Reports

Contents

Title: Tonzi Floor well sensor comparison
Date:2018-03-27
Data File: WellCompare_201802.csv
Refers to:TZ_fmet,57278,1260403156

From 2018-02-07 to 2018-03-27 the Old Floor well sensor was compared with a New sensor. 

The Old sensor is a Global Water WL16U sn 57278.  It has a 15ft range and outputs water level in feet and battery voltage.  This sensor was making one measurement at 20 and 50 minutes after the hour.

The New sensor is a METER CTD-10 sn 1260403156.  It has a 10m range and outputs water level (mm), temperature (C) and electrical conductivity (uS/cm).  This sensor was logged with a CR200 taking 10sec samples and saving 30min averages at the end of the half hour.

Figure 1. Initial time series comparison of the Old sensor with the new Sensor.  The Old sensor data has been converted from ft to mm: ft * 12 * 25.4 = ft * 304.8 = mm.

After converting the Old sensor to mm, a linear regression between the two shows the offset is 1681mm.  Since water level doesn't change very fast, the time differences between the measurements (10min) and methods (instantaneous vs 30min average) should not make much of a difference.

Regression Data

Residuals

Figure 2. Linear regression between Old sensor and New Sensor: New = Old * 0.9983 + 168.1, r2 = 1.0, n = 2297.  The offset here means the Old sensor is 168.1cm higher than the New sensor.

Additionally, manual measurements were made measuring from the top of the well pipe to the water level.  There were three manual measurements during the period of inter-comparison.  The raw manual measurements are in decimal feet from the top of the well pipe.  ft * 12 * -2.54 = ft * -30.48 = cm below the top of the pipe.

Regression Data

Residuals

Figure 3. Linear regression between the New sensor and Manual measurements.  The New sensor is reporting mm or water and the Manual measurements are cm below the top of the pipe as negative.  The offset here indicates the New sensor is 10.94m below the top of the pipe.  The slope here (m = 0.9765) is important in getting the two water levels derived from the pressure sensors to match the manual measurement.  However I am unsure why this difference appears.

With all the correct offsets and unit conversions applies the time series of the three measurements agree well.

Figure 4. Time series of all three measurements with correct offsets and unit conversions: New sensor *0.1*0.9765 - 1094, 0.9765 * (Old sensor*30.48+168.1) -1094, Manual * -30.48.  The Old sensor and New sensor are so similar that the differences are not visible unless the graph is zoomed in.

The New sensor also records electrical conductivity.  This may be interesting to see salinity decreasing as the refreshed in the spring and drawn down in the fall.

Figure 5. Electrical conductivity (salinity) from the New sensor.  Salinity falls as the ground water is recharged.

Conclusions:
The Old sensor is -930.3cm below the top of the pipe.
The New sensor is -1094cm below the top of the pipe.
It is unclear why there is a 2.35% difference between the pressure derived water levels and he manual tape measurement.