Climate-Vegetation Interactions over Arid and Semi-Arid Regions: A Multi-Scale Causality Analysis

Abstract:

This talk explores the mechanisms underlying global-scale feedbacks of vegetation on climate, with a special focus on arid and transitional (semi-arid) regions and cross-scale interactions.

Whether precipitation and temperature are known to be two of the major drivers of ecosystem dynamics, the inference of the forcing of vegetation on climate from observed data still leads to extremely contradictory results. This is mainly due to the intrinsic complex and nonlinear nature of climate-vegetation interactions, which is exerted over a wide range of space, temporal and frequency scales. Beside, traditional statistical tools applied to these feedbacks rely on linear correlation measures that can hardly distinguish the different components of these interactions.

We analyze monthly and sub-monthly globally gridded data of precipitation, temperature and NDVI (from both MODIS and AVHRR) by using an ensemble of different directional coupling statistics, and spectral metrics able to resolve cross-scale interactions.

Based on the concept of Granger causality, we assess the bi-directional causal influences between precipitation, temperature and NDVI. In particular, we focus on spectral causality measures, in order to infer sub-processes acting across different time and frequency scales.

Several examples from arid and semi-arid regions are introduced and examined. During the discussion of the result, we highlight the strength and weakness of the approach, also in the occurrence of nonlinear couplings.