GeoSight (Sematic-Vantage)

A geospatial intelligence platform for analyzing environmental signals from satellite data.

It enables users to explore vegetation health, wildfire activity, and water presence on a 3D globe, with derived intelligence that summarizes real-world conditions into actionable insights.

---

Overview

GeoSight is designed as a lightweight environmental intelligence system rather than a traditional map viewer.

It combines multiple satellite-derived signals into a unified interface where users can:

• visualize environmental data layers
• inspect specific locations or regions
• interpret conditions through a centralized intelligence panel

The system emphasizes signal → context → insight, bridging raw geospatial data with meaningful interpretation.

---

Core Features

Interactive Globe

• WebGL-powered 3D globe
• Smooth navigation, zoom, and exploration
• Layer-based visualization

---

Unified Intelligence Panel

A centralized control and analysis interface providing:

• active layer management
• real-time signal counts
• temporal controls (1d / 3d / 7d / 14d)
• derived summaries of environmental conditions

---

Inspect Panel

Click anywhere on the globe to analyze a location.

Displays:

• coordinates
• classification (NDVI, water, etc.)
• confidence / agreement
• imagery metadata
• contextual interpretation

---

NDVI (Vegetation Health)

• Satellite-based vegetation analysis
• Supports temporal comparison
• Detects vegetation decline patterns

---

Wildfire Detection

• Live wildfire detections (VIIRS)
• Confidence-based classification
• Integrated into intelligence summaries

---

Flood / Water Detection

• Surface water classification (MODIS)
• Permanent water baseline + detected water overlay
• Area-based summaries

---

Vegetation Loss Alerts

• Grid-based detection of vegetation decline
• Severity classification (severe / moderate / mild)
• Ranked and surfaced in the intelligence panel

---

Hazard Outlook (Derived Intelligence)

A rule-based system that synthesizes existing signals into:

• Fire Conditions
• Vegetation Stress
• Water Conditions

This is not a predictive model, but a derived situational summary based on:

• wildfire detections
• NDVI / vegetation context
• vegetation loss alerts
• flood/water state

---

Tech Stack

Frontend

• React + TypeScript
• Vite
• Cesium (WebGL globe rendering)
• Custom CSS (Palantir-style tactical UI)

Backend

• FastAPI (Python)
• REST-based geospatial services

Infrastructure

• Docker & Docker Compose (local development)
• Nginx (frontend serving & routing)
• Terraform (infrastructure-as-code)
• AWS-ready architecture (optional deployment path)

---

Deployment

Current Approach

The system is designed for containerized, self-hosted deployment:

• Local development: Docker Compose (see "Running Locally")
• Production: Docker containers on your own infrastructure
• Can be hosted via:
  • EC2 (Docker Compose or orchestration)
  • ECS / Fargate (future)
  • Any Docker-compatible platform
  • Nginx for frontend routing

API Considerations

The backend relies on external geospatial APIs (NASA GIBS, VIIRS, MODIS). Factor those costs into your hosting budget.

---

Running Locally

1. Clone the repo

git clone https://github.com/ntoptchi/Semantic-Vantage.git
cd geospatial

2. Run with Docker

docker compose up --build

3. Open in browser

http://localhost:5173

---

Design Philosophy

This project is built around:

• Clarity over clutter
• Derived insight over raw data
• Fast interaction loops
• Minimal but intentional UI

The interface is inspired by mission-critical intelligence systems, focusing on density, precision, and usability.

---

Future Work

• Region-based (state/province) analysis mode
• Temporal trend modeling for hazard outlook
• Expanded hazard inference (drought, storm systems)
• Full cloud deployment via Terraform + AWS

---

Author

Nicholas Toptchi
CS @ USF