NOAA RESTORE Science Program: A web-based interactive decision-support tool for Adaptation of Coastal Urban and Natural Ecosystems (ACUNE) in Southwest Florida: episodic water level monitoring in Clam Bay from 2018-06-10 to 2020-01-07 (NCEI Accession 0244397)

Methods:
To capture the episodic higher-than-normal water level events necessary for model verification and validation, 28 sensors were designed and built from scratch and then deployed/recovered twice. The sensors were deployed in four locations in Southwest Florida: Clam Bay (4), Goodland (8), Henderson Creek / Rookery Bay (8), and in the Ten Thousand Islands (Figure 1). These sensors were designed to deployed vertically out of the water with a triggered to enable them only when an event occurred (Figure 2 and Figure 3). This ensured maximum battery life (without the need of a solar panel) and minimal biologic growth both of which were highly useful as many of the deployment locations were remote and teeming with wildlife.

Once immersed, the sensors recorded the absolute pressure of the water column above the sensor (which is ultimately converted to water level). The sensors were encased in Schedule-40 PVC which incorporated a PVC union to enable them to be opened while remaining waterproof in the field. The pressure sensors were calibrated in a laboratory with the distance from the pressure sensor port to the top of the union collar determined. Immediately prior to a deployment, the internal highly accurate, temperature compensated RTC (~2 s/yr) was set and the sensor sealed.

The sensors were deployed by affixing them to strongly rooted, mature mangroves or trees to ensure they wouldn’t move even during extreme storm events (Figure 4). Pictures of the mounted sensors were taken and the sensors positions were surveyed to the top of the union collar with an RTK GPS system. This enabled to vertical position of the derived water level to be referenced to NAVD88. Sensors were configured to record at 128 hz which enables the resolution of waves, tides and surge. Locations for the sensors were determined during an exploratory visit to the region in mid-2018. Site spatial locations were chosen in a manner to attempt to provide both alongshore and cross-shore tracks. Vertical locations were chosen by estimating the MHHW lines and chose locations 10-20cm above that.

For the first deployment, the sensors were deployed at the beginning of the 2018 Atlantic Basin hurricane season and then recovered the following May. For the second deployment, the sensors were deployed several months later in the 2019 hurricane season and then recovered in Jan 2020. Preliminary analysis of the data for the first deployment determined that the vertical locations on several sensors were higher than expected astronomical tides as few events were recorded. Thus, the vertical locations of these sensors were lowered for the second deployment. The spatial positions of all sites remained the same for both deployments. Although no tropical storms were observed, several astronomically higher-than-normal water levels events were recorded.

Observed pressure data is encrypted and the stored on high-capacity SD card located inside the sensor with each detected event saved as a separate file. After a deployment, the sensors were opened, data recovered, and then post-processed. The first intermediate post-processing step includes unpacking and decrypting the data, determining the time of each observation, and conversion of observation units. Then as a final post-processing step, water pressure measurements were converted to a delta water level above the sensor using a hydrostatic approximation. Time varying air pressure used in the conversion were obtained from three nearby NCEI sites. Each sensor was provided with a preferential ordering of the site in case data for a particular time-period was missing. With the delta water level determined, this value was adjusted to be referenced to the collar of the union, the location at which the GPS fixes to Geoid12a (NAVD88) were made. As the sensors were mounted to potentially growing vegetation, fixes were obtained at both deployment and recovery. For the vertical reference datum conversion from the top of the collar to NAVD88, the vertical position of the collar was assumed to move linearly from deployment to recovery.

While the horizontal accuracy of the GPS fixes was good for the needed resolution, in some cases, the vertical accuracy was reported to be on the order of several cm, however, in other cases, the accuracy was 50 cm. Thus, the conversions to the NAVD88 become somewhat questionable until the point at which they can be further compared with simulated water levels. Unfortunately, vertical inaccuracies are a common probably with remote coastal sites as the ability to take sight lines or to triangulate the RTK is limited. In general, the post-processed GPS fixes were reported to be less accurate. As such, although nearly all the sensors report reasonable vertical heights, it is still recommended to demean the data and focus on relative change in water level and/or the magnitudes/periods of recorded waves. Due to the high sample rate, the events being individually recorded, and the large number of events recorded, an additional optional post-processing utility was developed to extract data for one or more sensors for all events occurring in a specific time-period. This utility then output data in a format compatible with Tecplot (see example plots in Figure 5), although it would be easily adaptable to output a different data format. Intermediate and final post-processed data has been included. Additionally, some example data plots as well as plots showing the time-period in which the events occurred have been supplied as well.