Analysis of crop species and varieties using airborne long-wave infrared hyperspectral imaging: a case study at Bernburg-Strenzfeld

Drought-induced stress of crops increasingly threatens agricultural yields and consequently food production security, which becomes even more challenging due to growing climatic instability. Consequently, the early detection of water-stress-related responses in crops is important to administer precise irrigation as well as for identifying varieties resilient to drought.

While multi- and hyperspectral remote sensing in the visible, near-, and short-wave infrared (VNIR/SWIR, 0.4–2.5 µm) is an established and robust tool for spatially assessing and monitoring vegetation vitality, less focus has been given to high-resolution spectral data covering the long-wave infrared (LWIR) so far. However, advancements in airborne sensors close this gap and allow for capturing detailed spectral information of vegetation components that are sensitive to water stress and show their fundamental vibrational features in the LWIR. Against this background, this case study evaluates the potential of airborne hyperspectral LWIR emissivity and temperature data to differentiate crop species and varieties.

The experimental setup is located at the Strenzfeld agricultural test site close to Bernburg, Central Germany, and comprises 32 plots, each approximately 67 x 9 m. The study includes three crop species (peas, winter wheat, and summer barley) with two varieties each, planted in four replicates, alongside eight bare soil plots. Hyperspectral LWIR data (7.4–11.8 µm, spectral resolution 6 cm^-1, spatial resolution 0.77 x 0.77 m) were recorded on 6 May 2025 using a Telops Hyper-Cam Airborne Mini. Data preprocessing, including geometric corrections and data cube mosaicking, was conducted using Reveal Airborne Mapper, while temperature-emissivity separation was employed via Reveal FLAASH-IR. Additionally, UAV-based broadband thermal data and RGB orthomosaics were acquired with DJI Zenmuse XT2 and DJI Zenmuse H20T sensors to coincide with the aircraft overpass.

Emissivity spectra and temperature data were analysed at the plot-level to identify crop-specific spectral features and assess inter- and intra-class variations. Principal Component Analysis (PCA) was used to explore clustering within the spectral data. To account for differences in vegetation cover and the background soil signal, (partial) unmixing approaches exploiting vegetation and bare soil emissivity spectra were used as well as spectral indices. Furthermore, an inter-comparison of the temperature values derived from the Hyper-Cam Airborne Mini and the DJI Zenmuse XT2 was performed.

The findings of this case study contribute to a better understanding of LWIR emissivity signatures of different crops and their variability. Initial results show that in addition to crop-specific traits, vegetation cover and thus the soil signal distinctively impact the observed emissivity and temperature values. This highlights the importance of selecting optimal phenological windows for data acquisition. A planned follow-up study will incorporate multi-temporal airborne LWIR data acquisitions and controlled irrigation experiments in order to identify crop varieties with increased drought-resilience.

This research is funded by the German Research Foundation (DFG, grant number: 514067990) and by the Federal Ministry of Agriculture, Food and Regional Identity (BMLEH, grant number: 28DE205A21).