AAAli A. Ameli

Ali A. Ameli

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Ali A. Ameli has a total of 47 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Catchment hydrologic function is widely time-variant

    by Ying Yan, Hamed Sharif, Ali A. Ameli

    Catchment hydrologic function describes how meteorological conditions are transformed into streamflow. Whether this transformation is stable or varies from event to event, and what controls that variability, remains unknown at the global scale, limiting confidence in flood-risk and water-security assessments. Here, we analyze about 1.7 million runoff-producing events across 4838 catchments on six continents. We classify each event by its water-input type (rain, snowmelt, or rain-on-snow) and by its antecedent soil moisture (dry, moderate, or wet). These two descriptors together define the event’s hydro-meteorological conditions. We summarize each catchment’s dominant hydro-meteorological event type in the dormant and growing seasons separately, and find a wide diversity of dominant types across the global catchment population. We then quantify, for each catchment, the extent of event-to-event time-variance in the precipitation–streamflow transfer function that represents catchment hydrologic function. Time-variant catchment hydrologic function is pervasive: 95.6% of catchments in the dormant season and 99.0% in the growing season show moderate to strong event-to-event variation, and even catchments experiencing highly similar hydro-meteorological conditions across events remain functionally time-variant. Within-catchment spatial heterogeneity in hydro-meteorological conditions modulates this variation. Greater time-variance in hydrologic function is associated with greater event-to-event variation in a catchment’s runoff response.

  • Open Access

    Interrogating the Functional Realism of Deep Learning Rainfall–Runoff Models: Diagnostic Insights and Mitigation Strategies

    by Ara Bayati, Ali A Ameli, Saman Razavi

    Deep learning rainfall–runoff models can achieve high predictive accuracy, yet still rely on correlation-driven shortcuts that are not defensible as catchment-scale mechanisms. This raises a central question: how far can correlation-driven learning be trusted to produce simulations that are hydrologically realistic, not just statistically accurate? To address this, we evaluate functional realism, defined as the extent to which a model’s internal functioning aligns with defensible mechanisms of streamflow generation. We propose a hydrology-specific Explainable AI (XAI) framework that extracts nonlinear, lag-dependent, time-varying impulse response functions (IRFs) describing how an LSTM internally maps isolated impulses in precipitation (P), temperature (T), and PET to simulated streamflow. Applied to 672 North American catchments where the LSTM demonstrated strong predictive skill, the IRFs reveal systematic functional inconsistencies masked by accuracy: in over 70% of rain-dominated catchments, short-term rises in T are associated with increased simulated streamflow and enhanced celerity even without rainfall; in snow-dominated catchments, PET is frequently treated as a proxy driver of snowmelt-related flow. We then discuss plausible origins of spurious functional learning, including seasonal confounding, heterogeneous regime mixing during training, simplicity bias (shortcut learning), and omitted drivers or missing processes. We also outline practical routes to reduce spurious learning by directly addressing these sources through input handling, regime-aware training, and targeted model adjustments.

  • Open Access

    Synthesizing Time-Variance in Catchment Hydrologic Function: Patterns, Controls, Methodological Advances and Global Extrapolation to Ungauged Basins

    by Ali Ameli

    Time-variance in catchment hydrologic function—how rainfall-runoff relationships shift across events, seasons, and years—remains a fundamental yet incompletely understood aspect of catchment behavior. Synthesizing this time-variance across scales is essential for advancing hydrological theory, improving predictions in ungauged basins, and guiding model development. Here, I present a synthesis of recent work spanning local to continental to global scales, integrating fine-resolution mechanistic models with large-sample data-driven methods.At the local scale, process-based modeling reveals how subsurface physical structure—including subsurface lateral permeability pattern—mediates the climate-induced time-varying partitioning of water between long-term storage, shallow and deep flow paths, and evapotranspiration. At continental to global scales, large-sample analyses across more than five thousand gauged catchments and 80,000 ungauged catchments expose systematic patterns in functional complexity (or time-variance): most catchments exhibit strongly time-varying rainfall-runoff behavior, with climate (particularly rainfall persistence and aridity) providing the dominant control, while geology and topography modulate outcomes locally.To enable these syntheses, we developed new data-driven methodologies for extracting catchment hydrologic function and quantifying its temporal variation from observational records. These methods provide a transferable framework for diagnosing functional behavior in gauged systems. These findings advance process-based explanations of hydrological phenomena across places and scales, connect event-scale dynamics to seasonal and long-term patterns, and offer new tools for identifying hydrological signatures in data. The implications extend to model structure selection, monitoring network design, and the development of a unifying hydrological theory that accommodates—rather than assumes away—functional time-variance.

