Large Language models in Earth Observation

If you've grasped the concepts discussed in the previous post, you're likely poised to explore how to effectively integrate them into your hydrological endeavors. However, you're not starting from scratch. Institutions like ESA and NASA have already outlined some of their applications through a series of posts and contributions. Machine Learning undoubtedly plays a significant role in Earth Observation and remote sensing analyses (see for instance here for an overview) or browse the activities of the RSLab for a perspective from the University of Trento. 

Here they are a list of interesting links that I collected:

• NASA and IBM Openly Release Geospatial AI Foundation Model for NASA Earth Observation Data
• Integrating Earth Observation (EO) with Large Language Models (LLMs): Towards A Multimodal EO-Language Model
• Linking Large Language Models for Space Domain Awareness (The topic can look like quite different from our central one but the connection is the management of large amount of data that should be merged for decision making)
• Large language model agent to interact with Earth Observation Data 
• Imagining the Future of Large Language Models and Open Science

After having read them I found that those links are give suggestions but at the end kind of work around the technical questions which I am looking for. 

I am sure that there exist many other resources on this topic, so, please do not hesitate to point my attention to them and I'll make the list grow. 

Can Large Language Models be useful in Hydrology ?

I've been thinking about the potential utility of Language Model ( LLM ) applications in the field of Hydrology. Understanding requires delving into relevant literature and gaining a deep knowledge of how the statistical principles behind LLM operate (because they are statstical tools). The Wikipedia link above, on  serves as an initial source of information, offering some foundational understanding. But let's say that people working on the topics were rediscovering from a different point of view things already known and relabeling them according to a new jargoon. 

For restablishing a little of reasonable context, I would delve into Cosma Shalizi's opinion to gain deeper insights. Careful reading and analysis ad zoom back to recover missing information is necessary to grasp the nuances. However, reading Percy Liang, lecture notes for CS324, Large Language Models (Stanford) [especially looking to "Introduction", "Modeling" and "Training"]  is a definitive settlement of the matter.

Next, I turn to a valuable resource: a work in progress authored by Sebastian Raschka. This book promises practical exercises to fixing the knowledge, albeit still in development.

Disclaimer: I am in a learning process too. So this page will be subject to modifications.