Search

The sources of uncertainty in wall-to-wall AGB maps propagate from the tree to pixel, but uncertainty due to forest cover mapping is rarely incorporated into the error propagation process. This study aimed to (1) elaborate an analytical procedure to incorporate forest-mapping-related uncertainty into the error propagation from plot and pixel predictions; (2) develop a stratified estimator with a model-assisted estimator for small and large areas; and (3) estimate the effect of ignoring the mapping uncertainty on the confidence intervals (CIs) for totals. Data consist of a subset of the Brazilian national forest inventory (NFI) database, comprising 75 counties that, once aggregated, served as strata for the stratified estimator. On-ground data were gathered from 152 clusters (plots) and remotely sensed data from Landsat-8 scenes. Four major contributions are highlighted. First, we describe how to incorporate forest-mapping-related uncertainty into the CIs of any forest attribute and spatial resolution. Second, stratified estimators perform better than non-stratified estimators for forest area estimation when the response variable is forest/non-forest. Comparing our stratified estimators, this study indicated greater precision for the stratified estimator than for the regression estimator. Third, using the ratio estimator, we found evidence that the simple field plot information provided by the NFI clusters is sufficient to estimate the proportion forest for large regions as accurately as remote-sensing-based methods, albeit with less precision. Fourth, ignoring forest-mapping-related uncertainty erroneously narrows the CI width as the estimate of proportion forest area decreases. At the small-area level, forest-mapping-related uncertainty led to CIs for total AGB as much as 63% wider in extreme cases. At the large-area level, the CI was 5–7% wider. [ABSTRACT FROM AUTHOR]

Estimating forest growing stock volume (GSV) is crucial for forest growth and resource management, as it reflects forest productivity. National measurements are laborious and costly; however, integrating satellite data such as optical, Synthetic Aperture Radar (SAR), and airborne laser scanning (ALS) with National Forest Inventory (NFI) data and machine learning (ML) methods has transformed forest management. In this study, random forest (RF), support vector regression (SVR), and Extreme Gradient Boosting (XGBoost) were used to predict GSV using Estonian NFI data, Sentinel-2 imagery, and ALS point cloud data. Four variable combinations were tested: CO1 (vegetation indices and LiDAR), CO2 (vegetation indices and individual band reflectance), CO3 (LiDAR and individual band reflectance), and CO4 (a combination of vegetation indices, individual band reflectance, and LiDAR). Across Estonia's geographical regions, RF consistently delivered the best performance. In the northwest (NW), the RF model achieved the best performance with the CO3 combination, having an R2 of 0.63 and an RMSE of 125.39 m3/plot. In the southwest (SW), the RF model also performed exceptionally well, achieving an R2 of 0.73 and an RMSE of 128.86 m3/plot with the CO4 variable combination. In the northeast (NE), the RF model outperformed other ML models, achieving an R2 of 0.64 and an RMSE of 133.77 m3/plot under the CO4 combination. Finally, in the southeast (SE) region, the best performance was achieved with the CO4 combination, yielding an R2 of 0.70 and an RMSE of 21,120.72 m3/plot. These results underscore RF's precision in predicting GSV across diverse environments, though refining variable selection and improving tree species data could further enhance accuracy. [ABSTRACT FROM AUTHOR]

The sources of uncertainty in wall-to-wall AGB maps propagate from the tree to pixel, but uncertainty due to forest cover mapping is rarely incorporated into the error propagation process. This study aimed to (1) elaborate an analytical procedure to incorporate forest-mapping-related uncertainty into the error propagation from plot and pixel predictions; (2) develop a stratified estimator with a model-assisted estimator for small and large areas; and (3) estimate the effect of ignoring the mapping uncertainty on the confidence intervals (CIs) for totals. Data consist of a subset of the Brazilian national forest inventory (NFI) database, comprising 75 counties that, once aggregated, served as strata for the stratified estimator. On-ground data were gathered from 152 clusters (plots) and remotely sensed data from Landsat-8 scenes. Four major contributions are highlighted. First, we describe how to incorporate forest-mapping-related uncertainty into the CIs of any forest attribute and spatial resolution. Second, stratified estimators perform better than non-stratified estimators for forest area estimation when the response variable is forest/non-forest. Comparing our stratified estimators, this study indicated greater precision for the stratified estimator than for the regression estimator. Third, using the ratio estimator, we found evidence that the simple field plot information provided by the NFI clusters is sufficient to estimate the proportion forest for large regions as accurately as remote-sensing-based methods, albeit with less precision. Fourth, ignoring forest-mapping-related uncertainty erroneously narrows the CI width as the estimate of proportion forest area decreases. At the small-area level, forest-mapping-related uncertainty led to CIs for total AGB as much as 63% wider in extreme cases. At the large-area level, the CI was 5–7% wider.

