Case study / 05

GeospatialCrop Analysis

A geospatial intelligence platform that transforms freely available satellite imagery into crop health, crop-type, and agricultural decision insights for Québec and beyond.

LIVE PLATFORM / AWS DEPLOYMENT

OPERATIONAL ANALYTICS / MULTI-USE-CASE SYSTEM

Case study / 05

Project context

Agriculture & Geospatial Intelligence

Plumfind contribution

Geospatial Analytics
AI & Intelligent Systems

Technology

Sentinel-2, Landsat, Prithvi AI, Geospatial Processing, AWS

Timeline

Use-case research, pipeline development, regional-scale deployment

BUSINESS CHALLENGE

Vast agricultural territory. Limited real-time visibility.

Farmers, insurers, lenders, and government stakeholders lack timely, scalable visibility into crop conditions, making critical decisions dependent on field surveys, historical data, and delayed reporting.

LACK OF TIMELY INSIGHTS / DECISION GAP

DISCOVERY & RESEARCH

01 / Business context

We began with the agricultural decisions that require timely visibility into crop health, crop types, and changing field conditions.

02 / Use case

Farmers, insurers, lenders, and government stakeholders informed the intelligence and reporting requirements.

03 / Decision layer

The solution prioritizes crop stress, health, maturity, distribution, area, and trends rather than exposing raw satellite imagery.

04 / Production plan

Tile-based processing and batch execution were designed to extend the workflow from individual plots to Québec-wide monitoring.

PRODUCTION REQUIREMENTS / AGRICULTURAL INTELLIGENCE

Different decisions. One shared intelligence layer.

We mapped the information needs of farmers, insurers, lenders, and policymakers to define a common geospatial intelligence framework that could support field-level decisions and province-wide analysis.

SOLUTION

From raw satellite imagery to actionable agricultural intelligence.

01

Automated satellite image acquisition

Sentinel-2 and Landsat imagery is automatically sourced and filtered by time period, cloud cover, satellite, and geographic tile.

02

AI crop detection

A geospatial foundation model (Prithvi) processes satellite scenes to generate georeferenced crop probability and crop-type maps.

03

Tile-based processing

Individual scenes are aligned, reprojected, and stitched into complete crop maps for each geographic tile.

04

Custom agricultural analytics

Crop areas, types, trends, and zonal statistics are transformed into reports, dashboards, and exportable datasets.

Architecture / system flow

Insights at scale thanks to AI and the cloud.

Each stage converts raw geographic data into a more structured layer of agricultural intelligence, while maintaining geospatial accuracy throughout the pipeline.

Implementation

From satellite data to a dependable production pipeline.

01

Data acquisition & processing

Automated downloads retrieve relevant Sentinel-2 and Landsat imagery based on defined geographic, temporal, and cloud-cover parameters. Images are aligned and georeferenced to enable consistent tile-level outputs.

02

Analytics & reporting

Custom analysis converts crop maps into area calculations, crop distributions, zonal statistics, trends, reports, and exports.

03

Scale & automation

Independent Tile ID processing enables repeatable execution across multiple geographic areas and batch-scale Québec coverage.

BUSINESS IMPACT

Agricultural intelligence people can act on.

Enable near-real-time crop health monitoring without new field hardware
Automate crop-type and field-level geospatial analysis
Reduce reliance on manual surveys and delayed assessments
Support earlier identification of crop stress and changing conditions
Enable insurers to assess agricultural risk and damage more efficiently
Create a scalable foundation for regional and province-wide monitoring
Turn freely available satellite data into operational agricultural intelligence

LESSON LEARNED

The value is not in producing more imagery or more model outputs. It is in transforming geospatial data into timely insights that farmers, insurers, lenders, and policymakers can act on.