Richer Metadata
STAC unlocked a far broader set of queryable parameters such as: nadir angle, sensor mode, processing level, cloud cover, and more. It gave users precise control over what data they wanted.
SIGMA Data: Tasking, Delivery & Data Products
SIGMA Data is the primary interface for Orbital Sidekick’s hyperspectral satellite constellation. Designed to service intelligence community analysts, commercial customers, and OSK’s own internal teams. One product. Three radically different mental models. The design challenge was making each audience feel like the platform was built specifically for them, while building a system and workflow that worked for the data pipeline.
In-depth case study available on request.
Overview
The Prime Directive
SIGMA Data serves as the primary interface for customers to interact with Orbital Sidekick’s constellation of hyperspectral satellites. The platform enables users to search & filter historical data archives, task satellites for new collections, analyze spectral intelligence from space, and deliver data products to customer S3 buckets.
As OSK’s designer, I was responsible for creating & managing an intuitive experience that makes complex satellite operations accessible to analysts, researchers, and decision-makers. Our goal was to empower the delivery of geospatial data to energy, agriculture, and defense markets, and allow additional industry-trending geospatial endeavors to succeed.
1.6Mkm²
Area of Operations
Monitored & Archived
SIGMA DATA
Three Audiences.
One Platform.
During discovery, we found that there were three audiences with three different mental models and reasons to access our data products. Tap a card to see which capabilities each one depends on most.
In-Q-Tel
An independent American not-for-profit VC firm based in Tysons, Virginia. It invested in OSK to keep the Central Intelligence Agency, and other intelligence agencies, equipped with the latest in hyperspectral data & insights, in support of United States intelligence capabilities.
Platform Capability
Intel
Commercial
Internal
Map-Based Search · AOI, coordinates, location names
Customisable Filter Ribbon · pin most-used parameters
Satellite Tasking · 4-step frontend, 7-stage pipeline
Data Delivery Dashboard · 8 pipeline statuses, S3 / browser
STAC / STAPI API Access · headless integration
Constellation Viewer · live fleet positions, pass planning
Raw L1A Sensor Data · full calibration control
Analysis-Ready Imagery · georectified, drop-in GIS
Refracting the Science
Into a Product Experience
A machine learning model turns a photon into an insight. But before users could search, task, and order those insights via a user interface, you had to build the data pipeline to fulfill each customer’s unique product experience.
Spectral Range · Visible → SWIR · drag to explore
RED-EDGE · vegetation stress
The Science
Not Your Average Camera
A normal camera records three broad colors. They are red, green, and blue. GHOSt records 472 narrow, contiguous bands from visible light through shortwave infrared. Because every material reflects light in a unique spectral signature, those bands don’t just show what a place looks like, they reveal what it’s made of.
SIGMA Data had to deliver a chemistry lab, to the user...
Standard Camera3 bands
Three wide buckets blended into a single visible color. Enough to see a scene — not enough to identify what’s in it.
GHOSt Hyperspectral472 bands
Hundreds of narrow bands across visible–SWIR. Each pixel carries a full spectral curve — a fingerprint of the materials it contains.
Signal to Decision
From Raw Potential to
Spectrally Intelligent Potential.
The same photon serves a scientist, an analyst, and a decision-maker very differently. Processing levels are a ladder: each rung trades raw fidelity for usability, climbing from calibrated radiance to ready-to-act intelligence that was ingested into the SIGMA Data Platform for ordering, browsing, and download.
Source
472-band
spectral cube
processed into →
LEVEL 1A/1BRadiance
Calibrated spectral data
raw radiance · full 472 bands · ENVI / GeoTIFF
Scientists
Run their own spectral models, train detection algorithms, and validate against ground truth.
LEVEL 1B GEOAnalysis-Ready
Georectified imagery Most customers start here
map-aligned · drop-in for GIS · cloud-masked
Analysts
Overlay directly in their GIS, measure change over time, and brief findings without preprocessing. The workhorse format most orders are delivered in.
DERIVED PRODUCTSIntelligence
Indices, detections & alerts
methane plumes · vegetation stress · burn ratio · change
Decision-makers
Act on a finished answer: a flagged methane leak, a fire-risk map, a structural-damage assessment.
Data Architecture
Product Levels & Data Delivery
Behind that ladder sits the technical architecture. Each processing level is a deliberate trade: how much OSK does to the data before delivery, versus how much the customer does themselves. Here's what each level actually contains.
Product Level
Description
Primary Use Case
Deliverable Assets
Format
Level 1A
No radiometric or geometric correction. Includes raw imagery with georeferencing data (IGM), TOA calibration coefficients, and bad pixel masks.
Advanced users needing full calibration control; researchers requiring raw sensor data.
- HSI Data Cube
- IGM (georeferencing)
- TOA Coefficients
- Dark Offsets
- Bad Pixel Mask
- RGB Preview
ENVI
Level 1B
Radiometrically calibrated to Top-of-Atmosphere (TOA). No geometric correction. Includes IGM for user-applied georeferencing.
