This is a one-day-ahead forecast of the plasma density in the plasmasphere provided with a 1-hour cadency. It is performed via a neural network, which takes the time history of the solar wind and the Kp index as input, combining the data sources present in PAGER with a hierarchy determined by their level of accuracy.

When there are too many missing values of solar wind and Kp index for a given date, the machine learning model misses input features and cannot produce an output. Therefore the plot for that date will not be produced.