Decoding AI & Hydrology for Water Management Decisions

In this series, hosts Laura Read and Eliza Hale discuss AI and how it can be used in hydrology. They talk strengths and limitations, and the role it can play in advancing water management science.

The series answers questions about machine learning and how it applies to the field of hydrology, including:

1. What are machine learning neural networks, and how do they apply to streamflow forecasting?
2. What’s the difference between machine learning vs. conceptual models, and why does it matter?
3. Why use machine learning for streamflow, and what does the current field of research say?
4. What is a theory-guided machine learning model?
5. And more!

Part 1: Decoding AI lingo and a primer on neural networks

Additional Reading from Part 1

1. A primer from Caltech that defines the difference between "artificial intelligence" and "machine learning." How Do AI and Machine Learning Differ?
2. A breakdown from Google that looks at AI vs. ML and explores how these two concepts are related and how they differ. Artificial intelligence (AI) vs. machine learning (ML).
3. A deep dive into what a neural network is and how we represent it in a machine learning model. Neural networks: representation.

Part 2: Machine learning vs. conceptual models for streamflow forecasting

Additional Reading from Part 2

1. A benchmark study from 2019 shows regionally trained long short-term memory (LSTM) networks outperforming basin-specific calibrations of several traditional hydrology models. Kratzert et al. Towards learning universal, regional, and local hydrological behaviors via machine learning applied to large-sample datasets.
2. A subsequent paper shows a LSTM model outperforming conceptual models, even when making ungauged/untuned predictions. Kratzert et al. Toward Improved Predictions in Ungauged Basins: Exploiting the Power of Machine Learning.
3. A third study compares frequently used hydrologic models in the Great Lakes region and Canada. LSTM was by far the most accurate type of model. Mai et al. Towards learning universal, regional, and local hydrological behaviors via machine learning applied to large-sample datasets.‍
   

Part 3: A unique approach to machine learning models for hydrology

Additional Reading from Part 3

1. Research Papers from the HydroForecast team.

‍

Thank you for viewing the virtual series: Decoding AI & Hydrology for Water Management Decisions.

Questions? Join us for a live Q&A session with our experts on December 6th. Register now, and send your questions to team@hydroforecast.com.