Estimating forest growing stock volume (GSV) is crucial for forest growth and resource management, as it reflects forest productivity. National measurements are laborious and costly; however, integrating satellite data such as optical, Synthetic Aperture Radar (SAR), and airborne laser scanning (ALS) with National Forest Inventory (NFI) data and machine learning (ML) methods has transformed forest management. In this study, random forest (RF), support vector regression (SVR), and Extreme Gradient Boosting (XGBoost) were used to predict GSV using Estonian NFI data, Sentinel-2 imagery, and ALS point cloud data. Four variable combinations were tested: CO1 (vegetation indices and LiDAR), CO2 (vegetation indices and individual band reflectance), CO3 (LiDAR and individual band reflectance), and CO4 (a combination of vegetation indices, individual band reflectance, and LiDAR). Across Estonia’s geographical regions, RF consistently delivered the best performance. In the northwest (NW), the RF model achieved the best performance with the CO3 combination, having an R2 of 0.63 and an RMSE of 125.39 m3/plot. In the southwest (SW), the RF model also performed exceptionally well, achieving an R2 of 0.73 and an RMSE of 128.86 m3/plot with the CO4 variable combination. In the northeast (NE), the RF model outperformed other ML models, achieving an R2 of 0.64 and an RMSE of 133.77 m3/plot under the CO4 combination. Finally, in the southeast (SE) region, the best performance was achieved with the CO4 combination, yielding an R2 of 0.70 and an RMSE of 21,120.72 m3/plot. These results underscore RF’s precision in predicting GSV across diverse environments, though refining variable selection and improving tree species data could further enhance accuracy.

Efforts to assess and understand changes in plant diversity and ecosystem functioning focus on the analysis of taxonomic diversity. However, the resilience of ecosystems depends not only on species richness but also on the functions (responses and effects) of species within communities and ecosystems. Therefore, a functional approach is required to estimate functional diversity through functional traits and to model its changes in space and time. This study aims to: (i) assess the accuracy of estimates of species richness and tree functional richness obtained from field data and Sentinel-2 imagery in tropical dry forests of the Yucatan Peninsula; (ii) map and analyze the relationships between these two variables. We calculated species richness and functional richness (from six functional traits) of trees from 87 plots of the National Forest Inventory in a semi-deciduous tropical forest and 107 in a semi-evergreen tropical forest. Species richness and functional richness were mapped using reflectance values, vegetation indices, and texture measurements from Sentinel-2 imagery as explanatory variables. Validation of the models to map these two variables yielded a coefficient of determination (R2) of 0.43 and 0.50, and a mean squared relative error of 25.4% and 48.8%, for tree species richness and functional richness, respectively. For both response variables, the most important explanatory variables were Sentinel-2 texture measurements and spectral bands. Tree species richness and functional richness were positively correlated in both forest types. Bivariate maps showed that 44.9% and 26.5% of the forests studied had high species richness and functional richness values. Our findings highlight the importance of integrating field data and remotely sensed variables for estimating tree species richness and functional richness. In addition, the combination of species richness and functional richness maps presented here is potentially valuable for planning, conservation, and restoration strategies by identifying areas that maximize ecosystem service provisioning, carbon storage, and biodiversity conservation. [ABSTRACT FROM AUTHOR]

