Research about Providence, Rhode Island

USGS2024

A predictive analysis of water use for Providence, Rhode Island

To explain the drivers of historical water use in the public water systems (PWSs) that serve populations in Providence, Rhode Island, and surrounding areas, and to forecast future water use, a machine-learning model (cubist regression) was developed by the U.S. Geological Survey in cooperation with Providence Water to model daily per capita rates of domestic, commercial, and industrial water use. The PWSs in this area form a connected network that sources water from the Scituate Reservoir in Rhode Island. The cubist regression model was trained and tested on daily per capita rates for three categories of water use (domestic, commercial, and industrial) that were developed from quarterly water sales data and U.S. Census Bureau population estimates within each PWS service area from January 2005 through December 2021. The model was then used to make forecasts of future water use under varying scenarios of climate change, population growth, and economic growth for the years 2030 and 2040. The resulting daily per capita rates, which were modeled from the historical data, had an r 2 value of 0.94 and root mean square error of 6.7 gallons per capita daily. Results of the model were used to estimate total water use (the product of daily per capita rates and population) for all public water systems over the historical study period. Daily per capita rates in the study area decreased from 2005 to 2021, while population increased during that same period. “Category of water use” was the variable with the greatest explanatory power for modeling daily per capita rates. Overall, both daily per capita rates and total water use were projected to decrease in 2030 and 2040, in comparison to historical values from 2005 to 2021. Daily per capita rates and total water use were forecasted to decrease as economic growth rates increase. Daily per capita rates were expected to decrease as population growth rates increase; however, total water use was less sensitive to population growth rates than daily per capita rates. Effects of climate change were minimal over the 2030 and 2040 forecasting horizon for the scenarios tested.

Rhode Island↗

USGS2023

Assessment of factors that influence human water demand for Providence, Rhode Island

To determine the most relevant climatic and economic factors driving water demand for Providence, Rhode Island, and to further the understanding of human interactions with water availability, linear regression models were developed to estimate single-family and multifamily residential, commercial, and industrial water demand for the service area of Providence Water for 2014–21. Monthly water use delivery data were provided by Providence Water. An array of climatic and economic data, the drought index, and binary variables to represent seasonal water use and the onset of the coronavirus (COVID–19) were investigated as possible explanatory variables for the water demand models. The water demand model with the best fit with the least amount of error was the single-family residential water demand followed in descending order of accuracy by the commercial, multifamily residential, and industrial water demand. Seasonal variables were significant in all models, and the COVID–19 binary variable was significant in the commercial and industrial models. One or two economic variables were significant in all models and one climatic variable was significant in all models except the commercial model. Overall residential, commercial, and industrial water demand in the Providence, Rhode Island, service area has decreased during the study period most likely because of widescale drought conditions and policies designed to improve water efficiencies. The linear regression models developed for single-family and multifamily residential, commercial, and industrial water use explained 94, 85, 91, and 77 percent, respectively, of the variability in monthly water use. Multifamily residential water demand displayed a less distinct seasonal trend than that observed for single-family residential customers, likely because multifamily homes tend to use less water outdoors. The commercial water-demand model included no climatic variables, one economic variable, the COVID–19 pandemic variable, and the high and low water use seasonal variables—the latter two variables indicating the importance of seasonal fluctuations in water use. The COVID–19 pandemic and a concomitant State executive order had the immediate effect of severely reducing commercial water use. The industrial water-demand model did not perform as well as the other models because industrial water delivery data display a greater range of values, both seasonally and for the overall study period.

Rhode Island↗