High-intensity rainfall flooding is an escalating global urban hazard, with exposure growing as cities expand and climate change intensifies. Increasing short-duration extremes are driving more frequent, severe flooding, raising damages, and disproportionately impacting vulnerable communities. These trends highlight the need for flood modeling approaches that are both high-resolution and computationally efficient to support real-time forecasting and operational decision-making. This study evaluates SynxFlow, a GPU-accelerated hydrodynamic model designed to deliver rapid, neighborhood-scale forecasts. Using gridded precipitation fields, SynxFlow simulated flood extent, depth, and velocity at fine spatial resolution across Cook County, Chicago, achieving short runtimes suitable for operational use. Validation against satellite-derived flood observations for a major storm event showed strong agreement, while a conventional CPU-based workflow substantially underestimated inundation. Overall, GPU-enabled hydrodynamic modeling can deliver accurate, near-real-time flood intelligence to strengthen warning systems, support equitable emergency response, and guide resilience investments.

Gauge-independent, multi-source precipitation data merging is a well-established approach for improving precipitation estimation across regions and scales. However, existing merging techniques often assume precipitation follows a unique, state-independent statistical distribution, overlooking the inherently mixed nature of rain and no-rain occurrence states. This assumption leads to two major limitations: (i) uncertainty in rain/no-rain classification and (ii) biased merging weights, resulting in suboptimal precipitation magnitude estimates. Here we introduce RainMerge, a novel two-stage merging framework designed to address these limitations. In Stage 1 (Binary Merging), we employ the Categorical Triple Collocation Merging (CTC-M) to enhance rain/no-rain detection by optimally combining three input datasets. In Stage 2 (Conditional Merging), we use the binarymerged product to conditionally group samples in time, and merge rain-flagged groups with at least two positive detections using Signal-to-Noise Ratio Optimisation (SNR-opt) to refine precipitation magnitudes. Using gauge observations from diverse hydroclimatic settings, we show that RainMerge improves daily rain/no-rain classification by 25-60 % in False Alarm Ratio (FAR) and Heidke Skill Score (HSS) relative to the parent products (ERA5, SM2RAIN, and IMERG-final). In terms of overall magnitude error (RMSE), RainMerge reduces the parent-product RMSE by 2-6 mm/day and outperforms the state independent (termed One-stage) framework, by 1.5-2.5 mm/day. Additionally, RainMerge reduces underestimation bias by 2-8 mm/day for extreme precipitation events (75th-99th percentiles) compared to One-stage, demonstrating its superior performance in capturing heavy events. These results highlight the potential of RainMerge for hydrological and climate applications, offering a globally applicable approach to improve detection and magnitude accuracy in multi-source precipitation products.

Abstract Effective water resource management under climate change is dependent on reliable modeling of future water availability, yet significant uncertainties remain due to shifting precipitation patterns and limitation in climate modeling. This study investigates a multiple‐lines‐of‐evidence approach aimed at reducing the uncertainty in projections of streamflow, specifically focusing on extreme flood events as proxies for total annual flows. Using subdaily rainfall‐runoff models calibrated for four climatically diverse Australian catchments, we compare projections from regional climate model (RCM) downscaling with a continuous precipitation generation approach conditioned on stable climatic covariates such as temperature. Our findings suggest that in wetter regions with pronounced extreme precipitation events, using flood events as proxies for total annual flows can reduce variance and bias, potentially offering more reliable inputs for water resource planning. Both continuous simulation and RCM downscaling demonstrated similar biases for rainfall when evaluated against historical observations. However, continuous simulation typically produced lower biases in modeled streamflow, and performed marginally better in representing low‐frequency climate variability; whilst also offering reduced computational demands, presenting a parsimonious alternative for climate impact assessment. Despite the improved precipitation representation by RCMs, compared to General Circulation Models (GCMs), epistemic uncertainty and sampling biases persist, limiting the confidence in projections of extreme streamflow events for water supply modeling. These results highlight the need for improved modeling of precipitation extremes under warming climates to refine and enhance the robustness of future water resource projections.

