ABAra Bayati

Ara Bayati

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Ara Bayati has a total of 7 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • 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

    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

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

    by Ara Bayati, Ali A Ameli, Saman Razavi

    We propose a hydrologic-specific Explainable AI (XAI) framework to extract nonlinear and nonstationary Impulse Response Functions (IRFs) used by Long Short-Term Memory (LSTM) models to simulate streamflow. IRFs reveal how LSTMs emulate streamflow generation and celerity propagation processes in response to impulses of precipitation (P), temperature (T), and potential-evapotranspiration (PET), enabling the evaluation of LSTMs' functional realism under short-term and long-term varying climate/weather conditions. Applying this framework to 672 catchments in USA and Canada, we found that while LSTMs achieve exceptionally high predictive accuracy, their extracted functionality often contradicts established hydrologic principles. Unexpectedly, the isolated effects of T or PET short-term variations on LSTM's simulated streamflow and celerity rate are often positive in direction. For example, in over 70% of raindominated catchments, particularly along the Pacific Coast, increased T within 1-14 days prior to streamflow events is associated with higher streamflow and celerity rates. Similarly, in the southeastern USA and California, LSTMs predict increased streamflow solely in response to PET rises. In snow-dominated catchments, particularly in the Rockies, LSTMs misinterpret PET (representing evaporation function) as a proxy for temperature, the primary snowmelt's driver. These behaviors, likely influenced by seasonality in dynamic inputs, data non-homogeneity, simplicity bias, and missing causal factors, undermine LSTMs' scientific reliability for streamflow forecasting and climate impact assessments. Our XAI framework integrates deep learning with hydrologic context through Differentiable modeling, providing a screening tool for the evaluation of hydrologic models' functional realism in short-and long-term forecasting, climate change studies, and generalization to ungauged basins.

  • Open Access

    Quantifying the contribution of wetlands drying to aerosol generation across Iran

    by Majid Bayati, Nooshdokht Bayat-Afshary, Mohammad Danesh-Yazdi

    Wetlands are accounted as important providers of ecosystem services, which yield several functionalities such as the support of biodiversity, flood control, soil stabilization to reduce dust generation, natural treatment of surface waters, groundwater replenishment, climate regulation and economic benefits. Over the past decades, the impacts of anthropogenic manipulations amplified by climatic changes have threatened both the quantity and quality of wetlands, worldwide. A continuous monitoring of wetlands is thus necessary to protect them from further destruction, as well as to devise and assess the success of any rehabilitation plans. The conventional methods of water body monitoring chiefly include field surveying, which is time consuming, costly, and limited in extent. Alternatively, remotely sensed data have facilitated a much less expensive and more extensive monitoring of water bodies over a wide range of spatiotemporal resolutions. In this study, we implemented a learning-based classification framework fed by remote sensing data to evaluate the historical trends of the most important wetlands across Iran using the Google Earth Engine cloud computing platform. To this end, we used Landsat imagery between 2000 and 2020 to extract the water body of wetlands in dry seasons to consider the most critical condition. We also examined different spectral indices to identify the best combination giving the largest classification accuracy for each wetland, separately, based on their distinct conditions of water depth and vegetation cover. We then quantified the contribution of wetlands drying to the generation of dust storms via a frequency-intensity index given the annual number of dusty days and the Aerosol Optical Depth (AOD) provided by MODIS. According to the results, the majority of the studied wetlands show significant descending trends with the average loss of 31% in surface area. The aerosol analysis also witnesses the expansion of dust generation sources around most of the retreated wetlands, particularly in those years when the wetlands areas were smaller than the long-term average. The above observations point out a potential threat for the agricultural activities and highlight serious consequences for the health of nearby urban and rural residents.Keywords: Wetland, Dust Storm, Remote Sensing, Environmental Monitoring, Ecosystem Protection


  • Open Access

    Coupling machine learning with high resolution satellite imagery to estimate spatiotemporal changes of salinity in water bodies

    by Majid Bayati, Mohammad Danesh-Yazdi

    The spatiotemporal dynamics of salinity in hypersaline lakes is strongly dependent on the rate of water flow feeding the lake, evaporation rate, and the phenomena of precipitation and dissolution. Although in-situ observations are most reliable in quantifying water quality variables, the spatiotemporal distribution of such data are typically limited or cannot be readily extrapolated for long-term projections. Alternatively, remotely-sensed imagery has facilitated less expensive and stronger ability to estimate water quality over a wide range of spatiotemporal resolutions. This study introduces a machine learning model that leverages in-situ measurements and high-resolution satellite imagery to estimate the salinity concentration in water bodies. To this end, 123 points were sampled in April and July of 2019 across the Lake Urmia surface covering the wide range of salinity fluctuations. Among the artificial neural networks, ANFIS, and linear regression tools examined to determine the relationship between salinity and surface reflectance, artificial neural networks yielded the best accuracy evidenced by R2 = 0.94 and RMSE = 6.8%. The results show that the seasonal change of salinity is linearly correlated with the volume of water feeding the lake, witnessing that dilution imposes a stronger control on the salinity than bed salt dissolution. The impact of disturbance in the lake circulation due to the causeway is also evident from the sharp changes of salinity around the bridge piers near spring when the mixing of fresh and hypersaline water from the southern and northern parts, respectively, takes place. The results of this study prove the promising potential of machine learning tools fed multi-spectral satellite information to map other water quality metrics than salinity as well.