Optimizing UAV-based uncooled thermal cameras in field conditions for precision agriculture

Wan, Quanxing; Smigaj, Magdalena; Brede, Benjamin; Kooistra, Lammert · 2024 · DOAJ

DOI: 10.1016/j.jag.2024.104184

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the challenges of obtaining accurate land surface temperature (LST) measurements using uncooled thermal cameras mounted on unoccupied aerial vehicles (UAVs) for precision agriculture. While UAV-based thermal imaging offers potential for monitoring crop water stress, measurement accuracy is compromised by ambient environmental conditions and intrinsic sensor characteristics. The research aims to quantify the influence of flight altitude, meteorological factors (air temperature, humidity, wind speed, net radiation), and camera-specific dynamics on temperature readings, while evaluating calibration methods to improve data reliability. The experimental design utilized two DJI Matrice 210 quad-rotor UAVs equipped with different thermal camera systems: a FLIR Tau 2 and a WIRIS 2nd Gen, both sharing the same sensor core but employing distinct calibration mechanisms. Field campaigns were conducted in an experimental maize field in the Netherlands across three dates in 2022. The UAVs hovered at altitudes of 5, 10, 20, and 40 meters above diverse temperature reference materials, including painted panels, soil, sand, and water containers maintained at specific temperatures. Ground-truth data were collected using thermocouples positioned on these materials, synchronized with UAV image captures. Real-time meteorological data were obtained from a nearby weather station. Data processing involved converting raw digital numbers to kinematic and radiometric temperatures using emissivity values from the MODIS UCSB library, followed by linear regression analyses to assess the impact of environmental and sensor variables. Results indicate that increasing flight altitude significantly reduces measurement accuracy, characterized by a decrease in the slope of the linear regression between UAV and ground measurements. Higher altitudes caused the underestimation of temperatures for warm objects and overestimation for cold objects, with the coefficient of determination ($R^2$) dropping from 0.53 at 5 meters to 0.36 at 40 meters. Environmental factors exerted complex and model-specific influences. For the WIRIS 2nd Gen, wind speed positively correlated with measured temperatures, contradicting initial hypotheses, while humidity showed a negative correlation. In contrast, the FLIR Tau 2 demonstrated a negative correlation with wind speed and a positive correlation with air temperature, aligning more closely with theoretical expectations. However, environmental metrics explained only a small portion of the variance (4.2% for WIRIS and 11.0% for FLIR). Additionally, the study highlighted a correlation between measurement errors and the focal plane array (FPA) temperature, underscoring the importance of intrinsic sensor dynamics. The findings emphasize that optimizing UAV-based thermal imaging requires a comprehensive understanding of both ambient conditions and camera-model-specific behaviors. The discrepancies between the two camera models, despite sharing the same sensor core, highlight the critical role of proprietary calibration techniques. The study concludes that standardized field calibration protocols and corrections for atmospheric attenuation and sensor temperature are necessary to enhance the reliability of LST data for agricultural monitoring. These insights provide a foundation for improving the precision of thermal remote sensing applications in variable field conditions.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• dms validation