Do water savings persist? Using survival models to plan for long-term responses to extreme drought

Abstract

Abstract Water utilities must maintain reliable service in a world where climate shocks and other socio-economic and health stressors are likely to disrupt water availability and demand more frequently. Understanding short- and long-term customer responses to these salient events is critical for infrastructure planning and capital investment. Although the short-term demand impacts of extreme droughts and related policy measures have been studied extensively, less is known about how these impacts persist—especially when driven by public awareness, media coverage, or other external drivers. Here, we introduce a novel approach combining survival models and change detection to assess water demand conservation ‘survival’ and rebound, using this method to analyze residential water demand in Costa Mesa, California after the state’s record-breaking 2012–2016 drought. We find that, of 54% of customers with detected savings in 2014–2015, just 25% rebounded to prior consumption levels after 5 years, implying mean conservation survival of 8 years. Survival was greater in young and politically progressive neighborhoods, smaller in residences with occupancy changes, and not significantly associated with water-efficiency rebates. Comparing the 2012–2016 drought to California’s milder 2007–2009 drought shows no evidence that drought severity associated with water savings persistence. This study presents an innovative approach to measure impacts of various stressors and their long-term water demand impacts. Our method enables utilities to more accurately discern structural changes in water demand, better informing strategic planning for short- and long-term water reliability and security.