Model-based estimation of industrial water intake and drinking water production, 2005 to 2021

In 2020, Statistics Canada started developing experimental model-based estimation methods for drinking water production and population served for the Biennial Drinking Water Plants Survey. The same procedure was done for the Industrial Water Survey, which includes thermal electric power generation, manufacturing and mining, including the coal mining, metal ore mining and non-metallic mineral mining industries. The primary objective of this project was to estimate the variables during the COVID-19 pandemic, as surveys did not collect data during that period, and to model the data gaps for non-survey years, given the biennial nature of the surveys.

Using statistical models, preliminary estimates of water intake (for the Industrial Water Survey) and estimates of potable water produced and the number of people served by drinking water plants (for the Biennial Drinking Water Plants Survey) are currently available at the national level dating back to 2005.

A comprehensive set of criteria has been applied to select auxiliary variables used to help modelling for water intake models. These criteria focus on industry-specific factors that influence water intake patterns. In industrial water intake models, variables such as energy consumption, gross domestic product and fuel usage are currently considered. These auxiliary variables are important for capturing the distinct dynamics within each sector, ensuring the models reflect specific water intake needs and behaviours across diverse industries.

Similarly, the potable water produced by drinking water plants and the population served are predicted using population estimates and other related variables.

The ultimate aim of the project is to refine these models by exploring each sector in greater detail and to build a more comprehensive and consistent dataset.

While national-level predictions of industrial water intakes, potable water produced and population served have already been successfully achieved, the next step involves modelling these predictions at higher temporal and spatial resolutions, such as monthly and provincial levels. This will allow for better tracking of provincial trends and support more effective data management.