I ground truth'd 3 GPS' today:

1. AMOD photo tracker (20 channel Sirf III with WAAS)
2. Garmin etrex 30 ( ? channel, GPS/WAAS + GLONASS ).
GPS was "D" at time of check.
3. iPhone 4.

etrex 30: 6.58 m
Photo tracker: 4.85 m
iphone 4: 3.53 m

So, for once, the iPhone beat "real" GPS's - though the amount was trivial.

In other areas (esp. in the woods) the etrex and photo tracker were 10 -
15 metres apart on some trail segments. (No ground truths to test
against), I didn't record for the iPhone on those segments.

No conclusion other than you can never trust GPS in the woods and

"A man with 3 GPS' doesn't know where he is."

--
"I see!" said the blind man as he picked up his hammer and saw.

On 12/11/2011 12:16 PM, Alan Browne wrote:
>
> I ground truth'd 3 GPS' today:
>
> 1. AMOD photo tracker (20 channel Sirf III with WAAS)
> 2. Garmin etrex 30 ( ? channel, GPS/WAAS + GLONASS ).
> GPS was "D" at time of check.
> 3. iPhone 4.
>
> etrex 30: 6.58 m
> Photo tracker: 4.85 m
> iphone 4: 3.53 m
>
> So, for once, the iPhone beat "real" GPS's - though the amount was trivial.
>
> In other areas (esp. in the woods) the etrex and photo tracker were 10 -
> 15 metres apart on some trail segments. (No ground truths to test
> against), I didn't record for the iPhone on those segments.
>
> No conclusion other than you can never trust GPS in the woods and
>
> "A man with 3 GPS' doesn't know where he is."
>
I assume this was without averaging.

I took the GPSMAP60 out to a reference point and got it to 4ft accuracy.

The emap was dear in canyons with heavy canopy. The helix on the
GPSMAP60 works under the same conditions, but the tests were don't at
different times. I think the quad helical is better than the patch if
you don't mind the size.

On 2011-12-11 18:03 , miso wrote:
> On 12/11/2011 12:16 PM, Alan Browne wrote:
>>
>> I ground truth'd 3 GPS' today:
>>
>> 1. AMOD photo tracker (20 channel Sirf III with WAAS)
>> 2. Garmin etrex 30 ( ? channel, GPS/WAAS + GLONASS ).
>> GPS was "D" at time of check.
>> 3. iPhone 4.
>>
>> etrex 30: 6.58 m
>> Photo tracker: 4.85 m
>> iphone 4: 3.53 m
>>
>> So, for once, the iPhone beat "real" GPS's - though the amount was
>> trivial.
>>
>> In other areas (esp. in the woods) the etrex and photo tracker were 10 -
>> 15 metres apart on some trail segments. (No ground truths to test
>> against), I didn't record for the iPhone on those segments.
>>
>> No conclusion other than you can never trust GPS in the woods and
>>
>> "A man with 3 GPS' doesn't know where he is."
>>
> I assume this was without averaging.

Not invoked. But from the Garmin's 5 second interval sampling there is
some filtering such as a moving average and minimum distance traveled.
I wish it was unfiltered so I could see what was going on - as I can
with the photo tracker.

>
> I took the GPSMAP60 out to a reference point and got it to 4ft accuracy.
>
> The emap was dear in canyons with heavy canopy. The helix on the
> GPSMAP60 works under the same conditions, but the tests were don't at
> different times. I think the quad helical is better than the patch if
> you don't mind the size.

In deep canyons or heavy forest I don't trust the horizontal with
consumer handhelds for any great accuracy - esp. if there's a likelihood
of multipath. Some years ago (late 90's) I had brief access to a
differential kinematic (not RT) system (recorder in backpack + recorder
at arbitrary reference). I didn't do any ground truths with it, but the
plot through the woods (post processing) was very clean and walk backs
were sub metre from the walk in. Wish I had that now.

--
"I see!" said the blind man as he picked up his hammer and saw.

