Kolloquium: Jon Cranko Page

A growing body of research has identified that climate extremes can dramatically impact terrestrial ecosystem functioning on timescales that far surpass those of the extreme weather itself. Similarly, prior work has shown that daily carbon and water fluxes can be dependent on antecedent climate drivers even during non-extreme periods - for instance, winter rainfall can be a strong predictor for spring growth. However, gaps remain in our understanding of the influence of antecedent climate on terrestrial fluxes. Whether this memory of prior climate identified in growth records is also present in flux measurements is an open question. Similarly identifying the interactions between these structural lagged responses and the legacies to extremes is an area which has not been fully explored.

In this talk, I'll present my PhD work where data assimilation and machine learning were used to explore the sensitivity of FLUXNET measurements to lagged rainfall at various eddy-covariance sites. First, I will discuss results from a hierarchical Bayesian model, used to explore the role of antecedent drivers on NEE and evapotranspiration at Australian flux sites. I will then present my latest work, which utilised a k-means clustering plus regression approach to identify the impact of memory effects at 65 flux sites. Each method has their strengths and weaknesses, and provide further important insight into the role that ecosystem memory plays in terrestrial fluxes.