Keeping up with current Machine Learning (ML) developments in hydrology can seem like a never-ending game of catch-up! Instead of drowning yourself in a heap of scientific publications, here are a few practical hints to help you stay ahead in the ever-evolving intersection of ML and hydrology.
Hint 1: Surround Yourself with Experts and Like-Minded ML Enthusiasts
When questioned about how they keep up with the latest ML developments, “speaking with experts” sprung as an almost involuntary response from my peers. Ironically, engaging with (human) experts and researchers remains one of the most effective ways to stay updated about ML.
Conversations with peers and specialists help filter through the vast ocean of information, making it easier to focus on themes relevant to your research ecosystem. More often than not, even senior and highly-cited researchers are very approachable. And what’s better than learning from experts themselves?
Hint 2: Be a Jack of All Developments and a Master of a Few
Since you have made your way to this blog, scientific publishing is presumably a familiar territory for you. However, not all publications require the same level of attention. Developing the skill to efficiently differentiate between these will make you an ML ninja in hydrology.
At times, briefly skimming studies suffices to get acquainted with the proposed idea. Some research, on the other hand, calls for a thorough read and even some hands-on implementation. A smart approach is to set aside a fixed amount of time, regularly, to stay updated without feeling overwhelmed.
Hint 3: Set a Thief to Catch a Thief – Use AI to Your Advantage
Harness the potential of Artificial Intelligence (AI) and other ML-based tools to track down the latest and most relevant research.
AI-powered answer engines are not only equipped to retrieve information, but ChatGPT, Gemini, Claude, or Copilot all have their unique style, so don’t hesitate to use them as your little helpers!
Can’t get on top of a paper? Ask AI to summarize. Encounter a text full of mysterious new terminology? Let AI explain them to you. Don’t know how to code something/ have errors in your code? Throw it at AI.
Using Chatbots allows you to sieve through the endless sea of information quickly, and what’s best: they are (mostly) free to use. So apply them to your advantage! But remember to always obey scientific integrity, so beware of overusing when writing texts.
Hint 4: Learn from YouTube – Your On-Demand Classroom
You are never too old to skip the conventional classroom and turn to the ultimate modern-day guru: YouTube! A variety of channels offer structured content, guiding you from the basics of ML to coding complex models.
These resources help reinforce your understanding by providing accessible and engaging explanations of the fundamental methodologies behind the latest developments. A few of my personal favorites are StatQuest with Josh Starmer, 3Blue1Brown, Machine Learning Street Talk and MIT Introduction to Deep Learning.
A well-chosen newsletter delivers relevant updates without requiring you to actively search for them, saving time and ensuring that important advancements don’t slip through the cracks.
Hint 5: Master the basics and get Hands-On Experience with Repositories Like Neural Hydrology and Hy2DL
While the field progresses rapidly, it is crucial to master its basics. Efforts made to understand and implement by hand, basic ML concepts and neural network architectures, surely pay great dividends in the long term.
Platforms like Neural Hydrology and Hy2DL provide access to cutting-edge models, allowing you to experiment, implement, and deepen your understanding of ML applications in hydrology. These repositories offer pre-trained models, datasets, and tutorials that help bridge the gap between theory and practice.
Coding from scratch does not only instil confidence, but is also fundamental in understanding the theoretical underpinnings of a model. Though coding a simple LSTM network might not seem very attractive, the experience is essential – only then can you make meaningful advances in your research.
Additionally, contributing to open-source projects or developing custom implementations based on these frameworks can significantly enhance your skill set and keep you at the forefront of advancements in the field.
Final Thoughts
In conclusion, keeping up with ML developments in hydrology need not be daunting. With a multifaceted approach, you can endeavor to be on top of everything new and even push interdisciplinary research. By integrating a combination of the above tricks, you can stay ahead of the curve and make meaningful contributions to the fast evolving field of ML in hydrology.
Image sources (part of header image): Stupariu, Mihai Sorin & Cushman, Samuel & Pleșoianu, Alin & Stupariu, Ileana & Fürst, Christine. (2022). Machine learning in landscape ecological analysis: a review of recent approaches. Landscape Ecology. 37. 10.1007/s10980-021-01366-9.
