Spectral Decomposition and Signal Separation of Climate Responses to Land Cover Changes

Abstract

While large-scale afforestation and reforestation are heavily discussed as strategies for nature-based climate change mitigation and adaptation, massive deforestation is ongoing. Such widespread land use and land cover changes (LULCCs) do not only alter the global climate through biomass carbon uptake or release but also through biogeophysical (BGP) processes related to changes in surface roughness, evaporation, transpiration, and albedo. These BGP effects act as local forcing to land-atmosphere interactions and lead to in situ climate responses. Caused by advection and spatio-temporal land-atmosphere-ocean interaction, they also generate non-local climate responses that occur remotely from the LULCC. How the non-local partition of climate response signals emerges at different spatial scales than its forcing is still object of ongoing research. Here, we present a spectral perspective on climate responses to LULCC forcing that aids in achieving systematic understanding. We introduce spectral decomposition of forcing and response fields into a sum of signals with different wavelengths based on spherical harmonics to compare the two fields across spatial scales. Building on this approach, we develop a novel tool called SCISSOR, a Spectral ClImate Signal SeparatOR to define the cross-scale response signal from the difference of response and forcing spectra. The cross-scale signal of BGP-driven temperature response to deforestation strongly resembles the non-local signal as estimated by the established moving window regression and checkerboard interpolation. We show the utility of spectral methods for model intercomparison and assess their potential for future studies aiming to understand the complex interaction between climate and land surface changes.