From water meter readings to forecasts – cloud services support water use planning

The Datasta tehokkuutta, tuottavuutta ja kilpailukykyä project has explored how data collected by wireless water meters can be used to forecast future water consumption. The pilot developed a demonstration solution based on Amazon Web Services (AWS) cloud services, combining measurement data storage and processing with forecasting models.

The solution used timestamped consumption readings transmitted by LoRa-based water meters. Measurement data stored in a database was processed into a format suitable for analysis and made available to forecasting models in Amazon SageMaker. This established a data processing workflow from water meter readings to forecasts of future consumption.

The forecasting experiments used one year of historical data to estimate how water consumption would develop over the following year. The methods tested included linear regression, ARIMA and Prophet. The forecasts differed between models, illustrating the importance of model selection and evaluating results in practical applications.

Forecasting water consumption and analysing measurement data can support applications such as:

• Water use planning: Forecasts can help estimate future consumption and anticipate resource needs.

• Detecting anomalies: Monitoring consumption trends can help identify unusual patterns of water use.

• Financial planning: Estimates of future consumption can support cost projections and the planning of water services and budgets.

The pilot provides a practical example of how data generated by IoT devices can be processed using cloud services to support decision-making. Similar workflows can also be applied in other settings where historical measurements are used to estimate future trends.