Forest above-ground biomass (AGB) is the basis of terrestrial carbon storage estimation, and making full use of the seasonal characteristics of remote sensing imagery can improve the estimation accuracy. In this study, we used multi-source time series and sample plots with the Random Forest (RF) model to estimate the AGB. The sources included Sentinel-1 (S-1), Sentinel-2 (S-2), and the S-1 and S-2 combination (S-1S-2). Time series included single season, annual, and multi-season. This study aims to (1) explore the optimal image acquisition season to estimate AGB; (2) determine whether the ability to estimate the AGB of multi-seasonal imagery exceeded that of annual and single-season imagery; (3) discover the sensitivity of different data to AGB according to phenological conditions. The results showed that: (1) images acquired in autumn were more useful for AGB estimation than spring, summer, and winter; (2) the S-1 multi-seasonal AGB model had higher accuracy than the annual or single-season one; (3) in autumn and spring, S-1 had higher estimation accuracy than S-2, and in autumn and spring, estimation accuracy from S-1S-2 was higher than that from S-1 and S-2; (4) in 16 AGB estimation models, the best estimation accuracy was achieved by the autumn AGB model from S-1S-2 (R² = 0.90, RMSE = 16.26 t/ha, p = 0.82, and rRMSE = 18.97). This study could be useful to identify the optimal image acquisition season for AGB estimation, thus reducing the economic cost of image acquisition and improving the estimation accuracy.

Net ecosystem productivity (NEP) serves as a key indicator for assessing regional carbon sink potential, with its dynamics regulated by nonlinear interactions among multiple factors. However, its driving factors and their coupling processes remain insufficiently characterized. This study investigated terrestrial ecosystems in Yunnan Province, China, to elucidate the drivers of NEP using 14 environmental factors (including topography, meteorology, soil texture, and human activities) and 21 remote sensing features. We developed a research framework based on “Feature Selection–Machine Learning–Mechanism Interpretation.” The results demonstrated that the Variable Selection Using Random Forests (VSURF) feature selection method effectively reduced model complexity. The selected features achieved high estimation accuracy across three machine learning models, with the eXtreme Gradient Boosting Regression (XGBR) model performing optimally (R² = 0.94, RMSE = 76.82 gC/(m²·a), MAE = 55.11 gC/(m²·a)). Interpretation analysis using the SHAP (SHapley Additive exPlanations) method revealed the following: (1) The Enhanced Vegetation Index (EVI), soil pH, solar radiation, air temperature, clay content, precipitation, sand content, and vegetation type were the primary drivers of NEP in Yunnan. Notably, EVI’s importance exceeded that of other factors by approximately 3 to 10 times. (2) Significant interactions existed between soil texture and temperature: Under low-temperature conditions (−5 °C to 12.15 °C), moderate clay content (13–25%) combined with high sand content (40–55%) suppressed NEP. Conversely, within the medium to high temperature range (5 °C to 23.79 °C), high clay content (25–40%) coupled with low sand content (25–43%) enhanced NEP. These findings elucidate the complex driving mechanisms of NEP in subtropical ecosystems, confirming the dominant role of EVI in carbon sequestration and revealing nonlinear regulatory patterns in soil–temperature interactions. This study provides not only a robust “Feature Selection–Machine Learning–Mechanism Interpretation” modeling framework for assessing carbon budgets in mountainous regions but also a scientific basis for formulating regional carbon management policies.

Forest above-ground carbon stock (AGCS) is one of the primary ecological evaluation indicators, so it is crucial to estimate the AGCS accurately. In this research, we added the climatic and topographic factors to the estimation process by a remote sensing approach to explore their impact and to achieve more precise estimations. We hope to develop a more accurate estimation method for AGCS based on remote sensing data and climate data. The random forest (RF) method has good robustness and wide applicability. Therefore, we modeled and predicted the AGCS by RF based on sixty field sample plots of Pinus densata pure forests in southwest China and the factors extracted from Landsat 8 OLI images (source I), Sentinel-2A images (source II), and combined Landsat 8 OLI and Sentinel-2A images (source III). We added the topographic and climatic factors to establish the AGCS estimation model and compared the results. The topographic factors contain elevation, slope, and aspect. Climatic factors contain mean annual temperature, annual precipitation, annual potential evapotranspiration, and monthly mean potential evapotranspiration. It was found that the R² and RMSE of the model based on source III were better than the R² and RMSE of the models based on source I and source II. Compared to the models based on source I and source II, the model based on source III improved R² by up to 0.08, reduced RMSE by up to 2.88 t/ha, and improved P by up to 4.29%. Among the models without adding factors, the model based on source III worked the best, with an R² of 0.87, an RMSE of 10.81 t/ha, an rRMSE of 23.19%, and a P of 79.71%. Among the models that added topographic factors, the model based on source III worked best after adding elevation, with an R² of 0.89, an RMSE of 10.01 t/ha, an rRMSE of 21.47%, and a P of 82.17%. Among the models that added climatic factors, the model that added the annual precipitation factor had the best modeling result, with an R² of 0.90, an RMSE of 9.53 t/ha, an rRMSE of 20.59%, and a P of 83.00%. The prediction result exhibited that the AGCS of the Pinus densata forest in 2021 was 9,737,487.52 t. The combination of Landsat 8 OLI and Sentinel-2A could improve the prediction accuracy of the AGCS. The addition of annual precipitation can effectively improve the accuracy of AGCS estimation. Higher resolution of climate data is needed to enhance the modeling in future work.

