Eucalyptus plantations play an important role in the timber supply and global warming mitigation around the world. Forest age is a critical factor for evaluating and modeling forest structure (e.g., diameter at breast height (DBH), height (H), aboveground carbon stocks (ACS)) and their dynamics. Recently, the spatial distribution of forest age at different scales based on time series remote sensing data has been widely investigated. However, it is unclear whether such data can effectively support the simulation and assessment of forest structure, especially in fast-growing plantation forests. In this study, the physiological principles in predicting growth (3-PG) model was firstly optimized and calibrated using survey and UAV lidar data at the sample plot (SP) scale, and was then applied at the forest sub-compartment (FSC) scale by designing different simulation scenarios driven by different forest age data sources and adjustments. The sensitivity of the simulated forest structure parameters to forest age was assessed at the SP and FSC levels. The results show that both the survey forest age data and the remote-sensing-derived forest age data could accurately estimate the DBH, H, and ACS of eucalyptus plantations with the coefficients of determination (R2) ranging from 0.87 to 0.94, and the relative root mean square error (RRMSE) below 20% at SP level. At the FSC level, the simulation results based on remotely sensed forest age data are significantly better than FSC forest age data from surveys by forestry bureaus, with R2 of ACS 0.7, RMSE 9.12 Mg/ha, and RRMSE 28.24%. The results of the sensitivity analysis show that the DBH, H, and ACS show different degrees of variation under different adjusted forest ages at SP and FSC level. The maximum difference in ACS is 82.91% at the SP scale if the forest age decreases 12 months and 41.23% at the FSC scale if the forest age increases 12 months. This study provides an important reference for future studies using forest age data obtained by remote sensing to drive the forest carbon model in a large spatial scale.
Tree species distribution is valuable for forest resource management. However, it is a challenge to classify tree species in subtropical regions due to complex landscapes and limitations of remote sensing data. The objective of this study was to propose a modified hierarchy-based classifier (MHBC) by optimizing the classification tree structures and variable selection method. Major steps to create an MHBC include automatic determination of classification tree structures based on the -score algorithm, selection and optimization of variables for each node, and classification using the optimized model. Experiments based on the fusion of Gaofen-1/Ziyuan-3 panchromatic (GF-1/ZY-3 PAN) and Sentinel-2 multispectral (MS) data indicated that (1) the MHBC provided overall classification accuracies of 85.19% for Gaofeng Forest Farm in China’s southern subtropical region and 94.4% for Huashi Township in China’s northern subtropical region, which had higher accuracies than random forest (RF) and classification and regression tree (CART); (2) critical variables for each class can be identified using the MHBC, and optimal variables of most nodes are spectral bands and vegetation indices; (3) compared to results from RF and CART, MHBC mainly improved the accuracies of the lower levels of classification tree structures (difficult classes to separate). The novelty in using MHBC is its simple and practical operation, easy-to-understand, and visualized variables that were selected in each node of the automatically constructed hierarchical trees. The robust performance of MHBC implies the potential to apply this approach to other sites for accurate classification of forest types.
Timely updates of carbon stock distribution are needed to better understand the impacts of deforestation and degradation on forest carbon stock dynamics. This research aimed to explore an approach for estimating aboveground carbon density (ACD) in the Brazilian Amazon through integration of MODIS (moderate resolution imaging spectroradiometer) and a limited number of light detection and ranging (Lidar) data samples using linear regression (LR) and random forest (RF) algorithms, respectively. Airborne LiDAR data at 23 sites across the Brazilian Amazon were collected and used to calculate ACD. The ACD estimation model, which was developed by Longo et al. in the same study area, was used to map ACD distribution in the 23 sites. The LR and RF methods were used to develop ACD models, in which the samples extracted from LiDAR-estimated ACD were used as dependent variables and MODIS-derived variables were used as independent variables. The evaluation of modeling results indicated that ACD can be successfully estimated with a coefficient of determination of 0.67 and root mean square error of 4.18 kg C/m2 using RF based on spectral indices. The mixed pixel problem in MODIS data is a major factor in ACD overestimation, while cloud contamination and data saturation are major factors in ACD underestimation. These uncertainties in ACD estimation using MODIS data make it difficult to examine annual ACD dynamics of degradation and growth, however this method can be used to examine the deforestation-induced ACD loss.
Eucalyptus plantations with fast growth and short rotation play an important role in improving economic conditions for local farmers and governments. It is necessary to map and update eucalyptus distribution in a timely manner, but to date, there is a lack of suitable approaches for quickly mapping its spatial distribution in a large area. This research aims to develop a uniform procedure to map eucalyptus distribution at a regional scale using the Sentinel-2 imagery on the Google Earth Engine (GEE) platform. Different seasonal Senstinel-2 images were first examined, and key vegetation indices from the selected seasonal images were identified using random forest and Pearson correlation analysis. The selected key vegetation indices were then normalized and summed to produce new indices for mapping eucalyptus distribution based on the calculated best cutoff values using the ROC (Receiver Operating Characteristic) curve. The uniform procedure was tested in both experimental and test sites and then applied to the entire Fujian Province. The results indicated that the best season to distinguish eucalyptus forests from other forest types was winter. The composite indices for eucalyptus–coniferous forest separation () and for eucalyptus–broadleaf forest separation (), which were synthesized from the enhanced vegetation index (EVI), plant senescing reflectance index (PSRI), shortwave infrared water stress index (SIWSI), and MERIS terrestrial chlorophyll index (MTCI), can effectively differentiate eucalyptus from other forest types. The proposed procedure with the best cutoff values (0.58 for and 1.29 for ) achieved accuracies of above 90% in all study sites. The eucalyptus classification accuracies in Fujian Province, with a producer’s accuracy of 91%, user’s accuracy of 97%, and overall accuracy of 94%, demonstrate the strong robustness and transferability of this proposed procedure. This research provided a new insight into quickly mapping eucalyptus distribution in subtropical regions. However, more research is still needed to explore the robustness and transferability of this proposed method in tropical regions or in other subtropical regions with different environmental conditions.