Multi-Condition Evaluation of Sentinel-1 for Rubber Canopy: Normal, Leaf-Fall, and Pestalotiopsis Outbreak Ariq Anggaraksa Riesnandar (a), Masita Dwi Mandini Manessa (a), Milla Putri Melina (a), Farida Ayu (b)
(a) Department Geography, Faculty Mathematic and Science, Universitas Indonesia, Indonesia
(b) Graduate School of Sustainability Studies, University of Toyama, Japan.
Abstract
We assess rubber (Hevea brasiliensis) canopy monitoring at the Sembawa Rubber Plantation, South Sumatra, Indonesia, using Sentinel-1 C-band Synthetic Aperture Radar (SAR) dual polarization (vertical-vertical, VV- vertical-horizontal, VH). The target variable is percent Green Canopy Cover (GCC%), derived from expert interpretation of multispectral Uncrewed Aerial Vehicle (UAV) orthomosaics. After radiometric calibration, speckle filtering, and terrain correction, we built Ascending (ASC) and Descending (DSC) time series and computed six dual-polarization indices: Normalized Difference SAR Index (NDSI), Radar Forest Degradation Index (RFDI), Radar Vegetation Index (RVI), volume-to-surface scattering ratio (VR), cross-polarization ratio (CR), and the VV-to-VH ratio. Window-specific Pearson correlations between GCC% and radar variables were statistically significant but low. The strongest signals occurred in March 2022 (normal, post outbreak): VV-to-VH ratio (ASC) r = 0.318, VR (ASC) r = 0.292, NDSI and RFDI (ASC) r = 0.273, RVI (ASC) r = −-0.273, CR (ASC) r = −-0.254. During July 2024 leaf fall, ASC features weakened (absolute r ≈- 0.14-0.15). The February 2023 outbreak showed only modest links (absolute r ≤- 0.14). November 2024 was largely not significant. June 2022 had very weak ASC links and no DSC availability. A complementary global analysis yielded absolute r ≈- 0.13-0.29 with all p < 0.001. These significant but low linear associations indicate non-linear scattering-canopy relations and possible geometry effects. The identified feature set supports development of linear mixed-effects models and machine-learning approaches for plantation-scale canopy estimation and early detection of change under seasonal defoliation and disease stress.