Accurate measurement of river discharge is central to flood risk management and water resources planning, operations and management. However, in data-scarce and information-restricted settings, including transboundary basins, satellite observations are widely used; nevertheless, consistent monitoring remains difficult for small and narrow rivers. Here we show that surface water dynamics observed by Sentinel-1 synthetic aperture radar (SAR) can be transformed into a hydrologically meaningful surrogate discharge signal, providing a reasonable estimation of streamflow without reliance on ground-based ancillary observations. We develop a scalable SAR-based framework that extracts temporally consistent surface water extent by combining adaptive thresholding with connectivity- and geometry-informed filtering, yielding a robust surrogate river discharge (SR) time series. Across 19 rivers selected across multiple continents (13 countries) spanning diverse hydroclimatic and geomorphic settings, SAR-derived SR exhibits strong and systematic relationships with observed streamflow, outperforming commonly used alternative surrogates particularly in small and complex river systems. As a downstream demonstration of hydrological utility, when integrated into a parsimonious rainfall–runoff model using satellite-derived precipitation and evapotranspiration, the SAR-based SR can help constrain physically defensible, continuous streamflow simulations, a desired necessity in data-scarce basins. Our results demonstrate when and where SAR observations can reliably inform river flow dynamics, highlighting both the potential and the limitations of surface water extent–based discharge inference for ungauged settings.

Simulation experiments in a high-resolution configuration of the Weather Research and Forecasting Model are used to test the hypothesis that land-surface vegetation inhibits propagation of rainstorms into the Upper Illinois River Basin (UIRB), thus decreasing associated total precipitation (Pt) and flooding. Two historical flood-generating rainstorms, representative of storm types in the UIRB, are selected and simulated: a mesoscale convective rainstorm in July 1996 and the remnant of the Hurricane Ike rainstorm in September 2008. For each rainstorm, three sensitivity experiments with differing land-surface vegetation configurations are simulated and compared with the reference experiment. Results show that vegetation changes inside the UIRB or inside a 1 degrees belt around it caused an increase in basin-average Pt during the 2008 rainstorm. However, these same vegetation configurations caused a decrease in basin-average Pt in the 1996 rainstorm. The largest decrease, between 18% and 32%, occurred in the experiment with a belt of trees. The mechanism for this decrease in Pt is windspeed decrease associated with increase in surface roughness owing to vegetation with high vertical extent, which inhibits propagation of the rainstorm into the basin. These findings highlight a linkage between land-cover characteristics outside the UIRB and precipitation inside it associated with mesoscale convective storms. Since precipitation is the key driver of flooding, this linkage can be combined with other structural approaches toward developing a long-term flood mitigation strategy for the UIRB.

Urban heat risk is intensifying, yet many cities continue to rely on static assessment approaches that inadequately capture how risk evolves spatiotemporally. We argue for a more dynamic and context-sensitive approach to urban heat assessment and action. We outline five shifts that recognize the interacting roles of hazard, exposure, vulnerability, and adaptive responses, and propose a decision-support framework to support more targeted, equitable, and actionable urban heat preparedness and adaptation.

Present-day road transportation is largely fueled by traditional fossil fuels and has unintended adverse effects through emissions of greenhouse gases, anthropogenic heat, and air pollutants, thereby affecting heat, air quality and public health. A thorough assessment of current environmental and health burdens helps clarify how future transport strategies could improve outcomes. Here, four case studies are presented – greenhouse gas emissions in the UK, anthropogenic heat emissions in Chicago (US), air pollutant emissions in Indian megacities, and air pollutant concentrations and associated health burden in the West Midlands (a major metropolitan area in the UK). By examining diverse regions worldwide, this chapter highlights that while mitigating road transportation impacts is a global challenge, its influence on emissions, air quality, heat waves, and public health is shaped uniquely by regional challenges and policy outcomes, and so addressing this requires region-specific targeted interventions.

The objective of this study is to investigate how large historical floods could become under "optimal" rainfall conditions, and which are the main factors that drive flood maximization. Observed storms are stochastically modified to create an ensemble that enables sampling flood extremes. This stochastic modification entails first the amplification of the intensities, followed by modification of the spatial patterns of precipitation. The rainfall intensities are modified using moisture maximization; the spatial patterns are changed by conditional rainfall simulation. The simulated precipitation is then used as input for a rainfall-runoff model to simulate corresponding floods. The methods are illustrated using flood event data from the Mulde river catchment in Germany. The results show that reasonable modifications on observed storms may lead to very extreme flood events. Additionally, the largest simulated precipitation events and floods can also be considered as plausible estimates of probable maximum precipitation and probable maximum flood, respectively.

