Chris Hopkinson

Vegetation class dependent errors in lidar ground elevation and canopy height estimates in a boreal wetland environment

An airborne scanning light detection and ranging (lidar) survey using a discrete pulse return airborne laser terrain mapper (ALTM) was conducted over the Utikuma boreal wetland area of northern Alberta in August 2002. These data were analysed to quantify vegetation class dependent errors in lidar ground surface elevation and vegetation canopy surface height. The sensitivity of lidar-derived land-cover frictional parameters to these height errors was also investigated. Aquatic vegetation was associated with the largest error in lidar ground surface definition (+0.15 m, SD = 0.22, probability of no difference in height P < 0.01), likely a result of saturated ground conditions. The largest absolute errors in lidar canopy surface height were associated with tall vegetation classes; however, the largest relative errors were associated with low shrub (63%, –0.52 m, P < 0.01) and aquatic vegetation (54%, –0.24 m, P < 0.01) classes. The openness and orientation of vegetation foliage (i.e., minimal projection of horizontal area) were thought to enhance laser pulse canopy surface penetration in these two classes. Raster canopy height models (CHMs) underestimated field heights by between 3% (aspens and black spruce) and 64% (aquatic vegetation). Lidar canopy surface height errors led to hydraulic Darcy–Weisbach friction factor underestimates of 10%–49% for short (<2 m) vegetation classes and overestimates of 12%–41% for taller vegetation classes.

The influence of flying altitude, beam divergence, and pulse repetition frequency on laser pulse return intensity and canopy frequency distribution

Eight airborne light detection and ranging (lidar) data collections were carried out over a forested and agricultural study site in Nova Scotia during 2005. The influences of flying altitude, beam divergence, and pulse repetition frequency on laser pulse return intensities and vertical frequency distributions within vegetated environments were investigated. Experimental control was maintained by varying each survey configuration setting independently while keeping all other settings constant. The land covers investigated were divided into highway, tall vegetation (mature and immature mixed wood regeneration stands), and short vegetation (hay field and potato crop). Laser pulse return data for 24 tall and 18 short vegetation plots were extracted, and the quartile heights of each vegetation profile were compared for each configuration. Observed laser pulse intensity values were found to be linearly related (coefficient of determination r2 = 0.98) to the peak pulse power concentration. A simple routine was developed to allow intensity data to be normalized and made comparable across datasets. By comparing the intensity and laser pulse return profiles it was found that reducing the peak pulse power concentration by widening the beam, increasing the flying altitude, or increasing the pulse repetition frequency tends to lead to (i) slightly reduced penetration into short canopy foliage by up to 4 cm, and (ii) increased penetration into tall canopy foliage (i.e., reduced maximum canopy return heights) by 15-61 cm. It is believed that a reduction in peak pulse power concentration delays pulse triggering within vegetation (i.e., increases penetration of the pulse into foliage) due to the need for increased surface area backscatter to raise the return pulse energy above some minimal threshold within the timing electronics of the sensor. Exceptions to these general observations were found in the high pulse repetition frequency data, where increased sample point density results in (i) increased noise and height range in the lidar distribution data, and (ii) increased likelihood of ground returns in the tall canopies sampled due to increased probability of pulses encountering canopy gaps. The implications of these results are that (i) laser pulse peak power concentration is the largest determinant of pulse return intensity and survey configuration based variations in canopy frequency distribution, and (ii) laser pulse height- and intensity-based models developed for vegetation structural or biomass assessment could be improved if they accounted for variations in peak power concentration.

Examining the Influence of Changing Laser Pulse Repetition Frequencies on Conifer Forest Canopy Returns

The distribution of laser pulses within conifer forest trees and canopies are examined by varying the rate of laser pulse emission and the inherent laser pulse properties (laser pulse energy, pulse width, pulse length, and roll-over or trigger time). In this study, an Optech, Inc. ALTM 3100 airborne lidar is used, emitting pulses at 50 kHz and 100 kHz, allowing for changes in laser pulse characteristics while also keeping all other survey parameters equal. We found that: 1. Pulses and associated characteristics emitted at 50 kHz penetrated further into the canopy than 100 kHz for a significant number of individual trees. 2. At tall tree plots with no understory, pulses emitted at 50 kHz penetrated further into the canopy than 100 kHz for a significant number of plots. 3. For plots with significant understory and shorter trees, pulses emitted at 100 kHz penetrated further into the canopy than 50 kHz. We suspect that this may be due, in part, to canopy openness. Laser pulse energy and character differences associated with different laser pulse emission frequencies are likely a contributing factor in laser pulse penetration through the canopy to the ground surface. Efforts to understand laser pulse character influences on canopy returns are important as biomass and vegetation structure models derived from lidar are increasingly adopted.

