Water quality measurements are made using two YSI 6-series water quality data logging sonde. maintained by the TIDE project at PIE-LTER. YSI Sonde sensors consist of Clark-type potentiometric Oxygen probe, a conductivity/temperature probe (thermistor) .

Data has been screened for obviously wrong data, including a period in the fertilized creek (Sweeney Right) during which the oxygen probe was buried by mud from a collapsed creek bank. Other than this period, conductivity readings did not indicate any blockages that might affect data quality. Oxygen saturation state (the measured variable using this type of sensor, not concentration) had a linear, time-dependent drift that was similar in each creek based on comparison with high precision handheld sondes (YSI ProDO). Oxygen saturation states were detrended, and then resulting saturation states converted to molal concentration using the oxygen solubility function of Garcia and Gordon 1992 (doi: 10.4319/lo.1992.37.6.1307 and 1993 erratum).

Typical sensor/data problems are:

For oxygen: Poor/inaccurate conductivity measurements will affect the dissolved oxygen accuracy, tears, nicks, or other wear on the oxygen selective membrane used on the Clark-type electrode.
For conductivity: Conductivity cell can have detrital material stuck in it, shorting out cell, resulting in lower than expected conductivity.
For depth: Depth strain gage pressure sensor may come out of the water at low tides resulting in many "zero" depth readings or sometimes negative values.

YSI Dissolved oxygen % saturation reporting can vary depending upon how the oxygen is calibrated.

For this data Oxygen percent saturation values is reported as DO%YSI at 1 ATM pressure relative to water-saturated air calibration, and additionally as DO%recalc, after detrending. Resulting metabolic fluxes are calculated from resulting oxygen concentrations relative to the saturation concentration at local atmospheric pressure (which in this case varies by less than 2% and can be neglected relative to much larger errors in advective and air-water transfer rates.

Water depth is logged in each creek using a HOBO U20 titanium water depth loggers rated to 30 ft, and combined with 2011 creek geometry transects by Will Kearney (Sergio Fagherazzi's group at BU) in order to calculate cross-sectional areas as well as surface areas through which air-water gas exchange occurs, and tidal exchange of creek water between timesteps. Windspeed from the cited additional database (the Marshview meteorology tower) is scaled to u10 using the wind profile power law approximation. Wind and estimated current velocities are used with a windspeed parameterization to calculate the air-water exchange coefficient of oxygen (kO2) after scaling the Schmidt number of oxygen of temperature and salinity.

Average values used to calculate metabolic fluxes are reported for the mean time period between the timestep at which they are reported and the timestep 10 minutes prior.

The reported values of kO2 are based on the parameterizations of:

kO2_1 Borges et al. 2004 with both wind and currents (doi: 10.1007/BF02907647)
kO2_2Borges et al. 2004 without wind
kO2_3 Nidzieko et al. 2014 with both wind and currents (doi: 10.1007/s12237-013-9765-2)
kO2_4 Nidzieko et al. 2014 without wind

Net ecosystem metabolism rates (oxygen production minus community respiration in mmol O2 m^-n min^-1) are reported as volumetric, areal, and per unit creek length rates using only kO2_1. Per unit length is likely the most appropriate approach to compare between timepoints in this case as the volume and surface area change rapidly over tidal cycles. Please see Kearns et al. 2016 (Nature Comm. ) for additional method details.

References:

Kearns et al. 2016, doi: 10.1038/ncomms12881

This data is also accessible through the LTER database, and has been assigned a database doi: 10.6073/pasta/fe47a9461bd332fae3ac7792af21c2b0