Assessment of high (diurnal) to low (seasonal) frequency variations of isoprene emission rates using a neural network approach
- ISSN: 1680-7316
- DOI: 10.5194/acp-8-2089-2008
Using a statistical approach based on artificial neural networks,\nan emission algorithm (ISO-LF) accounting for high to low frequency\nvariations was developed for isoprene emission rates. ISO-LF was\noptimised using a data base (ISO-DB) specifically designed for this\nwork, which consists of 1321 emission rates collected in the literature\nand 34 environmental variables, measured or assessed using National\nClimatic Data Center or National Centers for Environmental Predictions\nmeteorological databases. ISO-DB covers a large variety of emitters\n(25 species) and environmental conditions (10 degrees S to 60 degrees\nN). When only instantaneous environmental regressors (instantaneous\nair temperature T0 and photosynthetic photon flux density L0) were\nused, a maximum of 60% of the overall isoprene variability was assessed\nwith the highesTemissions being strongly underestimated. ISO-LF includes\na total of 9 high (instantaneous) to low (up to 3 weeks) frequency\nregressors and accounts for up to 91% of the isoprene emission variability,\nwhatever the emission range, species or climate investigated. ISO-LF\nwas found to be mainly sensitive to air temperature cumulated over\n3 weeks (T21) and to L0 and T0 variations. T21, T0 and L0 only accounts\nfor 76% of the overall variability.