Barra - Predictive Dam Monitoring Platform

Links:

Web app: http://barrafrom.space 
Github: https://github.com/frisibeli/barra-dam-monitoring

💎 Idea

Dam infrastructure across Bulgaria and Southeast Europe is monitored through manual processes with critical blind spots. Geodesic deformation surveys are typically performed twice a year, leaving 6-month gaps in structural monitoring. Reservoir state is tracked through stale, scattered data. Inflow forecasting - essential for managing drawdown, floods, and droughts - is largely absent.

The consequences are documented. Biser, 2012 - a known structural crack went unrepaired for nine years. Studena, 2019 - the data existed; nobody was watching.

Barra is a continuous satellite monitoring platform for dam infrastructure. Continuous InSAR displacement closes the structural monitoring gap. Automated water surface area extraction replaces stale volume estimates. A machine learning inflow forecast - battle-tested over the past week against observed reservoir behavior - gives operators days of lead time for drawdown decisions. 3D terrain simulations using digital elevation models (DEM) overlay forecasted water levels onto the surrounding landscape, making land area at risk of inundation visible and intuitive - for operators, civil protection, and downstream communities.

We have a pilot client. The dam management authority responsible for Ogosta reservoir in northwestern Bulgaria has agreed to be our launch partner. We validated operational pain points on-site with the team performing manual surveys today, and we are building the platform against their workflow.

🛰️ EU Space Technologies

Copernicus data is the core engine of the platform.

Sentinel-1 (SAR) - Structural displacement. We process interferometric SAR pairs with SBAS via MintPy to produce millimeter-precision time series of dam body movement. SAR penetrates cloud cover and operates day and night, delivering uninterrupted coverage on a 6–12 day cadence - no ground equipment required.

Sentinel-2 (optical) - Reservoir state and snow cover. We compute NDWI to extract reservoir surface area on 14-day cycles, replacing stale estimates with continuously updated state. We compute NDSI across the upstream catchment to track snowpack, which feeds directly into our ML inflow forecasting model.

Copernicus DEM - 3D terrain risk simulation. We use the Copernicus Digital Elevation Model to render reservoir and downstream terrain in 3D, then overlay forecasted water levels to simulate inundation extent. This translates abstract numbers into a visual that operators and authorities can act on immediately.

All Sentinel and Copernicus data is accessed through the Copernicus Data Space Ecosystem (CDSE). Each mission delivers a measurement the others cannot - together they form the platform.

🌊 Space for Water

Barra addresses two of the three challenges directly: Securing Equitable and Efficient Access to Water and Disaster Risk Monitoring.

On equitable and efficient access: better water management requires accurate inflow forecasts, up-to-date reservoir state, and transparency between operators and communities. Our ML inflow model - validated against observed reservoir behavior - combined with continuous water area monitoring gives operators the lead time to manage drawdown fairly across downstream users, anticipate scarcity, and avoid the kind of mismanagement that turned Studena into a 100,000-person water crisis.

On disaster risk monitoring: the most consequential dam failures in the region were enabled by monitoring gaps, not extreme events alone. Barra's InSAR layer makes structural change observable in near-real time, replacing inspection cycles measured in months with monitoring measured in days. The 3D DEM-based inundation simulation translates forecasted water levels into a concrete map of land area at risk - actionable input for operators, civil protection, and the communities living downstream.

🤼 Team

Kristiyan Krumov - Technology & AI. Built the satellite pipelines, the Sentinel-1 InSAR stack with SBAS/MintPy, the Sentinel-2 water and snow modules, the ML inflow forecasting model, the 3D DEM simulation layer, and the FastAPI backend. Grew up in Montana, Bulgaria - the city defined by the dam next to it.

Stefan Stoilov - Business & Product. Shapes go-to-market and product roadmap. Translates field-discovered pain points into product requirements and commercial structure.

Georgi Karastoyanov - Government Relations & BD. Our bridge to public sector clients. Led the Ogosta customer discovery engagement and is building the relationships to convert the pilot into a B2G contract.

All three of our grandfathers were workers or engineers behind Bulgarian dams. This mission is personal.

Presentation: https://docs.google.com/presentation/d/1PysrVrJRRRJpwmMXIbYEwLDrcHn5TJgFOdRI3sGoLRY/edit?usp=sharing