Hans-Erik Andersen

Accuracy of a high-resolution lidar terrain model under a conifer forest canopy

Airborne laser scanning systems (commonly referred to as light detection and ranging or lidar systems) can provide terrain elevation data for open areas with a vertical accuracy of 15 cm. Accuracy in heavily forested areas has not been thoroughly tested. In this study, a high-resolution digital terrain model (DTM) was produced from high-density lidar data. Vegetation in the 500-ha mountainous study area varied from bare ground to dense 70-year-old conifer forest. Conventional ground survey methods were used to collect coordinates and near-ground vegetation heights at 347 ground checkpoints distributed under a range of canopy covers. These points were used to check the DTM accuracy. The mean DTM error was 0.22 ± 0.24 m (mean ± SD). DTM elevation errors for four tree canopy cover classes were: clearcut 0.16 ± 0.23 m, heavily thinned 0.18 ± 0.14 m, lightly thinned 0.18 ± 0.18 m, and uncut 0.31 ± 0.29 m. These DTM errors show a slight increase with canopy density but the differences are strikingly small. In general, the lidar DTM was found to be extremely accurate and potentially very useful in forestry.

A rigorous assessment of tree height measurements obtained using airborne lidar and conventional field methods

Tree height is an important variable in forest inventory programs but is typically time-consuming and costly to measure in the field using conventional techniques. Airborne light detection and ranging (lidar) provides individual tree height measurements that are highly correlated with field-derived measurements, but the imprecision of conventional field techniques does not allow for definitive assessments regarding the absolute accuracy of lidar tree height measurements and the relative influence of beam divergence setting (i.e., laser footprint size), species type, and digital terrain model (DTM) error on the accuracy of height measurements. In this study, we developed a methodology for acquiring accurate individual tree height measurements (<2 cm error) using a total station survey and used these measurements to establish the expected accuracy of lidar- and field-derived tree height measurements for two of the most ecologically and commercially significant species in western North America, Douglas-fir (Pseudotsuga menziesii) and ponderosa pine (Pinus ponderosa). Tree height measurements obtained from narrow-beam (0.33 m), high-density (6 points/m2) lidar were more accurate (mean error ± SD = –0.73 ± 0.43 m) than those obtained from wide-beam (0.8 m) lidar (–1.12 ± 0.56 m). Lidar-derived height measurements were more accurate for ponderosa pine (–0.43 ± 0.13 m) than for Douglas-fir (–1.05 ± 0.41 m) at the narrow beam setting. Although tree heights acquired using conventional field techniques (–0.27 ± 0.27 m) were more accurate than those obtained using lidar (–0.73 ± 0.43 m for narrow beam setting), this difference will likely be offset by the wider coverage and cost efficiencies afforded by lidar-based forest survey.

Validation of the ICEsat vegetation product using crown-area-weighted mean height derived using crown delineation with discrete return lidar data

The Geoscience Laser Altimeter System (GLAS), a spaceborne light detection and ranging (lidar) sensor, has acquired over 250 million lidar observations over forests globally, an unprecedented dataset of vegetation height information. To be useful, GLAS must be calibrated to measurements of height used in forestry inventory and ecology. Airborne discrete return lidar (DRL) can characterize vegetation and terrain surfaces in detail, but its utility as calibration data for GLAS is limited by the lack of a direct relationship between the canopy height measurements collected by airborne and spaceborne lidar systems and coincident field data. We demonstrate that it is possible to use DRL to directly estimate the crown-area-weighted mean height (Hcw), which is conceptually and quantitatively similar to the Lorey’s height, which is calculated from forest inventory data, and can be used to calibrate GLAS without the use of field data. For a dataset from five sites in western North America, the two indices of height (Hcw from DRL and Lorey’s from forest inventory) are directly related (r2 = 0.76; RMSE of 3.8 m; intercept and slope of 0.8 m and 0.98, respectively). We derived a relationship between the DRL-derived Hcw and height information from coincident GLAS waveforms; the resulting equation explained 69% of variance, with an RMSE of 6.2 m.

