MLDTM

May 2026

The Machine Learning Thermospheric Density Model (MLTDM) is now available to the community through the CCMC Instant Run service via this URL.

MLTDM predicts neutral thermospheric mass density at 400 km altitude using a Random Forest model trained on GRACE and CHAMP satellite drag-derived density measurements (Murphy, K., Halford, A., Bard, C., Rae, J., 2024). The model uses solar EUV irradiance from FISM2 and geomagnetic indices from OMNI as inputs, retrieved automatically for the user-specified time period.

Predictions are available for dates between 2002 and early 2025. Output includes combined polar magnetic local time (MLT) and geographic projection plots of neutral density.

Model source