Skip to main content
U.S. flag

An official website of the United States government

Publication Date
26 November 2017

Modeling Stochastic Phenology in Earth System Models

Using probablistic mathematics in integral projection models.
Print / PDF

We derived equations that represent organismal maturation as a function of variable environmental conditions, such as variation in temperature that capture the effects of random variation but which do not require computationally expensive Monte Carlo simulations (replications of variable simulations to incorporate random variation).


Phenology models are important tools for predicting the impacts of climate change on living organisms---especially in seasonally fluctuating environments in which the consequences of phenological mismatch can be dire. Our new method will allow computationally efficient simulation of stochastic phenology process models in earth system models, enabling more realistic simulation of climate-dependent outbreaks of destructive and dangerous insects.


In biological systems, environmental and individual-level variability can have profound impacts on the phenology and demography of living organisms. Ignoring these types of variability in earth system models can produce model prediction error. Incorporating stochastic variability in earth system models in computationally efficient ways that do not include simulation stochastic individual based models remains challenging. Although individual-based models have become a popular tool among ecologists, they are often infeasible in the context of large earth system models due to the nature of parallelization of earth system simulations and due to the computational cost of running a multitude of stochastic individual-based simulations. Thus, many earth system models can benefit from a method of accounting for the effects of stochastic variability without resorting to Monte Carlo simulations of stochastic processes. Our paper describes our new approach for obtaining phenology predictions using computationally efficient algorithms that implicitly account for variability in rate parameters without requiring stochastic individual-based simulation.

Point of Contact
Devin Goodsman
Los Alamos National Laboratory (LANL)
Funding Program Area(s)