Using multilevel remote sensing and ground data to estimate forest biomass resources in remote regions: a case study in the boreal forests of interior Alaska

The emergence of a new generation of remote sensing and geopositioning technologies, as well as increased capabilities in image processing, computing, and inferential techniques, have enabled the development and implementation of increasingly efficient and cost-effective multilevel sampling designs for forest inventory. In this paper, we (i) describe the conceptual basis of multilevel sampling, (ii) provide a detailed review of several previously implemented multilevel inventory designs, (iii) describe several important technical considerations that can influence the efficiency of a multilevel sampling design, and (iv) demonstrate the application of a modern multilevel sampling approach for estimating the forest biomass resources in a remote area of interior Alaska. This approach utilized a combination of ground plots, lidar strip sampling, satellite imagery (multispectral and radar), and classified land cover information. The variability in the total biomass estimate was assessed using a bootstrapping approach. The results indicated only marginal improvement in the precision of the total biomass estimate when the lidar sample was post-stratified using the classified land cover layer (reduction in relative standard error from 7.3% to 7.0%), whereas there was a substantial improvement in the precision when the estimate was based on the biomass map derived via nearest-neighbor imputation (reduction in relative standard error from 7.3% to 5.1%).

Effects of lidar pulse density and sample size on a model-assisted approach to estimate forest inventory variables

Using lidar in an area-based model-assisted approach to forest inventory has the potential to increase estimation precision for some forest inventory variables. This study documents the bias and precision of a model-assisted (regression estimation) approach to forest inventory with lidar-derived auxiliary variables relative to lidar pulse density and the number of sample plots. For managed forests on the Lewis portion of the Lewis-McChord Joint Base (35025 ha, 23290 forested) in western Washington state, we evaluated a regression estimator for combinations of pulse density (.05–3 pulses/m2) and sample size (15–105 plots) to estimate five forest yield variables: basal area, volume, biomass, number of stems, and Loreys height. The results indicate that there is almost no loss in precision in using as few as .05 pulses/m2 relative to 3 pulses/m2. We demonstrate that estimation precision declined quickly for reduced sample sizes (as expected from sampling theory); but of more importance we demonstrate that sample size has a dramatic effect on the validity of inferences. Our investigations indicate that for our test dataset that central limit theorem based confidence intervals were too small on average for sample sizes smaller than 55. The results from this study can aid in identifying design components for forest inventory with lidar which satisfy users’ objectives.

Predicting live and dead tree basal area of bark beetle affected forests from discrete-return lidar

Bark beetle outbreaks have killed large numbers of trees across North America in recent years. Lidar remote sensing can be used to effectively estimate forest biomass, but prediction of both live and dead standing biomass in beetle-affected forests using lidar alone has not been demonstrated. We developed Random Forest (RF) models predicting total, live, dead, and percent dead basal area (BA) from lidar metrics in five different beetle-affected coniferous forests across western North America. Study areas included the Kenai Peninsula of Alaska, southeastern Arizona, north-central Colorado, central Idaho, and central Oregon, U.S.A. We created RF models with and without intensity metrics as predictor variables and investigated how intensity normalization affected RF models in Idaho. RF models predicting total BA explained the most variation, whereas RF models predicting dead BA explained the least variation, with live and percent dead BA models explaining intermediate levels of variation. Important metrics varied between models depending on the type of BA being predicted. Generally, height and density metrics were important in predicting total BA, intensity and density metrics were important in predicting live BA, and intensity metrics were important in predicting dead and percent dead BA. Several lidar metrics were important across all study areas. Whether needles were on or off beetle-killed trees at the time of lidar acquisition could not be ascertained. Future work, where needle conditions at the time of lidar acquisition are known, could improve upon our analysis and results. Although RF models predicting live, dead, and percent dead BA did not perform as well as models predicting total BA, we concluded that discrete-return lidar can be used to provide reasonable estimations of live and dead BA. Our results also showed which lidar metrics have general utility across different coniferous forest types.

Using Airborne Laser Scanning Data to Support Forest Sample Surveys

Forest surveys, in the form of both stand management and strategic inventories, have a long history of using remotely sensed data to support and enhance their design and estimation processes. By the use of airborne laser scanning data this capacity has emerged as one of its most important and prominent applications. The chapter includes a brief overview of forest inventory uses of remotely sensed data, a section on aspects of ground sampling that can be managed to optimize estimation of relationships between ground and airborne laser scanning (ALS) data, and a section on stand management inventories. The latter section reviews underlying and motivating factors crucial to the primary focus of the chapter, formal statistical inference for ALS-assisted forest inventories. Inferential methods are described for two primary cases, full and partial ALS coverage. Within each case, estimators for both design-based and model-based inference are presented.