The U.S. Forest Service, Forest Inventory and Analysis (FIA) program is tasked with making and reporting estimates of various forest attributes using a design-based network of permanent sampling plots. To make its estimates more precise, FIA uses a technique known as post-stratification to group plots into more homogenous classes, which helps lower variance when deriving population means. Currently FIA uses a nationally available map of tree canopy cover for post-stratification, which tends to work well for forest area estimates but less so for structural attributes like volume. Here we explore the use of new statewide digital aerial photogrammetric (DAP) point clouds developed from stereo imagery collected by the National Agricultural Imagery Program (NAIP) to improve these estimates in the southeastern mixed hardwood forests of Tennessee and Virginia, United States (U.S.). Our objectives are to 1. evaluate the relative quality of NAIP DAP point clouds using airborne LiDAR and FIA tree height measurements, and 2. assess the ability of NAIP digital height models (DHMs) to improve operational forest inventory estimates above the gains already achieved from FIA's current post-stratification approach. Our results show the NAIP point clouds were moderately to strongly correlated with FIA field measured maximum tree heights (average Pearson's r = 0.74) with a slight negative bias (−1.56 m) and an RMSE error of ~4.0 m. The NAIP point cloud heights were also more accurate for softwoods (R2s = 0.60–0.79) than hardwoods (R2s = 0.33–0.50) with an error structure that was consistent across multiple years of FIA measurements. Several factors served to degrade the relationship between the NAIP point clouds and FIA data, including a lack of 3D points in areas of advanced hardwood senescence, spurious height values in deep shadows and imprecision of FIA plot locations (which were estimated to be off the true locations by +/− 8 m). Using NAIP strata maps for post-stratification yielded forest volume estimates that were 31% more precise on average than estimates stratified with tree canopy cover data. Combining NAIP DHMs with forest type information from national map products helped improve stratification performance, especially for softwoods. The monetary value of using NAIP height maps to post-stratify FIA survey unit total volume estimates was USD 1.8 million vs. the costs of installing more field plots to achieve similar precision gains. Overall, our results show the benefit and growing feasibility of using NAIP point clouds to improve FIA's operational forest inventory estimates. [ABSTRACT FROM AUTHOR]

Remotely sensed maps of forest carbon stocks have enormous potential for supporting greenhouse gas (GHG) inventory and monitoring in tropical countries. However, most countries have not used maps as the reference data for GHG inventory due to the lack of confidence in the accuracy of maps and of data to perform local validation. Here, we use the first national forest inventory (NFI) data of the Democratic Republic of Congo to perform an independent assessment of the country's latest national spaceborne carbon stocks map. We compared plot-to-plot variations and areal estimates of forest aboveground biomass (AGB) derived from NFI data and from the map across jurisdictional and ecological domains. Across all plots, map predictions were nearly unbiased and captured c. 60% of the variation in NFI plots AGB. Map performance was not uniform along the AGB gradient, and saturated around c. 290 Mg ha−1, increasingly underestimating forest AGB above this threshold. Splitting NFI plots by land cover types, we found map predictions unbiased in the dominant terra firme Humid forest class, while plot-to-plot variations were poorly captured (R2 of c. 0.33, or c. 0.20 after excluding disturbed plots). In contrast, map predictions underestimated AGB by c. 33% in the small AGB woodland savanna class but captured a much greater share of plot-to-plot AGB variation (R2 of c. 0.41, or 0.58 after excluding disturbed plots). Areal estimates from the map and NFI data depicted a similar trend with a slightly smaller (but statistically indiscernible) mean AGB from the map across the entire study area (i.e., 252.7 vs. 280.6 Mg ha−1), owing to the underestimation of mean AGB in the woodland savanna domain (31.8 vs. 57.3 Mg ha−1), which was broadly consistent with the results obtained at the provincial level. This study provides insights and outlooks for country-wide AGB mapping efforts in the tropics and the computation of emission factors in Democratic Republic of Congo for carbon monitoring initiatives. [ABSTRACT FROM AUTHOR]

