cardidates: Methods for Peak Identification in Environmental Time Series
Phenology and seasonal succession in aquatic, and of course also terrestrial, ecosystems are strongly dependent on physical factors. In order to promote investigations into this coupling, objective and reliable methods of characterising annual time series are important. The cardidates package provides methods for fitting "peaks" of environmental data and approaches for an "objective" characterisation of what we call "cardinal dates", i.e. beginning, time of maximum and time of end of an identified peak. Objectivity means that it is not necessarily important to get good estimates for one particular time series but to get reproducible results independently of the person who performs the analysis.
The proposed methods were developed within the AQUASHIFT research program and used to determine the beginning, maximum and end of the spring mass development of phytoplankton in different lakes and water reservoirs. These time points, that we call "cardinal dates", can be analysed for temporal trends and relationships to climate variables. The complete methodology is described in Rolinski, Horn, Petzoldt, and Paul (2007). Until now we implemented only the most reliable approach using Weibull-functions (Method B in the article); other functions may follow.
This method consists of fitting four- resp. six parametric Weibull functions to environmental time series. Heuristics are applied for finding initial parameters in the six parametric case. As a second step, the fitted functions can be used to identify maximum, beginning and end of peaks by using a quantile-like approach.
Examples about the intention and use of the package can be found in the Quickstart Manual.
The methodology may also be useful for other ecological time series (e.g. bacteria, protozoa, insects or small mammals). Please don't hesitate to contact the authors if you feel that this package should be generalized to other processes.
You can download compiled packages for different operation systems (Linux, Mac, Windows) or inspect source code and documentation online before downloading.
Details can be found on the project summary page.