Juan Suarez

Vegetation height estimates for a mixed temperate forest using satellite laser altimetry

Data from the Geoscience Laser Altimeter System (GLAS) aboard the Ice Cloud and land Elevation Satellite (ICESat) offer an unprecedented opportunity for canopy height retrieval at a regional to global scale. The data also provide useful information for forest stand level assessment at coincident locations. In this study height indices from light detection and ranging (LiDAR) waveforms were explored as a means of extracting canopy height; these were examined with reference to a mixed temperate forest in Gloucestershire, UK, containing planted stands with a mean age of 51 years and mean maximum height of 26.6 m. A method based on using a terrain index (TI; maximum minus minimum elevations from a 7×7 subset 10‐m resolution digital terrain model (DTM)) to adjust the waveform extent (WE; signal begin minus signal end) produced an R 2 value of 0.89 when regressed against field measurements of maximum canopy height at footprint locations (field height = 0.91(WE−TI)+4.86; root mean squared error (RMSE) = 2.99 m, coefficient significance p<0.001, intercept significance p>0.01). Multiple regression performed on both WE and TI with field measurements produced an R 2 of 0.90 and an RMSE of 2.86 m (field height = 1.0208WE−0.7310TI; coefficient significance p<0.001, intercept not significant). Maximum canopy height estimates using an automated approach to ground return identification based on iterative fitting of Gaussian peaks (GP1_2MAXAMP) to the waveform explained 74% of variance when compared to field measurements (field height = 1.05(GP1_2MAXAMP); RMSE = 4.53 m, coefficient significance p<0.001, intercept not significant). The ability of satellite LiDAR to retrieve data for such a complex and diverse area further indicates the potential of this technique for both carbon accounting and forest management.

A Monte Carlo radiative transfer model of satellite waveform LiDAR

We present a method and results for a model of the interaction of waveform Light Detection And Ranging (LiDAR) with a three-dimensional vegetation canopy. The model is developed from the FLIGHT radiative transfer model based on Monte Carlo simulations of photon transport. Foliage is represented by structural properties of leaf area, leaf-angle distribution, crown dimensions and fractional cover, and the optical properties of leaves, branch, shoot and ground components. The model represents multiple scattering of light within the canopy and with the ground surface, simulates the return signal efficiently at multiple wavebands and includes the effects of topography. LiDAR-emitted pulse and spatial and temporal sampling characteristics of the instrument are explicitly modelled. Agreement is found between the integrated waveform energy and directly derived bidirectional reflectance factors from FLIGHT (root mean square error < 0.01), and between simulated and observed Ice, Cloud and land Elevation Satellite (ICESat) Geoscience Laser Altimeter System (GLAS) waveforms for a complex forest site. A sensitivity analysis gives expected effects of canopy parameters on the waveform, and indicates potential for retrieval of the canopy properties of fractional cover and leaf area, in addition to height. Where canopy and ground pulses can be separated, an index derived from the waveform shows theoretical retrieval of vertically projected plant area index with correlation coefficient R 2 = 0.87.

Uncertainty within satellite LiDAR estimations of vegetation and topography

This paper demonstrates the ability to identify representative ground elevation and vegetation height estimates within the Ice, Cloud and land Elevation Satellite/Geoscience Laser Altimeter System (ICESat/GLAS) waveforms for an area of mixed vegetation and varied topography. Estimating vegetation height within large-footprint Light Detection and Ranging (LiDAR) waveforms relies on the ability to estimate the uppermost canopy surface (signal beginning) and an elevation representing the ground surface, both of which are influenced by vegetation properties and topographic slope. We examined sources of uncertainty for vegetation height estimation from ICESat/GLAS data using airborne LiDAR data, field measurements and the FLIGHT radiative transfer model. In comparison with an independent 10-m resolution digital terrain model (DTM), a method using Gaussian decomposition of the satellite waveform produced a mean bias of −0.10 m when estimating ground elevation. A second method of estimating vegetation height using waveform extent and a terrain index effectively removed slope as an error source but produced a greater ground surface offset (−0.83 m). The two methods of estimating vegetation height compared well with airborne LiDAR estimates (correlation coefficient (R 2) = 0.68, root mean square error (RMSE) = 4.4 m and R 2 = 0.61, RMSE = 4.9 m, respectively). However, the complex interplay of the structural and optical properties of the intercepted vegetation and slope requires further understanding. A tool such as FLIGHT provides a useful means to explore the sensitivity of the waveform to both vegetation properties and topographic slope.

