Energy consumption from central heating has rapidly increased in the cities of the North China Plain (NCP). The increasing use of natural gas in the central heating supply system may have altered the spatial and temporal patterns of CO2 emissions from central heating, yet the quantitative impacts are poorly understood. Here we detect the spatio-temporal dynamics of CO2 emissions of central heating from 2012 to 2016 at the prefectural-city level in the NCP region, by using the satellite NPP-VIIRS nighttime light data and a panel regression model to estimate CO2 emissions on a 5 x 5 km(2) grid. We find that despite a slight decrease (2%) in 2014 under the "Natural Gas Utilization Policy", CO2 emissions continued to grow. Between 2012 and 2016, CO2 emissions from central heating in the NCP increased from 106 to 121 Tg, although CO2 emissions declined by 12% in Beijing due to the increasing contribution of natural gas boilers. The gridded CO2 emissions map shows that over 2012-2016 coal burning is the main driving force of CO2 emissions in both urban and non-urban regions, despite the increasing fraction of gas-based heating. Our results contribute to city-level policymaking on carbon reduction and climate change mitigation. The high-resolution gridded CO2 emissions can also be applied in physical models to facilitate carbon cycle studies.