On 12/11/11 2:16 PM, Alan Browne wrote:
>
> I ground truth'd 3 GPS' today:
>
> 1. AMOD photo tracker (20 channel Sirf III with WAAS)
> 2. Garmin etrex 30 ( ? channel, GPS/WAAS + GLONASS ).
> GPS was "D" at time of check.
> 3. iPhone 4.
>
> etrex 30: 6.58 m
> Photo tracker: 4.85 m
> iphone 4: 3.53 m
>
> So, for once, the iPhone beat "real" GPS's - though the amount was trivial.
>
> In other areas (esp. in the woods) the etrex and photo tracker were 10 -
> 15 metres apart on some trail segments. (No ground truths to test
> against), I didn't record for the iPhone on those segments.
>
> No conclusion other than you can never trust GPS in the woods and
>
> "A man with 3 GPS' doesn't know where he is."
>

I made the mistake of having two watches! :-o

On 12/11/2011 3:15 PM, Alan Browne wrote:
> On 2011-12-11 18:03 , miso wrote:
>> On 12/11/2011 12:16 PM, Alan Browne wrote:
>>>
>>> I ground truth'd 3 GPS' today:
>>>
>>> 1. AMOD photo tracker (20 channel Sirf III with WAAS)
>>> 2. Garmin etrex 30 ( ? channel, GPS/WAAS + GLONASS ).
>>> GPS was "D" at time of check.
>>> 3. iPhone 4.
>>>
>>> etrex 30: 6.58 m
>>> Photo tracker: 4.85 m
>>> iphone 4: 3.53 m
>>>
>>> So, for once, the iPhone beat "real" GPS's - though the amount was
>>> trivial.
>>>
>>> In other areas (esp. in the woods) the etrex and photo tracker were 10 -
>>> 15 metres apart on some trail segments. (No ground truths to test
>>> against), I didn't record for the iPhone on those segments.
>>>
>>> No conclusion other than you can never trust GPS in the woods and
>>>
>>> "A man with 3 GPS' doesn't know where he is."
>>>
>> I assume this was without averaging.
>
> Not invoked. But from the Garmin's 5 second interval sampling there is
> some filtering such as a moving average and minimum distance traveled. I
> wish it was unfiltered so I could see what was going on - as I can with
> the photo tracker.
>
>>
>> I took the GPSMAP60 out to a reference point and got it to 4ft accuracy.
>>
>> The emap was dear in canyons with heavy canopy. The helix on the
>> GPSMAP60 works under the same conditions, but the tests were don't at
>> different times. I think the quad helical is better than the patch if
>> you don't mind the size.
>
> In deep canyons or heavy forest I don't trust the horizontal with
> consumer handhelds for any great accuracy - esp. if there's a likelihood
> of multipath. Some years ago (late 90's) I had brief access to a
> differential kinematic (not RT) system (recorder in backpack + recorder
> at arbitrary reference). I didn't do any ground truths with it, but the
> plot through the woods (post processing) was very clean and walk backs
> were sub metre from the walk in. Wish I had that now.
>
Make that the emap was deaf. Fat finger typo!

The reference spot I use is in a canyon, but no canopy. I'll have to
check how many birds it got.

This is a blast from the past:
> http://www.gpsinformation.org/joe/gpsantennaspecs.htm

Emap uses a patch. GPSMAP60 a helix. But different chipsets.

On 11 Dic, 21:16, Alan Browne <alan.bro...@FreelunchVideotron.ca>
wrote:
> I ground truth'd 3 GPS' today:
>
> 1. AMOD photo tracker (20 channel Sirf III with WAAS)
> 2. Garmin etrex 30 ( ? channel, GPS/WAAS + GLONASS ).
> * * GPS was "D" at time of check.
> 3. iPhone 4.
>
> * * * etrex 30: 6.58 m
> Photo tracker: 4.85 m
> * * * iphone 4: 3.53 m

Interesting.
How you did this?
Placed at know position I suppose.
Did you recorded the istantaneous position or recorded for long time?

> So, for once, the iPhone beat "real" GPS's - though the amount was trivial.

Test execution may introduce some bias in the result.
For example if the distances are measured shortly after TTFF, the
iPhone4 should be greatly advantaged by A-GNSS

> In other areas (esp. in the woods) the etrex and photo tracker were 10 -
> 15 metres apart on some trail segments. *(No ground truths to test
> against), I didn't record for the iPhone on those segments.

