Altitude drops every 1-2 km of each run

Jim, the temperature sensor should not change the altitude reading. If anything, the temperature sensor should be used to correct for temperature dependent hardware errors, so the altitude reading is kept stable at the same value.

But I suspect that this is not working as it should on the F5, based on what I have seen on this forum earlier.

I am definitely not experiencing this and the temperature on pretty much every run has a similar trend as it's almost always colder outside than inside.

One other thing that comes so my mind is that (unless I am mistaken) the watch calibrates altitude in presence of a nearby waypoint with a known altitude. In case there is a waypoint near where OP starts his run but the altitude is incorrect, it can possibly lead to wrong starting altitude.

It is not possible to disable the barometric altimeter.

You could complain to Garmin that your altimeter is temperature sensitive. It shouldn't be. But I don't know if you would get any results out of that.

I am wondering if this is an error in the ambient pressure reading, or a calculation/assumption error in the conversion from ambient pressure to altitude. It is not entirely unthinkable that some developer thought he was clever and thought that he should correct the calculation for air density variations caused by ambient temperature - in which case your example would prove him wrong.

Could I persuade you to make a custom screen in an activity with the following four values:

• Ambient pressure
  
• Barometric pressure
  
• Elevation
  
• Temperature
  

Then report the values you see immediately after going out and the values you see when the temperature has settled.

yeah i'll do that on my next run, thanks!

For now the theory seems to check out. I went for a 25K run early this morning, consisting of a couple of loops of the same course, and as you can see the elevation drops with temp, but because the outside temp rises during my run, elevation rises aswell, even though it shouldn't have!ciq.forums.garmin.com/.../1362916.png

Jim, the temperature sensor should not change the altitude reading.

Temperature is used in the calculation, as in the same spot the ambient pressure will change as the temp does. If the temp changes because the internal temp of the watch is adjusting, you'll see a change in the altitude. That's why the temp needs to have stabilized. Pressure and temp are used for altitude, and when calibrating the initial altitude, it uses GPS for that elevation at that temperature and pressure. You'll see the formula for how temp/pressure/elevation play together on a number of sites on the web. (Weather sites with the calculations)

Jim, I am an engineer, specialized in thermodynamics. I very well know the relationship between pressure, temperature and altitude. I also know why the temperature in the watch has absolutely nothing to do with this relation.

Imagine this:
You are inside a building at ground level at elevation 500 meter. The ambient pressure is 1000 millibar. The temperature in the building is +25 °C. You calibrate the altimeter to 500 meter.
Then you go outside. The elevation is still 500 meter. The pressure is still 1000 millibar. The temperature outside is -15 °C.

If the watch in this situation decides that it must suddenly use another air density for the altitude calculation because it measures another temperature, it will make an error in the altitude calculation. Actually a pretty significant error. So it should not do that.

What you have seen on the web is correct. You are just misinterpreting it. The temperature of the watch is not what should be used for the calculation. The temperature of the immediate surrounding is not what should be used for the calculation.

The temperature which should be used for the calculation is the temperature of the air column between your altitude and sea level. This temperature doesn't suddenly change, just because you go outside the building. Because of that, the temperature in the altimeter calculation should stay constant, no matter what ambient temperature is measured.

AllanOlesen67 barometric altimeter in the VA3 with firmware 5.0 has this behaviour (confirmed by numerous users):
placed at 23°C on the table for an hour it shows stable altitude of -3 m and ambient air pressure 1002 mb. This is at 10 m GPS altitude (3rd floor).
After 15-20 min on my wrist the readings are stable pressure is 1004 mb and altitude -16 m.
Going for a run: the altitude drops 50 m after 15-17 min and stays +/- 3 m for the rest of the flat run (1 hour).

https://connect.garmin.com/modern/activity/2777180652

With firmware 3.50 the behaviour was similar, just the period before stabilising readings has been even longer, 25-30 min and the initial fall in altitude with the rising temperature has been even higher:

https://connect.garmin.com/modern/activity/2649045663

Now, I put the watch into the fridge for 10 min. the pressure is 1002,2 mb, altitude is 4 m.

After 10 min back at the room temp. the readings are: temp 21 degrees C, pressure 1002,7 mb, altitude 1,5 m.

