Using multispectral drones to predict water quality in a subtropical estuary

Drones are revolutionising earth system observations, and are increasingly used for high resolution monitoring of water quality. The objective of this research was to test whether drone-based multispectral imagery could predict important water quality parameters in an ICOLL (intermittently closed and opened lake or lagoon). Three water quality sampling campaigns were undertaken, measuring temperature, salinity, pH, dissolved oxygen (DO), chlorophyll (CHL), turbidity, total suspended sediments (TSS), coloured dissolved organic matter (CDOM), green algae, crytophyta, diatoms, bluegreen algae and total algal concentrations. DistilM statistical analyses were conducted to reveal the bands accounting for the most variation across all water quality data, then linear correlations between specific band/band ratios and individual water quality parameters were performed. DistilM analyses revealed the NIR band accounted for most variation in March, the Green band in April and the RE band in May, and showed that the most important contributors varied significantly among campaigns and variables. Significant linear correlations with R² > 0.4 were obtained for eleven of the water quality parameters tested, with the strongest correlation obtained for CHL and the green band (R² = 0.72). The relative importance of predictor bands and observed water quality parameters varied temporally. We conclude that drones with a multispectral sensor can produce useful 'snapshot' prediction maps for a range of water quality parameters, such as chlorophyll, bluegreen algae and dissolved oxygen. However, a single model was insufficient to reproduce the temporal variation of water parameters in dynamic estuarine systems.