Efforts to assess and understand changes in plant diversity and ecosystem functioning focus on the analysis of taxonomic diversity. However, the resilience of ecosystems depends not only on species richness but also on the functions (responses and effects) of species within communities and ecosystems. Therefore, a functional approach is required to estimate functional diversity through functional traits and to model its changes in space and time. This study aims to: (i) assess the accuracy of estimates of species richness and tree functional richness obtained from field data and Sentinel-2 imagery in tropical dry forests of the Yucatan Peninsula; (ii) map and analyze the relationships between these two variables. We calculated species richness and functional richness (from six functional traits) of trees from 87 plots of the National Forest Inventory in a semi-deciduous tropical forest and 107 in a semi-evergreen tropical forest. Species richness and functional richness were mapped using reflectance values, vegetation indices, and texture measurements from Sentinel-2 imagery as explanatory variables. Validation of the models to map these two variables yielded a coefficient of determination (R2) of 0.43 and 0.50, and a mean squared relative error of 25.4% and 48.8%, for tree species richness and functional richness, respectively. For both response variables, the most important explanatory variables were Sentinel-2 texture measurements and spectral bands. Tree species richness and functional richness were positively correlated in both forest types. Bivariate maps showed that 44.9% and 26.5% of the forests studied had high species richness and functional richness values. Our findings highlight the importance of integrating field data and remotely sensed variables for estimating tree species richness and functional richness. In addition, the combination of species richness and functional richness maps presented here is potentially valuable for planning, conservation, and restoration strategies by identifying areas that maximize ecosystem service provisioning, carbon storage, and biodiversity conservation.

The U.S. Forest Service, Forest Inventory and Analysis (FIA) program is tasked with making and reporting estimates of various forest attributes using a design-based network of permanent sampling plots. To make its estimates more precise, FIA uses a technique known as post-stratification to group plots into more homogenous classes, which helps lower variance when deriving population means. Currently FIA uses a nationally available map of tree canopy cover for post-stratification, which tends to work well for forest area estimates but less so for structural attributes like volume. Here we explore the use of new statewide digital aerial photogrammetric (DAP) point clouds developed from stereo imagery collected by the National Agricultural Imagery Program (NAIP) to improve these estimates in the southeastern mixed hardwood forests of Tennessee and Virginia, United States (U.S.). Our objectives are to 1. evaluate the relative quality of NAIP DAP point clouds using airborne LiDAR and FIA tree height measurements, and 2. assess the ability of NAIP digital height models (DHMs) to improve operational forest inventory estimates above the gains already achieved from FIA’s current post-stratification approach. Our results show the NAIP point clouds were moderately to strongly correlated with FIA field measured maximum tree heights (average Pearson’s r = 0.74) with a slight negative bias (−1.56 m) and an RMSE error of ~4.0 m. The NAIP point cloud heights were also more accurate for softwoods (R2s = 0.60–0.79) than hardwoods (R2s = 0.33–0.50) with an error structure that was consistent across multiple years of FIA measurements. Several factors served to degrade the relationship between the NAIP point clouds and FIA data, including a lack of 3D points in areas of advanced hardwood senescence, spurious height values in deep shadows and imprecision of FIA plot locations (which were estimated to be off the true locations by +/− 8 m). Using NAIP strata maps for post-stratification yielded forest volume estimates that were 31% more precise on average than estimates stratified with tree canopy cover data. Combining NAIP DHMs with forest type information from national map products helped improve stratification performance, especially for softwoods. The monetary value of using NAIP height maps to post-stratify FIA survey unit total volume estimates was USD 1.8 million vs. the costs of installing more field plots to achieve similar precision gains. Overall, our results show the benefit and growing feasibility of using NAIP point clouds to improve FIA’s operational forest inventory estimates.

