You are here

Harvard Forest >

Harvard Forest REU Symposium Abstract 2023

  • Title: Characterizing Uncertainty in Ecological Forecasts between Plant Functional Types
  • Author: Isa Marian Kazen (The University of Texas - Austin)
  • Abstract:

    In recent decades, the global community has widely acknowledged the ecological consequences tied to decision-making in human spheres such as industry, agriculture, and geopolitics. With this has come an increased interest in quantifying the efficiency of forests as carbon sinks. Ecological forecasting is a growing research area that has made it possible to predictively trace the terrestrial carbon cycle. Its success, however—in both creating reliable predictions and supporting improved decisional outcomes—depends on a robust understanding of uncertainty as it propagates from input to output. One source of such uncertainty for the SIPNET forecasting model arises from variability within parameters (distributions of size n=25), which differ between Plant Functional Types (PFTs). Here, we assess how the uncertainty associated with tree growth and soil respiration forecasts differs between temperate deciduous and conifer PFTs. To do this, we validated these forecasts against field measurements taken with dendrometer bands for tree growth, and a Picarro Mobile Gas Analyzer for soil respiration. Uncertainty was calculated using root mean squared error (RMSE) and mean absolute error (MAE). We then performed a sensitivity analysis of the tree growth and soil respiration forecasts to deciduous and conifer parameters. Using this information to identify the most influential parameters for each forecast and PFT, we calculated the coefficient of variation for each parameter distribution to determine its variability. Preliminary results indicate greater variability in SIPNET’s conifer parameters, and we anticipate that overall uncertainty will be higher in forecasts made for coniferous PFTs, compared to forecasts made for deciduous PFTs.

  • Research Category: Ecological Informatics and Modelling; Soil Carbon and Nitrogen Dynamics