Africa’s sovereign amber alert network

7-day intelligence gives farmers what satellites never did: time to act.

When floods strike Sub-Saharan Africa, communities don’t just lose homes — they lose harvests to waterlogging and lives to malaria weeks later. SHELTER watches with cloud-piercing radar, forecasts the whole cascade, and delivers the warning even when the network is down.

Activate alerts for my areaSee the live dashboard
How SHELTER turns satellite data into actionable intelligenceA satellite passes overhead and downlinks Earth Observation data. Radar penetrates cloud that blocks optical sensors. Machine-learning models measure standing water and crop stress, a risk model fuses them with rainfall and population, and a plain-language advisory is delivered to the household — over email, WhatsApp and SMS where there is a network, and by one-way NIGCOMSAT-1R satellite broadcast where there is not. Days later a fifth agent, Fahis, works backwards: it searches independent reporting to judge whether the intelligence was right, and records the verdict. Fahis can never alter an advisory.SENTINEL-1 · SENTINEL-2Radar and optical, every few daysCLOUD-PENETRATING RADARMonitoring continues through the stormSCOUTdiscoverANALYSTmeasureORACLEdecideHERALDdeliverFAHIS · WERE WE RIGHT?Independent reporting checked days later · verdict recorded, never fed back7-DAY INTELLIGENCEEmail · WhatsApp · SMS where there is a networkNIGCOMSAT-1R broadcast where there is not — one-way, no signal required

Pipeline ok · advisories generated · delivering on Email, Webhook

Four intelligence tracks

One platform, one satellite pipeline, four audiences. The MVP ships Agricultural Intelligence first — Environmental, Public Health and Financial & Credit Risk Intelligence follow as the next phase, on the same engine.

Agricultural Intelligence

Live now

Watch a plot through the season and know before a harvest is lost, not after.

  • Crop health monitoring
  • Soil moisture analysis
  • Crop stress scoring
  • Irrigation recommendations
  • Farm monitoring dashboards
  • Economic impact scoring
  • Climate impact scoring
  • Pest risk from degree-days

Environmental Intelligence

Next phase

The same cloud-piercing flood detection that already protects farms, surfaced for responders and planners.

  • Flood prediction
  • Flood vulnerability mapping
  • Community risk alerts
  • Emergency response dashboards
  • Infrastructure monitoring

Public Health Intelligence

Next phase

Standing water today is a malaria surge in roughly six weeks — and we already see the water.

  • Malaria environmental risk index
  • Standing water detection
  • Community health alerts
  • Climate-health monitoring
  • Disease surveillance support

Financial & Credit Risk Intelligence

Live now

A verified place is the missing half of a credit decision — and the same satellites already see it.

  • Smart Addressing — KYC/KYB verification with neighbourhood context
  • Proof of address with a QR anyone can check
  • Lender portfolio exposure
  • Agent & merchant network viability
  • Consent-governed, withdrawable address evidence

Why existing warnings miss

Optical satellites go blind exactly when it matters

Sentinel-2 cannot see through a rainstorm. The cloud that hides the ground is the same cloud that is causing the flood, so optical-only systems report nothing during the event they exist to detect.

The damage does not stop at the water line

Standing water becomes waterlogged roots in days, a lost harvest in weeks, and a malaria surge in roughly six. Each stage is predictable from the one before, and each is warned about separately or not at all.

Internet alerts arrive after the network fails

A flood takes out power and backhaul. The channel most warning systems depend on is the channel the disaster removes first.

The five-agent pipeline

01

Scout

Watches for new satellite coverage of every area under monitoring, and knows when it arrives. When cloud hides the ground it says so and switches to radar rather than reporting nothing — the moment most services go quiet is the moment ours has to keep working.

02

Analyst

Measures what changed on that specific piece of land: how much of it is holding water, and how much of the crop is under stress. Measurements only — no interpretation, no judgement, just the numbers the decision will rest on.

03

Oracle

Decides whether those measurements amount to a hazard, how severe it is, and how confident we are. Weighs the 7-day rainfall outlook and how many people are actually exposed, then names what the hazard is likely to trigger next — flooding leads to waterlogged crops, which lead to standing water and the health risk that follows.

04

Herald

Turns the decision into an advisory in the subscriber's own language, grounded strictly in what was measured, and delivers it on every channel they chose. When the ground network is down — which is when a flood warning matters most — it escalates to NIGCOMSAT-1R satellite broadcast.

05

Fahis

Runs backward, days later, and asks the one question nothing else asks: were we right? It looks for independent confirmation of the hazard we warned about and records a verdict. Where a remote area has no coverage either way, that is recorded as unverified — never quietly counted as a false alarm. No verdict can ever change an advisory, so accountability stays separate from the warning it judges.

The last mile is a satellite, not a signal bar

WhatsApp, Telegram, Signal, email and Slack all assume the recipient has working internet — the assumption a flood breaks first. At warning level and above, or when every terrestrial channel has failed, SHELTER escalates to a one-way NIGCOMSAT-1R broadcast that reaches the whole Ku-band footprint with no ground network at the receiving end.