Below is a periodically updated (every week or so) list of the projected start and end times of the SAAfree periods of the HST orbit as determined from the SOGS scheduling software for each SAA model.



SAA Model 02 (FGS guiding)  SAA Model 05 (FGS exposures)  SAA Model 23 (NICMOS) 
SAA Model 24 (STIS/CCD)  SAA Model 25 (STIS/MAMA)  
SAA Model 27 (ACS/SBC)  SAA Model 28 (ACS/CCD)  
SAA Model 29 (WFC3/UVIS)  SAA Model 30 (WFC3/IR)  
SAA Model 31 (COS NUV/MAMA)  SAA Model 32 (COS FUV/XDL) 
The 'flip' side of this would be the times when the telescope is in the SAAimpacted orbits. The table below gives the times when the telescope is projected to first enter the SAA and then the time which it last exits the SAA.



SAA Model 02 (FGS guiding)  SAA Model 05 (FGS exposures)  SAA Model 23 (NICMOS) 
SAA Model 24 (STIS/CCD)  SAA Model 25 (STIS/MAMA)  
SAA Model 27 (ACS/SBC)  SAA Model 28 (ACS/CCD)  
SAA Model 29 (WFC3/UVIS)  SAA Model 30 (WFC3/IR)  
SAA Model 31 (COS NUV/MAMA)  SAA Model 32 (COS FUV/XDL) 
Most Restrictive 
Model 26 
Model 04 + WFC3 + ACS 
Model 07 
NICMOS + FGS Science 
Model 03 + Model 06 + COS + STIS 
FGS Guiding 
Model 13 
Least Restrictive 
This data should be understood in the proper context. Changes to the ephemeris occur, and estimates of the SAA crossings become increasingly uncertain the further into the future the projection is made. Typically, the uncertainty is a function of the error in the nodal regression rate, with added uncertainty due to HST's intrack position. These two uncertainties can be considered in their effect independently (although, in fact, they are not and they are both manifestations of precession error):
The intrack uncertainty can directly be thought of as influenced by the variable rate of atmospheric drag on the telescope. The graph below shows the relationship between constant amounts of error in the dragrate, intrack error, and the duration between when the ephemeris was generated and when the projection is to be made for.
Note the loglog plot above expresses something fundamental about the intrack uncertainty: intrack error grows quadratically in time. If the uncertainty is 'x' at some time 'T', then at time '2*T', the uncertainty will be 4 times larger.
The HST extrapolated orbit file contains (currently) no altitude decay rate, so the intrack error for the times corresponds directly with the contours in the plot. For the 10week orbit file, the events do have a decay rate applied, so the contours in the plot would correspond to the error in the decay rate used to fit and produce the orbit file.
The decay rate is variable and not predictable beyond approximate levels. The decay rate is a function of the ballistic profile of HST (which changes each orbit) and the level of solar activity and its history. Generally, decay rates are highest during solarmaximum, and least during solar minimum. The decay rate uncertainty is a function of the projection time interval. In the plot below, the altitude decay rate history of HST is plotted from launch to the current timeframe with 56day and 1year rolling averages. To get an estimate of the solar activity level in some future year, subtract the approximate duration of the solar cycle (11.4 years) and compare with the plot. However, each solar cycle is different in intensity and duration (and even period), so such projections are only approximate.
Resolution:
Go to the 1st graph and note that a decay rate of 1.0 km/yr (0.5 nominal + 0.5 uncertainty) over a ~180 day time interval corresponds to ~ 1314 minute intrack error. The nominal decay rate of 0.5 km/yr would yield a shift of just over 6 minutes (so the events would be about 6 minutes earlier than those listed in the events tables), with the additonal 0.5 km/yr yielding an additional 7 minutes added error if the decay rate were higher than expected.
Resolution:
The first graph shows that event times at T120 days are expected to be earlier than shown in the tables (for the nominal decay rate) by about 25 minutes.
Which orbit and where in the orbit HST is are too uncertain at a range of 120 days. For the maximum decay rate variation, only at T60 days would one be able to guess within 30 minutes, and at the nominal decay rate, T90 days would be the horizon to get within about 30 minutes.
ijej 070220
ijej 131231 update.
ijej 140321 update.
ijej 150812 update to rate graph.