Seasonal dynamics of spectral vegetation indices at leaf, ecosystem and satellite scales for a boreal evergreen coniferous forest

The spectral vegetation indices (VIs) are widely used in ecology and ecosystem modelling to study carbon uptake and plant responses to climate change. VIs can potentially be used the learn about ecosystem processes at the large scale and used to inform and constrain mechanistic understanding and models. Key VIs such as Normalized Difference Vegetation Index (NDVI) reflects the chlorophyll contents, biomass, and canopy structural changes. The Photochemical Reflectance Index (PRI) and the Chlorophyll Carotenoid Index (CCI) relate to photosynthetic light-use efficiency (LUE) and also capture longer-term pigment changes of the vegetation at leaf and canopy scales, particularly for evergreen species. The Near-Infrared Reflectance of the vegetation (NIRv) relates to the canopy structure. The Water Index (WI) provides leaf water content information. However, the factors that control the seasonal changes of these VIs at different spatial-temporal scales is unclear, hence the question of whether VIs can successfully be scaled from leaf to satellite level remains to be answered. The main objective of this study is to examine, how and why the key VIs (NDVI, PRI, CCI, NIRv and WI etc.) change at the seasonal scale across leaf, ecosystem and satellite data.  

We use leaf-level measurements, continuous ecosystem observations and satellite data (atmospheric corrected MODIS products-MAIAC) across the spring recovery period of Scots pine (two years data) and Norway spruce (one year data) in a boreal site in Finland to answer: (1) how do VIs change during the photosynthetic spring recovery of the vegetation at leaf, ecosystem and satellite scales? (2) How do environmental and bio-physiological factors affect the seasonal dynamics of VIs? (3) do the main affecting factors change between canopy position and species? (4) whether the main factors change between spatial scales?   

Our preliminary results show that at the leaf level of Scots pine, both PRI and CCI are more strongly correlated with LUE at top-canopy (r = 0.92 and 0.93, respectively) than at low-canopy (r = 0.63 and 0.72) positions. At the leaf level in Norway spruce, only top-canopy PRI and CCI are significantly correlated with LUE (r > 0.75). When focusing on the correlations with PRI and CCI with pigments, we found that in Scots pine needles and for both top and low canopy, more than 80% of variation in PRI and CCI are explained by Car/Chl ratio and de-epoxidation state of xanthophyll cycle pigments (DEPS), respectively. However, in spruce for both canopy positions, the strongest correlation with PRI and CCI is lutein/Chl ratio (r is between -0.97 and -0.85), respectively), followed by Car/Chl ratio (r is between -0.84 and -0.72). At the ecosystem level, the PRI is correlated with GPP (gross primary productivity) when winter data and low PAR (<350 μmol m⁻² s⁻¹) is not considered (r = 0.63). The other VIs are under investigation and will also be presented. As a tentative conclusion, although optical properties covary with photosynthesis, mechanisms of variation appear species and light environment specific.