UrbanSplash is a decision support tool for urban water recreation that provides open data on water quality in a timely manner and forecasting capabilities. UrbanSplash uses sensors and technologies for near-real time detection of E. Coli bacteria, in-situ real-time data from low-cost IoT sensors, readily available meteorological data and geospatial data at catchment scale to provide near-real time data and forecasting. Briefly, UrbanSplash includes a base station & low-cost sensor networks distributed at catchment scale. The base station we propose is a smart floating pontoon or buoy-depending on the city infrastructure and if the application is for beaches in coastal areas, or inland water bodies (rivers, lakes, canals). The base station will use state-of-the-art sensors to monitor E. coli bacteria in near-real time and other sensors for physicochemical parameters. One such technology is the ColiSense system1 (TRL 6) developed in DCU Water Institute. ColiSense is a portable test kit that provides results in 75 min from sample collection. The 2nd technology, CS Sentinel is currently under development in RESTART2 project in DCU Water Institute. The base stations will be designed and scaled to support future technology deployment and will act as test beds for piloting new technologies. The low-cost IoT sensors will be used at catchment scale to provide real-time data on simple parameters like river discharge, turbidity and temperature. These parameters are the main predictor environmental variables used to forecast poor bathing water quality. The problem is that sensor networks are sparse and even this basic type of data is not available for the majority of local authorities. Low-cost IoT devices create a unique opportunity to provide critical real-time data at catchment scale at low capital and operational costs. Machine learning will be used to develop new forecasting tolls for bathing water quality from spatially distributed low-cost IoT sensors. The data collected during the piloting stage will be segmented into training and validation data sets while the reference data on E. coli will be collected through surveys and base station sampling. We will follow an integrated machine learning approach where we will integrate various information sources which can have potential impact on water quality. We will train, validate and test neural networks-based machine learning models capable of predicting & forecasting water quality and identifying patterns of the pollutants spread across wide geographical areas. As a complete solution, a digital platform will be developed to include 3 key components: Data platform – integrates real/near real-time in-situ data, real-time meteorological data and ML workflows for data validation and forecasting from low-cost IoT sensors; access to real/near-real time data and forecasting, historical data, and simple advice in terms of swimming risks (to be developed with key competent authorities); Resource centre/digital learning centre – bespoke to each city. Community can learn more about their waterways and their intrinsic link with arts, culture and heritage, water/environmental literacy (ecological and chemical indicators etc); Community engagement tool – this will enable citizens to take ownership of their waterways and empower the community by reporting pollution events, pollution sources and pressures, results of citizen science activities, cleaning campaigns, etc.