In Thailand, crop depredation by wild elephants intensified, impacting the quality of life of local communities and long-term conservation of wild elephant populations. Yet, fewer studies explore the landscape-scale spatiotemporal distribution of human–elephant conflict (HEC). In this study, we modeled the potential HEC distribution in ten provinces adjacent to protected areas in Eastern Thailand from 2009 to 2018. We applied the time-calibrated maximum entropy method and modeled the relative probability of HEC in varying scenarios of resource suitability and direct human pressure in wet and dry seasons. The environmental dynamic over the 10-year period was represented by remotely sensed vegetation, meteorological drought, topographical, and human-pressure data. Results were categorized in HEC zones using the proposed two-dimensional conflict matrix. Logistic regression was applied to determine the relevant contribution of each scenario. The results showed that although HEC probability varied across seasons, overall HEC-prone areas expanded in all provinces from 2009 to 2018. The largest HEC areas were estimated during dry seasons with Chantaburi, Chonburi, Nakhon Ratchasima, and Rayong provinces being the HEC hotspots.However, the HEC potential was reduced during severe and prolonged droughts caused by El Nino events. Direct human pressure caused a more gradual increase of HEC probability around protected areas. On the other hand, resource suitability showed large variation across seasons. We recommend zone-dependent management actions towards a fine-balance between human development and the conservation of wild elephants.

Rice cultivation is a major source of methane (CH₄) emission in the agricultural sector, with a significantly higher global warming potential than carbon dioxide. Accurate and scalable quantification of CH₄ from rice paddies is essential for carbon accounting. This study presents an automated framework for estimating rice CH₄ emissions from irrigated paddies in the central plain of Thailand, integrating multi-sensor Synthetic Aperture Radar (SAR) observations with the IPCC methodology. The framework combines Sentinel-1 C-band SAR time series for phenological detection, ALOS-2 PALSAR-2 L-band full-polarimetric SAR for water regime classification, and IPCC water-scaling factors corresponding to Continuous Flooding, Single Drainage, or Multiple Drainage regimes. Evaluated across five stratified holdout sets, the phenology detection algorithm achieved planting and harvesting date Mean Absolute Errors of 6.1 ± 1.4 and 8.3 ± 1.7 days, with a 97.0% ± 2.7% operational detection rate. Water regime classification employed rice growth stage-specific Support Vector Machine classifiers with Radial Basis Function kernels (SVM-RBF), achieving per-stage test Balanced Accuracy ranging from 0.59 to 0.89. End-to-end integration using a four-track counterfactual decomposition yielded a full-pipeline mean absolute error of 18.5 ± 4.5 $\mathrm{kg}{\mathrm{CH}}_4{\mathrm{ha}}^{-1}$ (21.4% of the mean ground-based CH₄ calculation) and a mean bias of 3.5 ± 5.8 $\mathrm{kg}{\mathrm{CH}}_4{\mathrm{ha}}^{-1}$. Water level classification was confirmed as the dominant algorithmic uncertainty source, while the IPCC Tier 1 emission factor structural range (−32% to +48% of the default) exceeded all algorithmic errors combined. The proposed framework provides a spatially explicit approach for integrating multi-frequency SAR data into IPCC-compliant methane estimation, supporting Monitoring, Reporting, and Verification applications.