# Utility Storm Response Intelligence

> A Databricks App combines existing outage forecasts with live weather, NWS alerts, and power-line data for utility maintenance teams.

Source: https://8fde.ai/case-studies/utility-storm-response-intelligence

---

case study

# Utility Storm Response Intelligence

A Databricks App combines the utility’s existing outage forecasts with Meteomatics weather, NWS alerts, and power-line data in a map-led view for maintenance teams.

TL;DR

- Serves the utility’s existing weather-driven incident forecasts through a Databricks App.
- Maps predicted risk across power lines and field infrastructure for rapid operational review.
- Adds Databricks Genie so teams can question the data and inspect what informs a prediction.
- Supports maintenance resource planning before severe weather reaches the service area.

## [Use Case](#use-case)

A U.S. electric utility in a southern state owned weather-driven models that predicted incidents across power lines and field infrastructure. Maintenance teams needed a practical way to use those forecasts when assigning crews and preparing resources ahead of storms.

The utility needed to combine forecast outputs with live weather, official warnings, and grid context. Maintenance planners needed one view that showed where risk was rising and which assets faced the greatest exposure.

## [Solution](#solution)

For this delivery, we built a Databricks App with a map-led dashboard that highlights forecast risk areas, combines incident context, and integrates Databricks Genie for questions about prediction data and forecast drivers.

We use Unity Catalog views to govern the model outputs, grid records, and alert context that feed the app. Maintenance teams can ask questions in plain English and inspect the factors behind a forecast.

Maintenance leaders use the app to compare forecasts with current conditions before assigning crews and resources.

### [Architecture](#solution-architecture)

We separated Meteomatics and NWS feeds from the utility’s client-owned risk models and ArcGIS grid data. We use Unity Catalog views to govern model, alert, and grid context. The Databricks App combines those sources in a map-led experience with Genie analysis and threshold alerts for maintenance planning.

External data sources

**Meteomatics** Live weather API Meteomatics provides live forecasts and 24 weather map layers for current and five-day weather context.

**NWS alerts** Watches & warnings National Weather Service feeds provide active watches and warnings, matched automatically to the towns and regions they affect.

Internal (client-owned) data

**Existing risk models** Client-owned forecasts The utility’s client-owned models predict incidents per town—day and night, five days ahead—along with the top factors driving each forecast.

**ArcGIS grid data** Power-line geometry The utility’s ArcGIS records provide power-line geometry for roughly 275,000 conductor sections.

**Unity Catalog views** Governed serving layer Curated, documented views govern predictions, power-line geometry, active alerts, and calibrated risk thresholds—one version of the truth for every surface, including Genie.

**Databricks App** Operational interface The Databricks App serves governed prediction outputs, live weather, official alerts, and grid context in one experience.

**Map, Genie & alerts** Operations experience A map-led five-day timeline, Genie-powered analysis, and stateful email alerts help teams act when a region or town crosses its calibrated risk threshold.

**Maintenance planning** Human decisions Maintenance leaders use the evidence to allocate crews and prepare resources ahead of severe weather.

### [How it works](#how-it-works)

The utility’s client-owned models combine weather forecasts with infrastructure data to estimate incident risk. The delivery uses those model outputs and adds current and five-day forecasts from Meteomatics. NWS feeds add watches and warnings, which the service matches to affected towns and regions.

Maintenance leaders use the Databricks App to review forecast results, weather, alerts, and power-line data on a map. They can compare areas of concern with the assets in each location.

Databricks Genie lets users ask about a location, incident likelihood, or forecast context and inspect the factors behind a prediction. Maintenance leaders decide how to allocate crews and prepare for storms.