Remotely sensed maps of forest carbon stocks have enormous potential for supporting greenhouse gas (GHG) inventory and monitoring in tropical countries. However, most countries have not used maps as the reference data for GHG inventory due to the lack of confidence in the accuracy of maps and of data to perform local validation. Here, we use the first national forest inventory (NFI) data of the Democratic Republic of Congo to perform an independent assessment of the country’s latest national spaceborne carbon stocks map. We compared plot-to-plot variations and areal estimates of forest aboveground biomass (AGB) derived from NFI data and from the map across jurisdictional and ecological domains. Across all plots, map predictions were nearly unbiased and captured c. 60% of the variation in NFI plots AGB. Map performance was not uniform along the AGB gradient, and saturated around c. 290 Mg ha−1, increasingly underestimating forest AGB above this threshold. Splitting NFI plots by land cover types, we found map predictions unbiased in the dominant terra firme Humid forest class, while plot-to-plot variations were poorly captured (R2 of c. 0.33, or c. 0.20 after excluding disturbed plots). In contrast, map predictions underestimated AGB by c. 33% in the small AGB woodland savanna class but captured a much greater share of plot-to-plot AGB variation (R2 of c. 0.41, or 0.58 after excluding disturbed plots). Areal estimates from the map and NFI data depicted a similar trend with a slightly smaller (but statistically indiscernible) mean AGB from the map across the entire study area (i.e., 252.7 vs. 280.6 Mg ha−1), owing to the underestimation of mean AGB in the woodland savanna domain (31.8 vs. 57.3 Mg ha−1), which was broadly consistent with the results obtained at the provincial level. This study provides insights and outlooks for country-wide AGB mapping efforts in the tropics and the computation of emission factors in Democratic Republic of Congo for carbon monitoring initiatives.

Integrating information about the spatial distribution of carbon stocks and species diversity in tropical forests over large areas is fundamental for climate change mitigation and biodiversity conservation. In this study, spatial models showing the distribution of carbon stocks and the number of species were produced in order to identify areas that maximize carbon storage and biodiversity in the tropical forests of the Yucatan Peninsula, Mexico. We mapped carbon density and species richness of trees using L-band radar backscatter data as well as radar texture metrics, climatic and field data with the random forest regression algorithm. We reduced sources of errors in plot data of the national forest inventory by using correction factors to account for carbon stocks of small trees (2) of 0.28 and 0.31 and a relative mean square error of 38.5% and 33.0% for aboveground biomass and species richness, respectively, at pixel level. Estimates of carbon density were influenced mostly by radar backscatter and climatic data, while those of species richness were influenced mostly by radar texture and climatic variables. Correlation between carbon density and species richness was positive in 79.3% of the peninsula, while bivariate maps showed that 39.6% of the area in the peninsula had high carbon stocks and species richness. Our results highlight the importance of combining carbon and diversity maps to identify areas that are critical—both for maintaining carbon stocks and for conserving biodiversity.

Forest growing stock volume (GSV) is an important parameter in the context of forest resource management. National Forest Inventories (NFIs) are routinely used to estimate forest parameters, including GSV, for national or international reporting. Remotely sensed data are increasingly used as a source of auxiliary information for NFI data to improve the spatial precision of forest parameter estimates. In this study, we combine data from the NFI in Poland with satellite images of Landsat 7 and 3D point clouds collected with airborne laser scanning (ALS) technology to develop predictive models of GSV. We applied an area-based approach using 13,323 sample plots measured within the second cycle of the NFI in Poland (2010–2014) with poor positional accuracy from several to 15 m. Four different predictive approaches were evaluated: multiple linear regression, k-Nearest Neighbours, Random Forest and Deep Learning fully connected neural network. For each of these predictive methods, three sets of predictors were tested: ALS-derived, Landsat-derived and a combination of both. The developed models were validated at the stand level using field measurements from 360 reference forest stands. The best accuracy (RMSE% = 24.2%) and lowest systematic error (bias% = −2.2%) were obtained with a deep learning approach when both ALS- and Landsat-derived predictors were used. However, the differences between the evaluated predictive approaches were marginal when using the same set of predictor variables. Only a slight increase in model performance was observed when adding the Landsat-derived predictors to the ALS-derived ones. The obtained results showed that GSV can be predicted at the stand level with relatively low bias and reasonable accuracy for coniferous species, even using field sample plots with poor positional accuracy for model development. Our findings are especially important in the context of GSV prediction in areas where NFI data are available but the collection of accurate positions of field plots is not possible or justified because of economic reasons.