Abstract Extreme sea levels resulting from storm surges are a major contributor to coastal flood risk and are often assessed using outputs from global storm surge models. However, these models exhibit systematic errors, particularly in data‐sparse regions. Here, we present a novel framework to improve simulated extreme water levels. The method applies quantile mapping to peaks‐over‐threshold extremes, is calibrated using paired observed and simulated annual maxima at tide gauges, and transfers the correction models from reference stations to nearby sites within a defined radius. Results show that the approach achieves consistent improvements within about 400 km of gauges, with median relative improvements of −0.84 in mean squared error and −0.60 in normalized root‐mean‐square error. Systematic errors are reduced most in high‐latitude, low‐wind regions. These findings highlight the transferability of the correction models and show that even sparse observations can improve extreme sea‐level estimates for global coastal hazard assessment.

Abstract Systematic biases in climate model outputs, particularly in extremes, limit their reliability for wind‐related applications. This study applies the Wavelet Bias Correction (WBC) method to daily near‐surface wind components (uwnd and vwnd) from three CMIP6 models (ACCESS‐CM2, HadGEM3‐GC31‐LL, and UKESM1‐0‐LL) over current (1979–2014) and future (2051–2086) periods, using National Center for Environmental Prediction reanalysis 2 reanalysis as a reference. The 1521‐grid study domain spans the western‐southern Pacific and Indian Oceans. Operating in the time‐frequency domain, WBC improves both magnitude and variability of simulated winds while explicitly preserving the physical co‐dependence of the u and v components, ensuring directional consistency. Percentile‐based evaluation shows that models generally underestimate median and extreme winds at the 99th percentile. Bias‐corrected projections indicate fewer extreme wind events (>49 km/hr) than in the current climate, consistent with observed declines in cyclone frequency. Results highlight WBC as a robust, physically consistent correction framework for wind simulations.

Understanding regional flood changes and their drivers is critical for risk management, yet challenges persist in flood regionalization due to uncertainties of catchment descriptors in defining hydrologically similar regions and integration of ungauged catchments. Here we develop a consensus-based machine learning framework to objectively identify homogeneous flood regions. The framework combines hierarchical and ensemble clustering while mitigating descriptor-dependent biases. We apply this framework to continuous observations of annual flood peaks (including peak magnitude and timing) from 1111 stream gauging stations across China during the period 1980-2017. We identify 20 homogeneous flood regions, with the similar flood regime within each region but distinct from the others. The indices characterizing climatologically mean flood regime are more influential (e.g., mean normalized flood discharge) than its interannual variability (e.g., coefficient of variation of flood discharge) in flood regionalization. Regional aggregation reveals stronger temporal persistence in annual flood peak series than seen at individual stations, highlighting the utility of regionalization in understanding regional flood changes. Chinese annual flood peaks demonstrate predominant trends of decreasing flood peak magnitudes (i.e., 15 out of 20 regions) and delayed occurrence (i.e., 15 out of 20 regions). The spatially coherent flood changes are primarily climate-driven, though dominant controls vary regionally between rainfall, snowmelt, and soil moisture dynamics. As the first comprehensive national-scale assessment of Chinese flood regimes, we provide a transferable framework for climate impact analysis and critical insights for adaptive flood management.

Urban weather and climate modeling is challenged by the highly heterogeneous and dynamic nature of cities. It exhibits a persistent trilemma between spatial granularity, spatiotemporal coverage, and physical interpretability. We articulate this challenge and propose a hybrid framework integrating physics-based models, urban observations, and machine learning. Framing this challenge as an integration problem across methods and scales, we provide a structured guide for next-generation, decision-relevant urban weather and climate modeling.

Understanding the dynamics of compound dry-hot extremes (CDHEs) is critical for climate risk assessment in semi-arid regions. This study over the state of Rajasthan in India investigates the evolution of rainfall deficits (defined by the 30th percentile precipitation threshold, P30) and temperature extremes (defined by the 70th percentile temperature threshold, T70), and their co-occurrence over six decades (1960-2024) using regridded district-level TerraClimate reanalysis data. Extreme thresholds were identified through percentile-based indices, and both frequency and persistence of events were analyzed. A first-order Markov chain framework developed based on exiting methodology was further applied to estimate stationary and transition probabilities, providing insights into long-term likelihoods and state-to-state shifts of compound extremes. Results show a fundamental spatio-temporal shift in extremes. Long-duration rainfall deficits that previously dominated the eastern and south-eastern regions have declined and are replaced by shorter but more frequent P30 events. Simultaneously, T70 events have increased sharply in frequency and persistence across the state, leading to more common cooccurrence of CDHE conditions. The early 2000s marked a peak in CDHE frequency and mean duration, particularly across central, south-western, and south-eastern Rajasthan. Stationary probabilities confirm that rainfall-only droughts since the 2000s occur infrequently, while heat-only and compound CDHE states persisted for all study period (1960-2024). Transition probabilities further indicate that both rainfall deficits and heat extremes have a high likelihood of evolving into compound states, with strong persistence once compounding begins. These findings highlight that Rajasthan's climate is moving into a regime defined by frequent and intensified compound extremes. The shift poses adverse implications on agriculture, water resources, and livelihoods, underscoring the urgent need for designing integrated monitoring and adaptation strategies for similar semi-arid regions across the globe to address concurrent rainfall variability and rising temperatures.

