Propagates uncertainty

Option	Description
Inherent capacity	Fully incorporates input and parameter uncertainty into output intervals.
Model-specific tools	Specialized extensions or package functions (e.g., bootstrap routines, sandwich variance estimators, delta method) are available to propagate uncertainty for that method.
Needs model-agnostic propagation	Requires external, generic workflows (e.g., Monte Carlo simulation, bootstrap wrappers) to track and quantify uncertainty across model outputs.

Definition

The method’s built-in ability to carry the quantified uncertainty of inputs (e.g., measurement error, parameter estimation variance) through the analysis to produce valid confidence or credible intervals on outputs.

Explanation

Proper uncertainty propagation prevents overconfident conclusions by ensuring that input errors and model uncertainties are reflected in the final inference. Some methods do this inherently; others require external tools or custom workflows. Methods lacking this aspect provide point estimates without a realistic appraisal of their reliability.

Tools/rationale for helping assessment

1. Decide if you require your RS measurement error (e.g. ±σ per pixel) to flow into your final effect estimates.
2. If you need full integration with no extra coding, choose Inherent capacity; if availability via method‐specific additional routines suits you (e.g. bootstrap), choose Model-specific tools; if you plan a separate Monte Carlo wrapper, choose Needs model-agnostic propagation.

Example

Your LiDAR canopy‐height σ is 0.3 m/pixel and you want those errors to propagate in your biomass estimate’s CI: i) with no time-consuming coding efforts: you pick Inherent capacity; ii) you are ok with some additional dedicated function calls and parameter tweaking: Model-specific tools.