Tracks citizens' behaviour during the haze crisis
Indonesia’s disaster management apparatus was operating on incomplete signal. Satellite hotspot data and static population density maps were the primary instruments available to agencies managing annual peatland fire crises affecting millions across Sumatra and Kalimantan. The gap was not in data volume. It was in the architecture of how that data was made legible, and to whom, and in what time.
Haze Gazer was designed as a real-time crisis analysis and visualization platform: a single-screen environment that synthesized hotspot locations, population mobility patterns, social media signal across Twitter and citizen journalism channels, image and video feeds, and radio broadcast transcripts into a unified situational picture. The core interface challenge was one of information hierarchy under emergency conditions. A dashboard built for monitoring a slow-moving crisis and one built for a decision-maker managing an acute event require fundamentally different visual grammars. Every design decision — spatial layout, color encoding, the treatment of anomalies in the data, the layering of map states — was made against the constraint of a single screen that had to function as a situation room.
The visualization system distinguished between data types that satellite infrastructure could not capture: population movement encoded as directional flow lines, behavioral signal extracted from social media metadata, and citizen-generated image documentation classified and surfaced in temporal sequence. A mobile version extended the platform to field actors including forest rangers, local government officials, and community organizers, requiring a structural redesign of the information architecture for constrained screen environments without degrading analytical utility.
The platform was eventually deployed in a presidential situation room during the 2016 crisis season.
Award: Early Stage Innovation, Harnessing Data for Resilience Award, USAID