A comparison of three methods for predicting wind speeds in complex forested terrain

Wind is one of the most important limiting factors for forestry in Britain. Most forestry plantations in this country have been established in upland areas on land marginal for agricultural use. These locations are commonly affected by high winds and poor soil conditions which make windthrow the most serious abiotic hazard in forestry (Miller, 1985). The effect of the wind is directly responsible for important losses of timber every year in Great Britain (NAO, 1993). Forest management options to improve the final crop quality are limited by windiness because it constrains the number of thinning regimes and shortens rotations. In addition, the quality of the timber is also affected by an increase in the proportion of compressed wood, poor stem straightness, repeated loss of leaders, which may developed crooked stems, and important alteration in the relation height‐diameter. Therefore sites affected by high levels of windiness tend to produce timber less suitable for higher value markets. Consequently, the determination of site vulnerability is important in order to minimise the risk of damage. Aspects influenced by site vulnerability include the choice of species for planting, the selection of silvicultural treatments (thinning and rotation length) and the estimation of the economic return on the investment in plantations (Miller, 1985).

A mixed pixel-and region-based approach for using airborne laser scanning data for individual tree crown delineation in Pinus radiata D. Don plantations

The aim of this study was to evaluate the use of high-resolution airborne laser scanner (ALS) data to detect and measure individual trees. We developed and tested a new mixed pixel- and region-based algorithm (using Definiens Developer 7.0) for locating individual tree positions and estimating their total heights. We computed a canopy height model (CHM) of pixel size 0.25 m from dense first-pulse point data (8 pulses m−2) acquired with a small-footprint discrete-return lidar sensor. We validated the results of individual tree segmentation with accurate field measurements made in 37 plots of Monterey pine (Pinus radiata D. Don) distributed over an area of 36 km2. Fieldwork consisted of labelling all of the trees in each plot and measuring their height and position, for posterior integration of the data from both sources (field and lidar). The proposed algorithm correctly detected and linked 59.8% of the trees in the 37 sample plots. We also manually located the trees by using FUSION software to visualize the raw lidar data cloud. However, because the latter method is extremely time-consuming, we only considered 10 randomly selected plots. Manual location correctly detected and linked 71.9% of the trees (in this subsample the algorithm correctly detected and measured 63.5% of the trees). The R2 values for the linear model relating field- and lidar-measured heights of the linked trees located manually and with the automatic location algorithm were 0.90 and 0.88, respectively.

A comparison of biophysical parameter retrieval for forestry using airborne and satellite LiDAR

This paper compares vegetation height metrics and fractional cover derived from coincident small footprint, discrete return airborne Light Detection and Ranging (LiDAR) scanning data (Optech Airborne Laser Terrain Mapper (ALTM)) with those estimated from large footprint, full waveform LiDAR profiling using the Geoscience Laser Altimeter System (GLAS). Estimates of maximum canopy height showed correspondence between the two methods with R2 = 0.68 (rms. error (RMSE) = 4.4 m). The relationship between 99th percentiles (often associated with forestry top height) showed R2 = 0.75, RMSE = 3.5 m. Detection of surface elevation limits corresponded well, (R2 = 0.71, RMSE = 5.0 m). Correlations between satellite waveform and airborne LiDAR canopy cover estimates gave R2 = 0.41 and R2 = 0.63 for dominant cover of conifers or broadleaf species, respectively. The results suggest that the broad Ice, Cloud and land Elevation Satellite (ICESat)/GLAS footprints can provide estimates of mixed vegetation canopy height which are comparable to those obtained from relatively high density airborne LiDAR data.

