Kern, A; Marjanovic, H; Barcza, Z (2020). Spring vegetation green-up dynamics in Central Europe based on 20-year long MODIS NDVI data. AGRICULTURAL AND FOREST METEOROLOGY, 287, 107969.

The present study focuses on the leaf unfolding dynamics of deciduous broadleaf forests in Central Europe. MODerate resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) was used to quantify green-up duration (GUD) for the wider Carpathian Basin located in Central Europe, covering the time period 2000-2019. GUD was calculated for similar to 170 000 pixels with deciduous broadleaf forest cover at 500 m spatial resolution. The GUD exhibited large interannual and elevation-dependent variability where the latter likely indicates the distribution of the different species. The longest mean GUD occurred in 2017 (32.7 days), while the shortest (14.5 days) was associated with 2018. The relationship between the start of leaf unfolding (SOS) and the GUD (R = -0.62) reveals that the timing of the bud break plays an important role in the leaf unfolding process. Multiple linear regression models were constructed to explain and forecast the GUD based on the date of SOS, the elevation and the meteorological variables. The main explanatory variable was the SOS date, explaining 38.3% of the GUD variability (RMSE= 8 days), while the addition of the elevation and its square to the model increased the explained variance to 47.8% (RMSE = 7.34 days). Further addition of meteorological variables covering periods prior to and after the SOS increased the explained variance (the best R-2 was 0.65). The results indicate the complexity of processes that drive the leaf unfolding. We propose that future studies should consider SOS, elevation and meteorological variables together to interpret GUD dynamics. Earlier SOS implies longer GUD, while delayed SOS is associated with short GUD, which means that the benefits of climate change possibly realized as a longer growing season could be smaller than anticipated. Accurate estimation of the SOS is a prerequisite for the successful modeling of the GUD.