Assignments#
Homework 1:#
In this assignment, you will explore recent advancements and challenges in climate modeling by reading the article Harnessing AI and computing to advance climate modelling and prediction by Tapio Schneider et al. [SBB+23]. Additionally, you will choose a second article related to this topic to broaden your understanding of the current state of climate modeling and the integration of AI.
This article, Harnessing AI and computing to advance climate modelling and prediction, discusses the current state and future directions of climate modeling. It emphasizes the challenges that traditional models face in accurately simulating climate processes due to limitations in computational resources and data resolution. The authors argue for the integration of advanced machine learning techniques, including AI-driven parameterizations, to improve model accuracy and efficiency. They highlight the need for global collaborations and the development of open-source platforms to accelerate advancements in climate modeling, addressing both scientific and societal needs in the face of climate change [SBB+23].
Instructions#
Read the Primary Article
Read Harnessing AI and computing to advance climate modelling and prediction by Tapio Schneider et al., which discusses the limitations of traditional climate models and suggests ways AI and machine learning can help improve model accuracy and efficiency.Select a Secondary Article
Choose one of the articles that either:Is cited within the primary article, or
Cites the primary article.
You can search for related articles that have cited this work on Google Scholar: Harnessing AI and computing to advance climate modelling and prediction.
High-Level Reading of the Secondary Article
Read the second article at a high level. Focus on understanding the key accomplishments, methods, data used, and main challenges or criticisms discussed in the article. This is a preliminary identification task, so don’t worry about understanding every technical detail.Presentation Preparation
Create a 4-slide presentation summarizing your findings. Each slide should address the following points:Slide 1: Summary of the primary article.
Slide 2: The most important accomplishments and contributions of the secondary article.
Slide 3: Any criticisms, challenges, or limitations highlighted in the article.
Slide 4: Data, methods, and any unique insights or “curiosities” that stood out to you.
The presentation should be concise and highlight only the most critical points.
Deliverables#
Submit your 4-slide presentation as a PowerPoint/PDF/Google Sheets link.
Be prepared to discuss your findings in class.
Homework 2:#
Overview#
The purpose of this homework is to help you engage with the academic literature relevant to your chosen problem area in atmospheric sciences. You will:
Select 5 to 10 academic papers that are directly relevant to your problem.
Write a structured markdown file summarizing these papers, following the format provided in the linked example.
Highlight key contributions, methodologies, and results of each paper while maintaining a consistent structure for clarity and ease of comparison.
By completing this assignment, you will develop a clear understanding of the current state of the field and identify gaps or opportunities for further exploration.
Instructions#
Step 1: Select Relevant Papers#
Choose between 5 and 10 papers that are highly relevant to your problem.
Prioritize papers that:
Are published in peer-reviewed journals or top conferences.
Offer diverse methodologies or perspectives.
Are foundational to your area of interest.
Step 2: Create a Markdown File#
Write your review in a Markdown file.
Ensure the structure of your file mirrors the following:
Example Structure#
# [Your Problem Topic]: A Literature Review
## Introduction
Provide a brief overview of the problem you are addressing, why it is important, and how the selected papers contribute to understanding or solving it.
## Paper Reviews
### Paper 1: [Title]
- **Authors:** [Author names]
- **Key Highlights:** [Summarize the paper’s primary contributions]
- **Data Used:** [What datasets were employed?]
- **Methodology:** [Summarize the techniques or models used]
- **Key Results:** [Highlight the main findings]
- **Strengths:** [What does the paper do particularly well?]
- **Limitations:** [What are the shortcomings?]
- **Relevance:** [How does this paper contribute to solving your problem?]
### Paper 2: [Title]
...
## Comparison of Approaches
| Methodology/Technique | Key Strengths | Key Limitations | Example Papers |
|------------------------|---------------------------------|-----------------------------------|---------------------------------|
| [Technique 1] | [Highlight key advantages] | [Highlight key disadvantages] | [Cite relevant papers] |
| [Technique 2] | ... | ... | ... |
## Challenges and Future Directions
- Highlight challenges identified from the papers reviewed.
- Provide your perspective on gaps in the field and potential future research directions.
## Conclusion
Summarize the key takeaways from your review. Discuss how these insights will guide your work.
## Questions About Methodologies and ML Approaches
This section identifies critical questions that guide understanding of the methodologies and machine learning (ML) techniques used in the reviewed papers. It covers aspects such as data preparation, model design, training processes, evaluation metrics, reproducibility, and practical applications. The purpose is to highlight areas where additional clarity or details are needed to replicate or extend the research, especially for those less familiar with the technical intricacies.
## References
1. [Full citation of Paper 1]
2. [Full citation of Paper 2]
...
n. [Full citation of Paper n]
Deliverables#
Self-contained markdown file.
Be prepared to discuss your findings in class.
Evaluation Criteria#
Your homework will be graded on the following:
Relevance: How well do the selected papers align with your stated problem?
Clarity: Structure of the Markdown file. Is it well-structured and easy to read?
Depth of Analysis: Have you thoroughly analyzed the methodologies, results, and contributions of each paper?
Comparative Insight: Does your comparison table offer meaningful insights across approaches?
Originality: Are your challenges and future directions reflective of your critical understanding of the field?
Potential Topics for Assignments#
Week ?: Complete a basic tutorial on Python and Jupyter Notebooks. Identify and summarize key AI/ML applications in atmospheric sciences by analyzing recent case studies.
Week ?: Analyze a simplified Navier-Stokes model to visualize atmospheric flow dynamics, using Python. Review a basic NWP model setup and discuss the role of physical laws and computational limitations.
Week ?: Write a brief report on a key atmospheric science challenge (e.g., extreme weather forecasting or climate downscaling) and propose how AI/ML could help address it. Use Jupyter Notebooks to visualize data from a modern atmospheric dataset (e.g., satellite or reanalysis data) and identify initial patterns relevant to atmospheric behavior.
Week 4: Implement regression and classification tasks using atmospheric data in Python.
Week 6: Build and evaluate a CNN for weather pattern recognition.
Week 8: Develop a model to predict temperature and precipitation using RNN or LSTM.
Week 10: Use a GAN to create synthetic weather scenarios and analyze model uncertainty.
Week 12: Implement a PINN to simulate a basic atmospheric process, like temperature advection.
Week 14: Apply UMAP on climate data for visualization; use SHAP to interpret a predictive model.
Final Project (Week 16).