Wavelet local multiple correlation analysis of long-term AOD, LST, and NDVI time-series over different climatic zones of India

Abstract

Atmospheric aerosols (aerosol optical depth, AOD) and green cover (normalized difference vegetation index, NDVI) significantly affect the radiation balance of a region and thereby modify the land surface temperature (LST). We have examined the long-term (2000–2017) temporal association between these variables using Wavelet Local Multiple Correlation (WLMC) analysis across six geographically separated areas representing different climatic zones of India. Spearman’s correlation between the variables indicates a mix of positive and negative correlations for varying seasons across the climatic zones. The non-stationary co-movement of multivariate correlation structure among the variables has been resolved by applying Maximal Overlap Discrete Wavelet Transform and WLMC analyses. Results show that the multivariate correlation integrates well beyond quarterly and biannual scales (16–32 weeks) for all zones. Daytime and nighttime LST explain the correlation structure in the data in zones from almost all climatic regions, except from central India where AOD and NDVI are the dominant variables. To some extent, NDVI plays an important role in eastern Indian region. The WLMC analysis confirms that the most reliable information in the multivariate spatial-temporal data at the regional scale can be suitably investigated. Regional climate models in this regard can further investigate the dynamics of the dominant variable in affecting the regional energy budget based on the WLMC analysis. The study has potential applications in forecasting extreme climate disasters and planning preemptive mitigation strategies.