Linking Remote Sensed Metrics to Field-collected Data: The Focal Data Site Streamer

A tool for field researchers to expand analysis possibilities

For ESIIL staff

Group Number: 15

Breakout Room #: S140

ESIIL staff edit in Markdown

People

Nyika Campbell

Learner · University of Colorado Boulder

Nyika works as a lab manager for the Mountain Ecology and Biogeography lab at CU Boulder.

Plant EcologyAlpine EnvironmentsData Science

RField Data CollectionDesign

View learner file

OR

Olivia Ross

Learner

Olivia Ross!

View learner file

EN

Emily Nagamoto

Learner

View learner file

Lauren Walker

Learner · University of Oregon · she/her

landscape modellingforest ecology

View learner file

Day 1

Team Norms and Decision Making

Our team norms:

• Usage of AI in our Project

  • During the ESIIL Summit: broad use of LLMs with disclosure for non-coding AI use
  • Following the ESIIL Summit: discuss use of AI for remainder of project

• How our Team Makes Decisions

  • Full transparency, err on the side of over communication
  • Small decisions (e.g., edits to the Github, new data sources): communication over slack
  • Larger decisions (e.g. sharing content, big updates): ask to meet as a group
  • Love it, live with it, hate it - if anyone is with the third option, continue the conversation

• Authorship

  • When sharing tool outside of group, asking consent from team first
  • Discuss sharing and authorship as project develops

• Communication

  • Contribute to a space where everyone feels comfortable to bring concerns, questions, and comments to the group early
  • Communicate early and often!

Our decision making strategy:

Full transparency, err on the side of over communication. Small decisions over Slack; larger decisions require meeting as a group. Use gradients of agreement — if anyone "hates it," continue the conversation.

Initial Inspiration

Day 2

Our question 📣

How can we leverage AI to more efficiently harmonize broad extent spatial datasets for use at the focal site scale?

Project Goals

Short term

• [ ] Make an agent.md file
• [ ] Make a prompt log
• [x] Make a data source list
• [x] Pick 3 data sources to do tests on
• [ ] Write an overall workflow for model selection/download

Long term

• GUI usable for land managers
• Add feature that helps select data types depending on the project goals/scale/extent etc.
• Make a light enough model to run locally to support data privacy and sovereignty

Project Phases

• Get group members competent in coding with IDE agents and github
• Start developing workflow for dataset pulling
• Create UI that allows land managers to do this

Hypotheses/Intentions

We aim to empower researchers working with field data at focal site scales to find, add, and utilize larger publically available remote-sense datasets in their workflows.

Why this matters (the “upshot”) 📣

This matters because: This data is important for planning and processing field data collection, but it can be difficult a repetitive to download and process.

People who could use this: We imagine land managers, graduate students, and ecological researchers using this tool.

Data sources we’re exploring 📣

Promising data sources:

• PRISM Climate Data (annual): High-resolution gridded climate data for the contiguous US
  • Annual Precipitation
  • Annual Min Temperature
  • Annual Max Temperature
• MODIS Land Surface Temperature (1 km): MOD11A1 daily land surface temperature and emissivity
• USGS Digital Elevation Model (DEM): National Elevation Dataset; write code with Elevation, then convert in R
• National Land Cover Database
• Open Street Map

Day 3

Team Photo

Team members and collaborators who contributed to this project.

Our process 📣

Plan for Jupyter notebook modules

Flowchart of data processing

Challenges identified

• Learning how to navigate GitHub and CyVerse (for some of us!)
• Deciding which LLMs to use, nice ones we have access to now or ones we are more likely to use in the future
• Running LLMs locally

What’s next? 📣

Short term:

• Improve skills in navigating and connecting VS Code, Python, Jupyter Notebook, and LLMs

Long term:

• Convert this into a GUI with the option to choose from more datasets
• Add a feature/flowchart that helps researchers select data types depending on the project goals/scale/extent (especially for climate data)
• Make a light enough model to run locally to support data privacy and sovereignty

Cite & Reuse

If you use these materials, please cite:

Summit Team. (2026). Summit Group 2026 Team 15 — Innovation Summit 2026. https://github.com/CU-ESIIL/Summit_group_2026_15

License: CC-BY-4.0 unless noted.