Event Based Water Sampling

This repository contains scripts to perform event-based sampling using an ISCO autosampler. Water levels are monitored via a Milone Technologies eTape and a Campbell Scientific datalogger is used to record water levels and trigger sampling. Sampling is not threshold triggered, but is based on deviation from streamflow trends. This provides flexibility to capture events during periods of high and mean flows, and importantly small events during dry or low flow periods. The exact approach was developed empirically using existing data for two small headwater wetlands, but it should be flexible enough to work across varied systems.

Approach

Water levels (WL) are read and recorded hourly along with the hourly change in water level (WL_Δ1). Rolling means of (WL_Δ1) are calculated for 2 (WL_Δ2), 6 (WL_Δ6), and 24-hr (WL_Δ24) periods as well as the hourly change in WL_Δ24 (ΔWL_Δ24). The 24-hr rolling standard deviation of WL_Δ1 is calculated and recorded (WL_SD-24). No sampling will occur within the first 24-hours of deployment to avoid skewed mean and standard deviation values. When the system is rebooted in situ sampling can resume with no pause. An event is triggered when (ΔWL_Δ24) is outside of the 99.5% confidence interval and WL_Δ6 is above a threshold, which can be defined in the constants table in the CR8 program.

Sample if
ΔWL_Δ24 >= WL_Δ24 + 2.58 * WL_SD-24 AND WL_Δ6 > 0.1

When an event is triggered a sample is taken, the hydrograph_limb variable is set from NA to 'rising', and the water level is stored (WL_Event. Samples are taken hourly and when WL_Δ2 is negative the hydrograph_limb is set to 'falling'. Sampling is ended when WL falls to 1.1 times WL_Event and hydrograph_limb is set to NA. During an event the previous change in daily water level (starting with the pre-event value) plus some noise within the standard deviation of daily water level changes is recorded rather than the actual water level changes. This is because within an event the standard deviation can become so large from the event flows that triggering another event less than 24 hours after the original event is essentially impossible.

Interval sampling can be triggered to perform regular time-based sampling if there have not be any storm events. Sampling will cease after 24 samples to avoid sample mixing.

Equipment

• ISCO Autosampler from Teledyne ISCO (this script has been tested using the 3700 Portable Sampler)
• Campbell Scientific Datalogger (this script is written for a CR800)
• Milone Technologies Standard eTape
  • I have tested sampling using the Voltage Divider and 0-5 V Resistance to Voltage output modules. Experience and personal communications with the manufacturer shows that the Voltage Divider is a more robust option that will not be immediately damaged from a shorted circuit or wiring error, however your precision will be reduced as the range of teh eTape is recorded from 2.5-5 V rather than 0-5 V.

Set up

Wiring guide for the CR800 and the two tested output module of eTapes

eTape	Datalogger (Voltage Divider)	Datalogger (0-5V Module)
Vin	5V	12V
Vout	1H	1H
Gnd	Ground	Ground

Wiring guide for the CR800 and the ISCO autosampler

ISCO	CR800
Flow Meter Port Pin A	SDI 12V
Flow Meter Port Pin C	SDI C1

Table Outputs

The CRBasic programs output three tables Diagnostics, ISCO, and Water_Level.
Diagnostics
-batteryVoltage
-loggerTemperature
ISCO
-bottleNumber
-waterLevel
-hydrograph_limb

Water_Level
-bottleNumber
-eTape
-waterLevel
-laggedWaterLevel
-deltaWaterLevel_1hr
-meanDeltaWL_2hr
-meanDeltaWL_6hr
-dailyWaterLevel
-laggedDailyWaterLevel
-deltaDailyWaterLevel
-meanDeltaDailyWL_24hr
-sdDeltaDailyWL_24hr
-samplingEvent
-hydrograph_limb
-startWaterLevel
-hoursAfterSample

To Do

• Set a minimum water level to avoid interval sampling when no flow is present
• Slow rising limbs are missed, not and quite event, but wet-up periods. Perhaps this is fine because it is captured by the interval sampling
• Set a pre-event period value to to fill in the zeroes in the array with the mean of the preevent_interval (would have artifically low sd). Also, first value of delta WL is first measured water level - 0. Could launch and then run 'reboot' after the first reading
• Add example data and figure
• Convert intermediate variables to declaration by Dim rather than Public