Evaluation of Growing Season Milestones, Using Eddy Covariance Time-Series of Net Ecosystem Exchange

Thursday, 18 December 2014
Boris Faybishenko1, Gilberto Pastorello2, Cristina Poindexter2, Olaf Menzer3, Deb Agarwal2, Dario Papale4 and Dennis D Baldocchi5, (1)Lawrence Berkeley National Laboratory, Berkeley, CA, United States, (2)Lawrence Berkeley National Laboratory, Advanced Computing for Science, Berkeley, CA, United States, (3)University of California Santa Barbara, Santa Barbara, CA, United States, (4)Tuscia University, Department for Innovation in Biological, Agro-food and Forest systems (DIBAF), Viterbo, Italy, (5)University of California Berkeley, Berkeley, CA, United States
Common methods for determining timing of plants’ developmental events, such as direct observation and remote sensing of NDVI, usually produce data of temporal resolution on the order of one week or more. This limitation can make observing subtle trends across years difficult. The goal of this presentation is to demonstrate a conceptual approach and a computational technique to quantify seasonal, annual and long-term phenological indices and patterns, based on continuous eddy covariance measurements of net ecosystem exchange (NEE) measured at eddy covariance towers in the AmeriFlux network. Using a comprehensive time series analysis of NEE fluxes in different climatic zones, we determined multiple characteristics (and their confidence intervals) of the growing season including: the initiation day—the day when canopy photosynthesis development starts, the photosynthesis stabilization day - the day when the development process of canopy photosynthesis starts to slow down and gradually moves toward stabilization, and the growing season effective termination day. We also determined the spring photosynthetic development velocity and the fall photosynthetic development velocity. The results of calculations using NEE were compared with those from temperature and precipitation data measured at the same AmeriFlux tower stations, as well as with the in-situ directly observed phenological records. The results of calculations of phenological indices from the NEE time-series collected at AmeriFlux sites can be used to constrain the application of other time- and labor-intensive sensing methods and to reduce the uncertainty in identifying trends in the timing of phenological indices.