A spatially explicit map of aboveground carbon stored in Mexico’s forests was generated from empirical modeling on forest inventory and spaceborne optical and radar data. Between 2004 and 2007, the Mexican National Forestry Commission (CONAFOR) established a network of ~26,000 permanent inventory plots in the frame of their national inventory program, the Inventario Nacional Forestal y de Suelos (INFyS). INFyS data served as model response for spatially extending the field-based estimates of carbon stored in the aboveground live dry biomass to a wall-to-wall map, with 30 × 30 m2 pixel posting using canopy density estimates derived from Landsat, L-Band radar data from ALOS PALSAR, as well as elevation information derived from the Shuttle Radar Topography Mission (SRTM) data set. Validation against an independent set of INFyS plots resulted in a coefficient of determination (R2) of 0.5 with a root mean square error (RMSE) of 14 t∙C/ha in the case of flat terrain. The validation for different forest types showed a consistently low estimation bias (15°) with an R2 of 0.34. A comparison of the average carbon stocks computed from: (a) the map; and (b) statistical estimates from INFyS, at the scale of ~650 km2 large hexagons (R2 of 0.78, RMSE of 5 t∙C/ha) and Mexican states (R2 of 0.98, RMSE of 1.4 t∙C/ha), showed strong agreement.

The prediction of growing stock volume is one of the commonest applications of remote sensing to support the sustainable management of forest ecosystems. In this study, we used data from the 4th Spanish National Forest Inventory (SNFI-4) and from the 1st nationwide Airborne Laser Scanning (ALS) survey to develop predictive yield models for the three major commercial tree forest species (Eucalyptus globulus, Pinus pinaster and Pinus radiata) grown in north-western Spain. Integration of both types of data required prior harmonization because of differences in timing of data acquisition and difficulties in accurately geolocating the SNFI plots. The harmonised data from 477 E. globulus, 760 P. pinaster and 191 P. radiata plots were used to develop predictive models for total over bark volume, mean volume increment and total aboveground biomass by relating SNFI stand variables to metrics derived from the ALS data. The multiple linear regression methods and several machine learning techniques (k-nearest neighbour, random trees, random forest and the ensemble method) were compared. The study findings confirmed that multiple linear regression is outperformed by machine learning techniques. More specifically, the findings suggest that the random forest and the ensemble method slightly outperform the other techniques. The resulting stand level relative RMSEs for predicting total over bark volume, annual increase in total volume and total aboveground biomass ranged from 30.8−38.3%, 34.2−41.9% and 31.7−38.3% respectively. Although the predictions can be considered accurate, more precise geolocation of the SNFI plots and coincide temporarily with the ALS data would have enabled use of a much larger and robust field database to improve the overall accuracy of estimation.

Multisource forest inventory methods were developed to improve the precision of national forest inventory estimates. These methods rely on the combination of inventory data and auxiliary information correlated with forest attributes of interest. As these methods have been predominantly tested over coniferous forests, the present study used this approach for heterogeneous and complex deciduous forests in the center of France. The auxiliary data considered included a forest type map, Landsat 8 spectral bands and derived vegetation indexes, and 3D variables derived from photogrammetric canopy height models. On a subset area, changes in canopy height estimated from two successive photogrammetric models were also used. A model-assisted inference framework, using a k nearest-neighbors approach, was used to predict 11 field inventory variables simultaneously. The results showed that among the auxiliary variables tested, 3D metrics improved the precision of dendrometric estimates more than other auxiliary variables. Relative efficiencies (RE) varying from 2.15 for volume to 1.04 for stand density were obtained using all auxiliary variables. Canopy height changes also increased RE from 3% to 26%. Our results confirmed the importance of 3D metrics as auxiliary variables and demonstrated the value of canopy change variables for increasing the precision of estimates of forest structural attributes such as density and quadratic mean diameter.

The German National Forest Inventory consists of a systematic grid of permanent sample plots and provides a reliable evidence-based assessment of the state and the development of Germany’s forests on national and federal state level in a 10 year interval. However, the data have yet been scarcely used for estimation on smaller management levels such as forest districts due to insufficient sample sizes within the area of interests and the implied large estimation errors. In this study, we present a double-sampling extension to the existing German National Forest Inventory (NFI) that allows for the application of recently developed design-based small area regression estimators. We illustrate the implementation of the estimation procedure and evaluate its potential for future large-scale operational application by the example of timber volume estimation on two small-scale management levels (45 and 405 forest district units respectively) over the entire area of the federal German state of Rhineland-Palatinate. An airborne laserscanning (ALS) derived canopy height model and a tree species classification map based on satellite data were used as auxiliary data in an ordinary least square regression model to produce the timber volume predictions. The results support that the suggested double-sampling procedure can substantially increase estimation precision on both management levels: the two-phase estimators were able to reduce the variance of the one-phase simple random sampling estimator by 43% and 25% on average for the two management levels respectively.