The Garmin SW site says: [h=4]Changes made from version 3.50 to 5.00:[/h]

• Fixed issues with elevation recording
  

Well, not much has been fixed. Do you think it's an error in calculations or a fault in hardware design?

I am definitely not experiencing this and the temperature on pretty much every run has a similar trend as it's almost always colder outside than inside.

One other thing that comes so my mind is that (unless I am mistaken) the watch calibrates altitude in presence of a nearby waypoint with a known altitude. In case there is a waypoint near where OP starts his run but the altitude is incorrect, it can possibly lead to wrong starting altitude.

This is a NEW feature of the Fenix 5 Plus, so definitely not the issue here;

– Added ability to use known elevation data from onboard maps + known GPS coordinates to calibrate altimeter

Btw, maybe i should have added from the start, but i live in the Netherlands, and actually only 2km from the beach, and a little below sealevel

AllanOlesen67 barometric altimeter in the VA3 with firmware 5.0 has this behaviour (confirmed by numerous users):
placed at 23°C on the table for an hour it shows stable altitude of -3 m and ambient air pressure 1002 mb. This is at 10 m GPS altitude (3rd floor).
After 15-20 min on my wrist the readings are stable pressure is 1004 mb and altitude -16 m.
Going for a run: the altitude drops 50 m after 15-17 min and stays +/- 3 m for the rest of the flat run (1 hour).

https://connect.garmin.com/modern/activity/2777180652

With firmware 3.50 the behaviour was similar, just the period before stabilising readings has been even longer, 25-30 min and the initial fall in altitude with the rising temperature has been even higher:

https://connect.garmin.com/modern/activity/2649045663

Now, I put the watch into the fridge for 10 min. the pressure is 1002,2 mb, altitude is 4 m.

After 10 min back at the room temp. the readings are: temp 21 degrees C, pressure 1002,7 mb, altitude 1,5 m.

The Garmin SW site says: [h=4]Changes made from version 3.50 to 5.00:[/h]

• Fixed issues with elevation recording
  

Well, not much has been fixed. Do you think it's an error in calculations or a fault in hardware design?

If the ambient pressure will vary between 1002 and 1004, just because you take the watch off, then something is wrong. It could be both a hardware and a software error. See a more elaborate explanation under item 2 below.

Anyway, I think we need to go back to some basics. Not only to reply to your post, but also because I have received PMs from people who felt a misplaced urge to educate me. So now I will try to explain it so thoroughly that I can point to this post whenever someone again got it wrong.

There are three different mechanics at play when discussing this issue, and it is important to discern between them:
1. The relationship between temperature and true ambient pressure inside/outside a building.
2. The relationship between temperature and true/measured ambient pressure.
3. The relationship between temperature, ambient pressure and calculated altitude and/or calculated pressure at sea level (aka Barometric Pressure).

1. The relationship between temperature and true ambient pressure inside/outside a building.
You can generally expect the ambient pressure inside and outside a building to be the same, even though the temperatures are different. If there is any significant pressure difference, you will not be in doubt when opening a door.

It is true that air expands when you heat it. But this will just cause air to escape through any small opening in the building. It will not cause the building to be pressurized.

Of course, there are special cases where you can get a pressure difference: Forced ventilation, wind load or ventilation openings placed at an altitude which is significantly (tens or hundreds of meter) above or below your elevation in the building. But in all these cases, you will not be in any doubt when you open a door to the outside. Just a few tenths of a millibar will cause a significant change to the force needed to open the door, and it will cause a rather strong gust of air going through the door.

I can actually back this up with an example: Currently, I am doing some informal measurements of ambient pressure in a building where the client has a process which is sensitive to ambient pressure. (The plant is quite new, cost around 500 million Euro, but they did not use the money for hooking an ambient pressure measurement into the DCS. So right now my Fenix 3 is his only source of ambient pressure information...) There is quite strong forced ventilation in some sections of the building, and when entering another section of the building (having the same ambient temperature), I can often feel both the change in force required to open a door and the strong gust of air coming through the open door. However, when comparing the ambient pressures on both sides of the open door, I have never seen a difference higher than 0.3 millibar.

2. The relationship between temperature and true/measured ambient pressure.
Even if the ambient pressure inside and outside the building is the same, we cannot be sure that our pressure measurement device doesn't have a temperature dependent measurement error. In fact, I would consider it quite likely that it has.

