Improving Lidar-based aboveground biomass estimation of temperate hardwood forests with varying site productivity

Shao, Gang; Shao, Guofan; Gallion, Joey; Saunders, Michael R.; Frankenberger, Jane; Fei, Songlin

doi:10.1016/j.rse.2017.09.011

Cited by 36 publications

(28 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The ALS metrics most strongly related to the C stocks in Pinus plantations were height metrics (Elev.P90, Elev.mode, Percentage.all.returns.above.mean, Elev.kurtosis, Elev.MAD.median, Elev.minimum, Percentage.all.returns.above.mode, Elev.maximum, Elev.P99), and other ALS-derived metrics concerning the horizontal distribution of the point cloud (Return.1.count, Canopy.relief.ratio, Return.2.count). Metrics generated from ALS data are commonly agreed to be highly correlated with C stocks, and many studies have successfully used height metrics (higher percentiles, mean, and maximum height) as important predictors for their quantification in Mediterranean pine species [23,40,63] and other conifers [14,[64][65][66]. In our study, all the species models used similar metrics, which demonstrates the consistency of the models.…”

Section: The Use Of Als Data and A Knn Model For C Stock Estimationsupporting

confidence: 70%

Assessment of the Carbon Stock in Pine Plantations in Southern Spain through ALS Data and K-Nearest Neighbor Algorithm Based Models

Navarrete-Poyatos

Navarro-Cerrillo

Lara-Gómez³

et al. 2019

Geosciences

View full text Add to dashboard Cite

Accurate estimation of forest biomass to enable the mapping of forest C stocks over large areas is of considerable interest nowadays. Airborne laser scanning (ALS) systems bring a new perspective to forest inventories and subsequent biomass estimation. The objective of this research was to combine growth models used to update old inventory data to a reference year, low-density ALS data, and k-nearest neighbor (kNN) algorithm Random Forest to conduct biomass inventories aimed at estimating the C sequestration capacity in large Pinus plantations. We obtained a C stock in biomass (W t -S) of 12.57 Mg·ha −1 , ranging significantly from 19.93 Mg·ha −1 for P. halepensis to 49.05 Mg·ha −1 for P. nigra, and a soil organic C stock of the composite soil samples (0-40 cm) ranging from 20.41 Mg·ha −1 in P. sylvestris to 37.32 Mg·ha −1 in P. halepensis. When generalizing these data to the whole area, we obtained an overall C-stock value of 48.01 MgC·ha −1 , ranging from 23.96 MgC·ha −1 for P. halepensis to 58.09 MgC·ha −1 for P. nigra. Considering the mean value of the on-site C stock, the study area sustains 1,289,604 Mg per hectare (corresponding to 4,732,869 Mg CO 2 ), with a net increase of 4.79 Mg·ha −1 ·year −1 . Such C cartography can help forest managers to improve forest silviculture with regard to C sequestration and, thus, climate change mitigation.In addition to biomass, C is also stored in litter and forest soils. In fact, soils are the largest reservoir of terrestrial C, for both organic and inorganic forms [6]. Soil organic carbon (SOC) can be stored in soils for thousands of years under, suitable conditions, and is a vital component of plant nutrient cycles [7]. Forest management aimed at increasing stand growth has been shown to be effective in increasing the C sequestration capacity [8,9]. Thinning treatments improve health and tree vigor, increasing forest productivity [10], while the soil C content shows a slight decrease in the first stages, recovering its level and increasing once the canopy is restored [11].Accurate estimation of forest biomass to produce spatially explicit mapping of forest C stocks over large study areas is of considerable interest nowadays [12]. In this regard, the field-based inventory is the most common method for evaluating the dendrometric characteristics and stand dynamics of forests. These classic forest inventories have a high demand for labor, are expensive, and require lots of field plots to obtain full inventory data for large areas on the ground [13,14]. Thus, with the advances in remote sensing techniques, these traditional methods are being replaced or supplemented with Light Detection and Ranging (LiDAR) techniques, which provide stand data over large areas, optimizing time and costs. LiDAR technology brings a new perspective to forest inventories by directly providing three-dimensional information on the entire surface [12].LiDAR (Laser Imaging Detection and Ranging) systems from an airborne platform (airborne laser scanning, ALS) are equipped with a scanning device th...

show abstract

Section: The Use Of Als Data and A Knn Model For C Stock Estimationsupporting

confidence: 70%

Assessment of the Carbon Stock in Pine Plantations in Southern Spain through ALS Data and K-Nearest Neighbor Algorithm Based Models

