Research 443: Remote Sensing of Irish Surface Waters

The main objective of the INFER project was to examine the potential of freely available Earth observation data for monitoring water quality in Irish lakes and transitional and coastal (TRAC) waters and to determine if prevailing weather conditions compromise the effective use of such remotely sensed data in an Irish context. Images from the twin Sentinel-2 satellites were used to estimate water quality parameters from a set of calibration lakes from which in situ water quality and atmospheric correction data were collected at times that coincided with satellite passes or were within a reasonable timeframe compared with ground truth acquisition. Field radiometry measurements were taken alongside water quality data to coincide with satellite imagery, which allowed glint-free reflectance from the water bodies to be determined. Based on the calibration with these field data, a coupled technique was developed to atmospherically correct and estimate chlorophyll a using two processers, C2RCC and ACOLITE, and this coupled technique was validated using historical water quality data collected for a range of lakes of different trophic status. A `real-time web-hosted platform was then developed that automatically ingests Ireland- specific Sentinel-2 data, processes the data based on the regionally optimised coupled approach that was developed, and renders predicted chlorophyll a and turbidity water quality maps for examination. A historical analysis of Earth observation data from the archived Landsat program, freely available from Google Earth Engine, was also conducted for 21 lakes regularly monitored by the EPA and local authorities. Observational data from 1984 to 2021 were reviewed, a statistical assessment of cloud coverage occurrence was determined, and those cloud-free observations matching in situ sampling events were curated. In all cases, while the spatial resolution of the Earth observation platforms was suitably high (for Sentinel-2, ~10 m2) to resolve and map water quality for the majority of water bodies of interest, the challenging meteorological environment principally the excessive cloud cover complicates the use of both the Sentinel-2 and the Landsat platforms, neither of which can see through cloud, limiting the availability of usable images that, in ideal conditions, could be taken as frequently as every 5 days (the Sentinel-2 constellation has a 5-day revisit time). Taken together, the findings of this project demonstrate the potential use of Earth observation imagery to estimate water quality in surface waters with the implementation of the coupled processor workflow found to be optimal for Irish conditions. The results show that the technique developed can be used to get a snapshot of water quality across the spatial scale to provide general guidance on trophic status. The results also show the need for additional work targeting the application of algorithms for different trophic statuses so that the algorithms are directly applicable to Water Framework Directive (WFD) monitoring. However, further calibration and validation work is needed, particularly in refining this work to take account of various lake typologies, trophic status and optical properties.