The phenology of deciduous forests is similar year to year. In it’s simplest form, modeling the influence of climate on phenology can therefore be modeled as discrete events, such as leave emergence in the spring. However, the appearance of leaves for some vegetation types is less well-defined seasonally, because changes are gradual: continuous leaf growth in grasslands or slow changes in the leaf physiology in evergreen forests (more on evergreens in a later post). Modeling this ongoing phenology is more challenging, as it requires an understanding of the biological and physical processes that govern the yearly trajectory rather than just a discrete event, such as leaf emergence.
In the past one and a half years, my focus has been largely on modeling and characterizing these ongoing processes. Below you can see a set of example time series showing the proportion of ground area covered by green foliage (fCover). Using a simple bucket model, where pools of energy or mass are represented as simple buckets which are filled or emptied based upon physical principles, I’ve been able to model the trajectory of grassland growth throughout the year for various grasslands in North America based upon precipitation, temperature, and sunlight. Black dots are actual measurements of the fraction of green foliage from PhenoCam images and the red line is the result of the model that only knows about precipitation, temperature, and sunlight (and not the PhenoCam images). The blue bars show the amount of rainfall (or snowfall). And the gray graphs on the left show the amount of agreement between the black dots (data) and the red line (model).
These model results illustrate nicely how PhenoCam images hold the potential to help develop vegetation growth models. More so, these continuous processes are difficult to observe visually, as day-to-day changes are small, further highlighting the value of an image-based approach.