AUTHOR=Hemming Michael P. , Roughan Moninya , Schaeffer Amandine TITLE=Daily Subsurface Ocean Temperature Climatology Using Multiple Data Sources: New Methodology JOURNAL=Frontiers in Marine Science VOLUME=7 YEAR=2020 URL=https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2020.00485 DOI=10.3389/fmars.2020.00485 ISSN=2296-7745 ABSTRACT=

The availability and accessibility of oceanographic data is critical to the sustainability of our oceans into the future. Ocean temperature climatology data products utilizing long time series provide context to ocean warming and allow the identification of anomalous environmental conditions. Here we describe a new methodology to create a daily subsurface temperature climatology using data from three different sources with varying spatial and temporal coverage. The Port Hacking National Reference Station off South East Australia is the site of bottle data collected typically every 1 to 4 weeks at discrete depths between 1953 and 2010, and since 2009 near-monthly vertical profiling CTD profiles and 5 min moored data at various depths. Calculating an unbiased climatology using temperature data sets obtained via different methods, with varying resolution and uncertainty, is challenging. To account for days with limited bottle data, and thus limit the bias from more recent higher temporal resolution data, a time-centered moving window of ±2 days was used to incorporate data collected on neighboring days. To account for different data sources measured on the same date, a date-averaging method was used. As moored data between 2009 and 2019 represented 70% of data for a given day of the year but approximately 1/7 of the 66 year temperature record, a novel data source ratio was implemented to avoid bias toward warmer recent years. Data were organized into their corresponding observed years, and a ratio of 6:1 between bottle and mooring observation years was enforced. To assess the methodology, the steps provided here were tested using synthetically-created temperature data with similar properties to the real observations. The lowest root mean square errors calculated between the known synthetic climatology statistics and the different solution-dependent synthetic climatology statistics confirmed the methodology. The resulting daily temperature climatology shows the seasonal cycle as a function of depth, related to changes in stratification and vertical mixing, and allows for the identification of temperature anomalies. The methodology presented in this paper is readily applicable to other sites across Australia and worldwide where long records exist consisting of multiple data sets with varying sampling characteristics.