Introduction

Everything you need to prepare for Hack the Harvest will be shared before the event. That includes background materials, prior work summaries, starter notebooks and access instructions. Use the time beforehand to form your team, read up and test your access, so you're ready to build at 10:30 a.m. Saturday.

Tools and Technologies

  • Devpost: registration
  • Keeboh: team formation, project submissions, the technical workspace and challenge resources
  • Discord: announcements, team coordination and scheduled mentor support
  • Livestream Q&A: Friday, November 13 at 5:00 p.m. Pacific, covering the challenge, Keeboh, how to submit and your questions
  • Data: public satellite imagery, crop and agricultural datasets, field boundaries and other geospatial data
  • Methods: time-series analysis, AI and machine learning, and visualization
  • NVIDIA Nemotron: optional

Inspiration

Silage corn is usually harvested early, while it's still green. Grain corn stays in the field longer and dries down first. That timing difference leaves a signal. Here are a few ways to look for it:

  • Changes over time: track vegetation and compare satellite imagery across harvest windows.
  • Patterns in the data: combine public datasets, build field-level time-series models, or look for spectral and temporal signatures.
  • An answer people can use: build a classification model or an interpretable visualization, or find a completely different signal.

Satellites let us observe thousands of fields without being there. Your job is to turn those observations into an answer.

Contact Us & Support Channels

Open data

Need

Source

Use it for

Which fields are corn

USDA Cropland Data Layer

A corn mask for the year. It maps crop type, not harvest type.

Season-long crop signal

Sentinel-2 L2A

10 m imagery for greenness over time and cloud masking.

Longer history

Landsat Collection 2 Level-2

30 m imagery going back before Sentinel-2.

Finding imagery

Microsoft Planetary Computer, Earth Search

Search catalogs instead of downloading scenes by hand.

Weather

gridMET, Daymet

Rain, temperature, growing-degree days and frost around harvest.

Regional totals

USDA NASS Quick Stats

Planted and harvested acres by county or state. Context, not field labels.

Boundaries and maps

Census TIGER/Line, USGS National Map

Counties, roads and reference geography.

Background reading (none of these classifies silage versus grain directly; they cover the pieces)