Improved space-time mapping of PM2.5 distribution using a domain transformation method

Abstract

The present work uses a new space-time projection (STP) technique to study the distribution of PM2.5 concentrations in one of the most populous, highly developed and highly polluted regions in China, the Shandong Province, during the period Jan 1–31, 2014. The theoretical and interpretational features of the STP technique are pointed out. A key feature of the technique is that it transfers the study of pollutant distribution from the original space-time domain R2×T onto a traveling domain of lower-dimensionality R2 that moves in the pollutant spread direction; analysis and computations are much easier in the R2 domain, avoiding the complexities of the R2×T domain; and the results are back-transformed to the original R2×T to generate space-time PM2.5 concentration maps over the entire region of interest. The Shandong study shows that the proposed STP technique has certain noteworthy analytical and computational advantages over mainstream mapping techniques of higher dimensionality (like space-time ordinary kriging, STOK): it avoids serious difficulties associated with space-time metric determination and variogram estimation in the original space-time domain, it allows the selection of more appropriate variogram models representing the PM2.5 variation, it generates more accurate PM2.5 maps, and it is also a computationally more efficient technique than the STOK technique.