Temperature Anomalies, Long Memory, and Aggregation (2021), Vera-Valdés, J.E., Econometrics, 9-1: 1-22. DOI: 10.3390/econometrics9010009 [Open Access].
This paper combines two of my research passions. It studies the long memory properties of climate data. The paper obtains the long memory estimates from individual grid temperatures and compares them against the long memory dynamics of global and regional averages. The results show that the long memory parameters in individual grid observations are smaller than the ones from regional averages. Thus, this paper supports the notion that aggregation may be exacerbating the long memory estimated in regional and global temperature data.
Temperature anomalies relative to 1961-1990 in the Northern Hemisphere (above) and its autocorrelation function along with the ones from short memory models (below). Source: Author’s own plot with data from HadCRUT4.