Aral Sea Early Warning
Water Crisis Intelligence
Aetheris already monitors Kazakhstan's water quality region by region. Pointed at the Aral Sea basin, that same Water Quality layer becomes an early-warning system for the world's most-cited water-scarcity collapse.
A sea that lost 90% of its volume — and the water it fed.
Six decades of irrigation drawdown on the Syr Darya and Amu Darya turned the Aral Sea into salt flat and dust. Communities in the Priaralye now draw from water that reads critical across the basin — but no one gets warned before it turns.
One signal chain, nation-wide.
Most environmental tools show you a slice. Aetheris closes the loop — from raw photons hitting a sensor to a decision made on the ground in Kazakhstan.
Sense
A national nervous system. Satellite-fed CAMS air-quality fields, Open-Meteo weather, and community reports stream 4,032 fresh readings a day — every region, hourly — into one coherent model.
Reason
Environmental AI that doesn't just chart the data — it explains it. Risk forecasting, anomaly detection, and causal analysis across every layer.
Act
From insight to intervention. Ranked sustainability actions, modeled impact, and a community that verifies change on the ground.
What's live, and what isn't.
Air quality readings are live today — fetched per city from Open-Meteo's CAMS satellite and forecast models, refreshed hourly. Water quality is currently a deterministic regional baseline, not a sensor feed; real-time ingestion is the next build milestone. Every number on the platform carries the badge that says which it is.
View live dataNext: SMS / USSD alerts for off-grid communities.
The fishing and farming communities of the Priaralye are exactly the people least likely to be watching a dashboard. The next build pushes WQI threshold alerts over SMS and USSD — no smartphone, no data plan required — driven by the same pipeline you see above. This step is on the roadmap, not yet shipped.
Known limitations
As of this submission- 01One of five layers is live. Air quality (AQI, PM2.5, PM10, NO₂) and weather come from Open-Meteo / CAMS at page load. Water, biodiversity, industrial load and environmental risk are a deterministic regional baseline — indicative, not measured.
- 02The Aral WQI figures above are modeled, not sampled. They encode known basin conditions rather than readings from instruments in the water. Treat them as a demonstration of the pipeline, not as survey data.
- 03Community reports are corroborated, never verified. Submissions persist to a shared Postgres database and are visible to every visitor in any browser, rate-limited to five per hour per device. Everything arrives as “awaiting review”; when two or more different devices report the same category in the same city within 72 hours, that group moves to “corroborated”. That is independent people describing the same thing — not proof they are right, and no instrument checks them. Nothing is ever auto-promoted to “verified”, and there is still no human moderator. The feed also carries 6 clearly marked illustrative entries from 6 example contributors.
- 04Shipped vs roadmap. Live now: 28 city stations across 17 regions, the Atlas, dashboard and assistant. Not yet built: real-time water sensor ingestion and the SMS/USSD alerting described above.
The whole platform speaks Kazakh, Russian and English.
Climate tools written only in English exclude the communities living the crisis. Every page, index and AI answer on Aetheris is fully localized — the Priaralye residents this work is for can read it in the language they actually speak, not a translated summary.
Built by Zhanbolat Izbanov, Alexander Tokarev and Insar Turtaev for the Oxford Saïd Global Climate Tech Challenge 2026.