The use of night-time lights satellite imagery as a measure of Australia's regional electricity consumption and population distribution

Abstract
Satellite imagery of night-time lights provided by the US Air Force Defense Meteorological Satellite Program (DMSP), using the Operational Linescan System (OLS), has been used to estimate the spatial distribution of electricity consumption throughout Australia. For the period 1997 to 2002, there was very high correlation between state electricity consumption and night-time lights with an R 2 value of 0.9346 at the state and territory spatial resolution. To increase the accuracy at which electricity consumption can be estimated at greater spatial resolution, an Overglow Removal Model (ORM) was developed to overcome the overglow effect caused by the dispersion of light into surrounding areas. The ORM makes use of the relationship between light source strength and the overglow/dispersion distance from the light source. As electricity consumption statistics at a greater spatial resolution than the state or territory level are not publically available in Australia, population statistics at the statistical local area (SLA) were used to demonstrate the increased accuracy of the ORM at returning the overglow light to its source, and, in turn, the accuracy of measuring electricity consumption. The ORM enabled an estimation of the electricity consumption of SLAs, greater than 10 km2, with an R 2 value of 0.8732, which is a 25.4% increase in accuracy over untreated data before applying the ORM. The increase in accuracy of the location of the origin of night-time lights can enable better georeferencing of satellite imagery of night-time lights and greater accuracy in locating population centres and centres of economic development, and assist with electricity infrastructure planning in regions of the world where statistics are not readily available. The result of the ORM is a map of Australian electricity consumption, and an estimation of the regional electricity consumption for all SLAs greater than 10 km2 in size is included.