Identifying influences on model uncertainty: an application using a forest carbon budget model

Area of Science:

• Environmental science
• Forestry
• Climate modeling

Background:

• Environmental simulation models require uncertainty quantification for reliable predictions.
• Estimating uncertainty for deterministic models is difficult with scarce foundational data.
• Forest carbon budget projections are crucial for climate change assessments.

Purpose of the Study:

• To demonstrate a probabilistic uncertainty analysis method for refining input uncertainty quantification.
• To assess uncertainties in forest carbon budget projections using the FORCARB model.
• To identify key factors influencing model uncertainty even with limited initial information.

Main Methods:

• Application of probabilistic uncertainty analysis.
• Monte Carlo simulations using the FORCARB model.
• Analysis of model sensitivity to probability density function characteristics (range, shape, covariability).

Main Results:

• The study successfully refined input uncertainty quantification with limited data.
• Model sensitivity was identified concerning the range, shape, and covariability of probability density functions.
• Covariability among model parameters significantly influenced uncertainty, particularly for average annual carbon flux estimates.
• The distributional form of probabilities was less critical than ranges and covariability.

Conclusions:

• Probabilistic uncertainty analysis is effective for refining estimates even with scarce data.
• Parameter covariability and ranges are critical drivers of uncertainty in forest carbon models.
• Understanding parameter interactions is essential for accurate carbon budget projections.