Monitoring forest productivity by means of remote sensing, dendrochronology and eddy covariance measurements

Remote sensing and dendrochronology are widely used for indirect monitoring of forest productivity, but neither fully capture the complex ecosystem with different plant life forms and stratification. Remote sensing better corresponds to canopy biomass, and dendrochronology to timber increment. Both methods are appealing due to the ease of data collection (Vicente-Serrano et.al., 2016). Analysis of joint dynamics of Normal Difference Vegetation Index (NDVI) and tree ring growth and its comparison to corresponding values of Net Ecosystem Production (NEP) derived from eddy covariance measurements should verify accuracy of this methods for estimation of the whole ecosystem productivity (Tchebakova et. al., 2015).
Timber sampling was carried on in Central Forest Nature Reserve, Russia, at four coniferous forest ecosystems in close proximity to eddy covariance towers. For further analysis the curve of annual tree increment was split into low-, middle- and high frequency fluctuations. NDVI was calculated for each site using Landsat images 4-5, 8. Twenty three years of NDVI (1985 to 2015 with seven gap years) and 16 years of NEP (1998 to 2014) were used from 15-25 July, the period of highest vegetation.
The research revealed a negative correlation (0.5-0.75) between middle frequency dynamics of NDVI, which declined over time, and annual radial increment, which increased. High frequency fluctuations of parameters were independent, and no correlation of radial growth with NDVI of previous vegetation periods was found. Pair correlations of NEP to NDVI and NEP to tree ring growth were sufficiently high, justifying multiple regression analysis. The joint contribution of NDVI and annual radial increment to NEP describes only 56% of its variation. This proves reliability but inconsistency of both indirect methods, while eddy covariance measurements is inaccessible for wider use.