Cross-correlating deforestation and atmospheric indexes to local rainfall and temperature patterns through detrended multiple cross-correlation analysis

Abstract

Deforestation negatively influences rainfall, soil and air temperature. The objective of this work was to investigate the multiscale association of deforestation, Pacific Decadal Oscillation, Southern Oscillation Index and East Atlantic Oscillation with rainfall and temperature dynamics. Monthly records of rainfall and temperature from Belem and Manaus (Brazilian Amazonia) were analyzed. The deforestation series corresponds to the whole Amazonia area. Data sets spanned from April 2008 to December 2019. The methodology was based on detrended cross-correlation analysis, multiple detrended cross-correlation and multiscale joint entropy. The significance was assessed by generating probability distribution functions from 5000 simulated Gaussian signals. For Belem, the association of deforestation/EAO/SOI rendered the maximal coefficient value with 18 months as the time scale of maximal cross-correlation. In Manaus, 18 months was the time scale of maximal association with all signal pairs influencing rainfall dynamics significantly. The significance region spanned from 12 to 24 months for both sites. The deforestation/EAO pair showed the highest multiple cross-correlations with air temperature at all time scales larger than 12 months for Belem. However, multiple cross-correlations and significance regions were smaller for Manaus. The joint entropy was smaller than the sum of its marginal values which indicates dependence between the coupled processes. Rainfall/SOI and temperature/SOI pairs portrayed the more complex dynamics in Belem. Rainfall/PDO and temperature/SOI were more complex in Manaus. The entropy correlation coefficient suggests m=15 months as the scale of the deforestation footprint on rainfall patterns. Multiscale methods are better descriptors of coupled atmospheric/anthropogenic processes than those derived from traditional statistics.