  • Open Access

    Wetlands set the pace of annual runoff in the northern Great Plains

    by Javad Rahmani, Irena F. Creed, Pascal Badiou, Ali A. Ameli

    Interannual variability in the runoff ratio—how much annual precipitation becomes streamflow—underpins water management, flood forecasting, and biogeochemical fluxes. In most hydrologic frameworks, this variability is attributed primarily to year-to-year climate drivers. Here, we show that in North America’s wetland-rich Prairie Pothole Region, wetlands play the dominant proximate role. Using 38 years of satellite-based inundation maps and hydroclimate data from 109 catchments, we find that annual wetland inundation extent explains interannual runoff and high-flow variability more strongly than any annual or intra-annual climate index in 69% of catchments. Climate, especially snow persistence, affects wetland inundation extent, but wetland inundation exerts a stronger net control on runoff and ultimately sets the pace of annual runoff through fill–spill hydrology. Catchments exhibit wetland inundation–runoff relationships ranging from linear to strongly threshold-like, with threshold-like behavior predominant—particularly where Geographically Isolated Wetlands are abundant. These findings reveal wetlands as active regulators of ecosystem water balance and provide a landscape-explicit basis for forecasting, conservation, and adaptive water management across the region.


  • Open Access

    Regionalization of Hydrologic Behavior and Pothole Water Storage Dynamics in the Prairie Pothole Region

    by Javad Rahmani, Chaopeng Shen, Ali A. Ameli

    In pothole‐dominated catchments, such as those in the Prairie Pothole Region (PPR), potholes strongly influence catchment hydrologic behavior through complex and dynamic fill–spill–connection mechanisms. This complexity—combined with the predominance of ungauged catchments and the lack of high‐resolution pothole inventories—poses challenges for both traditional hydrologic models and purely data‐driven deep learning approaches. To address this, we developed the δHBV‐Pot model within a differentiable modeling framework (δ). This physics‐informed deep learning model integrates the conceptual HBV model with a probabilistic algorithm that emulates the aggregate effects of pothole fill–spill–connection processes. Applied to 98 PPR catchments, δHBV‐Pot achieves stronger predictive accuracy and physical realism than a purely data‐driven Long Short‐Term Memory (LSTM) model and two conceptual hydrologic models. The PPR‐scale regional δHBV‐Pot model successfully simulates hydrologic behavior for the majority of pseudo‐ungauged (test) catchments withheld during model development, effectively regionalizing (a) high‐flow magnitude and interannual variability, (b) intra‐annual flashiness of high‐flow and normal flow conditions, and (c) interannual variability in pothole water storage dynamics. Moreover, the model identifies vulnerable catchments with large high‐flow magnitude and variability—even in the absence of streamflow data—and delineates catchments with varying temporal variability in pothole water storage without requiring detailed pothole inventories. Our findings highlight the value of combining conceptual hydrology with data‐driven deep learning models in pothole‐dominated regions. This integrated approach enables the regionalization of high‐flow and pothole storage characteristics to ungauged catchments, providing critical insights for vulnerability assessment and the design of sustainable water and ecological management strategies in pothole‐dominated landscapes.


  • Open Access

    Time Variance in Snowmelt Partitioning: A Mechanistic Modeling Approach to Explore the Role of Catchment Structure and Pre‐Snow Rainfall

    by Mahbod Taherian, Ali A. Ameli

    Understanding how snowmelt is partitioned into different hydrologic flowpaths/storages—and how this partitioning varies over time—is essential for predicting water availability and quality under climate variability. In this study, we examine the time‐variance of snowmelt partitioning patterns (SPP) in response to interannual variations in antecedent (Fall) rainfall before snowmelt seasons, across two snow‐dominated catchments in Canada and Sweden with contrasting geologic and topographic features. Using integrated subsurface–surface flow and transport modeling, combined with observational data, we simulate the partitioning of snowmelt into shallow flowpath, deep flowpath, evapotranspiration, and long‐term storage. To generalize our findings beyond the two case studies, we design a suite of virtual experiments that systematically vary catchment slope and the extent of the hydraulic conductivity's vertical and lateral heterogeneity. Results show that lateral heterogeneity in conductivity mediates the sensitivity of snowmelt partitioning to interannual variations in antecedent rainfall. While laterally homogeneous catchments display minimal sensitivity of snowmelt partitioning pattern to wet or dry Fall rainfall conditions, catchments with heterogeneous lateral structure store a significantly larger portion of snowmelt and reduce snow‐sourced shallow flow contributions in years with high pre‐snow rainfall than years with low pre‐snow rainfall. In contrast, while slope and vertical conductivity architecture govern SPP, they play a limited role in mediating SPP's temporal sensitivity to antecedent rainfall variability. These findings reveal that subsurface structure—including the extent of lateral subsurface heterogeneity—modulates the influence of climate variability on snowmelt partitioning and catchment hydrologic function. This has implications for predicting streamflow responses, groundwater recharge, and solute transport under changing climate regimes, and highlights the importance of representing time‐variable hydrologic behavior in hydrologic models.


