News Release

Sister climate cities, utility data predict future water, electricity demands

Peer-Reviewed Publication

Penn State

contemporary climate analogs


An international team of researchers used contemporary climate analogs and multi-outcome machine learning to match high-warming analogs and cities of interest. 

view more 

Credit: Courtesy Renee Obringer

UNIVERSITY PARK, Pa. – Modern-day Ciudad Mante, Mexico, could help Tampa, Florida, plan for shifting water and electricity demands due to climate change, according to an international team of researchers. Led by Renee Obringer, assistant professor of energy and mineral engineering at Penn State, the researchers used utilities data and climate analogs — contemporary cities with climates close to what other cities are predicted to experience in the future — to assess how climate change may impact residential water and electricity use across 46 cities in the United States.

Their computationally efficient model projected strong regional differences for future water and electricity demand, with some cities possibly experiencing increases in summer water and electricity demand of up to 15% and 20%, respectively, because of climate change. The researchers published their findings, which could inform how cities learn from each other in planning for climate change mitigation, in the journal One Earth.

“We’re trying to understand how future climate change scenarios might impact water-electricity demand in U.S. cities,” Obringer said. “What do these changes actually show in terms of how our bulk demand is changing, and how do we bridge the gap between research data and practice to help management agencies plan resilience to future change and better serve residents?”

According to the researchers, understanding how climate change impacts the way society uses water and electricity — from increased air conditioning use to more residential law irrigation — and how these impacts will continue to evolve is critical for building and managing resilient infrastructure systems. The issue has been a lack of available models to do such impact assessments at this regional scale, Obringer said, as most climate change models are more concerned with global trends and require significant computing power to process. Another difficulty is that water and electricity data is typically tracked and analyzed separately, with scientific papers isolating utilities in their analysis unless under specific hydropower and water-cooling scenarios. 

Despite the separation, Obringer pointed out that interconnection is usually an integral, built-in part of the system, as water is frequently used for cooling in electricity generation, and electricity is used to power the systems that treat, pump and distribute water. 

To train a model to predict future interactions, the researchers set out to compile past observational data on the utilities, but that proved to be the perfect microcosm of how separation creates barriers for collaboration. Obringer quickly found electricity usage data available through U.S. government agencies. However, water data is not beholden to the same regional transmission organizations, leading her to send Freedom of Act Information requests to utility companies; just over half replied.

“The isolation at the utility side is concerning, because the utility is making plans for the future understanding that climate change is likely going to shift their demand, but they might not be accounting for how it will also shift the water profile, which will, in turn, impact their demand profile,” Obringer said. “Understanding and planning for this interaction is critical for navigating climate change-induced impacts.”

Even when data is available, Obringer said, most studies utilize large-scale, global general circulation models that require sophisticated computer systems and produce technical data that is difficult for regional stakeholders to access and interpret, according to Obringer. To help address this gap, the researchers incorporated data from a previous study by Matt Fitzpatrick, associate director for research at University of Maryland Centre for Environmental Science, that analyzed how 540 U.S. cities, including State College, will look in 60 years based on several variables and then matched them to current-day regions.

With this information, in combination with observational data from the National Centers for Environmental Information’s North American Regional Reanalysis and the obtained coupled residential water and electricity demand for each city of interest, Obringer’s model could now project the water-energy nexus at an unprecedented scale.

“Nobody really knows what the temperature increasing by one and half degrees Celsius means,” Obringer said. “Instead, we can use the analogs to say the climate in New York City in 60 years will roughly look like northern Arkansas today. This provides a new methodology that allows communities to wrestle with any implications for how to manage their electricity grid or their water sources and account for what they may need in the future.”

The model projected a general increase in electricity demand for the country with slight increases in water demand, but strong variances in regional trends persisted. Some areas, like Tampa, were shown to potentially need less water because its analog of Ciudad Mante receives more annual precipitation. The western U.S., however, is projected to see increases in water consumption, potentially putting additional stress on the water system, given regional issues with drought. In Chicago, previous global general circulation model studies projected a 12% increase in electricity use, while this model’s water-energy nexus approach projected a potential 20% increase in electricity demand.

Regardless of the cause, the failure to accurately account for shifts in electricity demands can lead to disastrous results, Obringer said. In one worst-case scenario based on demand changes due to shared socioeconomic pathways, the team looked at Los Angeles County, which plans to achieve 100% renewable energy by 2050. If the predicted climate-induced increase in demand is not accounted for, the country could fall short of the projected 3.8 million megawatt hours of electricity needed. The difference would require 14,000 additional wind turbines, equivalent to approximately 20% of the current U.S. operational stock. Obringer stressed that when a city falls short of demand, not only could that lead to potential blackouts and compound social inequities, but the city may fall back on carbon-intensive electricity generation, exacerbating the climate crisis further.

“Planning for infrastructure, which usually happens 30 to 40 years in advance, requires a lot of funding and is supposed to last just as long with so much uncertainty,” Obringer said. “Simply, this model offers another methodology to help test different scenarios, see our ranges of outcomes and find the most optimal path to allow for the most flexibility going forward.”

While the study primarily focused on demand, the researchers said that others can tease out further analysis based on additional variations and scenarios specific to their region.

“If we can find a way to make those impacts more tangible — and I think this model is another tool in the toolbox that can do that — it will go a long way to improving communication, improving decision making and, hopefully, ensuring that any adaptation strategy that we have is scientifically backed and ultimately equitable and benefits the community,” Obringer said.  

Also contributing were Roshanak Nateghi, associate professor, and Jessica Knee, undergrad researcher, at Purdue University; Kaveh Madani, director of the United Nations University Institute for Water, Environment and Health; and Rohini Kumar, researcher at Helmholtz Centre for Environmental Research in Leipzig, Germany.

The National Science Foundation and the National Socio-Environmental Synthesis Center supported this research.

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.