NovaCast Weather Forecast

A modern weather forecasting application built for the NASA Space Apps Challenge 2025.

NovaCast utilizes advanced statistical ensemble methods combining NASA POWER and ERA5 data to deliver data driven long range weather forecasts up to 1.5 years ahead.

View Website  •  Report Bug

Key Features

Feature	Description
GPS Location Support	Automatically detect your current city with a single click using GPS & Reverse Geocoding.
Smart Forecasting	Uses weighted ensemble models combining NASA POWER historical data and ERA5 reanalysis.
Long Range Vision	Extended predictions up to 540 days (1.5 years) utilizing ensemble forecasting.
Smart Caching	Tiered backend caching (180/360 days) and multi-city frontend caching for instant load times.
Ensemble Engine	Hybrid blending of NASA POWER and ERA5 datasets for enhanced reliability.
Global Support	Seamless coordinate resolution for cities worldwide with localized matching priorities.
In-Memory Caching	1 hour TTL cache for instant repeat requests, optimized for low resource servers.

---

System Architecture

Technology Stack

Component	Technology	Description
Backend		Asynchronous API handling for high concurrency.
Data Engine		Vectorized data processing and statistical blending.
Frontend		High performance, zero dependency implementation.
Styling		Custom Glassmorphism design system.

Data Pipeline

graph LR
    A[User Request] --> B(NovaCast API)
    B --> C{Cache Check}
    C -- Hit --> D[Return Cache]
    C -- Miss --> E[Fetch External APIs]
    E --> F[NASA POWER]
    E --> G[ERA5 Reanalysis]
    F & G --> H[Ensemble Blending]
    H --> I[Store in Cache]
    I --> D

Loading

---

Deployment

Backend Setup

# Clone the repository
git clone https://github.com/AlpaganB/novacast.git

# Navigate to backend directory
cd novacast/backend

# Install dependencies
pip install -r requirements.txt

# Execute Server
uvicorn app:app --reload

Frontend Configuration

The frontend utilizes standard web technologies and requires no compilation.

1. Configure the API_BASE_URL in script.js to point to the operational backend.
2. Deploy the frontend/ directory via a static asset server.

---

Compliance and Support

License

This project is licensed under the MIT License.

Acknowledgments

• NASA POWER Project: Source for global solar and meteorological data.
• Open Meteo: Support for ERA5 reanalysis and Geocoding services.

---

NovaPulse © 2026