Biogeophysical effects of land use and land cover change not detectable in CMIP6 warmest month.

In addition to their changes to carbon pools, land use and land cover changes (LULCC) alter climates biogeophysically through their effects on surface fluxes for energy and water. These resulting climatic changes in temperature manifest differently across study type (observational or model-based) and spatial-temporal scales depending on the predominance of albedo, evaporative fraction and surface roughness as causal factors. With growing future demand for land-exhaustive activities to address societal needs and interest in mitigation strategies involving reforestation/afforestation, it is important to understand how past LULCC contributed to climate change. Here we assess the prevalence of the historical LULCC signal in the warmest average month of daily maximum temperatures using regularized optimal fingerprinting for detection and attribution. We use the simulations of four global climate models from CMIP6 historical and hist-noLu experiments and separate observations from the Climatic Research Unit and Berkley Earth. Aggregating data according to the new IPCC AR6 reference regions and regressing observations onto hist-noLu and lu (historical - hist-noLu) in a 2-way regression, we find that LULCC is not sufficiently detectable at continental and global scales for four GCMs and their multi-model mean. This is confirmed by the nearly unchanging detectability of historical climate change in separate 1-way regressions for historical and hist-noLu. To further explore lu, we confirm the predominance of noise and the non-local effects of LULCC over its local effects by finding insignificant signal-to-noise ratios for the fingerprint of forest cover on lu using a principal component analysis.