Methods for Assessing Historical Water Level Data Quality for the Texas Coast

High quality water level data sets are important as relative sea level rise is leading to increasing frequencies of inundation and coastal erosion. Coastal communities need to plan for adaptations and good quality local historic water level data are an invaluable resource, especially multiyear data sets including prior extreme events. In the late eighties, the Conrad Blucher Institute (CBI) installed and maintained tide stations in South Texas in collaboration with the National Oceanic and Atmospheric Administration (NOAA). In the early nineties the Texas General Land Office and the Texas Water Development Board along with CBI and the assistance of NOAA and the US Army Corps of Engineers, grew the network to a coast wide tide station network, the Texas Coastal Ocean Observation Network (TCOON). The network operations were interrupted in 2014 and have resumed since. Historical data is being evaluated prior to being integrated with the latest TCOON data. The team is using a set of four NOAA National Water Level Observation Network (NWLON) stations whose data was independently processed by NOAA (the standard) to assess and potentially improve the quality of the existing historical TCOON water level time series. While the majority of the data is identical, some discrepancies were found. These discrepancies, the subsequent processing decisions along with some identified causes of these differences, will be described and quantified.

The TCOON covers the Texas coastline from Port Isabel near Mexico to the Louisiana border, with a combined 90 presently operational and historical stations. The stations occupy a wide range of coastal settings including in different bays, water depths, distances from the open coast, and along ship channels and the Intracoastal Waterway. The water level records’ lengths vary significantly, with some stations offering data for as little as 2 months while others extending over 33 years and counting. This extensive and varied dataset will allow for better and more comprehensive analyses of long-term trends in water levels, making it a crucial resource for understanding coastal dynamics and addressing sea level rise and coastal erosion. The process of data acquisition from the TCOON tide stations to the data management and archiving in the historical CBI pharos database was an extensive multistep process which included preservation of incoming messages from the tide gauges for each measurement and subsequent alterations. The ongoing data quality improvement process is made possible by this comprehensive preservation of the data. The methods include developing tools to identify potential problems in data processing all the way from the rawest form of the data ingested. The process starts with the tide gauge collecting average water level every 6 minutes. Each 6 minute water level is calculated from 181 samples taken over a 3 minute interval at 1 Hz frequency. The data is then collected in two ways. transmitted via NOAA’s [NESDIS] GOES Data Collection System (DCS) satellites every 6 minutes (previously once every hour). NOAA’s ground stations receive this satellite data and CBI downloads it from NOAA’s DCS Administration and Data Distribution System (DADDS). The other data retrieval method is a direct connection to the station via cellular network modem (previously local radio networks). The data are decoded with software dependent on which communication path the data followed. Over the last 30+ years, the software has changed periodically to take into account evolving communication protocols and operational needs. A full historical TCOON water level data set is available to the public as part of a Texas Digital Library (TDL) collection. However, the existing TDL data is in relatively raw format and the present effort will further process the data to remove spikes and fill gaps, with methods tested and potentially improved for the four locations where both TCOON and CO-OPS processed data are available.

For these four test stations discrepancies were found in the data including: datum shifts, temporal offsets, missing values, flatlines, and spikes. Since NOAA data are the standard for water level data quality, the discrepancies are used to understand why TCOON data is different for these cases. The findings are then taken advantage of to improve the quality of the water level data for all the TCOON stations which data was not previously processed by NOAA. The algorithms developed to identify and quantify temporal and datum shifts will be described along with summary statistical differences. Metrics used to identify and characterize these discrepancies included the duration, magnitude, and frequency of datum shifts. Qualifying the behavior of datum and temporal offsets have led to identifying changes in the data decoders processing messages from the tide gauges. This work is part of the initial developmental steps for a new lighthouse software to process and provide access to historical TCOON water levels which will also include integrating automated spike detection and gap filling methods and potentially developing and implementing new ones. While focused on the Texas coast and the past operation of the TCOON network, the data flow process is relatively general and the authors hope that this work will be helpful to others working with historical water level time series.