Academic literature on the topic 'Multisource forest inventory'

Sustainable forest management requires information on the spatial distribution, composition, and structure of forests. However, jurisdictions with large tracts of noncommercial forest, such as the Northwest Territories (NWT) of Canada, often lack detailed forest information across their land base. The goal of the Multisource Vegetation Inventory (MVI) project was to create a large area forest inventory (FI) map that could support strategic forest management in the NWT using optical, radar, and light detection and ranging (LiDAR) satellite remote sensing anchored on limited field plots and airborne LiDAR data. A new landcover map based on Landsat imagery was the first step to stratify forestland into broad forest types. A modelling chain linking FI plots to airborne and spaceborne LiDAR was then developed to circumvent the scarcity of field data in the region. The developed models allowed the estimation of forest attributes in thousands of surrogate FI plots corresponding to spaceborne LiDAR footprints distributed across the project area. The surrogate plots were used as a reference dataset for estimating each forest attribute in each 30 m forest cell within the project area. The estimation was based on the k-nearest neighbour (k-NN) algorithm, where the selection of the four most similar surrogate FI plots to each cell was based on satellite, topographic, and climatic data. Wall-to-wall 30 m raster maps of broad forest type, stand height, crown closure, stand volume, total volume, aboveground biomass, and stand age were created for a ~400,000 km2 area, validated with independent data, and generalized into a polygon GIS layer resembling a traditional FI map. The MVI project showed that a reasonably accurate FI map for large, remote, predominantly non-inventoried boreal regions can be obtained at a low cost by combining limited field data with remote sensing data from multiple sources.

Analyzing forest monitoring costs and accuracy of forest carbon stock estimates are important criteria in the framework of Reducing Emission from Deforestation and Forest Degradation (REDD), because Monitoring, Reporting and Verification (MRV) system has been seen as an investment that aims to generate financial benefits to forest owners. Thus, comparisons of cost efficiency and accuracy were carried out between the LiDAR (Light Detection and Ranging) Assisted Multisource Programme (LAMP) and the field-based multisource Forest Resource Assessment (FRA) applied in the 23500 km2 Terai Arc Landscape (TAL) of Nepal in 2011 to estimate Above Ground Biomass (AGB). The model-based LAMP was applied by integrating 5% LiDAR sampling, wall to wall RapidEye satellite image and field sample plot inventory. The design-based FRA was carried out to generate comprehensive forest resource information. Administrative and initial variable costs of both approaches were calculated separately, and converted to unit costs for comparison. To compare the subsequent forest monitoring costs, cumulative costs were derived on the basis of the calculated present variable items and expenditures. The accuracies were calculated by using mean error of mean biomass estimates (tons/ha) at different spatial scales ranging from 1 to 350,000 ha forests. Design-based FRA was found to be cost-efficient (USD 0.22/ha) as compared to the LAMP approach (USD 0.28/ha) for baseline data collection, whereas administrative cost of multisource FRA (USD 0.26/ha) was significantly higher. Although a huge amount of data were generated through multisource FRA in each cycle, the LAMP approach appears to be cost-efficient to estimate AGB in subsequent forest inventory. The mean errors in the LAMP-derived mean biomass estimate were significantly smaller at all spatial resolutions than the FRA-plot-derived mean biomass estimate. The study concludes that spatial accuracy of LAMP is good enough to estimate biomass stock of Community Forests (CFs) where average size of CF is 150 ha in the study area. Banko Janakari, Vol. 23, No. 1, Page 12-22 DOI: http://dx.doi.org/10.3126/banko.v23i1.9463

Spatially detailed monitoring of forest resources is important for sustainable management but limited by a lack of field measurements. The increasing availability of multisource datasets offers the potential to characterize forest attributes at finer resolutions with regional coverage. This study aimed to assess the potential of mapping stem volume at a 30 m scale in eastern Texas using multisource datasets: airborne lidar, Landsat and LANDFIRE (Landscape Fire and Resource Management Planning Tools Project) datasets. Gradient-boosted trees regression models relating total volume, estimated from airborne lidar measurements and allometric equations, and multitemporal Landsat and LANDFIRE predictors were developed and evaluated. The fitted models showed moderate to high correlation (R2 = 0.52–0.81) with reference stem volume estimates, with higher correlation in pine forests (R2 = 0.70–0.81) than mixed forests (R2 = 0.52–0.67). The models were also precise, with relative percent mean absolute errors (pMAE) of 13.8–21.2%. The estimated volumes also consistently agreed with volumes estimated in independent sites (R2 = 0.51, pMAE = 34.7%) and with US Forest Service Forest Inventory Analysis county-level volume estimates (R2 = 0.93, pBias = −10.3%, pMAE = 11.7%). This study shows the potential of developing regional stem volume products using airborne lidar and multisource datasets, supporting forest productivity and carbon modeling at spatially detailed scales.