This can happen also on the road

On 2011-12-11 22:51 , miso wrote:
> On 12/11/2011 3:15 PM, Alan Browne wrote:
>> On 2011-12-11 18:03 , miso wrote:
>>> On 12/11/2011 12:16 PM, Alan Browne wrote:
>>>>
>>>> I ground truth'd 3 GPS' today:
>>>>
>>>> 1. AMOD photo tracker (20 channel Sirf III with WAAS)
>>>> 2. Garmin etrex 30 ( ? channel, GPS/WAAS + GLONASS ).
>>>> GPS was "D" at time of check.
>>>> 3. iPhone 4.
>>>>
>>>> etrex 30: 6.58 m
>>>> Photo tracker: 4.85 m
>>>> iphone 4: 3.53 m
>>>>
>>>> So, for once, the iPhone beat "real" GPS's - though the amount was
>>>> trivial.
>>>>
>>>> In other areas (esp. in the woods) the etrex and photo tracker were
>>>> 10 -
>>>> 15 metres apart on some trail segments. (No ground truths to test
>>>> against), I didn't record for the iPhone on those segments.
>>>>
>>>> No conclusion other than you can never trust GPS in the woods and
>>>>
>>>> "A man with 3 GPS' doesn't know where he is."
>>>>
>>> I assume this was without averaging.
>>
>> Not invoked. But from the Garmin's 5 second interval sampling there is
>> some filtering such as a moving average and minimum distance traveled. I
>> wish it was unfiltered so I could see what was going on - as I can with
>> the photo tracker.
>>
>>>
>>> I took the GPSMAP60 out to a reference point and got it to 4ft accuracy.
>>>
>>> The emap was dear in canyons with heavy canopy. The helix on the
>>> GPSMAP60 works under the same conditions, but the tests were don't at
>>> different times. I think the quad helical is better than the patch if
>>> you don't mind the size.
>>
>> In deep canyons or heavy forest I don't trust the horizontal with
>> consumer handhelds for any great accuracy - esp. if there's a likelihood
>> of multipath. Some years ago (late 90's) I had brief access to a
>> differential kinematic (not RT) system (recorder in backpack + recorder
>> at arbitrary reference). I didn't do any ground truths with it, but the
>> plot through the woods (post processing) was very clean and walk backs
>> were sub metre from the walk in. Wish I had that now.
>>
> Make that the emap was deaf. Fat finger typo!
>
> The reference spot I use is in a canyon, but no canopy. I'll have to
> check how many birds it got.
>
> This is a blast from the past:
>> http://www.gpsinformation.org/joe/gpsantennaspecs.htm
>
> Emap uses a patch. GPSMAP60 a helix. But different chipsets.

That's fine for units with external antennas. A choke ring antenna may
be better in canyons (I'm not sure). (And the choke ring can likely
support helical or patch antennas). But they're big mothers.




--
"I see!" said the blind man as he picked up his hammer and saw.

On 2011-12-12 04:36 , claudegps wrote:
> On 11 Dic, 21:16, Alan Browne<alan.bro...@FreelunchVideotron.ca>
> wrote:
>> I ground truth'd 3 GPS' today:
>>
>> 1. AMOD photo tracker (20 channel Sirf III with WAAS)
>> 2. Garmin etrex 30 ( ? channel, GPS/WAAS + GLONASS ).
>> GPS was "D" at time of check.
>> 3. iPhone 4.
>>
>> etrex 30: 6.58 m
>> Photo tracker: 4.85 m
>> iphone 4: 3.53 m
>
> Interesting.
> How you did this?
> Placed at know position I suppose.

Knownish. Google Earth in an area where the registration seems to be on
the better than 1 m level. (I know places where there's a distinct 15
meter west registration error).

> Did you recorded the istantaneous position or recorded for long time?

Hard to tell. The only one of the three that has continuous
unfiltered/unaveraged positions is the phototracker. The etrex was
sampling at 1 per 5s. But seems to also have a minimum distance
traveled filter going on (even though that option was not selected - I
used a period of 5 seconds).

The result is I get a scatter plot (random walk) for the photo tracker
(over 1 or 2 minutes) and pinned positions for the iPhone and etrex.
Really wish they would free run.

The iPhone also seems to be filtered, but Apple keep the internals deep
beyond API's that could get more interesting data.

