Social Memory vs Abrupt Behavior Changes. Modelling water consumer behavior change in non-stationary climate and social scenarios

Abstract:

Urbanization, population growth, and changing climate conditions have altered urban water demand patterns and increased the pressure on water resources. Understanting changing water consumption behaviors at different time scales is key to inform water governance, network systems operations, water demand management actions, and supply development both in normal conditions, as well as in anomalous/emergency situations. The current COVID19 pandemic and the global Shelter in Place orders is one of those anomalous conditions that has been and is expected to transform lifestyles and work patterns across the world. Further understanting the consequences of such systematic shocks on our water demand patterns would provide more behavioral insights to inform the operations and management of water infrastructure under such unprecedented events.

In this work, we investigate how water consumption behavior changes during the 2012-16 California drought compare with behavior changes during the COVID-19 lockdowns. We identify social memory effects - due to prolonged behavior change and demand management - in comparison with abrupt behavior changes – due to the sudden change in lifestyles and habits during the COVID-19 lockdowns. For our analysis, we rely on state-of-the-art customer segmentation techniques that exploit water meter data coupled with data dimesionality reduction, advanced correlation, and unsupervised clustering techniques. This research contributes behavioral insights on urban water consumption in non-stationary climate and social scenarios. Such insights can be ultimately exploited by utilities and decision makers to design and implement customer-tailored water demand management strategies under various drought scenarios, along with informing planning and management strategies to improve short-term and long-term resilience of urban water systems under rapidly changing water demand patterns due to unprecedented events such as the current pandamic.