A multisource inventory method has been applied in the Finnish National Forest Inventory (NFI) since 1990.The method utilizes satellite images and digital map data,in addition to field measurements, and produces estimates of allfield parameters for computation units as well as thematic maps. Information from base mapsis employed in delineating forestry land from other land use classes.The map data are not necessarily up-to-date and often containsignificant errors. This paper introduces a statistical calibration method aimed atreducing the effect of map errors on multisource forest resourceestimates. The correction is based on theconfusion matrix between land use classes of the field sample plots and corresponding map information.The proposed method is illustrated in a realistic setting using data from the ninth NFI.

Multisource forest inventory with two-phase sampling offers several advantages in the forest management planning when compared with the traditional visual inventory by stands. For example, by combining data from remote sensing imagery with field measurements, it is possible to estimate the forest characteristics of large areas at a more reasonable cost than by using the traditional visual inventory by stands. In this study, the k-nearest-neighbours estimation (k-nn), stand inventory data, and geostatistical interpolation were combined for estimation of five forest variables (mean diameter, mean height, mean age, basal area, and volume) per sample plot and stand. Digitized aerial photograph features, visually interpreted aerial photograph features, and updated stand inventory data were used as the auxiliary data sources in the estimation of forest variables. The results show that, at the sample plot level, the k-nn estimates based on the auxiliary data sources were more accurate than the updated stand inventory data transferred to the plot level. At the stand level, the updated stand inventory data were more accurate than the k-nn estimates. When the k-nn estimates were combined with the updated stand inventory data, the accuracy of the estimates was significantly improved at both the sample plot and stand level. The geostatistical interpolation, which was tested on the stand level estimation, did not result in any further improvement in the accuracy of the estimates.

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.

Forest biomass reflects the material cycle of forest ecosystems and is an important index to measure changes in forest structure and function. The accurate estimation of forest biomass is the research basis for measuring carbon storage in forest systems, and it is important to better understand the carbon cycle and improve the efficiency of forest policy and management activities. In this study, to achieve an accurate estimation of meso-scale (regional) forest biomass, we used Ninth Beijing Forest Inventory data (FID), Landsat 8 OLI Image data and ALOS-2 PALSAR-2 data to establish different forest types (coniferous forest, mixed forest, and broadleaf forest) of biomass models in Beijing. We assessed the potential of forest inventory, optical (Landsat 8 OLI) and radar (ALOS-2 PALSAR-2) data in estimating and mapping forest biomass. From these data, a wide range of parameters related to forest structure were obtained. Random forest (RF) models were established using these parameters and compared with traditional multiple linear regression (MLR) models. Forest biomass in Beijing was then estimated. The results showed the following: (1) forest inventory data combined with multisource remote sensing data can better fit forest biomass than forest inventory data alone. Among the three forest types, mixed forest has the best fitting model. Forest inventory variables and multisource remote sensing variables can match each other in time and space, capturing almost all spatial variability. (2) The 2016 forest biomass density in Beijing was estimated to be 52.26 Mg ha−1 and ranged from 19.1381–195.66 Mg ha−1. The areas with high biomass were mainly distributed in the north and southwest of Beijing, while the areas with low biomass were mainly distributed in the southeast and central areas of Beijing. (3) The estimates from the RF model are better than those from the MLR model, showing a high R 2 and a low root mean square error (RMSE). The R 2 values of the MLR models of three forest types were greater than 0.5, and RMSEs were less than 15.5 Mg ha−1, The R 2 values of the RF models were higher than 0.6, and the RMSEs were lower than 13.5 Mg ha−1. We conclude that the methods in this paper can help improve the accurate estimation of regional biomass and provide a basis for the planning of relevant forestry decision-making departments.