Calibrating conceptual rainfall-runoff models to predict a catchment's response to extreme rainfall events is a significant challenge, especially when historical data is limited and the event exceeds recorded extremes. This challenge is amplified in non-stationary environments where the frequency and intensity of extreme rainfall and floods are increasing. Traditional parameter estimation methods thus may not adequately capture the behavior of extreme events. To address this, we propose an Adaptive Markov Chain Monte Carlo (MCMC) algorithm that incorporates a model robustness-likelihood function; a novel approach in the context of hydrologic parameter sampling. Our two-stage sampling method first uses an automated scheme to identify the posterior distribution of parameter sets, then in the second stage, we sample robust parameter sets from this posterior using a likelihood function based on hypothetical (design) storm events. We demonstrate this approach using the GR4H model applied to a catchment in Victoria, Australia. The robust parameter sets selected through this process exhibit reduced variability across simulations, providing more reliable predictions for extreme events not observed in historical data. This methodology increases confidence in simulating hydrologic responses to unprecedented events. While use is made of probabilistic design storms in defining the robust parameter space, this rationale can be extended to incorporate other alternatives including site analogs, future climates or paleo reconstructions across the broad realm of hydrological applications.

Urban flooding poses a growing global challenge, disrupting urban transportation systems and triggering cascading impacts on mobility, infrastructure, and economies. In this study, historical flood patterns across the Chicago metro area were identified using flood depth at a neighborhood scale. This book chapter offers insights into how historical and future flood significance levels can help planners and commuters to predict and manage transportation infrastructure in urban environments. The proposed framework integrates crowdsourced validation with real-time data from residents and hydrodynamic models to provide information on flood significance on node and route-levels. Using pattern recognition algorithms and GIS-based feature extraction, the framework classifies intersections and traffic signal nodes into low, medium, and high flood-risk categories, providing actionable insights for infrastructure and emergency planning. By integrating flood risk levels on nodes, the proposed approach will help urban planners to determine the safest travel path that minimizes flood risks.

Process-based build-up/wash-off models underpin urban stormwater quality management, yet structural simplifications and environmental non-stationarity often induce systematic biases that erode decision reliability. This "Making Waves" article explores a pivotal question: Can biases, long treated as model "defects", be reframed as information assets that strengthen decision-grade predictive credibility? This paper proposes a model-agnostic machine learning error correction strategy. The core idea is to recast model-observation errors not as random noise but as structured, information-rich signals carrying "memory bias". By learning the dynamic patterns within recent errors, this correction layer enables real-time, non-intrusive adjustments to the physical model's output. In the study catchment, it raised pollutant NSE compared baseline by roughly 20 % on average with hydrologic inputs and 66 % with lagged errors. Testing across two catchments demonstrates that the proposed framework achieves superior accuracy and stability compared with traditional and pure machine-learning baselines. This approach evolves traditional models from static tools requiring periodic recalibration into responsive systems capable of dynamic output correction. By making systematic biases 'learnable' and 'usable', the model can effectively respond to environmental fluctuations, ensuring its long-term validity and reliability in a constantly changing environment.

