4.1 | Strengths and weaknesses of PRELES
The key advantages of PRELES are that it is a compromise between predictive accuracy and model complexity. It can be calibrated with a few variables derived from EC measurements and hence run with a few environmental variables. The required EC data are available from many sites around the world (Baldocchi, 2003). PRELES has been reported to work well in all boreal forests (Minunno et al., 2016, Tian et al., 2020). Based on this assessment, we felt justified in using it, with calibration from 2014-2017 and environmental data from 2012-2013, to model carbon fluxes in 2012-2013.
Although the availability of EC data is an advantage for PRELES, EC data must be viewed with caution. In particular, at our site, preliminary analyses of the data revealed significant problems in the data despite the flat ground surface, uniform canopy, and low leaf area index. A careful study of the problem revealed significant decoupling of the above- and below-canopy air masses, which often led to advection (Jocher et al., 2018). It is common for EC studies to use a vertical wind speed cutoff, the u* filter, to detect and remove such events (Aubinet et al., 2001; Papale et al., 2006). We found that the u* filter was insufficient and that a measurement relying on the comparison of below-canopy and above-canopy vertical wind speeds was required (Jocher et al., 2018). This concern was earlier raised in another boreal forest in Finland (Alekseychik, Mammarella, Launiainen, Rannik, & Vesala, 2013). We used a decoupling filter (Thomas, Martin, Law, & Davis, 2013), which is still unusual in the EC community, to correct the EC data that were used to parametrize PRELES in this study. Unless this correction was performed, PRELES would have been parameterised incorrectly if we wished to quantify total ecosystem fluxes; it would only have described the decoupled fluxes.