Agriculture and food security
Crop area, crop health and harvest progress measured from multispectral imagery at national scale. This is the analytics behind our Tonga crop-estimation programme, offered as a product.
Sensors
- Sentinel-2: 10 m multispectral, red-edge bands built for vegetation, roughly 5 day revisit
- Landsat 8/9: 40-year archive for baselines and drought comparison
- Sentinel-1 SAR: rice and flood-irrigated crops under monsoon cloud
What multispectral imagery actually measures
Healthy vegetation reflects near-infrared light strongly and red light weakly; stressed vegetation narrows that gap. Indices built on this difference, tracked across a season against the archive, separate a normal year from a failing one district by district, weeks before harvest figures reach a ministry through paperwork.
Crop-type classification runs on the same data: planting calendars, spectral signatures and field geometry distinguish rice from maize from cane with useful accuracy at 10 metres. The output is a map of what is actually planted, not what was reported.
Proven where it is hardest
Our crop-estimation programme for the Kingdom of Tonga pairs multispectral and hyperspectral observation over root crops, vanilla and coconut across three island groups, processed inside the kingdom and delivered to its Ministry of Agriculture. Small plots, frequent cloud, subsistence staples: agriculture monitoring in its least forgiving form. The same pipeline scales upward more easily than it scaled down.
What the product delivers
Season-long monitoring per administrative district: planted area, condition scoring against archive baselines, harvest-progress estimates, drought and flood flags. Delivered to the ministry's own systems, with the processing chain documented so the numbers survive scrutiny at budget time.