HR: 0800h
AN: A51B-0049    [Abstracts]
TI: Assessment of Isoprene Emission Variations: Toward a New Modelling Approach
AU: * Boissard, C
EM: boissard@lisa.univ-paris12.fr
AF: Laboratoire Interuniversitaire des Systèmes Atmosphériques, Universités Paris 7 & 12, 61, avenue du Général de Gaulle, Créteil, 94010 France
AU: Chervier, F
EM: chervier@lisa.univ-paris12.fr
AF: Laboratoire Interuniversitaire des Systèmes Atmosphériques, Universités Paris 7 & 12, 61, avenue du Général de Gaulle, Créteil, 94010 France
AU: Dutot, A
EM: dutot@lisa.univ-paris12.fr
AF: Laboratoire Interuniversitaire des Systèmes Atmosphériques, Universités Paris 7 & 12, 61, avenue du Général de Gaulle, Créteil, 94010 France
AB: MCT model actual biases are believed to mainly rely on the uncertainties in assessing accurate and highly resolved emissions of, in particular, volatile organic compounds (VOC). Biogenic VOC (BVOC) emissions exceed their anthropogenic inputs and have been shown to be highly variable in space and in time, reflecting both, immediate and long term plant adaptations to highly variable ambient environmental conditions. Due to the complexity of the processes involved into the seasonal regulation of BVOC emissions, our current understanding remains uncertain. Thus, available BVOC emission models, such as the G95 model, mainly describe high (minutes to hour) frequency variations, although lower frequency variations (e.g. seasonal) were shown to significantly account for in the overall observed variability. Using a neural network approach, a broad parameterization of isoprene emissions accounting for high to low frequency variations was assessed. Based on previously published data, a specific isoprene data base (ISO-DB, n=1332) was created for this work. ISO-DB covers emission rate values obtained for a large variety of emitters (mainly high emitters) and measured under a large diversity of environmental conditions, together with 34 measured and/or assessed low to high frequency environmental regressors. Medium (day d-1) and high (instantaneous) frequency environmental parameters (temperature and light intensity) alone were shown to be significantly insufficient in describing the overall isoprene emission variability. Assessing best nonlinear regressions between the emission rates and 19 of the environmental regressors, an isoprene algorithm (ISO-LF) accounting for high to low frequency variations of ambient air and soil conditions was obtained. ISO-LF was shown to describe more than 90% of the considered isoprene emission variability (r2=0.91) compared to 41% for the G95 model. ISO-LF was found to be mainly sensitive to long term air and soil temperatures and to long term deep soil water content. Moreover, validation of ISO-LF on non stored monoterpene emissions was shown to give poor results.
DE: 0315 Biosphere/atmosphere interactions (0426, 1610)
SC: Atmospheric Sciences [A]
MN: Fall Meeting 2005