Variants on these models include the need for accumulated chilling in autumn
(again with a variable start date and threshold temperature) for vernalisation or the
balance of chill and heat units. A summary of models is given in Chuine et al.
(2003). In practice, temperatures are taken from a nearby met station and will be
recorded in a screen at a given height above the ground. As such, they will at best
approximate the temperatures the plant experiences. In woodland environments
there will be considerable differences in temperature from the woodland floor to
the canopy. Improvements to models may be achieved by using soil temperatures in
some circumstances (e.g. Sparks et al., 2005). Optimisation may reveal a range of
starting dates and threshold temperatures with similar properties, and the selection
of model parameters with wider applicability (‘portability’) should be sought. A
pragmatic attitude to modelling is essential.