Abstract The ability of weather, ocean and climate forecasting models to accurately predict climate trends, extremes and variability is dependant on accurate and reliable observations of past integrated Earth Systems. However, accurate, long-term measurements are expensive and the spatial domain needing to be measured is large, which often leads to disparate and fragmented datasets of variable reliability. This paper describes the results and lessons learned from an experimental ocean monitoring station located for a twelve month period on the 20m contour of the inshore shelf of the South East coast of Australia. An overview of the dynamic processes of the ocean and coastal systems that were to be measured are described with illustrations of the subsequent data collection systems developed utilising readily available materials and low cost instrumentation. Generalised additive modeling (GAM) can be used to unify fragmented datasets and give insight into potentially nonlinear relationships occurring within the data. In this paper, the efficacy of using low cost software is demonstrated using GAM techniques to evaluate a year of noisy but repeatable turbidity data culminating in an inverse relationship between turbidity and wave height. The paper concludes with a discussion of the research station's results and lessons learnt to date that have assisted in resolving data accuracy and repeatability issues. This has enabled a reliable long term data base of the past to be established based on medium quality data being recorded utilising low cost instrumentation and software. Recommendations of further work and designs are also described that could be of benefit to researchers worldwide attempting to capture on record a vast and noisy environment with low or limited operating budgets.