Calibration of multi-spectral photosynthetically active radiation sensor

Photosynthetically active radiation (PAR) sensors are widely used in plant stress monitoring, ecophysiology, and forest growth modeling. Recently, multi-spectral PAR sensors have emerged as practical alternatives to established single-channel systems. These sensors offer significant potential, given their competitive accuracy to commercial systems and additional capability to obtain spectral information. This work introduces a new calibration framework specifically tailored for multi-spectral sensors. Previous approaches were only accurate and reliable under identical light conditions. Our calibration setup within a controlled environment ensures consistency and reproducibility of the sensor calibration in every ambient light condition. A dedicated data acquisition and preprocessing framework accounts for spectral characteristics of each component, forming the basis for the Partial Least Squares (PLS) calibration model. This fits the multi-spectral data to the ideal quantum response and enables robust modeling even in case of high collinearity among input variables. With only one calibration coefficient for each channel, a highly efficient implementation of the PAR calculation is possible. Moreover, the method compensates for the spectral leakage of the NIR in the visual range channels. The spectral accuracy of the calibration model is demonstrated by presenting, for the first time, the quantum response curve of a calibrated multi-spectral PAR sensor. In addition, a field validation experiment is conducted under variable weather conditions to assess the absolute performance. With an error of 2.43% relative to a commercial sensor, this work highlights the accuracy achieved with the proposed calibration method. For further investigation, this work provides a repository with data from our calibration setup. • Calibration setup for multi-spectral sensors regardless of environmental conditions. • Presentation of the sensors competitive quantum response curve. • Validation of the calibration in a field experiment. • Development of preprocessing framework that considers the spectral characteristic of each device. • Implementation of an algorithm designed to mitigate spectral leakage in NIR.