oah
FeederSync Aqua - OneAquaHealth 2026 Proposal Package
October 4, 2026
FeederSync Aqua
Turning freshwater ecosystem observations into irrigation decisions
One-liner: FeederSync Aqua is a water-energy resilience intelligence platform that fuses live weather, groundwater reference data and real OneAquaHealth freshwater ecosystem observations into deterministic irrigation and resilience decisions.
The Problem
Irrigation does not end at the field edge. Runoff carries nutrients, sediment and pathogens into streams, and withdrawals change the dilution available to the ecosystem. Yet irrigation scheduling and freshwater ecosystem health are managed as two separate worlds: one runs on weather forecasts and pump constraints, the other lives in research datasets nobody operationalizes.
- Water managers schedule pumping blind to ecosystem conditions. A degraded stream reach gets the same irrigation policy as a healthy one.
- Environmental data stays academic. The OneAquaHealth research network has real stream monitoring observations, but there is no pipeline turning them into operational water-use decisions.
- Resilience scores are black boxes. Most dashboards show a number nobody can audit. When a score cannot be traced to its inputs and weights, it cannot drive policy.
FeederSync Aqua closes this loop: ecosystem observations become deterministic inputs to the same resilience engine that schedules irrigation.
How It Works
A five-step pipeline, all computed, nothing simulated for the camera:
- 1Observe. The platform pulls real OneAquaHealth public API data: 5 European research cities, 106 stream monitoring sites, 96 health-risk records with pathogen, fecal and antimicrobial-resistance indicators. Every record is served with source, retrieval timestamp and sampling date, always labeled historical.
- 2Assess. A deterministic eco freshwater index combines the observed risk factors with fixed published weights (pathogen 0.40, fecal 0.35, ARG 0.25) into a 0-100 score and a policy band (LOW / MODERATE / HIGH / ELEVATED). The same input always yields the same output. No model invents it.
- 3Connect. The eco observation feeds the resilience engine as an optional deterministic factor in its freshwater-risk component, alongside live water stress (Open-Meteo), groundwater extraction stage (CGWB 2025 reference), energy stress and irrigation flexibility.
- 4Optimize. The irrigation optimizer generates real pumping windows respecting solar availability, pump demand, transformer limits, soil moisture and environmental conditions.
- 5Adapt. Change the scenario and the decision changes: heavy rain delays unnecessary irrigation, drought raises water stress and the schedule responds. The published schedule reaches farmers through the Farmer view.
Measured Results (Working Prototype)
Real outputs from the running application, verified during recording:
| Check | Result |
|---|---|
| Benevento site BN1, observed 2023-06-28 | Pathogen 20.5, fecal 24.3, ARG 47.9 -> eco index 28.7 -> policy MODERATE |
| Coimbra site C1, same pipeline | Eco index 20.1 -> policy LOW (different site, different decision) |
| Toulouse site T1 | Eco index 41.6 -> policy MODERATE |
| Tumakuru irrigation optimization | Windows 15:00-18:00 + 08:00-14:00, 800 pump-hours, 100% coverage |
| Heavy rain scenario | Windows shrink to 15:00-16:00 + 08:00-11:00, 12 slots rain-gated out |
| Drought scenario | Resilience 42/100 (stressed), water stress 57/100, schedule recomputed |
Data Honesty
What the system does not do, stated plainly:
- The OneAquaHealth observations are European research-site records from 2023-2024, served as reference ecosystem data. They are never presented as live readings or as Karnataka measurements.
- Live pH, dissolved oxygen, turbidity, conductivity and nitrate feeds are not integrated. The water-quality endpoint reports this honestly instead of fabricating values.
- When an upstream source is unreachable, the API returns 503 with a clear message, never a plausible-looking fallback number.
- A documented data contract defines how future local monitoring stations plug in: POST observations as {siteCode, sampledAt, indicators, source}, and the eco index recomputes deterministically with no engine changes.
Business Model
- Utilities and irrigation departments pay SaaS per managed feeder or command area for the resilience and scheduling intelligence layer.
- Performance tier takes a share of verified value: reduced peak power cost and avoided water waste, measured against baselines the utility already owns.
- Farmers pay nothing. They receive optimized schedules and keep full manual override, enrolled through FPOs as the trust anchor.
Pilot Plan
- Phase 1 (months 1-6): 2 feeders in the Tumakuru/Kolar belt, Karnataka. Validate schedules against actual pumping, confirm the resilience scores track observed stress, integrate the first local freshwater sampling station via the data contract.
- Phase 2 (months 7-18): 100 feeders across Karnataka and Maharashtra via DISCOM partnership, aligned with PM-KUSUM-C solarized feeders.
- Phase 3 (year 3): Multi-state rollout; ecosystem observations from Indian urban stream monitoring folded into the same deterministic pipeline.
Live Prototype
Nine-page dark command-center app: Overview, Water, Freshwater, Energy, Resilience, Optimization, Alerts, Simulation, Data Sources, Farmer. Real Open-Meteo weather/solar/soil-moisture/ET0 with 15-minute cache and stale-cache fallback, CGWB 2025 district groundwater reference, deterministic engines throughout, SQLite cache, Leaflet map and charts. The existing FeederSync Agri simulator and RBAC console keep working untouched.
Risks & Mitigations
- EU reference data is not local ground truth -> labeled historical everywhere; Karnataka pilot adds local stations through the documented contract.
- No live water-quality sensors yet -> honest unavailable states today; the integration contract is already in the codebase.
- Farmer trust in scheduled pumping -> manual override always available; FPO as trust anchor; schedules published transparently with the facts behind them.
- Upstream API downtime -> 6-hour cache plus stale-cache fallback, then explicit 503. The demo video shows the failure path working.
Team
