Altitude gain and loss no where near the same.

If you look at the elevation graph, does it actually show the start and the finish at the same altitude? The difference could be due to weather change. Although I agree that 150 meters is a fairly massive difference to be explained by a weather change. 

In the time you took to complete the climb, was there a significant change in weather conditions?

If your Watch Mode is set to Auto, try setting it to Altimeter and see if that makes a difference. 

Long press Menu | Sensor & Accessories | Barometer | Sensor Mode: Auto/Altimeter Only/Barometer Only

What was the elapsed time for this activity ?.

How did the weather conditions change during the ascent ? More specifically did the pressure change over and above the change you would expect from altitude gain. 

What mode do you have the barometer in ?

There can be several reasons: 

If you have auto-calibration enabled, during climbs you can get faulty altitude data. For example if the climb is steep, and the GPS lock is off, auto calibration will be to wrong altitude. This can accumulate. (I have experienced this several times.)

A shift in weather is an unlikely explanation. 200 meters in altitude is equal to some 25 mbar in barometric change. 1 mbar per hour (equals 8 meters per hour) indicates a storm warning and is a rapid shift. 

You may have had strong winds directly in to the sensor. This may throw off the ascent. (I notice this especially when climbing/mountain running and suddenly meeting strong winds; say 15+ m/s). 

Personally I never use auto calibration during my sessions, especially in alpine/mountain settings. I always calibrate at start to a known altitude and then use altitude setting. I did a 3,5 hour session yesterday with some 530 meters ascent and descent. When I was back at the starting point, the altitude was 12 meters lower, equalling an increase of 1.5 mbar in barometric pressure during the period. (Which was verified by the barometer at home).

With the exception of strong head winds into the sensor port, I find the sensors quite accurate. But you must be prepared to understand how barometric shifts impacts the readings  (and also temperatures), as well as using manual settings and calibration, perhaps even re-calibration to known altitudes during longer sessions.

n.y.a.n said:

Personally I never use auto calibration during my sessions, especially in alpine/mountain settings. I always calibrate at start to a known altitude and then use altitude setting.

Same. I've tried every combination of settings and techniques many times, and have always found this to be the best way to do it. I always get good results this way.

Even if you don't know the altitude at the start, just guesstimating it (or letting the watch estimate it using DEM or GPS) is usually good enough. Even if it's not completely correct, your high and low elevation might just be shifted one way or the other, but your ascent and descent and overall elevation plot curve will be pretty accurate.

Auto altimeter calibration can be useful for some scenarios, but I've found that 99% of the time I get much better data when I l turn it off.

Back when you experimented with various settings, did you find any difference in elevation gain between Smart and 1 Second recording? According to a reply from Garmin Support in Taiwan, One Second recording should log more gains but I have not seen that to be the case. After reading a long post from a site where the author did exhaustive One Second vs Smart recording tests, he concluded that Smart is just as accurate as One Second. I've since changed to Smart in favour of smaller files and short sync time.

Yes the start and finish was at the same location. It was only a short trip 7hrs.
I think the problem may be by using auto calibration and taking a reading every second (which presumably exaggerated the errors accumulatively).

Thanks for all your excellent feedback.
I have concluded that recording every second and using autocalibration and auto altimeter/barometer settings may have contributed to the large error in ascent and descent starting from and finishing at the same location.
my start altitude read 2819m and finish altitude at 2859m which does agree with the ascent recorded at 1050m and descent at 855m (a 195m discrepancy).

I have now adjusted my settings to manual calibration and using smart recording intervals. I am not sure why this system is outdated as stated in one reply though.

My errors between start and finish height gain with are now around 20m to 50m in 1100m within an 8 hour window.

I think that this shows that we need to be careful on how we set up our fenix 7's!

Maybe something for Garmin for feedback on this device.

tralphs said:

I have concluded that recording every second and using autocalibration and auto altimeter/barometer settings

By auto altimeter/barometer settings, do you mean not calibrating before activity at all? I usually just calibrate the altimeter using DEM before my trail runs or ride. We have no known altitudes to use for manual calibration.