The National Forest Inventory of Finland (NFI) produces national- and regional-level statistics for sustainability assessment and strategical-level decision making. So far, the regional-level statistics are based on a systematic sampling design with geographical stratification. Auxiliary information such as remote sensing is not used for design or estimation at the regional level, but it is used at the small-area level, i.e., for municipality-level results. To improve the cost efficiency of the NFI, possibilities for using auxiliary data in both the design and estimation are of interest. We assessed the improvements obtainable by using an interpreted satellite image — the multisource NFI result from a previous NFI — as auxiliary information in the design phase. The results show that even though the multisource NFI map is not very accurate, significant improvements in efficiency can be obtained by using either the local pivotal method (LPM) or stratification. LPM improves efficiency by matching the sample distribution to population distribution. These results encourage us to further investigate (i) what would be the improvement with more accurate auxiliary information, for example, laser scanning data, and (ii) how LPM fits in a real-life situation where part of the plots are permanent and it would be used to select only the temporary plots.

Forests are crucial in carbon sequestration and oxygen release. An accurate assessment of forest carbon storage is meaningful for Chinese cities to achieve carbon peak and carbon neutrality. For an accurate estimation of regional-scale forest aboveground carbon density, this study applied a Sentinel-2 multispectral instrument (MSI), Advanced Land Observing Satellite 2 (ALOS-2) L-band, and Sentinel-1 C-band synthetic aperture radar (SAR) to estimate and map the forest carbon density. Considering the forest field-inventory data of eastern China from 2018 as an experimental sample, we explored the potential of the deep-learning algorithms convolutional neural network (CNN) and Keras. The results showed that vegetation indices from Sentinel-2, backscatter and texture characters from ALOS-2, and coherence from Sentinel-1 were principal contributors to the forest carbon-density estimation. Furthermore, the CNN model was found to perform better than traditional models. Results of forest carbon-density estimation validated the improvements effectively by combining the optical and radar data. Compared with traditional regression methods, deep learning has a higher potential for accurately estimating forest carbon density using multisource remote-sensing data.

