Hubble Space Telescope reentry prediction

Here's an attempt to estimate the earliest possible HST reentry date.
The air density (the most significant unknown in determining the reentry date) is calculated by means of NRLMSISE-00 atmosphere model along with two data files for the current and predicted solar and geomagnetic indices. Please, see here for the current value and short range prediction of the solar and geomagnetic indices and here for the long range prediction.
The Marshall Solar Cycle Forecast site gives the best estimate value of the F10.7 and Ap indices along with an approximate ±2 sigma interval. Although the forecast extends to the year 2041, it is still not enough for the reentry (in July 2041 the simulation gives an average altitude of about 440 km). But if the 95^th percentile of the indices is used (simulating a very high solar activity), a high confidence (approximately 97%) lower bound of the reentry date can be obtained from the simulation (roughly speaking, a reentry before that date is completely impossible, by natural decay).

Simulation date: 2020-06-13

The initial state of the satellite has been calculated using the following TLEs: .

Earliest reentry: 2029-01-29 (averaged the 5 TLEs)
Best fit average ballistic coefficient: 140 kg/m²
NRLMSISE-00 data files: "SW-All.txt" updated 2020 Jun 13 10:08:48 UTC, "jun2020f10_prd.txt" updated 11 Jun 2020
Gravity model: SGG-UGM-1 truncated to degree and order 15

The graph (obtained with the 95^th percentile of the indices) shows the predicted minimum, mean and maximum radius vector scaled to a reference sphere with radius of 6371 km (just to show an approximate altitude), but when the instantaneous radius vector drops below 130 km, the blue plot no longer shows the mean radius vector (or semi-major axis), it shows the instantaneous radius vector. The popup label shows the terrestrial time, the radius vector (km) and the inertial speed (m/s).