  • Open Access

    Evaluating the Functional Realism of Deep Learning Rainfall‐Runoff Models Using Catchment Hydrology Principles

    by Ara Bayati, Ali A. Ameli, Saman Razavi

    Deep learning (DL) models such as Long‐Short‐Term‐Memory (LSTM) networks have achieved exceptional predictive accuracy in rainfall–runoff modeling. Yet these models learn from statistical correlations rather than hydrologic insights, raising the question of whether their internal functional reasoning is physically reliable. Despite previous studies highlighting unexpected outcomes from LSTMs under long‐term climate shifts, functional realism—defined as the extent to which a model's internal functioning aligns with defensible mechanisms of streamflow generation—remains largely underexplored. We introduce a hydrology‐specific Explainable AI (XAI) framework that opens the black‐box of LSTM. It extracts nonlinear, lag‐dependent, and time‐varying Impulse Response Functions (IRFs) which quantify the functional relationships that LSTM uses to reflect the isolated influence of precipitation ( P ), temperature ( T ), and potential evapotranspiration ( PET ) on simulated streamflow. IRFs reveal how LSTMs internalize streamflow generation during events, offering a catchment hydrology perspective for evaluating model realism. Applying this framework to 672 North American catchments with strong LSTM predictive skill, we find that high accuracy often masks hydrologically implausible reasoning: in over 70% of rain‐dominated basins, short‐term temperature rises unexpectedly raise simulated streamflow and enhance celerity rate even without rainfall; in snow‐dominated regions, PET is misattributed as a driver of snowmelt‐related flow and enhances the catchment's celerity rate. We conclude that correlation‐driven learning can compromise the robustness of LSTM‐based forecasts under weather extremes and short‐term and long‐term climatic shifts. Our framework bridges deep learning with hydrologic understanding and offers a scalable diagnostic for assessing the functional realism of DL models across diverse catchment types.


  • Open Access

    Nonstationarity in Snowmelt Partitioning: A Mechanistic Modeling Approach to Explore the Role of Catchment Structure and Antecedent Climate

    by Mahbod Taherian, Ali Ameli

    Understanding how snowmelt is partitioned into different hydrologic flowpaths/storages—and how such partitioning varies over time—is essential for predicting water availability/quality under climate variability. In this study, we examine the (non)stationarity of snowmelt partitioning patterns (SPP), in response to interannual variations in antecedent (Fall) rainfall prior to snowmelt seasons, across two snow-dominated catchments with contrasting geologic and topographic features: one in Canada and the other in Sweden. Using integrated subsurface–surface flow and transport modeling (ParFlow-CLM-EcoSLIM), combined with observational data, we simulate monthly to annual partitioning of snowmelt into shallow flowpaths, deep flowpaths, evapotranspiration, and long-term storage. Further, to generalize our findings beyond the two case studies, we design a suite of virtual experiments that systematically vary catchment slope, conductivity’s vertical contrast, and conductivity’s lateral heterogeneity. Results show that lateral heterogeneity in conductivity strongly mediates the sensitivity of snowmelt partitioning to antecedent rainfall: catchments with heterogeneous lateral structure store significantly more snowmelt and reduce shallow flow contributions during wetter years, while laterally homogeneous catchments display minimal sensitivity to wet or dry Fall rainfall condition. In contrast, slope and vertical conductivity architecture govern SPP but play a limited role in mediating SPP’s temporal sensitivity to antecedent rainfall variability. These findings reveal that subsurface structure—particularly lateral heterogeneity—mediates the extent to which climate variability alters snowmelt partitioning. This has implications for predicting streamflow responses, groundwater recharge, and solute transport under changing climate regimes, and highlights the importance of representing time-variable hydrologic behavior (or functions) in hydrologic models