage alone produced precipitation with unrealistically inflated inter-annual variance, yielding a projected rainfall trend far steeper than that of the driving climate model — a known artifact of applying quantile mapping to overly smooth deterministic outputs. The diffusion stage corrected this, damping the spurious variance and aligning the global average trend with the GFDL projection while refining the spatial pattern of change. Crucially, when the pre-trained framework was applied without retraining to a second, independent model — the Max Planck Institute’s MPI-ESM-HR — it preserved that model’s own climate signal rather than forcing the data toward the ERA5 historical trend, demonstrating that the framework adapts to the input climate model rather than overfitting to its training distribution. The authors caution, however, that the approach implicitly assumes the statistical relationship between large-scale circulation and small-scale precipitation, learned from present-day observations, remains valid under future warming.

The computational economics of the method are a major part of its appeal. Running a full Earth system model at quarter-degree resolution for large ensembles remains prohibitive; generating a single global high-resolution precipitation field with the new framework takes roughly two seconds on one GPU, and each of the two networks contains fewer than 25 million parameters. This opens the door to the large ensembles of high-resolution precipitation projections that impact assessments, flood management and water resource planning urgently need but have never had. It also enables a practical decoupling: climate models can continue running at coarse resolution while precipitation is generated afterwards as a post-processing step, coupling directly to land surface and impact models without inflating the cost of the underlying simulation.

The authors are careful to frame the work as a proof of concept rather than an operational replacement for fully coupled three-dimensional parameterizations. The model is trained frame by frame, so the stochastic fine-scale details it generates are not explicitly temporally consistent from one day to the next — a limitation that matters less for daily data than for sub-daily applications, but one the team hopes to address with video diffusion architectures. ERA5 itself carries known biases, including a tendency toward wet conditions and an underestimation of extreme intensities, and the framework would need retraining on other datasets for specific operational uses. Integrating the scheme directly into a running climate model, extending it to predict three-dimensional tendencies, enforcing physical conservation laws, and accelerating the sampling process through distillation all remain open challenges, as does the risk of instability and drift that has plagued earlier machine learning parameterization efforts.

Even so, the study marks a significant step in the ongoing convergence of generative artificial intelligence and climate science. It follows a wave of machine learning successes in weather forecasting, but tackles a harder problem: not predicting the atmosphere’s next state, but reconstructing the hidden fine structure of rainfall from its coarse, resolved circulation — and doing so in a way that reduces bias, quantifies uncertainty, and respects the climate change signal. If the approach can be hardened for fully coupled deployment, the blurry rain maps of tomorrow’s climate projections may acquire the crisp, physically plausible texture of the real thing.

Quantifying global water-cycle components such as river discharge and land evapotranspiration with high accuracy has posed a persistent challenge in hydrological and climate sciences. Conventional Earth system models, while invaluable, have shown biases that skew projections critically. The research team applied an emergent constraint method, a sophisticated technique that combines predictions from various models with real-world observational data to refine estimates and reduce uncertainties. This methodology facilitates an unprecedentedly reliable quantification of past and future global river flow and evapotranspiration under a warming climate.

Between 1980 and 2014, global river flow was estimated to be approximately 39.1 ± 5.4 thousand cubic kilometers per year, revealing lower values than previous estimates. The ratio of river flow to precipitation was found to be 0.35 ± 0.03, also falling below earlier projections. These revised figures suggest that previous models may have systematically overestimated the contributions of river systems to global water turnover, shedding light on potential overpredictions in hydrological budgets used in climate impact assessments.

Simultaneously, the global land evapotranspiration—the combined process of evaporation from land surfaces and transpiration from vegetation—was evaluated at approximately 73.4 ± 6.2 thousand cubic kilometers per year. This updated figure enhances the accuracy of water flux estimates across continental landscapes, which are critical for understanding terrestrial water availability and ecosystem health. The dual constraints on river flow and evapotranspiration together paint a clearer picture of terrestrial water dynamics and their sensitivity to climatic variations.

The study’s projections for the future indicate a nuanced yet concerning trajectory. Under climate change scenarios, global river flow is expected to rise by 7.8 ± 5.5 millimeters per year per degree Celsius of warming. This figure is approximately 9.3% lower than the mean increase projected by the ensemble of Earth system models without emergent constraints. The reduction in expected river flow increase suggests that the hydrological response to global warming may be less intense than previously assumed, although still significant enough to warrant close attention.

