Abstract
High average temperatures lead to high regional electricity demand for cooling buildings, and large populations generally require more aggregate electricity than smaller ones do. Thus, future global climate and population changes will present regional infrastructure challenges regarding changing electricity demand. However, without spatially explicit representation of this demand or the ways in which it might change at the neighbourhood scale, it is difficult to determine which electricity service areas are most vulnerable and will be most affected by these changes. Here we show that detailed projections of changing local electricity demand patterns are viable and important for adaptation planning at the urban level in a changing climate. Employing high-resolution and spatially explicit tools, we find that electricity demand increases caused by temperature rise have the greatest impact over the next 40 years in areas serving small populations, and that large population influx stresses any affected service area, especially during peak demand.
Climate change modelling predicts rising temperatures and consequent increases in storm intensity, flooding and inundation. These changes pose risks to infrastructure and neighbourhoods as well as disruptions to the energy supply and its dependent infrastructure. Direct effects have already included damage to power plants, roads, bridges and communication towers, and resultant interruption of electrical energy, transportation and communications sectors in cities1. As climate conditions continue to change, local communities and their critical infrastructure will respond, adapt and evolve. Population will shift in response to these changes2,3, for example, as services that generate new economic activity in more environmentally stable locations will attract new workers and associated households. This shift will force demand locations for electricity to change. As a result, networked infrastructures may be required to accommodate new load centres and to minimize vulnerability to natural disasters4. To provide information about the complex interactions among climatic conditions, population shifts, and energy supply and use, new tools informed by consistent spatially disaggregated data are needed5.
Although a recent report began to quantify increases in electricity demand based on regional warming6, little research has been conducted to quantify the potential impacts of combined climate and population stresses on the operating limits of electrical substations within the power grid. There are two main reasons for this gap. First, until recently, climate projections at high enough resolution to be compatible with infrastructure modelling have been unavailable. Second, spatially explicit population migration data and analysis techniques capable of providing estimates of future energy demand have been missing and/or difficult to characterize.
Previous research has predicted increases in electricity demand in response to increases in global temperature expected with climate change. For instance, using the Electricity Information (EIA) National Energy Modelling System, Hadley et al.7 showed that for most US locations, the savings of electricity in the winter months due to fewer cooling degree days do not offset the added expenditures on electricity in the summer for the increase in heating degree days. The climate inputs used in the Hadley study, however, were global model output at 2.5∘ (latitude, longitude) spatial resolution, so while the study was able to capture electricity customer response to some general trends in future temperature, it was unable to resolve regional differences in temperature in both base and future cases. To improve on such predictions, the California Energy Commission8 employed a constructed analogues method for statistically downscaling global climate model output at coarse spatial resolution (2.5∘) to finer (∼13 km) resolution, using analogues from present regional climate. With this method, along with electricity utility billing data, and demographic information, they made projections regarding increases in residential electricity use.
Although much research regarding population movement has been performed and weaknesses in various methods identified, none of the methods has been applied to the assessment of changes in electricity demand due to spatially explicit changes in population. It has been noted9 that the growth of every small area is linked to causes and forces located elsewhere in the region and by demographic and economic interactions; thus, the process of regional growth and resulting electricity demand changes is hard to characterize. Yet strides have been made in relating population changes to changes in electricity demand. A variety of commercial models for forecasting electric load based on land use projections have been used to generate scenarios for long-range planning. Among the first were linear urban models based on a gravity model for population movement10 and generalized to a set of matrix computations11. While these models take into consideration the projected development of various types of electricity customer (residential, commercial, industrial), they do less well at predicting the ways in which redevelopment of land use will occur. Therefore, the US Geological Survey12, Duke Energy13 and the California Energy Commission14, among others, have expanded such models to incorporate economic and demographic weights for the determination of various redevelopment types using agent-based modelling. For each of these models, the critical components are those of the spatial distribution of the population among their places of residence and their places of business, and the way changes in those distributions are projected for the future. However, these models do not evaluate electricity usage by customers as organized within utility service areas or how service area organization might change as a result of new urbanization.
Vulnerability of the electric grid to climate impacts is ultimately a function of its exposure, its sensitivity and its adaptive capacity to stress15. Thus, in this study we examine human response to exposure to changes in regional temperature16 and increases in landfall hurricane intensity7,17,18, and electrical grid sensitivity to these human- and climate-induced stresses. We employ a spatial methodology for electric load forecasting for grid planning in which a cost–distance algorithm based on regional population is used to determine the service area for a given substation. We apply satellite population observations and predictions based on Census and Internal Revenue Service (IRS) data to project spatial shifts in electricity demand in the southern US region (Supplementary Fig. 1) and we incorporate several sudden redistributions of population in response to a 2005-like hurricane season at the predicted return period for such events in the region based on recent climate science. We also calculate local percentage change in electricity demand given temperature changes from dynamically downscaled (12 km resolution) climate model projections for the region to analyse further the effect of increases in global temperature and resulting regional electricity demand consequences. From this analysis, we make a substation-service-area-level projection of substation capability in the southern US to support changes in demand due to temperature rise and sudden population shifts in response to intense storms. This procedure allows us to demonstrate a viable tool for making high-resolution predictions in the absence of address-specific data regarding electricity use.
