search

GEUS-Glaciology-and-Climate / GC-Net-level-1-data-processing

This repository contains scripts used to flag, adjust and interpolate GC-Net data
GNU General Public License v3.0
7 stars 3 forks source link

Swiss camp + JAR2: Humidity sensor swap in 2005? #110

Closed maikenrevheim closed 1 year ago

maikenrevheim commented 1 year ago

Hi :)

it looks like the humidity sensor 1 and 2 at Swiss camp and JAR2 got swapped in 2005:

RH_SC_issue

The large increase in RH over time at Swiss camp is also suspicious

RH_JAR2_issue

Maiken

maikenrevheim commented 1 year ago

Looks like the same thing happened at NASA-U in 2016:

RH_NASA-U_issue

Do you want to fix these issues @BaptisteVandecrux, or should I try? I have not changed anything in the repository in a while, so I am a bit afraid of messing something up.

Maiken

jasonebox commented 1 year ago

Hi Baptiste. Can the fix be made sooner than later so Maiken can use these two sites/years in her thesis?

jasonebox commented 1 year ago

I think specific humidity would show the swap better, but would have to assume or test whether the temperature wiring were also swapped.

BaptisteVandecrux commented 1 year ago

-[] NASA-U: confirmed. image

BaptisteVandecrux commented 1 year ago

After swap RH1 is now generally higher than RH2. image image image

jasonebox commented 1 year ago

NASA-U the humidity difference looks suspicious after 2018, see green line.

BaptisteVandecrux commented 1 year ago

I see these jumps in upper/lower RH difference quite regularly (see #111 ). It could be one of the following reason: 1) the spacing changed leading to a greater difference between RH low and upper 2) existing hygrometers de-calibrated (drift) 3) one or two new hygrometers were installed and introduced a bias

At least I don't see it as a sensor swap issue.

BaptisteVandecrux commented 1 year ago

The latest commits should have solved most of the sensor swaps. So I am closing this issue.

To spot the swapped instruments I assumed winter atmospheric inversion at all sites and therefore: RH1 higher than RH2 (cold air has a higher relative humidity) and TA1/3 lower than TA2/4 during that time.

For some sites, important shifts appear in the RH1 - RH2 curves. This could be a change in the spacing, new instruments or a drift from the existing sensor. I cannot know at this point even though it is crucial for methods using both levels.

I am hoping that in the future, we could cross-calibrate the RH sensors on the two levels by identifying periods where the two levels should have the same RH value (e.g. cloudy windy days?). By adjusting one sensor to the other, we would make sure that the difference between levels are due to humidity gradients and not to instrument miscalibration.