Land surface temperature and emissivity are essential for understanding ecosystem energy balance, evapotranspiration, vegetation health, and water use. Yet satellites do not directly measure true surface temperature; they record radiance that must be converted through inversion models. This becomes especially difficult in forests, where 3-dimensional (3D) canopy structure makes thermal signals strongly directional. New high-resolution thermal infrared satellite missions are increasing the need for more reliable emissivity estimates, because viewing-angle effects alone can introduce large uncertainties, sometimes several kelvin. Based on these challenges, deeper research is needed on how forest structure controls directional thermal emissivity.
Published (DOI: 10.34133/remotesensing.0738) on 19 February 2026 in Journal of Remote Sensing, the study was conducted by researchers from Université de Toulouse, Beijing Normal University, The University of Hong Kong, Jilin University, Hong Kong Polytechnic University, and partner institutions. The team investigated how forest architecture affects directional thermal infrared emissivity, a key variable for converting satellite radiance into reliable land surface temperature. Their work addresses a major challenge for next-generation thermal missions that seek more precise forest monitoring and climate-relevant observations.
The researchers used the DART 3D radiative transfer model to simulate directional emissivity across eight realistic RAMI-V forest scenes. Two independent DART-based methods produced nearly identical results, confirming the model's reliability. By contrast, three commonly used analytical models were less accurate; FR97 performed best among them, but still showed notable errors at some sites. Across the eight forest types, directional emissivity ranged from 0.972 to 0.996, a spread large enough to produce forest temperature retrieval errors greater than 1 K. The study also found that emissivity generally increased with view zenith angle and remained azimuthally symmetric unless forests were sparse or trees were arranged asymmetrically, such as in rows.
The analysis revealed that canopy architecture strongly controls thermal behavior. Forests with higher leaf area index and more homogeneous tree distribution, such as HET07, HET09, and HET51, showed relatively high mean emissivity values around 0.994. In contrast, row-planted or more open stands such as HET14 and HET16 were closer to 0.985, while winter pine stands with low leaf area index fell to about 0.980 or even 0.974. Reducing tree density in one birch forest case caused emissivity to drop sharply, from about 0.994 to 0.986 or 0.976, depending on removal intensity. The model also showed that neglecting trunks and branches can misrepresent directional behavior, especially at oblique viewing angles. Jacobian analysis further quantified how leaves, wood, and ground each contribute to emissivity changes under different canopy structures.
No direct author quotation is provided in the paper, so the following is a press-style paraphrase based on the authors' discussion: Our results show that forest thermal emissivity cannot be treated as a simple surface constant. It is shaped by canopy architecture, viewing geometry, and the relative contribution of leaves, wood, and ground, making accurate 3D modeling essential for remote sensing and climate applications.
The team modeled eight 100 m × 100 m RAMI-V forest scenes with contrasting structures, including pine, birch, citrus orchard, poplar, savanna, and temperate forest cases. They used DART-Lux, a Monte Carlo version of the DART model, to simulate directional radiance, reflectance, and emissivity over the thermal infrared domain. Simulations were run at 10 μm using fixed emissivity values of 0.98 for leaves, 0.95 for soil, and 0.94 for wood. The study also computed Jacobian maps to measure how changes in reflectance from leaves, ground, and woody components affect directional emissivity.
This work could help improve land surface temperature retrieval for upcoming thermal satellite missions by better accounting for forest structure and observation angle. It also provides a stronger physical basis for multi-angular emissivity correction, especially in forests with sparse cover, low leaf area, or row planting. In the longer term, the approach may support more reliable ecosystem monitoring, climate modeling, and precision observation of forest stress, energy exchange, and land-atmosphere interactions at high spatial resolution.
###
References
DOI
10.34133/remotesensing.0738
Original Source URL
https://spj.science.org/doi/10.34133/remotesensing.0738
Funding information
This work is funded by the National Key Research and Development Program of China (grant no. 2023YFF1303601), the National Science Foundation of China (grant nos. 42090013, 42130104, and 42271338), the China Scholarship Council, and the TOSCA program of the French Space Agency (CNES).
About Journal of Remote Sensing
Journal of Remote Sensing, an online-only Open Access journal published in association with AIR-CAS, promotes the theory, science, and technology of remote sensing, as well as interdisciplinary research within earth and information science.