Full historic record true colour changes in Greater Sydney drinking water supply systems in Australia

• Constrained and full historic data provide substantial differences in true colour trend. • Robust equation developed for the conversion of apparent colour into true colour data. • Inclusion of full historic colour data substantially reduced the rate of true colour change. • Full historic trends showed no or a decrease in ∼ 90% of Greater Sydney’s water supply. • Catchment characteristics showed no substantial effect on colour trend categories. True colour in drinking water poses no direct health risk; however, visibly coloured water (>15 TCU) often leads to consumer complaints. Long-term analyses of colour trends in drinking water reservoirs are limited by historical industry-wide shift from apparent to true colour monitoring techniques. This study established a conversion equation to estimate true colour (400 nm), from apparent colour, iron and turbidity, and applied this to data from 1931 to 1990 across eight reservoirs in New South Wales, Australia. A Bayesian hierarchical generalised linear model was then used to analyse full historic record trends (∼80 years) and compare the findings to constrained historic record trends derived only from direct sampled true colour values only. While observed true colour constrained historic record trends showed substantial increases across seven of eight reservoirs, incorporating full historic record converted true colour data resulted in either decreases or no change in four major reservoirs that collectively supply ∼ 90% of the region’s total drinking water volume. Notably, full historic record trends substantially reduced the rate of change compared to constrained historic record trends. These contrasting results highlight the opportunity of applying methods to make use of the full historic record for trend analysis. Qualitatively, we found that reservoirs showing decreasing colour trends were the largest in both surface area and storage volume, providing evidence that morphometry can be a driver of colour, through processes such as photooxidation, settling and mixing. Future research should prioritise understanding catchment condition change analysis, as well as long-term rainfall cycles and their relationship with reservoir colour, and integrating lake-specific biogeochemical processes to better explain in-lake true colour dynamics.