Forecasting Worldwide Local Carbon Dioxide Levels Using Spectral Techniques and Big Data Sources

Abstract

This paper presents a method for the estimation and forecast of worldwide local emission levels based on a space-time extension of a discrete thin-plate smoothing model in the frequency domain. The proposed method can potentially be applied where there is an interest in forecasting the time evolution of large geographical or spatial databases at all global grid locations simultaneously. In particular, the paper illustrates the method in the estimation and forecast of underlying local levels of carbon dioxide around the globe using the 0.1ox 0.1o grid from the EDGAR database. The results highlight the world distribution of carbon dioxide emission levels and their variations during the historic period recorded in the database as well as their forecasts up to 2025. Global carbon dioxide will reach 53,161 mill tons by 2025, representing an increase of 29.4% from 2018 levels and of more than 52% since 2012. Shanghai, China, is the location leader in carbon dioxide emission levels since 2009; reaching 46.13 mill tons in its surrounding 10 km square region by 2025. China-Korea, North America, India, Europe and Japan, in this order, will be the highest emitting regions in the world by 2025, with Africa-Middle East, due to the oil and gas fields industry, and South America rising fast behind. In fact, most developing countries show a high rate of increase in their carbon dioxide levels during the 2012--2025 period. In comparison, North America and Japan show stable levels while Europe may see its levels actually decrease. However, even within these more developed regions it is of environmental interest to identify high emitting locations at a local level.