DMSP Cusp Crossing Database (1987-2014) and Trained XGBoost Models for Cusp Boundary Prediction

This archive accompanies the paper:

Zhu, Y., Michael, A. T., & Toffoletto, F. R. (2026). Predicting Ionospheric Cusp Location from Solar Wind: An XGBoost Model Trained on 27 Years of DMSP Data. Journal of Geophysical Research: Space Physics (under review).

Contents

• cusp_crossings_1987_2014.csv.zip: 48,056 DMSP cusp crossings (F06-F18, 1987-2014) with equatorward and poleward MLAT/MLT boundaries and 74 matched solar wind/IMF features.
• cuspML_models.zip: Trained XGBoost model weights for four prediction targets in XGBoost universal binary JSON (.ubj) format, plus metadata and README.

Companion analysis code: https://github.com/YiwenZhu77/cuspML