Analysis of spatiotemporal household water consumption patterns and their relationship with meteorological variables

Abstract

The spatiotemporal patterns of household water consumption can provide valuable information for urban water management. Thus, this paper introduces a novel method based on the Bayesian Estimator of Abrupt Change, Seasonal, and Trend (BEAST) algorithm to extract these patterns. The proposed method divides the study area into different regions using cells of equal sizes. Then the BEAST algorithm estimates the trend and the seasonality of the consumption time series of the buildings within each cell as spatiotemporal patterns. The proposed method is evaluated using 5-year consumption data of 566 residential buildings in Shabestar City, Iran. We also studied how changes in trend and seasonality of average wind speed, precipitation, relative humidity, and temperature affect consumption patterns. The results showed that household water consumption had a linear trend, with only three change points, one of which, because of its synchronicity, can be ascribed to the COVID-19 outbreak. The water consumption data also showed a relatively weak seasonality. The correlation analysis showed no meaningful relationship between the trends of considered meteorological variables and water consumption. However, the seasonality of average wind speed and temperature showed a weak correlation of 0.71 and 0.51 with the seasonality of the water consumption data, respectively.