Assessing the influence of flight parameters, interferometric processing, slope and canopy density on the accuracy of X-band IFSAR-derived forest canopy height models

High resolution, active remote sensing technologies, such as interferometric synthetic aperture radar (IFSAR) and airborne laser scanning (lidar) have the capability to provide forest managers with direct measurements of 3‐dimensional forest canopy surface structure. While lidar systems can provide highly accurate measurements of canopy and terrain surfaces, high resolution (X‐band) IFSAR systems provide slightly less accurate measurements of canopy surface elevation over very large areas with a much higher data collection rate, leading to a lower cost per unit area. In addition, canopy height can be measured by taking the difference between the IFSAR‐derived canopy surface elevation and a lidar‐derived terrain surface elevation. Therefore, in areas where high‐accuracy terrain models are available, IFSAR may be used to economically monitor changes in forest structure and height over large areas on a relatively frequent basis. However, IFSAR flight parameters and processing techniques are not currently optimized for the forest canopy mapping application. In order to determine optimal flight parameters for IFSAR forest canopy measurement, we evaluated the accuracy of high resolution, X‐band canopy surface models obtained over a mountainous forested area in central Washington state (USA) from two different flying heights (6000 m and 4500 m), from different look directions, and with different interferometric processing. In addition, we assessed the influence of terrain slope and canopy density on the accuracy of IFSAR canopy height models. High‐accuracy lidar‐derived canopy height models were used as a basis for comparison. Results indicate that sensing geometry is the single most important factor influencing the accuracy of IFSAR canopy height measurements, therefore acquiring IFSAR from multiple look directions can be critically important when using IFSAR for forest canopy measurement applications, especially in mountainous areas.

Accuracy of an IFSAR-derived digital terrain model under a conifer forest canopy

Accurate digital terrain models (DTMs) are necessary for a variety of forest resource management applications, including watershed management, timber harvest planning, and fire management. Traditional methods for acquiring topographic data typically rely on aerial photogrammetry, where measurement of the terrain surface below forest canopy is difficult and error prone. The recent emergence of airborne P-band interferometric synthetic aperture radar (IFSAR), a high-resolution, microwave remote sensing technology, has the potential to provide significantly more accurate terrain models in forested areas. Low-frequency, P-band radar energy physically penetrates through the vegetation canopy and reflects from the underlying terrain surface, allowing for accurate measurement of the terrain surface elevation even in areas with dense forest cover. In this study, the accuracy of a high-resolution DTM derived from P-band IFSAR data collected over a mountainous forest area in western Washington State was rigorously evaluated through a comparison with 347 topographic checkpoints measured with total station survey equipment and collected under a variety of canopy densities. The mean DTM error was –0.28 ± 2.59 m (mean ± standard deviation), and the root mean squared error (RMSE) was 2.6 m. DTM elevation errors for four canopy cover classes were –0.67 ± 1.20 m (RMSE = 1.38 m) for clearcut, –0.62 ± 1.00 m (RMSE = 1.18 m) for heavily thinned, –0.41 ± 2.32 m (RMSE = 2.36 m) for lightly thinned, and 0.20 ± 3.31 m (RMSE = 3.32 m) for uncut.

Impact of Topographic Correction on Estimation of Aboveground Boreal Biomass Using Multi-temporal, L-Band Backscatter

Synthetic aperture radar (SAR) has been shown to be a useful tool for estimating aboveground biomass (AGB), due to the strong correlation between the biomass and backscatter. In particular, L-band SAR is effective for estimating the lower range of biomass that characterizes most boreal forests. Unfortunately, the topographic impact on backscatter can dominate the normal forest signal variation. Since many boreal environments have significant topography, we investigate several topographic correction techniques to determine their effect upon AGB prediction accuracy. Different approaches to addressing the topography include: 1) no correction, 2) local incidence angle (LIA) correction, 3) pixel-area correction, and 4) a novel empirical slope correction. The investigation was performed for a data-rich experimental area near Tok, Alaska, for which Advanced Land Observing Satellite Phased Array type L-Band Synthetic Aperture Radar (ALOS PALSAR), field plots, lidar acquisitions, and a high-quality digital elevation model (DEM) existed. Biomass estimations were performed using both single- and dual-polarization (HH and HV) regressions against field plot data. The biomass estimation for each of the topographic corrections was compared with the field plot biomass, as well as more extensive lidar biomass estimations. The results showed a clear improvement in AGB estimation accuracy from no correction, to LIA, to pixel-area, to the novel pixel-area plus empirical slope correction. Using the field plot data for validation, the SAR root mean square error (RMSE) derived from the best approach was found to be 37.3 Mg/ha over a biomass range of 0-250 Mg/ha, only marginally less accurate than the 33.5 Mg/ha accuracy of the much more expensive lidar technique.