Intelligent calibration of hyperspectral systems: The Adaptive Calibration Cycle (ACC) concept

Modern unmanned aerial vehicles (UAVs) equipped with hyperspectral cameras are becoming key tools in precision agriculture and the monitoring of agricultural ecosystems. However, despite the increasing accuracy of sensors, a methodological issue remains unresolved – the static nature of field calibration procedures. Traditional approaches based on one-time reference measurements fail to ensure data reliability under variable conditions of illumination, soil moisture, and atmospheric factors. This article aimed to present a conceptual model – the Adaptive Calibration Cycle (ACC) – a self-learning system that integrates the stages of data acquisition, calibration, and processing into a unified closed-loop framework with continuous feedback. The research methodology was based on simulation of calibration processes using secondary empirical data, a comparative analysis of static and adaptive approaches, and an evaluation of ACC performance according to key metrics such as reflectance error, radiometric stability, and data reproducibility. The algorithmic implementation of the cycle employed online learning mechanisms, a Kalman filter, and an edge computing architecture for real-time correction. Modelling results demonstrated that implementing ACC reduces average reflectance error by more than 70%, increases radiometric stability by 20-25%, and shortens response time to 0.25 seconds. In agricultural applications, this ensures more accurate determination of vegetation indices (NDVI, PRI), timely detection of plant stress, and optimisation of irrigation and fertilisation. The proposed methodology represents a transition from a static to an adaptive approach in field spectrometry and opens up new opportunities for intelligent remote monitoring systems in the agro-industrial sector, ensuring high precision, reproducibility, and data stability