A spatially explicit map of aboveground carbon stored in Mexico's forests was generated from empirical modeling on forest inventory and spaceborne optical and radar data. Between 2004 and 2007, the Mexican National Forestry Commission (CONAFOR) established a network of ~26,000 permanent inventory plots in the frame of their national inventory program, the Inventario Nacional Forestal y de Suelos (INFyS). INFyS data served as model response for spatially extending the field-based estimates of carbon stored in the aboveground live dry biomass to a wall-to-wall map, with 30 × 30 m2 pixel posting using canopy density estimates derived from Landsat, L-Band radar data from ALOS PALSAR, as well as elevation information derived from the Shuttle Radar Topography Mission (SRTM) data set. Validation against an independent set of INFyS plots resulted in a coefficient of determination (R²) of 0.5 with a root mean square error (RMSE) of 14 t·C/ha in the case of flat terrain. The validation for different forest types showed a consistently low estimation bias (<3 t·C/ha) and R²s in the range of 0.5 except for mangroves (R² = 0.2). Lower accuracies were achieved for forests located on steep slopes (>15°) with an R² of 0.34. A comparison of the average carbon stocks computed from: (a) the map; and (b) statistical estimates from INFyS, at the scale of ∼650 km² large hexagons (R² of 0.78, RMSE of 5 t·C/ha) and Mexican states (R² of 0.98, RMSE of 1.4 t·C/ha), showed strong agreement. [ABSTRACT FROM AUTHOR]

Today, European forests face many challenges but also offer opportunities, such as climate change mitigation, provision of renewable resources, energy and other ecosystem services. Large-scale analyses to assess these opportunities are hindered by the lack of a consistent, spatial and accessible forest structure data. This study presents a freely available pan-European forest structure data set. Building on our previous work, we used data from six additional countries and consider now ten key forest stand variables. Harmonized inventory data from 16 European countries were used in combination with remote sensing data and a gap-filling algorithm to produce this consistent and comparable forest structure data set across European forests. We showed how land cover data can be used to scale inventory data to a higher resolution which in turn ensures a consistent data structure across sub-regional, country and European forest assessments. Cross validation and comparison with published country statistics of the Food and Agriculture Organization (FAO) indicate that the chosen methodology is able to produce robust and accurate forest structure data across Europe, even for areas where no inventory data were available. [ABSTRACT FROM AUTHOR]

Integrating information about the spatial distribution of carbon stocks and species diversity in tropical forests over large areas is fundamental for climate change mitigation and biodiversity conservation. In this study, spatial models showing the distribution of carbon stocks and the number of species were produced in order to identify areas that maximize carbon storage and biodiversity in the tropical forests of the Yucatan Peninsula, Mexico. We mapped carbon density and species richness of trees using L-band radar backscatter data as well as radar texture metrics, climatic and field data with the random forest regression algorithm. We reduced sources of errors in plot data of the national forest inventory by using correction factors to account for carbon stocks of small trees (<7.5 cm DBH) and for the temporal difference between field data collection and imagery acquisition. We created bivariate maps to assess the spatial relationship between carbon stocks and diversity. Model validation of the regional maps obtained herein using an independent data set of plots resulted in a coefficient of determination (R2) of 0.28 and 0.31 and a relative mean square error of 38.5% and 33.0% for aboveground biomass and species richness, respectively, at pixel level. Estimates of carbon density were influenced mostly by radar backscatter and climatic data, while those of species richness were influenced mostly by radar texture and climatic variables. Correlation between carbon density and species richness was positive in 79.3% of the peninsula, while bivariate maps showed that 39.6% of the area in the peninsula had high carbon stocks and species richness. Our results highlight the importance of combining carbon and diversity maps to identify areas that are critical—both for maintaining carbon stocks and for conserving biodiversity. [ABSTRACT FROM AUTHOR]

