En ▼

Proven Impact

Landslides under control with early action

Hyperlocal forecasting and operational alerts anticipate landslide risk windows in Petrópolis, guiding teams and cutting impact and cost.

Up to 12h

lead time

94%

accuracy

10,313

events analyzed

94%

accuracy

10,313

events analyzed

Precisa de ajuda?

Operational Challenge

Hillside storms: decide earlier

Intense hillside rainfall. With a hiperlocal forecast and AI alerts by community and critical windows, we’re ready to act.

Introduction

Petrópolis is our R&D living lab (funded by FAPERJ) against landslides. Combining hyperlocal forecasting, radar, rain gauges and AI, we turn rainfall and susceptibility into community-level alerts in Petrópolis so teams act before the event.

Challenge

Orographic rain and intense downpours trigger flash floods and landslides on slopes and valley floors. City teams receive generic, late forecasts that don’t support timely decisions. Achieving community-level granularity remains a daily challenge for civil protection.

Solution

A 1-km WRF forecast plus rain sensors and susceptibility maps feed an AI model that anticipates risk through an operational dashboard. With >94% accuracy, we deliver community-specific alerts that drive pre-positioning and preventive road closures.

Gil Kempers

Director of Emergency Management for the City of Petrópolis

"Cittua is a technology platform that uses artificial intelligence to generate more accurate information, enabling Petrópolis to issue rain warnings and alerts to local residents and visitors. The entire process is managed via geolocation through the mobile app, aiming for prevention and, consequently, saving lives."

Introduction

Petrópolis is our R&D living lab (funded by FAPERJ) against landslides. Combining hyperlocal forecasting, radar, rain gauges and AI, we turn rainfall and susceptibility into community-level alerts in Petrópolis so teams act before the event.

Challenge

Orographic rain and intense downpours trigger flash floods and landslides on slopes and valley floors. City teams receive generic, late forecasts that don’t support timely decisions. Achieving community-level granularity remains a daily challenge for civil protection.

Solution

A 1-km WRF forecast plus rain sensors and susceptibility maps feed an AI model that anticipates risk through an operational dashboard. With >94% accuracy, we deliver community-specific alerts that drive pre-positioning and preventive road closures.

Gil Kempers

Director of Emergency Management for the City of Petrópolis

Landslides under control in Petrópolis

94% accuracy: 1km prediction + AI, local boundaries, and alerts provide the basis for preventive actions.

Petrópolis is our “living lab” for landslide R&D, supported by FAPERJ. The city faces intense rainfall on steep slopes and valley floors, demanding practical lead time, territorial granularity and plain operational language. Cittua built a pipeline that turns risk into decision, blending data science with field routines to reduce human and financial impact.

Our methodology fuses a 1-km WRF forecast, rain sensors monitoring and susceptibility maps (geology, slope, soil, vegetation) into an AI model trained on the local incident inventory. The pipeline ingests data continuously, applies forecast bias-correction and computes community-level risk. Using operational thresholds (e.g., 50–60 mm/6h), we issue a dashboard and proactive alerts in critical windows, with checklists that standardize response.

Results show robustness and real-world value: 10,313 events (2015–2024) underpinned training/validation, and the top classifier achieved ~94% accuracy. In assisted operations, we typically deliver 3-12 hours of lead time for intense rain with landslide potential, enabling resilience actions such as pre-positioning and preventive closures at sensitive spots. In operational snapshots, alerts guided decisions before the rainfall peak, cutting reactive responses and false positives and improving time-to-action.

Beyond tech, our edge is decision language: a simple dashboard with neighborhood-level risk, severity tiers and objective recommendations (“when to act, where, and with what”). FAPERJ’s support was decisive to move applied research into field testing. Next steps: scale to full operational use across the Serra region, deepen community integration, and evolve short-term forecast keeping the focus on what matters most: anticipate risk and protect lives.

Faça como Teresópolis: antecipe, alerte, proteja.

Lado a lado com a Defesa Civil: oficinas práticas, simulações e alertas hiperlocais no plano de contingência operacional da cidade. Previsão que Salva Vidas. Dados que viram decisão. Mais tempo para agir, mais segurança para a sua cidade.

Talk to Cittua

Ready to write the next success story with Cittua?

Turn climate challenges into opportunities. Talk to our experts and discover how Cittua technology can protect lives, reduce risks and generate results for your business.