Batticaloa Flood Early Warning System

Hourly streamflow forecasts · Floodplain inundation maps · Threshold-based alerts
Batticaloa District · Eastern Province · Sri Lanka
Project overview

What is BFEWS?

The Batticaloa Flood Early Warning System (BFEWS) is an operational, physically-based flood forecasting system covering the basins that drain into the Batticaloa lagoon on Sri Lanka's east coast. It combines real-time rain and water-level observations with ECMWF rainfall forecasts and a chained HEC-HMS / HEC-ResSim / HEC-RAS model suite to produce 9-day streamflow forecasts, floodplain inundation maps and threshold-based flood alerts.

4
River basins covered
10+10
River gauges + rain gauges
9days
Forecast horizon · hourly
3
Alert tiers · auto-distributed
Catchments

Basins draining into the Batticaloa lagoon

Four major river basins feed the Batticaloa floodplain. Each is modelled end-to-end — from upstream rainfall–runoff, through reservoir operations, to 2-D floodplain hydraulics around the lagoon.

Batticaloa catchment terrain map
Terrain DEM of the Batticaloa catchment with the four basins and major reservoirs.

Basins covered

  • Maduru OyaLargest basin in the system. Maduru Oya reservoir operations routed through HEC-ResSim.
  • Mundeni AruUpstream rainfall–runoff routed to the lagoon; alert-monitored at Sink_Mundeni.
  • Magilawattuwan (Mahawelliganga downstream)Routed to Wadamunei — a primary alert-monitored junction.
  • Andella OyaDrains directly to the lagoon; routed to Manduur for alerting.
Hydrological network

Real-time observation network

A dedicated network of telemetered rain and river gauges feeds the system every few minutes. Live data is hosted at rivernet.lk and ingested into BFEWS through the obs-collector service.

Network composition

10
River gauges
10
Rain gauges
  • Telemetered & 2-minute polledobs-collector service polls the RiverNet API every 2 minutes.
  • ECMWF IFS forecast feedGridded numerical weather prediction refreshed every 6 hours, 9-day horizon.
  • Validated against satelliteInundation outputs verified against UNOSAT 10 m flood maps.
Gauge network map showing rain and river gauge locations
Rain gauges, river gauges and reservoirs across the Batticaloa basins.

Field installations

Gauges are deployed on bridges and tank bunds across the four basins, with solar-powered telemetry feeding real-time data into the central database. Stations are accessible publicly at rivernet.lk.

Rain gauge installation on a bridge
Rain gauge on a bridge crossing
Water level gauge installation
Water-level gauge at a reservoir
Model setup

A chained HEC modelling suite

BFEWS chains three USACE Hydrologic Engineering Center models — HEC-HMS, HEC-ResSim and HEC-RAS — each handling one stage of the rainfall-to-flood pathway. Models run on a dedicated server while only the observations and live outputs are exposed through the cloud.Sri Lanka's Irrigation Department set up these models and continuously refines them based on field measurements and observations.

Step 1 · Catchment hydrology

HEC-HMS

Rainfall–runoff modelling for each of the four upstream basins. Produces streamflow time series at every junction in the network.

Step 2 · Reservoir routing

HEC-ResSim

Routes upstream inflows through the Maduru Oya and Unnichchai reservoirs using operational rule curves, producing controlled releases.

Step 3 · Floodplain hydraulics

HEC-RAS (2-D)

Two-dimensional unsteady simulation of the Batticaloa floodplain — generates hourly inundation maps with depth and velocity at every grid cell.

INPUTS PROCESSING MODEL CHAIN OUTPUTS Observed Rainfall gauge network Rainfall Forecast ECMWF · every 6 h Operational Database obs · runs · outputs Rainfall Merger gap-fill · merge · bias-correct Model Orchestrator runs the model chain Upstream catchment HEC-HMS Reservoir routing HEC-ResSim Downstream catchment HEC-HMS Floodplain hydraulics HEC-RAS (2-D) Hydrographs Flood Alerts Inundation Maps
Operational sequence — automated on a 6-hour cycle.
Decision support

Alert thresholds

Each basin has explicit trigger thresholds combining reservoir water-level and accumulated rainfall. When any trigger fires for a forecast cycle, the Distributor service generates a bulletin and pushes it to subscribed operators by email and FTP.

Basin Reservoir Alert level
Andella Oya With reservoir WL > 33 m (msl) and 24 h rainfall 100–135 mm; or WL < 33 m and rainfall > 190 mm in 24 h
Magilawattuwan With reservoir WL > 27 m and 24 h rainfall 100–135 mm; or WL < 24 m and rainfall > 190 mm in 24 h
Mundeni Aru Without reservoir High likelihood of 24 h rainfall > 135 mm
Maduru Oya Without reservoir High likelihood of 24 h rainfall > 135 mm
WL = water level above mean sea level. Thresholds set by the project hydrology team and configurable per junction.
Live tools

Open the operational dashboards

Four web tools sit on top of the live database — accessible to operators, partners and decision-makers.

Project

Clients & implementing team

Client
UN FAO
Food and Agriculture Organization of the United Nations
Project management
LHI
Lanka Hydraulic Institute
System development
STEMCONNECTX
STEMCONNECTX (Pvt) Ltd