[GSoC Project Proposal]: Enhancing Daily Skill Assessment Workflows for NOAA’s Surge and Tide Operational Forecast System (STOFS)

[JackReevesEyre-NOAA]

Project Description

This project focuses on enhancing the AUTOVAL package, which provides daily skill assessments and statistical summaries for different STOFS components. Currently, AUTOVAL generates static HTML reports evaluating model performance across different cycles and locations. The primary goal is to transform this existing Python-based evaluation logic into an interactive Chatbot interface. This interface will allow users to query general model performance and station-specific skill, and request on-demand plots or summary tables. By the end of the GSoC period, the applicant will have developed a functional, conversational bridge between AUTOVAL's data and the end-user.

Expected Outcomes

Integrated Chatbot Interface: A functional chatbot (prototype React SPA and modular component) ready for deployment on the nowCOAST website, allowing users to interactively query STOFS data and produce on-demand skill assessments.
Expanded Data Integration: Integration of new observational data sources, broadening the package’s validation capabilities beyond current limitations.
3D Variable Assessment (Stretch Goal): If time allows following the successful integration of the first two steps, the project will aim to extend the evaluation workflow from 2D variables to 3D oceanographic parameters, including temperature, salinity, and currents.

Skills Required

Python; Chatbot & NLP Frameworks; Data Visualization; Scientific Data Handling (especially Xarray, Numpy, Pandas, NetCDF files). Preferred additional skills: Statistical Knowledge of skill assessment metrics; Version Control; Web Development.

Expected Project Size

350 hours

Project Difficulty

Intermediate

Mentor Name(s)

Jack Reeves Eyre, Atieh Alipour, Adam Gibbons, Gregory Seroka

Mentor GitHub Handle(s)

@JackReevesEyre-NOAA, @AtiehAlipour-NOAA, @agibbons-NOAA, @gseroka

Mentor Contact Email(s)

jack.reeveseyre@noaa.gov, atieh.alipour@noaa.gov, adam.m.gibbons@noaa.gov, gregory.seroka@noaa.gov

Contributor Evaluation/GSoC GitHub Issue Links

Applicants are expected to demonstrate their technical proficiency by completing the following two-step qualification task:

Environment Setup & Validation: Follow the instructions in the project’s README.md to install the AUTOVAL package and execute the STOFS-2D-Global test case. Applicants must successfully generate the validation output and share a link to their results (e.g., a Google Drive or Cloud Storage link) as proof of setup.

Prototype Development (Mini-Chatbot): Develop a lightweight Python-based chatbot that interacts with STOFS-2D-Global data stored in the cloud. The prototype should demonstrate the ability to:

• Fetch data for a specific date from the provided S3 bucket.
• Process a user query for specific stations (e.g., "Show me the skill for Station X").
• Return a simple summary and visualization in response.

The STOFS-2D-Global data bucket can be accessed here: https://noaa-gestofs-pds.s3.amazonaws.com/index.html

Example STOFS-2D-Global data for 01/25/2026: https://noaa-gestofs-pds.s3.amazonaws.com/index.html#stofs_2d_glo.20260125/

Additional Project Background/Links

No response

Previous Project Idea Link(s)

No response

[vivekdevaa124]

Dear IOOS Mentors,

I am a developer specializing in Python, React, and LLM integration with a strong interest in building conversational interfaces for scientific data. I am eager to contribute to the AUTOVAL Chatbot project, as it aligns perfectly with my experience in bridging complex backend logic with interactive, AI-driven web tools.

I am proficient in the MERN stack and scientific Python libraries, and I am currently working on the STOFS-2D-Global qualification task to demonstrate my ability to process cloud-stored data for on-demand visualization. I look forward to the possibility of collaborating with the IOOS team.

Best regards,

Vivekdevaa

[AtiehAlipour-NOAA]

@vivekdevaa124 Thank you so much for your interest in the AUTOVAL project. Sounds great! At this moment, we are just building the idea pages. We need to wait until we know if we will be a selected organization for GSoC.

[asadjan4611]

Hello IOOS Team,
Thank you for sharing this project idea. I am excited to contribute and have experience in Python, React, and LLM integrations. I am beginning the qualification steps and look forward to collaborating with the team.

[AtiehAlipour-NOAA]

Hello IOOS Team, Thank you for sharing this project idea. I am excited to contribute and have experience in Python, React, and LLM integrations. I am beginning the qualification steps and look forward to collaborating with the team.

@asadjan4611 Thank you so much for your interest! We’re happy to hear you have the right skills for the project. At the moment, we are still finalizing our idea pages. Please wait until we know if we have been selected as a GSoC organization before you start working on the qualification tests.

[asadjan4611]

@AtiehAlipour-NOAA
Thank you for your quick response. I will wait for the confirmation regarding your GSoC selection. In the meantime, could you please explain what the qualification tests will include? I am very interested in contributing to this project.

Thank you

[JackReevesEyre-NOAA]

@asadjan4611 -- the qualification tasks will include some simple steps to make sure you have the technical abilities to fully engage with the project. For now, we'd suggest exploring the code and data (linked above) and making sure you read and understand the project description and related topics. Thanks again for your interest.

[asadjan4611]

@JackReevesEyre-NOAA Okay Thanks

[NobleCoder69]

Hi @JackReevesEyre-NOAA,

I’m very interested in this GSoC 2026 project. My background includes working with LangChain, LangGraph, MCP concepts, and web technologies (HTML/JavaScript/TypeScript), and I’m comfortable with Python for data and AI-related workflows.

The idea of transforming AUTOVAL’s evaluation outputs into an interactive chatbot interface is particularly exciting. I’d like to begin with the qualification tasks outlined in the description.

Before I proceed, could you please clarify:
• Are there any recommended cloud access methods or authentication requirements for interacting with the STOFS S3 bucket?
• And do you have preferences regarding chatbot frameworks (e.g., pure Python vs LLM/NLP-assisted approaches)?

Looking forward to contributing and learning more about AUTOVAL.

[AtiehAlipour-NOAA]

Hi @NobleCoder69, thank you for your interest in the AUTOVAl project! Please hold off until Feb 19, as we don't yet know if we’ve been accepted. To answer your question: you can use anonymous access for the cloud to download data, and a mix of Python and LLMs for the chatbot.

[ManikyaPant]

Hello mentors,

I am interested in the AUTOVAL chatbot project and have started reading the repository to understand the current validation workflow and STOFS outputs.

Before beginning the qualification tasks I had a few quick questions:

1. Should the prototype chatbot use a simple rule based parser or is an LLM approach expected?
2. Are static matplotlib plots sufficient for the prototype or is interactive plotting preferred?
3. For station skill queries should we reproduce the full AUTOVAL metrics or a reduced subset?

I will proceed with the environment setup and STOFS 2D Global test case and share the results once complete.

[JackReevesEyre-NOAA]

Hi @ManikyaPant -- thank you for your interest. To answer your questions:

1. LLM is expected (with python-based implementation).
2. Static plots is fine.
3. A subset of metrics is fine.

When it comes to sharing results, please get in touch by email, rather than posting here.

[JackReevesEyre-NOAA]

IOOS has been accepted as a mentoring organization for GSoC 2026! We ask that potential contributors work through the "Contributor Evaluation" (described in the project description at the top of this page) and get in touch by email to discuss application ideas.

Note that IOOS also has a contributor guide to help potential contributors understand the process and our expectations.

And further details of the timeline are here: https://developers.google.com/open-source/gsoc/timeline