These images show the impact of coronavirus on electricity demand in Chinese cities

The novel coronavirus (SARS-CoV-2) and the disease it causes (COVID-19) have caused significant disruptions to markets around the world since the virus was first identified in Wuhan City in China in late 2019.

In the energy sector, the impact has been most apparent in the dramatic fall in oil demand in China. It has also impacted the global price of oil. The International Energy Agency predicts the overall demand will lower by about 350,000 barrels per day over 2020. Likewise, the US Energy Information Administration has revised downwards its projections for oil and products like gasoline.

Recently, some analysis of the impacts on electricity demand, especially in Italy — one of the hardest hit countries in the world — have emerged. Italian power demand dropped by up to 18% as of mid-March. Power demand gives a reasonable indicator of economic activity, as it is used in all sectors from households to businesses and industry.

Satellite data has been used to look at emissions from the transport sector, and has found large reductions in major US cities. We used a NASA/NOAA satellite instrument to take a look at the decrease in electricity in some key Chinese cities.

Lights at night

Electric power disruptions can occur anywhere in the world. Many of these are associated with natural disasters such as typhoons, wind storms, tornadoes and earthquakes. War can be another cause of power outages. In addition, there are outages associated with the breakdown of electric grids. These can arise from fuel shortages, ageing power plants or faulty delivery systems. The frequency of such outages likely increases when growth in consumption outpaces the power generation and delivery systems. In some cases, power companies rotate power outages as an equitable means of distributing a scarce commodity. A rarer occurrence is a slowdown in economic activity — in this case due to a pandemic.

The vast majority of studies conducted with satellite-observed night-time lights are based on monthly or annual cloud-free composites produced by the Earth Observation Group (EOG) at the Payne Institute at the Colorado School of Mines. These are favoured by researchers for a number of reasons: the files are available in a generic format; the input data have been screened to exclude solar, lunar and stray light contamination; and the annual night-time lights are filtered to remove detections from biomass burning, aurorae, and other background noise.

We briefly explore electric power demand using night-time lights recorded by the Visible Infrared Imaging Radiometer Suite (VIIRS) — a billion-dollar instrument onboard the Suomi satellites. This instrument's extremely low detection limits make it possible to detect and characterize lighting from bright urban areas to dimly lit rural settlements. The VIIRS' primary mission is weather. The sensor is flown on polar-orbiting satellites and, with a 3000 km swathe, is capable of collecting a complete set of day and night images every 24 hours.

By 'subtracting' one month from another it is possible to identify areas where lighting has either dimmed or brightened. To control for annual cycling in lighting brightness levels, we compared image difference pairs from before and after the Chinese new year months in 2019 (February) and 2020 (January). Specifically, December 2018 minus March 2019 was compared with December 2019 minus February 2020.

Wuhan (Hubei Province), Changsha (Hunan Province), and Xi’an (Shaanxi Province) all show profound differences in lighting demand between the beginning of 2019 and the beginning of 2020. The red colour shows lighting was dimmed (a proxy for lower power demand), while the cyan shows the opposite. In the three Chinese cities explored (Figures 1-3 below), the red dominates the images.

For comparison, we considered another large urban centre in an emerging country. Delhi, the capital of India, shows precisely the reverse trend of the Chinese cities. As of 25 March, India has reported 562 confirmed cases of COVID-19 (as opposed to 81,218 cases in mainland China).

The satellite data records the location and brightness of electric lighting worldwide, and can be used as a proxy for power demand levels. The EOG makes monthly global cloud and moon-free average radiance products from DNB data. This was used to explore demand loss in China, which in turn reflected a loss in economic activity in several Chinese cities, including Wuhan. It should be possible to use this methodology to monitor the return of normal activity levels as the virus disruptions dissipate.