So when the pressure sensor is at 1000 millibar true ambient pressure and 25 °C, it may correctly show 1000 millibar. But when the same sensor is at 1000 millibar true ambient pressure and 0 °C, it may incorrectly show 995 millibar. (These figures I have just made up for the sake of an example.)

Now, this change in measured ambient pressure would be equal to an altitude change of more than 40 meter. So that would of course be unacceptable, and we would need to correct it. This will be quite simple if we know the measurement error as a function of temperature, and we know the temperature of the pressure sensor.

This is probably the reason that our watches have the temperature sensor and the pressure sensor built into the same physical component (as stated by someone in another thread): The measured temperature is used for correcting a temperature dependent error in the measurement of ambient pressure.

Such a correction is in my opinion the only correction which should take place. Any other correction would be a misunderstanding. See item 3 below for elaboration.

Coming back to your description of the 1002/1004 millibar ambient pressure depending on whether you carry the watch on your wrist or not:
This is a clear error, and in my opinion it is unacceptable. I have done similar tests of my Fenix 3 where 20-30 °C of temperature change would only cause a change of a few tenths of a millibar to the measured ambient pressure.

However, without any knowledge of the raw, uncorrected signal from the ambient pressure sensor, we can't really know the reason for the error you are seeing. It could be any case of:

• The ambient pressure sensor has a temperature dependent error which is not corrected in software.
  
• The ambient pressure sensor does not have a temperature dependent error, but the software thinks it has and wrongly applies correction (perhaps because someone reused code from an earlier watch model with another sensor type...).
  
• The ambient pressure sensor has a temperature dependent error, and the software is trying to correct it, but is using a wrong correction function.
  

3. The relationship between temperature, ambient pressure and calculated altitude and/or calculated pressure at sea level (aka Barometric Pressure).
This is something else which people often like to point out: When the temperature is changed, the density of air is also changed, and consequently the pressure decrease per meter of altitude will be different.

While they are technically correct in this claim, let us consider what would happen if we actually tried to actively correct our altitude correction based on the temperature in the watch:

We are at an altitude of 1000 meter. The ambient pressure is 920 millibar. The temperature in the watch is 25 °C. We calibrate the altimeter to 1000 meter. When we calibrate the altimeter, what actually happens is that the watch calculates the expected ambient pressure at sea level, based on measured ambient pressure and our manual input of altitude. For this calculation, the watch will use an assumed density of the air column between sea level and our altitude.

Now we go outside. We are still at 1000 meter. The ambient pressure is still 920 millibar. The temperature of the watch has dropped to 0 °C. We look at the altimeter to see our altitude. When the watch calculates the altitude, what actually happens is that the watch compares the measured ambient pressure to the earlier calculated ambient pressure at sea level. The altitude is then calculated, based on the difference between these two pressures and an assumed density of the air column between sea level and our altitude.

It is easy to see that if we suddenly change the assumed density of the air column between sea level and our altitude while we are actually still at the same altitude, and the ambient pressure at out altitude is actually still the same, then we will get an altitude correction error which can be rather huge.

It is also easy to see that neither the density nor the temperature of the air column between sea level and our elevation will change, just because we are walking out the door. The air doesn't know that we are walking out the door.

The logical consequence of this is that the watch should not try to use measured temperature to make assumptions regarding the density of the air column between sea level and our elevation. It can only go wrong.

Nevertheless, it has on several occasions been claimed in these forums that the watch will and should do exactly that. I can only repeat that it would be wrong. I can also add that I have actually tried to replicate such behaviour with my Fenix 3 at an altitude of 600 meter without seeing any signs of it happening. If the claims were correct, I should have been able to see one of two effects of a changed temperature:

• The difference between ambient pressure and barometric pressure remains unchanged while the displayed altitude is changed.
  
• The difference between ambient pressure and barometric pressure is changed while the displayed altitude remains unchanged.
  

The actual result of this experiment was that both the displayed altitude and the difference between ambient pressure and barometric pressure remained unchanged. Just as they should.

I guess that someone will now say: "But we need some temperature for the calculation. Otherwise we will not know the air density." And they will be right. But a standard model for this exists. It is called the ISA model and describes the typical relation between altitude and air density/temperature/pressure. This model is used in barometric altimeters in airplanes, and I would assume that it is also used by Garmin.