Premium
Global trends of aerosol optical thickness using the ensemble empirical mode decomposition method
Author(s) -
Zhang Zhao Yang,
Wong Man Sing,
Nichol Janet
Publication year - 2016
Publication title -
international journal of climatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.58
H-Index - 166
eISSN - 1097-0088
pISSN - 0899-8418
DOI - 10.1002/joc.4637
Subject(s) - hilbert–huang transform , climatology , moderate resolution imaging spectroradiometer , environmental science , spectroradiometer , empirical orthogonal functions , linear regression , mode (computer interface) , aerosol , trend analysis , autoregressive model , meteorology , geography , mathematics , geology , reflectivity , econometrics , statistics , satellite , optics , aerospace engineering , engineering , operating system , physics , white noise , computer science
ABSTRACT Aerosol trends and rates of change were analysed between 2003 and 2013, over both land and ocean, using MODerate resolution Imaging Spectroradiometer (MODIS) monthly aerosol optical thickness ( AOT ) products ( MYD08 ). Unlike previous research, the ensemble empirical mode decomposition ( EEMD ) was implemented in this study. Results show that sustained positive or negative trends are globally observed in most areas during the study period. However, increasing rates were decelerated and even became downward trends over western North America, central South America, East China Sea and southeastern China. Comparing EEMD results with linear regression, it is evident that the increasing and decreasing rates from the EEMD method are much stronger. Zonally averaged trends clearly indicate an opposite trend between the southern and northern hemispheres. In addition, this study demonstrates that linear regression may not fit trends statistically in some areas, such as central South America and part of the Indian Ocean. Around 32.74% (12 816) of pixels exhibit low correlation ( r 2 = 0.5) between linear and nonlinear trends from EEMD . Approximately 12.46% (4877), 6.56% (2567) and 1.85% (724) of pixels experience significant variations against the F ‐test, autoregressive process of the first‐order for EEMD and linear regression, respectively. The rates of change observed in this study can be used in analysing the long‐term effects of aerosols on climate change and earth's radiative budget.