Forest growing stock volume (GSV) is an important parameter in the context of forest resource management. National Forest Inventories (NFIs) are routinely used to estimate forest parameters, including GSV, for national or international reporting. Remotely sensed data are increasingly used as a source of auxiliary information for NFI data to improve the spatial precision of forest parameter estimates. In this study, we combine data from the NFI in Poland with satellite images of Landsat 7 and 3D point clouds collected with airborne laser scanning (ALS) technology to develop predictive models of GSV. We applied an area-based approach using 13,323 sample plots measured within the second cycle of the NFI in Poland (2010–2014) with poor positional accuracy from several to 15 m. Four different predictive approaches were evaluated: multiple linear regression, k-Nearest Neighbours, Random Forest and Deep Learning fully connected neural network. For each of these predictive methods, three sets of predictors were tested: ALS-derived, Landsat-derived and a combination of both. The developed models were validated at the stand level using field measurements from 360 reference forest stands. The best accuracy (RMSE% = 24.2%) and lowest systematic error (bias% = −2.2%) were obtained with a deep learning approach when both ALS- and Landsat-derived predictors were used. However, the differences between the evaluated predictive approaches were marginal when using the same set of predictor variables. Only a slight increase in model performance was observed when adding the Landsat-derived predictors to the ALS-derived ones. The obtained results showed that GSV can be predicted at the stand level with relatively low bias and reasonable accuracy for coniferous species, even using field sample plots with poor positional accuracy for model development. Our findings are especially important in the context of GSV prediction in areas where NFI data are available but the collection of accurate positions of field plots is not possible or justified because of economic reasons. [ABSTRACT FROM AUTHOR]

The prediction of growing stock volume is one of the commonest applications of remote sensing to support the sustainable management of forest ecosystems. In this study, we used data from the 4th Spanish National Forest Inventory (SNFI-4) and from the 1st nationwide Airborne Laser Scanning (ALS) survey to develop predictive yield models for the three major commercial tree forest species (Eucalyptus globulus, Pinus pinaster and Pinus radiata) grown in north-western Spain. Integration of both types of data required prior harmonization because of differences in timing of data acquisition and difficulties in accurately geolocating the SNFI plots. The harmonised data from 477 E. globulus, 760 P. pinaster and 191 P. radiata plots were used to develop predictive models for total over bark volume, mean volume increment and total aboveground biomass by relating SNFI stand variables to metrics derived from the ALS data. The multiple linear regression methods and several machine learning techniques (k-nearest neighbour, random trees, random forest and the ensemble method) were compared. The study findings confirmed that multiple linear regression is outperformed by machine learning techniques. More specifically, the findings suggest that the random forest and the ensemble method slightly outperform the other techniques. The resulting stand level relative RMSEs for predicting total over bark volume, annual increase in total volume and total aboveground biomass ranged from 30.8–38.3%, 34.2–41.9% and 31.7–38.3% respectively. Although the predictions can be considered accurate, more precise geolocation of the SNFI plots and coincide temporarily with the ALS data would have enabled use of a much larger and robust field database to improve the overall accuracy of estimation. [ABSTRACT FROM AUTHOR]

Multisource forest inventory methods were developed to improve the precision of national forest inventory estimates. These methods rely on the combination of inventory data and auxiliary information correlated with forest attributes of interest. As these methods have been predominantly tested over coniferous forests, the present study used this approach for heterogeneous and complex deciduous forests in the center of France. The auxiliary data considered included a forest type map, Landsat 8 spectral bands and derived vegetation indexes, and 3D variables derived from photogrammetric canopy height models. On a subset area, changes in canopy height estimated from two successive photogrammetric models were also used. A model-assisted inference framework, using a k nearest-neighbors approach, was used to predict 11 field inventory variables simultaneously. The results showed that among the auxiliary variables tested, 3D metrics improved the precision of dendrometric estimates more than other auxiliary variables. Relative efficiencies (RE) varying from 2.15 for volume to 1.04 for stand density were obtained using all auxiliary variables. Canopy height changes also increased RE from 3% to 26%. Our results confirmed the importance of 3D metrics as auxiliary variables and demonstrated the value of canopy change variables for increasing the precision of estimates of forest structural attributes such as density and quadratic mean diameter. [ABSTRACT FROM AUTHOR]