One of the most impactful outcomes of this research is the 66% reduction in inter-model uncertainty achieved through the emergent constraint approach. This dramatic narrowing of confidence intervals bolsters the reliability of future projections and serves as a methodological template for refining other climate and environmental models. By incorporating real observational data systematically, model outputs become not only more precise but also better aligned with physical realities observed on the ground.

The implications of these refined estimates extend beyond academic interest, influencing water resource management, agricultural planning, and flood risk assessment. As climate change progresses, accurately predicting the availability and distribution of freshwater resources is pivotal for mitigating adverse impacts on societies and ecosystems worldwide. Overestimated projections can lead to misallocation and inefficient management, whereas underestimated ones may yield risks unmitigated. This study provides a balanced recalibration necessary for informed policy and adaptation strategies.

It is well-known that river systems act as vital integrators of terrestrial hydrological processes, linking precipitation, surface runoff, and groundwater flow into coherent discharge patterns. This research capitalizes on the magnitude and diversity of river flow observations from major basins, harnessing their integrative nature to constrain model outputs robustly. The global reach—spanning diverse climatic zones and catchment characteristics—lends robustness and generalizability to the emergent constraint findings.

Evapotranspiration, too, is a critical water-cycle component closely tied to vegetation dynamics and energy fluxes. The updated evapotranspiration estimates contribute to a more accurate global water balance, crucial for modeling climate feedbacks such as soil moisture deficits and drought severity. Understanding these interactions aids in predicting how ecosystems will adapt or degrade under future climatic stressors, potentially influencing carbon cycling and biodiversity.

Methodologically, the emergent constraint approach used here represents a sophisticated fusion of theoretical model ensembles and empirical observation, designed to leverage the complementary strengths of each. This innovative statistical technique identifies consistent relationships—emergent constraints—that allow observed variables to narrow the range of model outputs, improving predictive skill. Its successful application to global river flow marks a significant stride in hydrological modeling, offering a pathway for refinement in other complex Earth system components.

Equally noteworthy is the study’s treatment of historical variability to anchor projections more firmly. By cross-validating model ensembles against observed river flows during several decades, the researchers establish a baseline that captures natural climate variability alongside long-term trends. This aspect is crucial for avoiding biases arising from transient anomalies and enhances confidence in attributing observed changes to anthropogenic climate influences.

These refinements collectively highlight that Earth system models alone may not fully capture the complexity and regional heterogeneity inherent in hydrological cycles. Integrating observational data, particularly at the basin scale, provides critical checks and balances, addressing over-simplifications and improving spatial and temporal resolution. This hybridized approach underscores a paradigm shift toward model-observation synergy in climate and hydrological sciences.

Beyond scientific precision, the study’s implications resonate in sectors reliant on reliable water availability projections. From agriculture, which depends on sustained water supplies for crops, to urban planning focused on flood defenses and infrastructure resilience, accurate forecasts are indispensable. This robust recalibration of the global water-cycle components informs adaptive management strategies to better safeguard human and ecological well-being against uncertain climatic futures.

The research also reminds us of the challenges inherent in projecting complex environmental systems amid climate change. It emphasizes the need for continuous refinement of models and observational networks, as well as the importance of integrating diverse data streams to address uncertainties and biases. Advancing these integrative methodologies will be paramount to maintaining accurate and actionable forecasts as climate change accelerates.

Looking ahead, the findings invite further investigations into the mechanisms behind model discrepancies and the representation of hydrological processes, such as soil moisture dynamics, groundwater flow, and vegetation feedbacks. Enhanced Earth system models, informed by emergent constraint techniques, will be better positioned to anticipate regional impacts and extremes, ultimately guiding more resilient water resource governance frameworks.

Overall, this study marks a pivotal contribution in hydrological science, balancing the scales of model projections with real-world observations and setting a new standard for accuracy in global water-cycle estimation. By critically reappraising past and future estimates of river flow and evapotranspiration, the research offers a vital recalibration for climate impact assessments and resource management in a warming world.

Subject of Research: Global water cycle quantification, river discharge, land evapotranspiration, Earth system model validation and refinement.

Article Title: Overestimation of past and future increases in global river flow by Earth system models

Article References:
Zhang, Y., Blöschl, G., Wei, H. et al. Overestimation of past and future increases in global river flow by Earth system models. Nat. Geosci. (2026). https://doi.org/10.1038/s41561-025-01897-9

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41561-025-01897-9

Keywords: global water cycle, river flow, land evapotranspiration, Earth system models, emergent constraint, climate change projections, hydrological uncertainty reduction, freshwater resources