TerraVysers

HRT Challenge: Best Use of Data for Predictions and Decision-Making

Inspiration

Environmental data exists at massive scale, but decision makers rarely have tools that turn it into region-specific, actionable insight. Research outputs, satellite products, and soil models often remain locked behind technical barriers.

TerraVysers was built to bridge that gap by converting raw geospatial and climate data into predictive, operational intelligence for land management, policy planning, and environmental investment.

What It Does

TerraVysers is a geospatial decision platform that allows users to select any region on a map and receive predictive and analytical outputs, including:

Soil erosion risk forecasting using a RUSLE-based environmental model
Carbon sequestration potential for reforestation and carbon credit assessment
Crop yield risk indicators to support food security planning

The system is modular by design. New datasets, predictive models, or policy layers can be added without changing the core pipeline.

How We Built It

We structured the team around functional roles:

ML and AI: predictive models for carbon potential and crop yield risk
Backend: geospatial processing, raster pipelines, and data fusion
Frontend: interactive map interface and regional selection tools
Presentation and coordination: problem framing, time management, and narrative

The platform integrates open-access satellite imagery, soil databases, and climate data into a unified prediction and reporting system.

Technical Approach

Climate input: Precipitation-based erosion drivers from global rainfall datasets
Soil input: Soil composition and erodibility from open soil property APIs
Vegetation input: NDVI from Sentinel-2 satellite imagery
Prediction layer: Modular pipeline for carbon sequestration and crop yield models
Output layer: GeoTIFF risk surfaces and region-level summary statistics for decision support

All computations are aligned to a consistent geospatial grid to ensure reliable overlays and comparisons.

Challenges

Several high-quality global datasets require manual approval or restricted access, limiting real-time integration
Large raster and satellite products created compute and performance constraints within the hackathon timeframe
The team had no prior working history of working together, requiring rapid alignment across technical and non-technical roles

Accomplishments

Integrated satellite imagery, soil data, and climate signals into a unified prediction pipeline
Deployed a carbon sequestration model for reforestation potential
Built crop yield risk prediction for wheat varieties across European regions
Delivered a map-driven interface that converts scientific data into policy-readable outputs

What We Learned

Time coordination is as critical as technical execution
Clear communication of skill gaps improves overall system reliability
Modular system design enables fast iteration under tight constraints

What’s Next

Expand crop yield prediction to additional crops and continents
Validate erosion and carbon models against ground-truth datasets
Add policy and investment layers such as conservation priority scoring and carbon market readiness
Deploy regional dashboards for government and NGO use cases

TerraVysers is designed as a decision engine, not a visualization tool. The goal is to convert environmental data into operational guidance for land use, climate strategy, and food security planning.