Standard analysis workflows; users with GIS expertise who prefer custom georeferencing.
- Calibrated HSI Cube
- IGM (georeferencing)
- RGB Preview
ENVI, NumPy (.npy)
Level 1B Georectified
Radiometrically calibrated to TOA with georectification applied. Analysis-ready imagery.
Operational users; customers needing immediate deployment; minimal GIS processing.
- Georectified HSI Cube
Zipped ENVI
Dark Subtract
Internal
Internal
Imagery with only dark subtraction applied plus georeferencing. Intermediate processing level.
Internal QA/QC; pipeline validation; troubleshooting.
- Dark-subtracted Cube
- IGM (georeferencing)
ENVI
Data Standards & Infrastructure
From Atlas to STAC / STAPI
In a workshop with Element 84, engineers and designers side by side. We mapped out OSK's full data pipeline, constellation architecture, and hyperspectral data types.
The outcome was converting OSK's entire data catalog to the SpatioTemporal Asset Catalog (STAC) and SpatioTemporal Asset Tasking Interface (STAPI) standards. This was a foundational infrastructure decision that cascaded into almost every customer-facing capability in SIGMA Data.
It unlocked even more richer & powerful abilities for SIGMA Data's data products, capabilities, and user interface elements.
STAC/STAPI Workshop between Element 84 and OSK's software engineering team. Together, we made OSK's data STAC/STAPI configurable for commercial use.
System Architecture · STAC/STAPI Order Flow
07 services · 02 actors
STAPI defined the order contract between SIGMA, the new Order Service, and the GHOSt constellation's Operations Manager. Order Operations UI gives Customer Success a way to manage order status, while STAC provides the catalog backbone.
What STAC / STAPI Unlocked
Unlocking Commercial & Government Power
Global Filter Bar Expansion
STAPI-compliant metadata directly powered the customizable filter ribbon in SIGMA's map interface, allowing analysts to pin their most-used parameters and configure the ribbon to match their workflows.
Platform Interoperability
STAC compliance meant customers could plug OSK's API directly into their own tools, pipelines, and platforms without custom integration work, unlocking commercial and government partners at scale.
Standardized Tasking via STAPI
STAPI defined the order contract between users, OSK's services, and the GHOSt constellation. This enabled product configurations like geography type, time range, nadir angle, cloud cover threshold, delivery location, and processing priority.
STAPI Product Configurations
Eight Parameters Per Order
When submitting a tasking order, users configure their selected product with a set of defined parameters. These were translated directly from the STAPI specification into SIGMA's order form UI. Each one was mapped to a real physical or operational dimension of a satellite GHOSt pass.
Order Anatomy - Where Each Parameter Lives
8 Inputs
Environmental & National Impact
Data from Orbit. Impact on the Ground.
SIGMA Data powered some of the most consequential applications of commercial hyperspectral remote sensing to date, reaching federal intelligence programs, environmental missions, and national security customers around the world.
Geopolitical Intelligence
Damage Assessment of the Fordow Fuel Enrichment Plant, Iran.
Following Operation Midnight Hammer on June 22, 2025, OSK analysts used hyperspectral data from the GHOSt-5 satellite to assess damage to the underground Fordow facility — a site previously inaccessible to standard optical analysis.
PCA analysis revealed a linear subsurface structure measuring approximately 315m × 275m beneath the mountain — interpreted as ground shift from potential collapse of the underground facility. Spectral fingerprinting positively identified three of four concrete signatures adjacent to bomb entry points, confirming the gray surface material as concrete ejected from the destroyed underground structure.
This demonstrated hyperspectral sensing's ability to reveal material and structural intelligence invisible to traditional electro-optical imagery — a level of precision previously limited to government programs.
Read the full analysis →472
Spectral bands
315m
Subsurface anomaly
3/4
Signatures confirmed
Environmental Intelligence
Fire Risk & Damage Assessment — Eaton Canyon & Palisades, California.
In November–December 2024, OSK captured HSI data over Eaton Canyon revealing that south-facing slopes near Altadena showed significantly drier, less healthy vegetation than north-facing slopes — months before the fires ignited. When the Eaton Fire burned, the correlation between pre-fire vegetation health indices and fire perimeter was striking.
Post-fire, the Normalized Burn Ratio (NBR) derived from OSK's SWIR bands showed near-perfect correlation with CAL FIRE's Palisades Fire perimeter. Analysis of HSI-derived RGB composites overlaid with the DINS structural damage database mapped burned and surviving areas at a resolution and fidelity unavailable with traditional multispectral sensors.
The data was subsequently used by Federal programs to aid fire risk assessment and post-fire damage analysis across the affected areas.
Read the full analysis →OSAVI
Pre-fire vegetation
NBR
Burn severity
DINS
Damage overlay