>
>> So, for once, the iPhone beat "real" GPS's - though the amount was trivial.
>
> Test execution may introduce some bias in the result.
> For example if the distances are measured shortly after TTFF, the
> iPhone4 should be greatly advantaged by A-GNSS

I repeat this test about every 2 weeks. Usually the iPhone is off by
10m or more on any given test at any open sky location. It just hit the
spot yesterday. Chance.

>> In other areas (esp. in the woods) the etrex and photo tracker were 10 -
>> 15 metres apart on some trail segments. (No ground truths to test
>> against), I didn't record for the iPhone on those segments.
>
> This can happen also on the road

Given masking (buildings or the vehicle) yes indeed. Some car GPS's
(such as TomTom) correlate to the street/road they think they're on, so
occasionally the display will show you on an off ramp or service road.
This sometimes leads to re-routing instructions. You really need to
think for yourself. Though in a strange city it can be harder.




--
"I see!" said the blind man as he picked up his hammer and saw.

Alan Browne <(E-Mail Removed)> writes:

> Knownish. Google Earth in an area where the registration seems to be
> on the better than 1 m level. (I know places where there's a distinct
> 15 meter west registration error).

I'm pretty suspicious of claiming that's right closer than a few meters.
But, tracks from an Oregon match up well with Bing imagery when editing
for openstreetmap.

Another source of ground truth that seems far more trustworthy is, in
the US, NGS passive control points, which have published coordinates.
Beware the horizontal accuracy of vertical controls can be lacking.

I am not clear on the situation in Canada, but NGS seems to do joint
adjustments with their Canadian counterparts, so surely it's similar
(but maybe not as accessible to randoms).

Then, figure out what datum you get when you are using WAAS. I believe
it's no longer WGS84, but an ITRF variant which is (with navigation
solution GPS) indistinguishable. I dimly remember that with
differential with Coast Guard beacons one got NAD83 instead.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (NetBSD)

iEYEARECAAYFAk7mbjYACgkQH9p66AmO1g7xSQCdFcjWCHL9Hv hUHgQ8W4qYOBxh
iQEAnjX25Gp+XXYs78l87OBAnj8hDxcH
=zqI3
-----END PGP SIGNATURE-----

On 2011-12-12 16:12 , Greg Troxel wrote:
>
> Alan Browne<(E-Mail Removed)> writes:
>
>> Knownish. Google Earth in an area where the registration seems to be
>> on the better than 1 m level. (I know places where there's a distinct
>> 15 meter west registration error).
>
> I'm pretty suspicious of claiming that's right closer than a few meters.
> But, tracks from an Oregon match up well with Bing imagery when editing
> for openstreetmap.

If one argued it was 2 metres at best I wouldn't fight too hard.

> Another source of ground truth that seems far more trustworthy is, in
> the US, NGS passive control points, which have published coordinates.
> Beware the horizontal accuracy of vertical controls can be lacking.

The Canadian NRC site has conversion tools online.

> I am not clear on the situation in Canada, but NGS seems to do joint
> adjustments with their Canadian counterparts, so surely it's similar
> (but maybe not as accessible to randoms).

Google Earth uses WGS-84 uniformly (AFAIK). So I reference to that.
Survey marks I use are converted to WGS-84 using online tools.

> Then, figure out what datum you get when you are using WAAS. I believe
> it's no longer WGS84, but an ITRF variant which is (with navigation
> solution GPS) indistinguishable. I dimly remember that with
> differential with Coast Guard beacons one got NAD83 instead.

The FAA settled on NAD-83. It remains unclear to me if WAAS is WGS-84
or NAD-83 even though WAAS is for aviation, esp. in the US.

However, if you read this recent memo (LAAS) then ITFR00/WGS-84 =or=
NAD-83/NAVD-88 can be used as the reference datum. Which is in use must
form part of the certification document.
http://www.nsma.org/recommendation/WG3-92-036.pdf
But read the recommendations for more entertainment.

Then this one:
http://www.faa.gov/about/office_org/...etic_Datum.pdf

Specifies both NAD-83 and ITRF00 (WGS-84) for GBAS (LAAS) and specifies
that (naturally) all references need to be in the same datum. (no
brainers need memos too).

Good points. I'm going to review the "truth" of the "truths" I use.

--
"I see!" said the blind man as he picked up his hammer and saw.