Continuous wavelet transformations of the first principal component of the central Siberian MXD data (top) and TRW (bottom) show significant power at decadal and multidecadal periods over the period 1880–1998

<p><strong>Figure 4.</strong> Continuous wavelet transformations of the first principal component of the central Siberian MXD data (top) and TRW (bottom) show significant power at decadal and multidecadal periods over the period 1880–1998. For instance, the red colors indicate persistent variability in tree growth at a period of decades (from about 16 to 32 years) in both TRW and MXD while the cooler colors indicate little in the way of power at higher frequencies (e.g., interannual). Color scale is by 10% quantiles.</p> <p><strong>Abstract</strong></p> <p>Different methods have been developed for measuring carbon stocks and fluxes in the northern high latitudes, ranging from intensively measured small plots to space-based methods that use reflectance data to drive production efficiency models. The field of dendroecology has used samples of tree growth from radial increments to quantify long-term variability in ecosystem productivity, but these have very limited spatial domains. Since the cambium material in tree cores is itself a product of photosynthesis in the canopy, it would be ideal to link these two approaches. We examine the associations between the normalized differenced vegetation index (NDVI) and tree growth using 19 pairs of tree-ring widths (TRW) and maximum latewood density (MXD) across much of Siberia. We find consistent correlations between NDVI and both measures of tree growth and no systematic difference between MXD and TRW. At the regional level we note strong correspondence between the first principal component of tree growth and NDVI for MXD and TRW in a temperature-limited bioregion, indicating that canopy reflectance and cambial production are broadly linked. Using a network of 21 TRW chronologies from south of Lake Baikal, we find a similarly strong regional correspondence with NDVI in a markedly drier region. We show that tree growth is dominated by variation at decadal and multidecadal time periods, which the satellite record is incapable of recording given its relatively short record.</p>