Evaluating Prospects for Improved Forest Parameter Retrieval From Satellite LiDAR Using a Physically-Based Radiative Transfer Model

A space-based full-waveform LiDAR system, optimised for vegetation analysis, offers the opportunity for global biophysical parameter retrieval of the worlds forests. However the conditions under which signals from the ground and vegetation can be detected will vary as a result of sensor specifications, vegetation characteristics and underlying surface properties. This paper demonstrates the utility of a ray tracing radiative transfer model for assessing sensitivity to site-specific conditions (e.g., topography, canopy and ground reflectance) that will improve our ability to estimate structural parameters in forest ecosystems.

Lidar Remote Sensing for Biomass Assessment

Optical remote sensing provides us with a two dimensional representation of land-surface vegetation and its reflectance properties which can be indirectly related to biophysical parameters (e.g. NDVI, LAI, fAPAR, and vegetation cover fraction). However, in our interpretation of the world around us, we use a three-dimensional perspective. The addition of a vertical dimension allows us to gain information to help understand and interpret our surroundings by considering features in the context of their size, volume and spatial relation to each other. In contrast to estimates of vegetation parameters which can be obtained from passive optical data, active lidar remote sensing offers a unique means of directly estimating biophysical parameters using physical interactions of the emitted laser pulse with the vegetation structure being illuminated. This enables the vertical profile of the vegetation canopy to be represented, not only permitting canopy height, metrics and cover to be calculated but also enabling these to be related to other biophysical parameters such as biomass. This chapter provides an overview of this technology, giving examples of how lidar data have been applied for forest biomass assessment at different scales from the perspective of satellite, airborne and terrestrial platforms. The chapter concludes with a discussion of further applications of lidar data and a look to the future towards emerging lidar developments.

Forestry Applications for Satellite Lidar Remote Sensing

This paper presents a method to estimate forest parameters and surface topography from NASAs Geosciences Laser Altimeter System (GLAS). Their potential use as observational inputs to models is demonstrated using a wind-risk model for the UK, ForestGALES. Relative heights above ground were used as biophysical parameter estimators. Top Height was estimated with R 2 = 0.73, RMSE = 4.5 m. Diameter at breast height estimates differed for conifer-dominated stands (R 2 = 0.72, RMSE = 0.07 m) and for stands containing mostly broadleaves (R 2 = 0.41, RMSE = 0.11 m). Ground elevation estimation produced R 2 = 0.997, RMSE = 2.2 m. These three parameters were applied to F orestGALES for stand-level assessment of wind-throw risk. Stability is sensitive to small differences in tree dimensions, and therefore vegetation parameters require greater accuracy than those currently retrievable from GLAS to more reliably determine risk of wind-throw. Future satellite lidar missions such as NASAs DESDynI sensor aim to produce improved vegetation parameter estimation plus greater spatial coverage which would offer more appropriate inputs for forestry models.

See the forests with different eyes

In this paper, biomass estimations are compared using quadpol ALOS PALSAR, TerraSAR-X and LIDAR data over Glen Affric, Scotland. Biomass was retrieved from the ALOS PALSAR and TerraSAR-X data using Yamaguchi decomposition to obtain useful information about the scattering mechanisms. A regression equation is obtained from the establishment of a relationship between the ratio of volume and surface scattering mechanism and the biomass obtained from fieldwork. Since the study site is a mountainous region, the terrain slope effects need to be compensated before retrieving the biomass. For LIDAR, the vertical structure of both the underlying topography and the forest structure were generated to estimate the biomass allometrically. Validation on the results of the biomass estimation was done by comparing the biomass estimated using ALOS PALSAR, TerraSAR-X and LIDAR. The results suggest that in some areas the biomass retrievals are broadly comparable.