Reports

Contents

Title: GT Water Level Comparison and Correction
Date:2021-09-09 - 2021-10-11
Data File: GT_WaterLevelCompare_20211011.csv
Refers to:GT,20010026

A strong diurnal fluctuation in the water level signal on the order of 0.25 psi (18cm) was clearly evident before the breech on Oct 4th.

Figure 1. Raw water level in psi, 24hr (48 point) running mean, and "Noise" left over by subtracting the running mean from the raw signal.

The pressure sensor measures water temperature too but being about 1m below the ground surface there is very little fluctuation is this temperature.  The pressure sensor is vented through a desiccant tube in the data logger box, so I looked at air pressure and box temperature (PanelT).  There seems to be no relationship with air pressure which is good but there seemed to be a slight relationship with box temperature - maybe the box getting hot is some how pressurizing the desiccant tube?  The sensor is a digital SDI-12 sensor so there should not be any kind of analog noise.  But then there are two signals in the water pressure signal: the slow multi-day change in water level and this diurnal noise.  So I removed a 24 hour running mean from the water pressure which just left the diurnal fluctuation that had a pretty good relationship with box temperature

Regression Data

Residuals

Figure 2. Regression of Noise in the pressure sensor against the data logger panel temperature.  This is only for the period before the 10/04 breech.

I then used this relationship to calculate a "noise" signal based on the box temperature and removed this from the raw pressure signal.  This did a good job of flattening out the pre-breech data but still reflected the post breech tides.  And it seems to compare well with the USGS water level data from their EXO sensor near the breech.

Figure 3. Corrected Tower water level compared to the USGS EXO water level.  The USGS EXO pressure has been offset by -14.86psi here to account for differences mounting height with respect to sea level.

However, our Tower sensor doesn't seem to be catching the low tide water levels.  This isn't surprising because our sensor is in a well which is likely retaining water as the tide goes out. 

So, all is well?  Maybe.  Correcting the water level pressure with the PanelT is not a great fix.  Buying a new pressure sensor to compare is a good idea as well as moving/adding a pressure sensor in a deeper part of the channel instead of in the well.