Extended Data Fig. 4: Illustration of the issue with comparatively low measurement resolution for X-ray MXD.
From: Fennoscandian tree-ring anatomy shows a warmer modern than medieval climate
a) X-ray image with analysis track path indicated within the solid white rectangle, and examples of the effect of different effective measurement resolutions. b) The photosensors in a) build up measurement profiles, where the blue sensor builds the blue profile corresponding to a 20-micron effective measurement resolution, and the orange sensor builds up the orange profile corresponding to a 60 micron effective measurement resolution, approximating the effective measurement resolution of the X-ray methodology29. Note how the time series of MXD reflect inverse variations if developed using high-resolution or low-resolution equipment, i.e., the middle ring exhibits the lowest or highest value depending on resolution. The explanation for this is that very narrow latewood widths are associated with comparatively lower MXD values even though the “true” MXD value may be high. c) Relationships between TRW and A-FEN and d) LWW and A-FEN. e) Relationships between TRW and anatomical MXD (MXDCWT) and f) LWW and MXDCWT. g) Relationships between TRW and X-FEN and h) LWW and X-FEN. All datasets display correlations and using datapoints covering 850–2005 CE. Note how the X-FEN always is stronger correlated with TRW and LWW than the MXDCWT. A higher correlation is expected if TRW or LWW is affecting the measurement. Spearman rank correlation coefficients were used due to the possibly non-linear relationships between width and density. Rraw and rdiff. refers to untreated and first differenced data prior to correlations, respectively.
