Next Generation Data Model for Phenology Observations

In most, if not all, previous work to collect phenological observation data, the emphasis has been on recording the single date on which a particular phenological event (or “phenoevent”), such as the first flower or full leaf drop, occurred. However, to understand the uncertainty in this date, it is also necessary to record the dates on which observations were made and the phenoevent had not yet occurred. As discussed in other presentations, the USA-NPN has created a data model and user interface where observers report the state of specific life stages (or phenophase) as occurring or not occurring for each plant being observed. The date for a particular phenoevent can then be calculated from the observations, along with an estimate of the uncertainty in the date of the event. This change in the way the observational data is collected and recorded makes substantial changes in the underlying data model for phenological observations and enables a number of new capabilities. If an observation is viewed as a set of yes/no/didn’t look answers to the questions of whether the life stages are present, then the responses represent the state of the system, and mathematics and software patterns of finite state theory apply. Biological knowledge that certain states do not occur for particular plants can be used as an input validation/quality control method. Likewise a particular species will typically transition from state to state in particular patterns. An observer reporting a transition which does not normally occur can be used to prompt for further questions (such as whether an anticipated phenoevent was missed) or to provide insight into extreme weather events (such as the severe frost in the Midwestern and Southeastern US in April 2007). In this poster, we will present further details of the underlying data model, the ways in which finite state theory and software patterns can be used, and the methods for comparing phenological data collected under phenoevent-based protocols with this state-based phenophase protocol.