Uncertainty Assessment of NDVI Measurement with Radiometric Compensation by means of Monte Carlo Analysis

This paper introduces a workflow for evaluating uncertainty in Normalized Difference Vegetation Index (NDVI) measurements, enhanced by the integration of radiometric compensation to minimize the effects of atmospheric conditions. The workflow incorporates key factors impacting NDVI accuracy, such as solar irradiance, atmospheric conditions, and camera signal-to-noise ratio. A Monte Carlo simulation is employed to analyze sensitivity across two atmospheric scenarios: dry-clear, and humid-hazy. By integrating radiometric compensation, which uses Permaflect sheet reflectance curves and solar irradiance data, the model effectively compensates biases caused by varying weather conditions and light fluctuations. This compensation adjusts spectral data for the red and near-infrared bands, ensuring consistent NDVI calculations regardless of environmental conditions. The results show that the primary contributors to NDVI uncertainty are atmospheric effects, particularly humidity-haze, as well as camera wavelength tolerance. By compensating atmospheric influences, this workflow demonstrates that compensating for NDVI measurements enables the distinction between healthy and unhealthy plants under different environmental conditions. Additionally, the proposed methodology assesses the uncertainty of NDVI measurements, allowing for the estimation of expected results in advance. This provides a useful tool for designing flight missions and selecting the appropriate camera to be mounted on UAVs. Furthermore, the impacts of uncertainty sources on the radiometric compensation model are assessed, and calculated bounds quantify variations in its slope and offset.