Use of smart meter readings and nighttime light images to track pixel-level electricity consumption

Abstract

ABSTRACTDespite a wealth of night remote sensing studies on electricity consumption, some unsolved issues limit future studies. These include the coarse resolution of nighttime light (NTL) imagery, the mismatching acquisition time between current NTL imagery and electric power data, and unknown data quality of public power records with a low update frequency and at a regional/national level. Thus, we attempt to address these well-known issues by using readings from smart meters. Such a new dataset provides high spatial and temporal resolution electricity consumption information, which enables the tracking of electricity consumption at the fine scale. To this end, all available NTL images from space and over eight million records of electric power consumption from ground smart meters with matching acquisition time were collectively employed for comprehensive examinations. Results of our analyses suggest two major findings. First, positive linear correlation exists between NTL images and smart meter readings at the pixel level. Second, their relationship is changing in different seasons, due to vegetation seasonality and snow cover. With the high update frequency of power consumption data from smart meters, these findings provide guidance for electricity consumption estimation at the fine-grained resolution, which will be informative and useful for local power planning.