A5084460034- Hydrological Forecasting Using AI
- Hydrology and Watershed Management Studies
- Energy Load and Power Forecasting
- Flood Risk Assessment and Management
- Water resources management and optimization
- Hydrology and Drought Analysis
- Reservoir Engineering and Simulation Methods
- Computational Physics and Python Applications
- Meteorological Phenomena and Simulations
- Solar Radiation and Photovoltaics
- Neural Networks and Applications
- Groundwater flow and contamination studies
- Machine Learning and ELM
- Water Quality Monitoring Technologies
- Groundwater and Watershed Analysis
- Soil erosion and sediment transport
- Model Reduction and Neural Networks
- Precipitation Measurement and Analysis
- Network Traffic and Congestion Control
- Traffic Prediction and Management Techniques
- Urban Heat Island Mitigation
- Network Security and Intrusion Detection
- Mobile Agent-Based Network Management
- Plant Water Relations and Carbon Dynamics
- Mining Techniques and Economics
University of Waterloo
2020-2024
McGill University
2014-2019
Ericsson (Ireland)
2017
Abstract Data assimilation allows for updating state variables in a model to represent the initial condition of catchment more accurately than OpenLoop simulation. In hydrology, data is often prerequisite forecasting. According Hornik (1991, https://doi.org/10.1016/0893-6080(91)90009-T ) artificial neural networks can learn any nonlinear relationship between inputs and outputs. Here, we hypothesize that could simulated streamflow (from hydrological model) corresponding variables. Once...
A novel ensemble-based conceptual-data-driven approach (CDDA) is developed where a data-driven model (DDM) used to "correct" the residuals from an ensemble of hydrological (HM) simulations. The CDDA respects processes via HM and it benefits DDM's ability simulate complex relationship between input variables. can accomodate any DDM, allowing for different configurations be tested. tested streamflow simulation in three Swiss catchments HM, HBV (Hydrologiska Byråns Vattenbalansavdelning),...
Abstract The input variable selection problem has recently garnered much interest in the time series modeling community, especially within water resources applications, demonstrating that information theoretic (nonlinear)‐based algorithms such as partial mutual (PMI) (PMIS) provide an improved representation of modeled process when compared to linear alternatives correlation (PCIS). PMIS is a popular algorithm for problems considering nonlinear selection; however, this method requires...
Abstract In water resources applications (e.g., streamflow, rainfall‐runoff, urban demand [UWD], etc.), ensemble member selection and weighting are two difficult yet important tasks in the development of forecasting systems. We propose test a stochastic data‐driven framework that uses archived deterministic forecasts as input results probabilistic forecasts. addition to data (ensemble) model output uncertainty, proposed approach integrates both uncertainties, using variable methods,...
In a companion paper, Sikorska-Senoner and Quilty (2021) introduced the ensemble-based conceptual-data-driven approach (CDDA) for improving hydrological simulations. This consists of an ensemble model (HM) simulations (generated via different parameter sets) whose residuals are 'corrected' by data-driven (one per HM set), resulting in improved simulation. Through case study involving three Swiss catchments, it was demonstrated that CDDA generates significantly streamflow when compared to HM....
Abstract. This study proposes a novel hybrid method that substantially accelerates and improves deep learning (DL) model development for streamflow prediction. The leverages combination of long short-term memory (LSTM) network random forests. A encoder-decoder is designed, where pre-trained LSTM utilized as an encoder to extract temporal features from the input data. Subsequently, forest decoder processes encoded information make predictions. Our was tested on 421 catchments in continental...
Abstract. This study proposes a novel hybrid method that substantially accelerates and improves deep learning (DL) model development for streamflow prediction. The leverages combination of long short-term memory (LSTM) network random forests. A encoder-decoder is designed, where pre-trained LSTM utilized as an encoder to extract temporal features from the input data. Subsequently, forest decoder processes encoded information make predictions. Our was tested on 421 catchments in continental...