En France, la méthode de l'Inventaire Forestier National (IFN) répond à des besoins de politique publique aux échelles nationales et régionales. Sur des plus petits territoires, la précision est souvent insuffisante pour répondre aux besoins des activités de gestion. Les méthodes IFM peuvent répondre à ce besoin en combinant des données d'inventaire et des données de télédétection. La thèse vise à améliorer la précision de l'IFN à des échelles subrégionales à locales en intégrant les données du système lidar spatial GEDI dans des approches multisources.Cependant, cette intégration se heurte à un verrou majeur, lié à l'absence de correspondance spatiale entre les échantillons sur le terrain (placettes d'inventaire) et les empreintes GEDI. Par ailleurs, les données GEDI sont mal géoréférencées, ce qui complexifie leur intégration dans certaines approches d'IFM. Cette thèse se concentre sur ces problématiques et est divisée en trois parties principales.Premièrement, une méthode d'amélioration du géoréférencement de GEDI a été développée en se basant uniquement sur un modèle numérique de terrain (MNT) de référence à haute résolution spatiale. Cette méthode compare, pour une série de positions autour de la localisation indiquée dans les produits GEDI, les élévations du terrain des empreintes GEDI avec celles du MNT de référence, générant une carte d'écarts en fonction des décalages en X et Y. En utilisant un algorithme d'accumulation de flux sur cette carte, une position améliorée qui minimise l'écart avec le MNT est proposée pour chaque empreinte GEDI.Ensuite, deux approches d'utilisation des données GEDI avec les données de l'IFN ont été élaborées. Les zones d'étude se situent dans les Vosges et utilisent environ 500 placettes IFN et plus de 100,000 empreintes GEDI. La première approche est une approche d'échantillonnage double pour la stratification (2SS), reposant sur des variables communes entre GEDI et IFN, sans nécessiter de coïncidence spatiale entre les deux sources de données. Les approches 2SS reposent généralement sur des échantillons de données probabilistes, ce qui n'est a priori pas le cas de l'échantillonnage de GEDI. Ainsi, une analyse préliminaire a été nécessaire pour comprendre les caractéristiques spécifiques de l'échantillon des mesures GEDI. La pertinence de la variable commune choisie, la hauteur maximale des arbres, a également été vérifiée. Par rapport aux estimations basées uniquement sur les données IFN, l'approche 2SS a amélioré la variance des estimations de volume de 56%.La deuxième approche utilise un lien entre données GEDI et données IFN établi indirectement en utilisant les images Sentinel-2 et Sentinel-1, avec la méthode des k-plus proches voisins (kNN) combinée avec du bagging (bootstrap aggregation). Il s'agit de propager l'information des placettes terrain au niveau des empreintes GEDI pour densifier les placettes IFN en tirant parti des mesures de structure forestière GEDI, bien corrélées aux attributs forestiers d'intérêt (ex. le volume de bois). Tout d'abord, en utilisant un kNN-bagging, on cherche pour chaque placette IFN les empreintes GEDI ayant les caractéristiques les plus proches de celles du point IFN pour des variables de lien Sentinel, complétées ou non avec une variable de lien supplémentaire de hauteur. On estime ainsi l'ensemble des variables GEDI pour chaque placette IFN. Ensuite, un modèle de régression est établi par kNN-bagging pour estimer le volume de bois à partir des variables GEDI les mieux prédites à l'étape précédente et les variables Sentinel. Le volume est estimé au niveau de toutes les empreintes GEDI. La stratégie complétée par une variable de lien de hauteur a atteint un coefficient de détermination de 58%. Par la suite, sur la base du réseau dense de placettes avec volume ainsi obtenu, des méthodes standards d'estimation sur de petites surfaces (small area estimation) ou de cartographie haute résolution, pourront être implémentés
The thesis focuses on the contribution of spaceborne lidar to the development of Multisource Forest Inventory (MFI) methods. In France, the National Forest Inventory (NFI) method addresses the requirements of public policies at regional and national levels. However, on smaller territories, precision is often insufficient to meet the needs of management activities. MFI methods better address these needs by combining inventory data with remote sensing data. This thesis aims to improve NFI accuracy at sub-regional to local scales by integrating data from the spaceborne lidar GEDI into multisource approaches.Unfortunately, this integration is complicated due to the lack of spatial correspondence between field samples (inventory plots) and GEDI footprints. Additionally, GEDI data are poorly georeferenced, making them difficult to integrate into certain MFI approaches. This thesis focuses on these issues and is divided into three main parts.As a first step, a method for improving GEDI georeferencing, based on a high-resolution reference digital elevation model (DEM) was developed. This method compares, for a series of positions around the location indicated in the GEDI products, the ground elevations of the GEDI footprints with those of the reference DEM, generating an error map according to X and Y offsets. Using a flow accumulation algorithm on this error map, an improved position minimizing the distance from the DEM is proposed for each GEDI footprint.Next, two approaches for using GEDI data with NFI data were developed. The study sites are located in the Vosges and use ∼ 500 IFN plots and over 100,000 GEDI footprints.The first approach is a double sampling for stratification (2SS) approach, based on common variables between GEDI and NFI, without requiring spatial correspondence of the two data sources. 2SS approaches are generally based on probabilistic data samples, which is not a priori the case for GEDI's sampling pattern. Thus, a preliminary analysis was required to understand the characteristics of the spatial distribution of the GEDI sample. The relevance of the chosen common variable, i.e. the maximum tree height, was also verified. Compared with estimates based only on NFI data, the 2SS approach improved the variance of growing stock volume estimates by up to 56%.The second approach is based on a link between GEDI data and NFI data, established indirectly by using spatially exhaustive data sources, the Sentinel-2 and Sentinel-1 images. To establish the model linking the different data sources, we chose to use the k-nearest neighbor (kNN) method combined with bagging (bootstrap aggregation). The aim is to propagate information from field plots to GEDI footprints in order to "densify" NFI plots by taking advantage of GEDI forest structure measurements, which are well correlated with the forest attributes of interest (e.g. growing stock volume). First, for each NFI plot, we looked for the GEDI footprints with the characteristics of the Sentinel link variables, supplemented or not with a height link variable, that are closest to those of the NFI point. Using a kNN-bagging approach, the set of GEDI variables is therefore estimated for each NFI plot. Next, a regression model is established by kNN-bagging to estimate the volume using the best predicted GEDI variables from the previous step and the Sentinel variables. The volume is estimated at the level of all GEDI footprints. The strategy supplemented by a height link variable performed best and reached a coefficient of determination of 58%. Subsequently, using the resulting dense sample of volume plots, standard methods for small area estimation (scale of the municipality or district) or high-resolution volume mapping can be implemented