Checking Space Weather...

Analyzing the latest data from NASA satellites.

Solar Wind Speed

-- km/s
Live from Satellites

AI Forecast (1 Hour)

-- km/s
Range: --

Storm Risk

--
Storm Probability: --%

Magnetic Shield (Bz)

-- nT
Lower is more dangerous

Prediction Rationale

Analyzing...

Space Physics Parameters

Particle Density -- p/cc
Temperature -- K
Wind Pressure -- nPa
Total Magnetism (Bt) -- nT

Automated Telemetry Assessment

System Log
Generating AI situation report based on live parameters...

Prediction Accuracy Tracker

This table compares the AI's 1-hour forecast against reality.

Prediction Time Target Time (+1H) Predicted (km/s) Actual (km/s) Error Predicted Risk

Error Analysis

Average Error (MAE) --
Worst Error --

System Architecture & ML Pipeline

HeliosCast utilizes a hybrid Machine Learning pipeline designed to predict solar wind speeds and classify geomagnetic storm risks. The models are trained on the high-resolution OMNI dataset, which contains decades of near-Earth solar wind magnetic field and plasma parameters. Data is resampled to 5-minute intervals, missing values are imputed using forward-filling interpolation, and the target variables are shifted to create a strict T + 1 Hour forecast horizon.

Training Corpus Stats

Total Records--
Input Features--
Primary Engine--

Feature Engineering Space

Features are grouped into Raw Physics (e.g. Density, Bz), Derived Parameters (e.g. Plasma Beta, Alfven Mach), and Temporal Lags (1h/3h delays, Rolling Averages, Derivatives).

Regression Benchmark (T+1H Speed)

Predicting the exact solar wind speed (km/s). Evaluated using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R² Variance Score.

Algorithm MAE RMSE R² Score

Classification Benchmark (Storm Risk)

Categorizing risk into Normal, Elevated, or High based on predicted solar activity. Evaluated using multi-class Accuracy and weighted F1-Score.

Algorithm Accuracy F1 Score

Global Feature Importance (SHAP Approximation)

What is this? This chart uses SHAP (Explainable AI) to reveal which factors the AI relies on most to predict space weather. The longer the bar, the more critical that feature is for the AI's decision.

Feature Glossary:
  • bx, by, bz: Magnetic Field Vector Components (nT)
  • bt: Total Magnetic Field Strength
  • speed: Solar Wind Plasma Speed (km/s)
  • density: Proton Density (p/cc)
  • temperature: Plasma Temperature (K)
  • dynamic_pressure: Force exerted by solar wind
  • ma_1h: 1-Hour Moving Average (Trend)
  • std_1h: 1-Hour Standard Deviation (Volatility)
  • lag_1h / 3h: Past data from 1 or 3 hours ago