Navarrete-Poyatos

Navarro-Cerrillo

Lara-Gómez³

et al. 2019

Geosciences

View full text Add to dashboard Cite

show abstract

“…Destructive methods require cut samples of trees, and the wood is weighted after it has dried (Parresol, ). Indirect methods use allometric equations, in which the conventional variables used include the field‐measured normal diameter and height of the trees (Moore, ; Shao et al, ). The ability of LiDAR to collect a large amount of densely sampled elevation data promises a more efficient and inexpensive tool—developed by training data‐driven expert systems—for forest biomass management (Shao et al, ).…”

Section: Introductionmentioning

confidence: 99%

“…Indirect methods use allometric equations, in which the conventional variables used include the field‐measured normal diameter and height of the trees (Moore, ; Shao et al, ). The ability of LiDAR to collect a large amount of densely sampled elevation data promises a more efficient and inexpensive tool—developed by training data‐driven expert systems—for forest biomass management (Shao et al, ). Time series of LiDAR measurements have already been used for forest dynamic monitoring, that is, for monitoring tree growth (Zhao et al, ).…”

Section: Introductionmentioning

confidence: 99%

Estimation of forest biomass from light detection and ranging data by using machine learning

2019

View full text Add to dashboard Cite

The use of data driven predictive systems is becoming widespread as innovations in machine learning techniques have allowed the training of increasingly sophisticated models via the available data. The light detection and ranging (LiDAR) remote sensing technique is being increasingly applied to obtain informative terrain maps, due to its ability to collect large amounts of data with satisfactory accuracy. This paper focuses on the application of machine‐learning‐based predictive systems for the extraction of biomass information from LiDAR data. Biomass information has inmense ecological and economical value. We demonstrate the estimation of the Pinus radiata biomass in the Arratia‐Nervión region (Spain). Biomass estimation is considered a regression problem in which the ground truth for some specific sample sites is available. The promising results obtained in this study indicate that LiDAR data can be used to carry out detailed biomass mappings by the extrapolation of the models trained in this study.

show abstract

“…Common prediction methods in model-based or model-assisted forest inventory are e.g., linear models [1][2][3][4][5], non-linear models [6,7], random forest methods [8] and k nearest neighbour (k-NN) [2,[9][10][11] methods. Part of these methods include also information of the uncertainty in plot level predictions of the attributes.…”

Section: Introductionmentioning

confidence: 99%

“…This can be caused e.g., by poor correlation between extreme values of the attribute and used remotely sensed data based predictors, insufficient number of field measurements of the extreme attribute values in the available training set, or excessive averaging behaviour of the used model. For instance, the above ground biomass (AGB) can generally be predicted using LiDAR or satellite image data with good precision up to some limit, but for AGB values larger than that the models tend to systematically under-estimate the biomass compared to the real values, see e.g., [3,7,12].…”

Section: Introductionmentioning

confidence: 99%

Distribution Statistics Preserving Post-Processing Method With Plot Level Uncertainty Analysis for Remotely Sensed Data-Based Forest Inventory Predictions

Junttila¹,

Kauranne²

2018

Remote Sensing

View full text Add to dashboard Cite

Remotely sensed data-based models used in operational forest inventory usually give precise and accurate predictions on average, but they often suffer from systematic under- or over-estimation of extreme attribute values resulting in too narrow or skewed attribute distributions. We use a post-processing method based on the statistics of a proper, representative training set to correct the predictions and their probability intervals, attaining corrected predictions that reproduce the statistics of the whole population. Performance of the method is validated with three forest attributes from seven study sites in Finland with training set sizes from 50 to over 400 field plots. The results are compared to those of the uncorrected predictions given by linear models using airborne laser scanning data. The post-processing method improves the accuracy assessment linear fit between the predictions and the reference set by 35.4–51.8% and the distribution fit by 44.5–95.0%. The prediction root mean square error declines on the average by 6.3%. The systematic under- and over-estimation are reduced consistently with all training set sizes. The level of uncertainty is maintained well as the probability intervals cover the real uncertainty while keeping the average probability interval width similar to the one in uncorrected predictions.

show abstract

Improving Lidar-based aboveground biomass estimation of temperate hardwood forests with varying site productivity

Cited by 36 publications

References 39 publications

Assessment of the Carbon Stock in Pine Plantations in Southern Spain through ALS Data and K-Nearest Neighbor Algorithm Based Models

Assessment of the Carbon Stock in Pine Plantations in Southern Spain through ALS Data and K-Nearest Neighbor Algorithm Based Models

Estimation of forest biomass from light detection and ranging data by using machine learning

Distribution Statistics Preserving Post-Processing Method With Plot Level Uncertainty Analysis for Remotely Sensed Data-Based Forest Inventory Predictions

Contact Info

Product

Resources

About