The multi-source challenge
PlanetSentry ingests data from sources that were designed independently, use different data models, publish on different schedules, and describe overlapping sets of events in different ways. EONET uses a category-based event model with geometry collections. USGS publishes GeoJSON earthquake feeds with detailed seismic parameters. GDACS uses XML-based alert documents with impact scoring. NOAA provides tropical cyclone advisories in structured text formats.
The first challenge is normalizing these disparate inputs into a common internal representation that captures the essential information from each source — what happened, where, when, how severe — without losing the source-specific detail that makes each feed valuable. This normalization must handle missing fields, inconsistent coordinate systems, and varying temporal precision.
Event categorization and severity mapping
Different sources use different severity scales. EONET categorizes events by type but doesn't assign severity. USGS provides earthquake magnitude. GDACS assigns alert levels (green, orange, red) based on impact estimates. Tropical cyclone advisories include Saffir-Simpson category or maximum sustained winds.
PlanetSentry maps these heterogeneous severity indicators onto a display system that allows users to quickly assess relative severity across event types. An M7.0 earthquake, a Category 4 hurricane, and a GDACS red alert all indicate serious events — but expressing that equivalent severity across fundamentally different measurement scales requires deliberate design choices rather than simple numerical comparison.
Geographic display and clustering
Displaying thousands of events on a 3D globe without overwhelming the user requires intelligent geographic handling. Events that are geographically close are clustered at lower zoom levels and revealed individually at higher zoom levels. Event markers are sized and colored to convey severity, allowing users to scan the globe and immediately identify the most significant active events.
The 3D globe projection introduces display challenges that flat maps avoid. Events near the poles can appear distorted. Events at the date line must be handled correctly. The globe's curvature means that marker sizes that look appropriate at one zoom level may be too large or too small at another. These are solved through dynamic scaling that adjusts marker rendering based on the current view state.
Refresh cadence and data freshness
Different sources update at different rates. USGS earthquake data updates within minutes of an event. EONET events may be added or updated on a daily basis. GDACS alerts publish within hours of a significant event. Weather overlay data refreshes multiple times daily. PlanetSentry manages these different cadences to present the freshest available data without overwhelming source APIs with excessive requests.
The platform implements caching strategies that balance data freshness with performance and API courtesy. Earthquake data, which changes most rapidly, is cached for shorter periods. Event catalog data, which changes less frequently, is cached longer. Weather overlay imagery is refreshed on its natural publication cycle. The result is a monitoring surface that feels current without generating unnecessary load on upstream data providers.