All data were processed through a combination of manual and automated quality checks. Quality control procedures were based on protocols adapted from the U.S. Integrated Ocean Observing System (IOOS) Quality Assurance/Quality Control of Real-Time Oceanographic Data (QARTOD) recommendations (IOOS, 2018; Palevsky et al., 2024; Chua et al. (2025) related dataset in this project).

First, all deployments were merged into a continuous time series and retimed to 10-minute intervals. Depth was calculated using pressure measurements in the ponds and barometric pressure from a central location at SMIIL using the Gibbs Seawater Toolbox (GSW, McDougall and Barker 2011). Time series from each pond were manually inspected with “human-in-the-loop” (HITL) annotations to identify and document suspect depth data. Points flagged for review were compared to service trip logs in order to diagnose the reason for flagged data (e.g., periods where sensors were not fully submerged in seawater). The time periods for flagged depth data that were associated with service trips were then removed for all parameters.

In each pond, each individual HOBO sensor measured temperature in addition to pressure, conductivity, pH, or oxygen. The temperature time series presented for each pond is the median temperature collected by the four sensors in each pond. We conducted gross range tests on the median temperature, calculated salinity, pH, and oxygen to ensure data fell within reasonable ranges. Ranges for each parameter are kept consistent across all ponds. For the time points flagged by temperature or salinity gross range tests in each pond, the corresponding oxygen and pH values were flagged and removed from the pond’s dataset.

Salinity (unitless) was calculated from conductivity and temperature using the Gibbs Seawater Toolbox. Salinity data was observed to be “spikier” than other parameters with acute changes in salinity occurring without the expected corresponding changes in temperature. As such we identified these spikes using the “isoutlier” function in MATLAB that determines outliers as a point that is over three standard deviations from the moving mean calculated using a three hour window. If the identified “spikes” resulted from points greater than 0.05 salinity units from the moving mean then the point was removed. Time points flagged as salinity “spikes” were removed from the corresponding oxygen and pH time series.

Onset HOBO Conductivity data loggers are not capable of internal calibration, so data were compared with nearby channel station salinity, water level, and temperature data (Gull Platform, ~1 km away, see Chua et al. related dataset) for salinity calibration and validation. During certain high tides, water from the channel rises above the marsh platform and enters the ponds. During these high tides, pond depths were observed to increase and decrease tidally, and pond temperatures were similar to that in the channel, consistent with pond flushing. However, the preliminary salinity data from the HOBO loggers in the ponds were offset from the channel salinity data during this period, indicating offsets in the HOBO Conductivity measurements. In order to correct for these calibration offsets, we identified high tide periods when channel water entered the ponds based on water depth and temperature data and determined calibration offsets for the pond salinity measurements by correcting the pond salinity to match that of the channel during these high tide periods. Salinity data reported here for the Ring, Shark, and Long Reach ponds were corrected using this approach. For the final two ponds (White and Drum), the HOBO Conductivity sensors deployed at these sites malfunctioned and did not report any usable data. Salinity data reported for White and Drum ponds are the median calibrated salinity values from the three ponds that did not experience sensor malfunctions.

DO concentrations (μmol/L) that passed quality control procedures were adjusted for temperature, salinity (calibrated as above) and pressure using the Aanderaa Data Instruments (2017) compensation equations to provide temperature, salinity and pressure compensated oxygen concentrations (μmol/L). We calculated the density using the GSW Toolbox with the calibrated salinity, temperature and depth variables to provide DO concentrations in μmol/kg. DO equilibrium saturation (DOeqsat) was calculated using the GSW Toolbox according to Garcia & Gordon (1992). DO percent saturation (DOsat) was then calculated as the ratio of the observed DO concentration (DOconc) to the equilibrium saturation concentration (DOeqsat), and multiplied by 100.

Calibration accuracy for both DO and pH was validated based on discrete samples collected at the beginning and end of each deployment (see related dataset). For comparison, pH, DIC, and TA data were converted to hydrogen ion space ([H+]) using the CO2SYSv3 MATLAB package (Lewis & Wallace, 1998; Sharp, 2023; Van Heuven et al., 2011).