PhenoForecast: Ecological forecasting of leafing and flowering phenology during climate change to inform public health

Yiluan Song, Stephan B. Munch, Kai Zhu
AGU presentation 2021, EFI 2022 presentation (EFI Futures Outstanding Presentation Award)

Highlights

  • We aimed to deliver near-term forecasts of vegetative and reproductive phenology of wind-pollinated plants that induce allergy.
  • We leveraged diverse phenology data including remote sensing, phenophase observations, and airborne pollen concentrations.
  • We used spatiotemporal Gaussian Process Empirical Dynamic Modelign (GPEDM) to forecast maps of phenology in the continental US.
  • We implemented infrastructure to deliver forecasts every day for the following 35 days on our web app, PhenoForecast.

Background

MY ALT TEXT

Synthesizing phenology data from multiple sources.

Methods

MY ALT TEXT

Infrastructure to collect data, fit models, make predictions, and deliver forecasts.


MY ALT TEXT

Four levels of the model behind PhenoForecast and key data sources.


MY ALT TEXT

Steps in each level of the model.

Product

Evaluation

Home