[Top] [All Lists]

[TowerTalk] Combining ant elevation pattern data with propagationpredict

To: <>
Subject: [TowerTalk] Combining ant elevation pattern data with propagationpredictionsto produce a contest bandplan
From: (Jim Lux)
Date: Thu May 22 10:10:50 2003
> 1.  The models used for predictions contain statistical data, thus there
> is no possibility of guaranteeing that a particular path's predicted
> characteristics will be accurate or even that the path will exist at all.

> I understand and accept this.  However, it's kind of like reading
> semaphore signals in the fog.  If the fog is too thick, you won't see
> the signaller.  If it's somewhat thick you may be able to see him but
> not make out 100% of what he is sending.  No fog, no problem.  BUT if
> you're facing the wrong way you will never see the signaller, fog or no
> fog.  I'm looking to the predictions to tell me which way to face.

Exactly so.. just so you recognize the limitations of your modeling (which,
by the way, I too am interested in... looking forward to HFTA, which will do
a lot of this)

> 2. The models are based on smoothed numbers.  The daily SFI/sunspot
> numbers you get are not smoothed so aren't the right values to use.
> On a 6 hourly basis you can get Effective SSN based on real time FoF2
> observations from
> The USAF provides forecasts about a month ahead of K and SFI values.
> Presumably these forecasts improve on the model, else why would they
> bother.  I think the forecast is for a longer period than the 27 day
> rotation period of the sun.
> IPS in Australia provides Hourly Area Prediction charts centred on any
> location you choose.  It shows what I presume to be Optimum Working
> Frequency (it tells you on their website) to any point on the chart.
> Why are they doing this if it doesn't mean anything on a particular day?
> What I'm trying to say here is that it looks to me as if people are
> doing meaningful path predictions (and others are buying them).

IPS uses a different modeling approach more suited to short time range
predictions.  Not that the IONCAP models won't work, it's just that they
don't have the statistical validity that models more suited to hour-to-hour
predicts might have.  Think of it like predicting the stock market
behavior... If your interest is in behavior over 20 years, a month or year
at a time, you'd use a different technique than if you were interested in
day trading.  The day trading model might work real well in the short run,
but have large divergences in the long run.

Compare also, for example, climatology predictions and weather predictions.
IONCAP is more like a climatological program (We predict that we'll get 7.5
inches of rain this year, but we don't really know on what days)... Short
run forecasting is more like weather prediction (it will be 95 degrees
tomorrow afternoon).

Different models, different analysis, different validations.

> 5.  Just because a path is open doesn't mean that there are Qs to be
> made because a) there's no one there  b) they all have their beams
> pointing sideways to you working a different opening
> a)  I didn't mention it because I was trying to keep things simple.  The
> FOM includes the number of stations reachable via that path who entered
> the contest last year.
> b)  Sad but true.  I guess experience is the only teacher here.  Say, I
> could also plot paths for each location I'm trying to reach to see what
> they're likely to be working.... oops, sorry, got carried away there.

Actually, that IS a clever idea... As you mention, the "other end of the
link" might have a gain antenna, pointed elsewhere.  More practically, the
band they'll be on is the band that's "hot" for them.  However, bear in mind
a path that's good from you to them, is probably also good from them to you,
and that's where they'll be. (hope springs eternal).

Jim, W6RMK

<Prev in Thread] Current Thread [Next in Thread>