Retrospective gradient factors on real dives

Rick Walsh rickmwalsh at gmail.com
Thu Dec 15 15:00:38 PST 2016


On 10 December 2016 at 09:43, Robert Helling <helling at atdotde.de> wrote:

>
> John,
>
> Am 09.12.2016 um 22:51 schrieb John Van Ostrand <john at vanostrand.com>:
>
> On Thu, Dec 8, 2016 at 9:36 AM, Willem Ferguson <willemferguson@
> zoology.up.ac.za> wrote:
>
>>
>> Would it be feasible to calculate retrospective gradient factors on real
>> dive profiles? We sometimes use a VR3 dive computer which does give this
>> information, but it would be very useful to to have it being calculated
>> automatically for each dive.
>>
>
>  Do you mean you want to see the decompression stops for a different set
> of GFs? Or do you want to automatically find the most liberal GFs that fit
> a dive profile?
>
> For the former questoin set up GFs in the Preferences and go back to view
> the dive. For the latter question I iterate through GFs manually to find
> the closest one.
>
>
> I guess what Willem meant was to do something along the lines we do now
> for VPM-B planned dives: There we compute GFhigh/low such that you would
> get a similar decompression profile.
>
> I have not yet answered to Willem’s original mail as I do not really know
> how one would implement this. Here are my thoughts: At each instant of time
> during the dive, we could compute a gradient factor such that the current
> depth would be the ceiling. That part is easy. One could then try to find
>  a line in the gradient factor vs depth plot that best approximates these
> points. The problem is that this makes only sense during the decompression
> phase of the dive, i.e. for the time where it makes sense to consider the
> current depth to be at least roughly the ceiling. During the bottom part of
> the dive as well as during the first part of the ascent this would give you
> far too low gradient factors.
>
> So we would need to find out for which part of the dive to apply this. I
> don’t know what a good criterium would be: Something like the last 25% of
> max depth? Some part where the ascent is slower than the current ascent
> rate? We could take user input (maybe in form of the ruler) or we could
> find that last part of the dive where a linear approximation for the
> gradient factor vs depth plot is best (measured in chi squared or something
> similar).
>
>
> I know I'm a bit late to the conversation here, but we could take a
slightly more conservative approach where the calculated retrospective
high/low gradient factors represent a linear bound rather than best fit.
In other words, find the lowest pair of gradient factors that such that at
no point in the dive is the ceiling exceeded.  In effect it's like looking
at the ceiling plotted on the profile, and increasing the gradient factors
as much as possible without the diver breaking the ceiling.

The process could be:
1) calculate instantaneous gradient factors for each sample (just need the
max for all tissues)
2) Take GF-high as the maximum gradient factor (yes, this assumes that
GF-high occurs upon surfacing, but this is realistic in most cases, and in
any case it should be conservative)
3) Loop through each calculated per-sample gradient factor, and linearly
extrapolate to find the GF-low (at depth of first stop, assuming the GF-low
at max depth preference is not selected) such that the current tissue's
gradient factor sits on the GF-low/GF-high line
4) Take GF-low as the maximum of all the per-sample GF-low extrapolated
values.

Whether it is better to calculate retrospective gradient factors as
best-fit or as a bound is another question, but this method should be
easier.

Cheers,

Rick
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.subsurface-divelog.org/pipermail/subsurface/attachments/20161216/b0b23a06/attachment.html>


More information about the subsurface mailing list