🌧️ RainSeekers

RainSeekers is a web application developed for the NASA Space Apps Hackathon 2025.
It predicts temperature, pressure, humidity, and wind speed for a given location and date, then evaluates whether specific activities (like going on a picnic, visiting the beach, or hosting a concert) are suitable for that day.

Our mission: Help people plan better, safer, and more enjoyable activities by making NASA’s weather data accessible to everyone.

---

🚀 Features

• NASA Data Integration – Uses NASA’s M2T1NXSLV v5.12.4 dataset (1980–1983, 1991–2011).
• Deep Learning Predictions – Forecasts:
  • Temperature (T2M)
  • Humidity (QV2M)
  • Wind speed (U10M, V10M)
  • Surface pressure (PS)
• Supported Cities – Cairo, Washington DC, Rio de Janeiro, Tokyo, Sydney.
• Activity Scoring System – Rates how suitable the weather is for different activities.
• Historical & Future Data – Explore trends and predict upcoming conditions.
• Data Visualizations – Interactive charts for a clear, user-friendly experience.
• Accessible Web Platform – No technical background needed.

---

🛰️ Data Pipeline & Processing

We designed an optimized pipeline to handle large NASA climate datasets efficiently:

1️⃣ Data Collection

• Downloaded from NASA using eatchaccess.
• Extracted only the 5 supported cities: Cairo, Washington DC, Rio de Janeiro, Tokyo, Sydney.
• Selected daily averages for key variables:
  • T2M (temperature)
  • QV2M (humidity)
  • U10M, V10M (wind components)
  • PS (surface pressure)

⏱️ Performance:

• Sequential processing: 3–5 hours per year of data.
• With multithreading: 40–60 minutes per year.
• Saved intermediate results to avoid repeating this step.

---

2️⃣ Preprocessing

• Converted raw NetCDF data to Pandas DataFrames.
• Engineered new features to represent cyclic variables:
  • Day: day_sin, day_cos
  • Month: month_sin, month_cos
  • Latitude: lat_sin, lat_cos
  • Longitude: lon_sin, lon_cos

---

3️⃣ Data Optimization

• Saved preprocessed data in Feather format (.feather) for instant loading.
• Before: Loading NetCDF files for 25 years = ~40 minutes.
• After: Loading Feather files = < 1 second.
• Massive improvement in training & prediction efficiency.

---

⚡ Efficiency Gains (Before vs After)

Step	Before	After (Optimized)	Improvement
Downloading and Processing (1 year)	3-5 hours	40-60 minutes	~4x faster
Data Loading (25 yrs)	~40 minutes	< 1 second	~2400× faster

---

🌍 Scalability & Reproducibility

• Scalable: Can be extended to more cities or even global coverage.
• Reproducible: Intermediate and final datasets are saved and versioned, ensuring consistent results without repeating heavy computations.
• Hackathon-Ready: Preprocessed data allows training and predictions in real time during demos.

---

🛠️ Tech Stack

• Backend + Frontend: Flask (serves both the web interface & API)
• Machine Learning: Deep Learning Regression Model
• Data: NASA MERRA-2 (M2T1NXSLV v5.12.4)
• Deployment: (TBD – local/demo for hackathon use)

---

📈 Model Performance

We trained a deep learning regression model to predict 5 climate variables. Below are the Mean Absolute Errors (MAE) for each target:

Variable	Description	MAE
T2M	Temperature (K)	1.577
QV2M	Humidity (kg/kg)	0.001
U10M	Wind (east-west) (m/s)	2.654
V10M	Wind (north-south) (m/s)	2.855
PS	Surface Pressure (Pa)	418.554

✅ The errors are within acceptable ranges for climate prediction, making the model reliable for activity planning.

---

📂 Project Structure

RainSeekers/
│── data/                    # Dataset (for historical data)
│── model/                   # Trained regression model
│── scaler/                  # The scaler used with the data
│── static/                  # Assets (CSS, JS, images)
│── templates/               # HTML templates
│── README.md                # Documentation
│── app.py                   # The flask app
│── data_handler.py          # The component responsible for handling data efficiently
│── weather_predictor.py     # The component responsible for predicting future/unavailable data
│── requirements.txt         # The required python libraries

---

⚙️ Installation

Clone the repository and install dependencies:

git clone https://github.com/Aya-Abdellatif/rain-seekers.git
cd rain-seekers
pip install -r requirements.txt

Run the Flask app:

flask run

Open in your browser at:
👉 http://localhost:5000

---

📊 How It Works

1. Input: User selects a city and a date.
2. Prediction: The deep learning model forecasts weather variables.
3. Activity Scoring: The system assigns a score (e.g., "Good for picnic 🌳", "Bad for outdoor concerts 🎤").
4. Visualization: Charts display historical trends and predicted values.
5. Recommendation: Users receive a simple, actionable suggestion.

---

🌍 Impact

RainSeekers makes NASA’s weather data accessible to everyone.

• Helps individuals plan safe outdoor activities.
• Supports event organizers (e.g., concerts, sports).
• Provides insights for communities and small farmers.
• Educates students and enthusiasts about data-driven climate forecasting.

---

🌐 Web App Interface

Home Page

Weather Insights

About Us

Clothing Recommendation

---

👩‍💻 Team

Developed by Team RainSeekers for the NASA Space Apps Hackathon 2025:

Pavly Samuel

Aya Abdullatif

Aalaa Ayman

Ahmed Essam

---

📜 License

This project is licensed under the MIT License – free to use and modify.

---

✨ “Exploring data, predicting tomorrow.” – RainSeekers