Satgen140  Using and Abusing Keplerian Elements   by GM4IHJ     1st Dec 91
Example 1
As W3IWI recently pointed out. Using NASA/NORAD elements to follow the
very interesting orbital changes of Oscar 13, needs care. NASA compile
these elements starting from launch date , and then update them very
frequently using radar measurements. The problem with this is that
atmospheric drag to quote just one cause of orbit change, is highly
unpredictable and, very spasmodic. If say Ao13 has just experienced
several days of increased drag due to the upper atmosphere being
supplemented by the output of a Solar storm, any attempt to pull the Keps
into line with the radar will probably grossly overcorrect. NASA
compensate for this by taking frequent up dates but this has the result
that if you try to use their every bulletin for plotting AO13, you get an
effect like trying to walk a straight line when you are drunk. Sober
observers are aware that individual NASA bulletins contain a lot of random
fluctuations . So taking a smooth view of AO13 using new keps at least a
month apart. AO13 seems to be slowing it suicide dive. The decrease of
perigee low point altitude (and complementary increase of Apogee high
point altitude)is perhaps bottoming out. The important point over the next
few months will be what happens to Apogee  now that AO13 is suffering some
drag at perigee. If apogee declines markedly, AO13 is still in trouble.
Example 2
Another use to which more and more radio amateurs are putting Keplerian
elements, is Orbit Modelling. Eg is Sara Oscar 23 in a good orbit for
viewing Jupiter? Decrease the mean motion to give the sat some
"theoretical height". Does Jupiter viewing time increase ? Please note
that SARAs orbit is clearly not optimum for its task , because its low
altitude means it is often viewing Jupiter through the Ionosphere. So its
reception of 2 to 15MHz signals is highly variable. Did anyone model this
before SARA was built? This is certainly an area in which Amsat/IARU 
could have helped. Yes , SARA had to take the orbit it was given. But why
restrict its reception to a frequency band which would often be obscured.
Modelling during design would have pin pointed this conflict of purpose.
Example 3
What ever you do with Keplerian elements remember their limitations.NASA
warn ,they are not very accurate.The NASA method of updating individual
units from launch , even when those units join up subsequently, produces
some hilarious results , if you take them at face value.Current Keps for
Kristall/Mir show these units which are firmly docked together as having
different mean motions  ( K= 15.584979613 , M= 15.5854001 ) A non existent
difference of about 0.0004 of an orbit per day = 16 kms approx. So who can
believe a Keps statement to more than 3 places of decimals ? 

Software for orbit modelling, must be flexible, so that mean motion,
station location, orbit perturbation , can be varied whilst the program is
running, permitting print out of changes and their effects. Several
stations report difficulty COMPILING my SPW.bas SARA PLANET WATCHER
software (as transmited file no.3352 on UO3 ). Problems arise because
COMPILED BASICS use reduce COMMAND sets and restrict program variable
names. If you compile my INTERPRETED BASIC software, you destroy its
modelling flexibility because you must DELETE the several instances where
I use commands which are not in the COMPILED command set (EDIT,LIST )to
extract/alter terms whilst the program is running. 73 de GM4IHJ @ GB7SAN