Identification of Seasonal to Decadal Controls on Phenology by Contrasting and Integrating Models, Datasets and Detection Methods

Friday, 19 December 2014: 2:25 PM
Matthias Forkel1, Nuno Carvalhais1, Mirco Migliavacca1, Kirsten Thonicke2, Sibyll Schaphoff2, Werner von Bloh2, Martin Thurner3 and Markus Reichstein1, (1)Max Planck Institute for Biogeochemistry, Jena, Germany, (2)Potsdam Institute for Climate Impact Research, Potsdam, Germany, (3)Stockholm University, Department of Applied Environmental Science, Stockholm, Sweden
Global land surface phenology regulates the climate system through the exchange of carbon, water and energy. Unfortunately, the climatic and ecosystem controls for seasonal, inter-annual and decadal dynamics of phenology are poorly understood. This lack of understanding is reflected in current dynamic global vegetation models that misrepresent vegetation phenology in comparison to satellite datasets of vegetation greenness. However, uncertainties on the spatial patterns and temporal dynamics of land surface phenology arise from the variety of datasets, and methods for time series smoothing and extraction of phenological events. Consequently, the application of dynamic global vegetation models to identify seasonal to decadal controls on phenology requires firstly model improvement, and secondly the consideration of uncertainties from datasets and detection methods. Here, we improved the LPJmL dynamic global vegetation model by implementing a new phenology scheme and by optimizing the model parameters against satellite datasets of vegetation greenness, albedo and gross primary production. We evaluated the phenology of LPJmL globally against three satellite datasets of FAPAR (fraction of absorbed photosynthetic active radiation) and by using ten methods for time series smoothing and phenology detection. Our results (1) demonstrate an improved performance of LPJmL with the new and optimized phenology model over the previous model version, and (2) show that the agreement between estimated start and end of season dates from LPJmL and the different satellite datasets is higher than the agreement between different datasets. Based on the improved LPJmL phenology, we quantify the effect of seasonal temperature, light and water controls and of processes as land use and land cover change, permafrost dynamics, fire disturbance and CO2 fertilization on the timing of start and end of the growing season. Our results demonstrate that water availability is an important seasonal phenological control not only in water-limited biomes but also in permafrost-dominated boreal forests and the arctic tundra. Besides seasonal cold temperature, seasonal light and water availability, and land use and land cover change were the prevailing controls on inter-annual variability and trends in land surface phenology.