SAR L-Band Sensitivity: Detecting Early Deforestation
Figure 3. Early Deforestation detection results by sensor type, SAR (L-band) Detection, Optical (Landsat and Sentinel-2), and SAR and Optical combined. (Flores-Anderson et al., 2025)
|
Early Deforestation Detection in the Tropics using L-band SAR and Optical multi-sensor data and Bayesian Statistics
As described in the paper, we utilized L-band Synthetic Aperture Radar (SAR) to overcome the limitations of optical sensors in cloud-prone tropical regions, specifically targeting the detection of ''early deforestation''. This initial stage of forest loss—where trees are logged but left on the ground—is often invisible to optical sensors due to persistent cloud cover and to C-band radar due to its limited canopy penetration. By integrating ALOS-2 PALSAR-2 data into our Bayesian framework, we developed a two-tier approach using harmonic curve fitting and Z-scores to derive change probabilities. Our findings indicate that the Radar Forest Degradation Index (RFDI) is uniquely sensitive to these structural changes, allowing for detection with a mean time lag of just 16 days and a user’s accuracy of 99.19%. This methodology provides a critical foundation for using high-frequency L-band data from upcoming missions like NISAR to prevent illegal activities by identifying disturbances well before subsequent burning or complete biomass removal occurs. Paper :
|
SAR C-Band vs. L-Band: Distinguishing Disturbance Stages
Figure 4. Forest change processes as seen by optical data (Sentinel-2), Sentinel-1 C-band and ALOS-2 PALSAR-2 L-band. ALOS-2 Data ©JAXA [2018–2020]. All rights reserved. (Flores-Anderson et al., 2026)
|
On the sensitivity of SAR C- and L-band dual-polarized data for detection of early deforestation in the tropics
In this paper, our research specifically compared the sensitivities of C-band (Sentinel-1) and L-band (ALOS-2) SAR to different stages of tropical forest loss to guide more effective multi-sensor monitoring. We found that dual-polarized C-band has low sensitivity to the initial transition from stable to logged forest because its shorter wavelength cannot distinguish felled trees from intact canopies. However, C-band is highly effective at detecting later stages of deforestation, such as biomass burning or mechanical removal, where the signal becomes clearly differentiated. In contrast, L-band RFDI was able to correctly discriminate logged forest with 67% power, compared to only 34% for C-band. These insights demonstrate that an optimal forest monitoring system must be a multi-component framework that leverages L-band for early alerts and C-band for characterizing final land-cover conversion and vegetation regrowth. Paper:
|
SAR Fusion for National-Scale Cropland Tracking
Figure 7. The areas where cropland expanded in Zambia between 2000–2015, color-coded by the year of expansion, shown in relation to cropland established prior to 2000 (white), as mapped by the GSFAD cropland layer. (Xiong, S. et al., 2022)
|
Probabilistic Tracking of Annual Cropland Changes over Large, Complex Agricultural Landscapes Using Google Earth Engine
To overcome the lower density of imagery due to high cloud cover while leveraging the capabilities of SAR for improving cropland classifications, we applied a modified BULC-U (Unsupervised) workflow to monitor annual cropland expansion across the complex landscapes of Zambia, integrating a fusion of SAR and optical data to solve the problem of data sparsity in cloudy regions. This approach constructed a detailed satellite record, combining Landsat imagery with SAR data from Sentinel-1 and ALOS PALSAR, allowing for a more robust characterization of agricultural fields that are often spectrally similar to surrounding natural vegetation. By combining the structural information from SAR with optical vegetation indices, we identified cropland gain and used a shapelet-based thresholding method to estimate the precise year of expansion between 2000 and 2015. This study highlights that including active sensors is essential for creating spatially continuous, annual accounts of agricultural growth in remote or under sampled regions where traditional surveys are lacking. Paper:
|
