Skip to content

Management Practices Prevent Compound Disturbance

✏️

Aerial view of irrigated rangelands and riparian buffers Raw photo location: hero.jpg

One sentence on impact: We are going to use an LLM to identify the management actions that primarily take place following disturbances that facilitate fire, such as wind, drought, insect/path, and fire.

Sprint brief · View shared code · Explore data

About this site: This public log captures our rapid work at the Innovation Summit. Update it directly in your browser (open a file → pencil icon → Commit changes) so teammates and partners can follow along.

How to use this page (for the team)

  • Edit this file: docs/index.md → ✎ → change text → Commit changes.
  • Add images: upload to docs/assets/ and reference like assets/your_file.png.
  • Keep text short and visuals first. Think “slide captions,” not essays.

Day 1 — Define & Explore

*Focus: Exploring management actions around linked disturbances

Our product 📣

  • A synthesis manuscript and high level dataset of forest management actions (maybe a map or something)

Our question(s) 📣

  • What management actions take place after drought, windstorm, fire, and insect/path to prevent future wildfires? -Ie. how are managers purposefully or inadvertently breaking disturbance links in order to prevent future catastrophic disturbances and tipping points?

Hypotheses / intentions 📣

-We expect to have a ton of fun and learn a lot :_

Why this matters (the “upshot”) 📣

Keeping working lands intact depends on catching tipping points early. Reliable, transparent indicators help conservation districts invest in the right practices before recovery becomes costly or impossible.

Inspirations (papers, datasets, tools)

Field notes / visuals

Whiteboard brainstorm summarizing threshold indicators Raw photo location: day1_whiteboard.jpg Caption: Scoping session mapping management levers (grazing, irrigation, restoration) to measurable early warning signals.

Different perspectives: Ranching partners emphasized socio-economic thresholds (labor, funding) alongside ecological ones; we parked those for future integration.

Day 2 — Data & Methods

Focus: what we’re testing and building; show a first visual (plot/map/screenshot/GIF).

Data sources we’re exploring 📣

  • Soil moisture anomalies from GridMET regridded to allotment boundaries.

Pattern revealed during exploration Raw photo location: explore_data_plot.png Snapshot: Standardized anomaly series highlights persistent drying in 2022–2023.

  • Practice implementation log compiled by Western Water Conservancy (rotational grazing, riparian buffers, rest periods).
  • Vegetation vigor (NDVI) from Landsat 8/9 30 m composites created with Google Earth Engine.

Methods / technologies we’re testing 📣

  • Rolling Kendall trend test to flag monotonic declines in vegetation vigor.
  • Bayesian change point detection on moisture anomalies to identify abrupt state shifts.
  • Simple rules engine that pairs detections with recommended management actions from partner playbooks.

Challenges identified

  • Management logs mix qualitative notes with numeric entries; we need lightweight coding for summit pace.
  • Cloud cover limits Landsat availability during the growing season—evaluating Sentinel-2 fallback.
  • Alert thresholds must balance early warnings with false positives to maintain partner trust.

Visuals

Static figure

Early pattern we’re seeing Raw photo location: figure1.png Figure 1. Change-point detection flagging a 2021 soil moisture shift aligned with drought declarations.

Animated change (GIF)

Seasonal/temporal change animation Raw photo location: change.gif Figure 2. NDVI animation showing patchy recovery following targeted rest rotations.

Interactive map (iframe)

Open full map

If an embed doesn’t load, put the normal link directly under it.

Final Share Out — Insights & Sharing

Focus: synthesis; highlight 2–3 visuals that tell the story; keep text crisp. Practice a 2-minute walkthrough of the homepage 📣: Why → Questions → Data/Methods → Findings → Next.

Team photo at start of Day 3 Raw photo location: team_photo.jpg

Findings at a glance 📣

  • Soil moisture breakpoints matched partner-observed forage crashes within ±2 weeks in three test allotments.
  • Dashboards combining NDVI trends and practice adherence reduced false alarms by 30% compared with NDVI alone.
  • Managers want alert text that pairs each trigger with “next steps” and a confidence score.

Visuals that tell the story 📣

Lead conclusion visual placeholder Raw photo location: fire_hull.png Visual 1. Prototype dashboard panel showing combined moisture/NDVI alerts and suggested management actions.

Supporting panels for key insights Raw photo location: hull_panels.png Visual 2. Comparative view of allotments with and without timely rest rotations, highlighting vegetation recovery potential.

Complementary result figure placeholder Raw photo location: main_result.png Visual 3. Time-to-threshold simulation illustrating how proactive management extends resilient conditions by ~18 months.

What’s next? 📣

  • Integrate partner-collected socio-economic indicators to broaden dashboard scope.
  • Package code into reproducible notebooks with parameter toggles for different districts.
  • Share prototype with Colorado Coalition of Conservation Districts for external feedback.
Read the sprint brief
Read the brief 		View shared code
View code 		Explore data workflow
Explore data

Team

Name Role Contact GitHub
Sofia Martinez Lead scientist sofia.martinez@colorado.edu @sofiamartinez
Marcus Li Data scientist marcus.li@colostate.edu @mlisystems
Talia Yazzie Community partner liaison talia.yazzie@ntuniv.edu @taliy
Arjun Patel Decision support engineer arjun.patel@wwc.org @apatel

Storage

Code Keep shared scripts, notebooks, and utilities in the code/ directory. Document how to run them in a README or within the files so teammates and visitors can reproduce your workflow.

Documentation Use the docs/ folder to publish project updates on this site. Longer internal notes can live in documentation/; summarize key takeaways here so the public story stays current.

Cite & reuse

If you use these materials, please cite:

Martinez, S., Li, M., Yazzie, T., Patel, A. (2025). Management Practices Prevent Thresholds — Innovation Summit 2025 (Group 17). https://github.com/CU-ESIIL/management-practices-prevent-thresholds-innovation-summit-2025__17

License: CC-BY-4.0 unless noted. See dataset licenses on the Data page.