Abstract Traditional residual post‐processing methods have demonstrated substantial improvements in simulated hydrological response; however, they typically aggregate multiple sources of uncertainty into a single lumped residual and apply uniform statistical treatment across all hydrologic conditions. This uniform treatment overlooks the fact that error characteristics can vary systematically across different rainfall and flow regimes. Given the fundamentally different processes governing catchment storage accumulation and depletion, mainly driven by the presence or absence of rainfall, there is a strong rationale for employing distinct residual models tailored to specific hydrologic states. We introduce here a novel two‐stage residual modeling framework that represents residuals into their hydrological‐state dependent components. Using hydrologic understanding to define the states and storage‐related information from the pre‐existing hydrologic model to inform residual prediction, we use deep learning to characterize residual behavior in each state and improve streamflow simulations via state‐dependent error correction. We illustrate the usefulness of the proposed model using selected catchments across the varying climatic zones in Australia. In the Warren River Catchment, the proposed model reduces RMSE by 59%–91% overall and by 65%–86% in the upper‐tail flow regimes, while also yielding consistent gains in NSE. These results suggest that hydrologically conditioned residual modeling can improve both bulk predictive performance and the representation of hydrologically important high‐flow events. The framework also retains utility under limited hydrologic model calibration, which may be valuable in operational settings where pre‐existing hydrologic models continue to be used.

Lake breezes can strongly influence the environment in coastal areas. As the most populous city along Lake Michigan, Chicago frequently experiences lake breezes, yet their impact on air quality remains poorly understood. Using one-year measurements from meteorological and the Microsoft Eclipse low-cost sensor networks, we identify 42 lake breeze events in 2022 and assess the net effects of lake breezes on the urban environment in Chicago. We find that lake breezes occurring on weekdays generally reduce ozone (O3) mixing ratios on average up to ∼6 ppb along the northeast shoreline, contrary to many other coastal regions, where sea/lake breezes drive O3 exceedances. Reduced O3 during lake breeze events likely result from a combination of processes, including enhanced NO titration in a shallow boundary layer, especially during the early stage of the event and in areas with heavy emissions, the inflow of cleaner air, and the lower temperatures that suppress the O3 formation by reducing precursor emissions and slowing-down reaction rates. In contrast, on average, lake breezes enhance fine particulate matter (PM2.5) concentrations by up to ∼3 μg m-3, due to the accumulation of local emissions within a shallower boundary layer and enhanced partitioning of semivolatile species at lower temperatures. Additionally, we identify 51 weak events during which lake breezes remained mostly within several kilometers of the lakeshore. These weak lake breeze events show broadly consistent effects on temperature and O3, albeit with smaller magnitudes. These results highlight the interactions of emissions, atmospheric chemistry, and meteorological conditions during the lake breeze events and their impact on the urban environment, underscoring the need for region-specific evaluations of sea/lake breeze effects.

Recent flood risk assessment studies exhibit two main limitations: (1) insufficient attention to building-level analyses, despite the spatial heterogeneity of flood hazards, and (2) restriction to a single spatial scale. To address these gaps, this study proposes a comprehensive framework for building-level flood risk assessment by integrating hazard, exposure, and socio-economic vulnerability (SEVI) using a multiplicative index-based approach. While previous studies have advanced building-level hazard and population exposure estimation, vulnerability assessment at this fine scale remains challenging. In this study, hazard is represented by inundation depths associated with a 100-year return period. Population exposure is downscaled from 100 m gridded data to individual buildings using the Cadastral Expert Dasymetric System (CEDS) method. SEVI is downscaled from the subdistrict level to buildings using housing prices. The upstream Yeongsan River Basin of South Korea is selected due to the presence of major agricultural dams and densely populated regions. Results show that (1) the area containing the highest number of high-risk buildings is also the only high-risk subdistrict; (2) the dominant proportion of buildings in a given risk class does not necessarily reflect subdistrict-level risk; and (3) high-risk buildings may occur in less severe flood zones, driven by high population density and low adaptive capacity.

Encapsulating soil water drying attributes in derived remote sensing products is essential for their application to various environmental studies. Systematic differences are observed for soil moisture (SM) drying phases (drydown) in Soil Moisture Active Passive (SMAP) level 4 (SMAP L4) product when compared to ground observations due to differences in measurement processes. An algorithm (BRF: Bivariate Recursive Filter) is able to translate drydown parameters, namely the drydown recession coefficient and initial wetness, to in-situ scale to enhance consistency with regional in-situ observations. Here, we propose a reconstruction procedure to structure the complete SM time series so that it is compatible with the bivariate transformed drydown attributes. The present study develops the reconstructed SM time series at any grid location SMAP L4 data is available at. The method is validated with 6 global in-situ networks and improvements have been observed in the reconstructed SMAP SM series. Significant enhancements are observed particularly during the drydown periods. Additional improvement is noticed when observations from spatially nearby in-situ stations for a target grid is used; this can assist in regional hydrological studies whenever ground data is available.