Forest above-ground carbon stock (AGCS) is one of the primary ecological evaluation indicators, so it is crucial to estimate AGCS accurately. In this research, we added the climatic and topographic factors to the estimation process by a remote sensing approach to explore their impacts and to achieve more precise estimations. We model and predict AGCS by Random Forest (RF) based on sixty field sample plots of Pinus densata pure forests in southwest of China and the factors extracted from Landsat 8 OLI images (Source I), Sentinel-2A images (Source II), combined Landsat 8 OLI and Sentinel-2A images (Source III). We added the topographic and climatic factors to establish AGCS estimation model and compare the results. The topographic factors contain elevation, slope and aspect. Climatic factors contain mean annual temperature, annual precipitation, annual potential evapotranspiration and monthly mean potential evapotranspiration. It was found that the model based on Source III was better than Source I and Source II. Among the models without adding factors, the model based on Source III worked the best, with an R2 of 0.87, an RMSE of 10.81 t/ha, an rRMSE of 23.19%, and a P of 79.71%. Among the models that added topographic factors, the model based on Source III worked best after adding elevation, with an R2 of 0.89, an RMSE of 10.01 t/ha, an rRMSE of 21.47%, and a P of 82.17%. Among the models that added climatic factors, the model that was added the annual precipitation factor had the best modeling result, with an R2 of 0.90, an RMSE of 9.53 t/ha, an rRMSE of 20.59%, and a P of 83.00%. The prediction result exhibited that the AGCS of the Pinus densata forest in 2021 was 9,737,487.52 t. The combination of Landsat 8 OLI and sentinel-2A could improve the prediction accuracy of AGCS. The addition of annual precipitation can effectively improve the accuracy of AGCS estimation. Higher resolution of climate data is needed to enhance the modeling in the future work.

This study proposed an RF-guided heuristic feature-selection framework that integrates multi-source remote-sensing data for estimating Pinus densata aboveground carbon stock (AGCS) in Shangri-La, Yunnan Province, China. Compared with four baseline feature-selection methods, the Random Forest–Alpha Evolution (RFA) and Random Forest–Markov Chain Monte Carlo (RFM) algorithms generated more informative feature subsets and improved model performance, with the Optuna-optimized AdaBoost model based on RFM features achieving the highest accuracy (R² = 0.71, RMSE = 10.53 t/ha). These results suggest that RF-guided heuristic feature selection can effectively improve AGCS estimation in complex mountainous environments. Vegetation indices and texture features were consistently prioritized across different feature-selection methods. Shapley Additive Explanations (SHAP)-based interpretation revealed that the most influential predictors were the Sentinel-2A green normalized difference vegetation index (S2_GNDVI) and precipitation of the wettest month (bio13) in the RFA Method, and the Sentinel-2A red-edge normalized difference vegetation index (S2_NDVI45) and bio13 in the RFM Method. These findings underscore the critical importance of canopy greenness, moisture availability, and structural complexity in regulating carbon accumulation in montane conifer forests. The final AGCS maps yielded total estimates of 9.83 Mt (RFA) and 10.46 Mt (RFM), and revealed a consistent spatial pattern, with moderate AGCS values dominating the landscape and a general tendency for higher values in the northwest and lower values in the southeast. In summary, the combination of RF-guided heuristic feature selection, Optuna-optimized machine learning and SHAP provides an effective and interpretable framework for AGCS estimation in mountain forests.

The Greater Khingan, a key cold-temperate coniferous forest region in northern China, is frequently affected by forest fires with severe ecological and economic impacts. The study investigates the influence of key environmental and anthropogenic drivers on forest fire susceptibility and evaluates multiple machine-learning approaches for regional fire assessment. Using 2001–2018 fire point data and multi-source remote sensing data, we integrated 13 driving factors across four dimensions: meteorology, topography, vegetation, and human activities. Collinear variables were screened using the Variance Inflation Factor (VIF). Three machine learning models—Logistic Regression (LR), Random Forest (RF), and Support Vector Machine (SVM)—were constructed to assess the long-term potential risk of forest fire occurrence. Driving mechanisms were analyzed using standardized regression coefficients and the SHapley Additive exPlanations (SHAP) interpretable algorithm, and spatial distribution maps of regional forest fire risk were generated based on the optimal model. Among the three models, RF achieved the highest predictive accuracy, with an accuracy of 0.919 and an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.966, significantly outperforming LR and SVM. SHAP analysis reveals that forest fires are primarily driven by climatic factors (Pres and Prec as core drivers), regulated by topographic factors, and weakly affected by human factors. The proposed framework provides an effective tool for long-term forest fire susceptibility assessment by combining robust predictive performance with interpretable model outputs. The findings provide scientific support for long-term strategic forest fire risk zoning, regional firefighting resource allocation, and the formulation of differentiated prevention and control strategies, and also offer methodological references for forest fire prediction in other cold-temperate forest regions in China.