Towards a universal lidar canopy height indicator

A light detection and ranging (lidar) canopy height study was conducted with 13 datasets collected using four different models of airborne laser terrain mapper (ALTM) sensors over 13 widely variable vegetation types ranging in average height from <1 m to 24 m at five sites across Canada between 2000 and 2005. The study demonstrates that the vertical standard deviation of all topographically detrended first and last laser pulse returns (LSD) is a robust estimator of canopy height (Ht) for a wide variety of vegetation types and heights and lidar survey configurations. After regressing Ht against LSD for 77 plots and transects, it was found that Ht could be predicted as a simple multiplication (M) of LSD (M = 2.5, coefficient of determination (r2) = 0.95, root mean square error (RMSE) = 1.8 m, tail probability (p) < 0.01). For forest plots only, LSD was found to better predict average tree height (r2 = 0.80, RMSE = 2.1 m, p < 0.01) than Loreys height (r2 = 0.59, RMSE = 3.0 m, p < 0.01). A test of the LSD canopy height model was performed using stand heights (HtFRI) from an independent forest resource inventory (FRI) for four vegetation classes. Results from the raw FRI and modelled stand height comparison displayed close to a 1:1 relationship (HtFRI = 0.97HtLSD, r2 = 0.73, RMSE = 4.7 m, p < 0.01, n = 38). All plot and transect canopy heights were also compared with the localized maxima of laser pulse returns (Lmax). For individual surveys over homogeneous vegetation types, Lmax generally provides a better canopy height indicator. Across all surveys and site types, however, LSD was almost always shown to have a more consistent relationship with actual canopy height. The only observed exception was in the case of forest plot level Loreys mean tree height. The advantages of using a multiplier of LSD to estimate canopy height are its apparent insensitivity to survey configuration and its demonstrated applicability to a range of vegetation types and height classes.

Mapping Snowpack Depth beneath Forest Canopies Using Airborne Lidar

An evaluation of airborne lidar (Light Detection And Ranging) technology for snow depth mapping beneath different forest canopy covers (deciduous, coniferous, and mixed) is presented. Airborne lidar data were collected for a forested study site both prior to and during peak snowpack accumulation. Manual field measurements of snow depth were collected coincident with the peak snowpack lidar survey, and a comparison between field and lidar depth estimates was made. It was found that (1) snow depth distribution patterns can be mapped by subtracting a “bare-earth” DEM from a “peak snowpack” DEM, (2) snow depth estimates derived from lidar data are strongly related to manual field measures of snow depth, and (3) snow depth estimates are most accurate in areas of minimal understory. It has been demonstrated that airborne lidar data provide accurate snow depth data for the purpose of mapping spatial snowpack distribution for volume estimations, even under forest canopy conditions.

Lidar plots * a new large-area data collection option: context, concepts, and case study

Forests are an important global resource, playing key roles in both the environment and the economy. The implementation of quality national monitoring programs is required for the generation of robust national statistics, which in turn support global reporting. Conventional monitoring initiatives based on samples of field plots have proven robust but are difficult and costly to implement and maintain, especially for large jurisdictions or where access is difficult. To address this problem, air photo- and satellite-based large area mapping and monitoring programs have been developed; however, these programs also require ground measurements for calibration and validation. To mitigate this need for ground plot data we propose the collection and integration of light detection and ranging (lidar) based plot data. Lidar enables accurate measures of vertical forest structure, including canopy height, volume, and biomass. Rather than acquiring wall-to-wall lidar coverage, we propose the acquisition of a sample of scanned lidar transects to estimate conditions over large areas. Given an appropriate sampling framework, statistics can be generated from the lidar plots extracted from the transects. In other instances, the lidar plots may be treated similar to ground plots, providing locally relevant information that can be used independently or integrated with other data sources, including optical remotely sensed data. In this study we introduce the concept of “lidar plots” to support forest inventory and scientific applications, particularly for large areas. Many elements must be considered when planning a transect-based lidar survey, including survey design, flight and sensor parameters, acquisition considerations, mass data processing, and database development. We present a case study describing the acquisition of over 25 000 km of lidar data in Canadas boreal forests in the summer of 2010. The survey, which included areas of managed and unmanaged forests, resulted in the production of more than 17 million 25 × 25 m lidar plots with first returns greater than 2 m in height. We conclude with insights gained from the case study and recommendations for future surveys.

The effect of glacier wastage on the flow of the Bow River at Banff, Alberta, 1951-1993

A surface area/volume relationship was used to estimate total glacier volumes for the highly glacierized Hector Lake Basin (281 km 2 ) in the Canadian Rockies in the years 1951 and 1993. The change in volume was calculated and this value then extrapolated up to the Bow Basin at Banff (2230 km 2 ) based on relative proportions of glacier cover. The mean net glacier volume loss estimate of 934 × 10 6 m 3 was divided into annual proportions of glacier wastage and storage using a local mass balance record collected at Peyto Glacier in the Mistaya Valley, contiguous to the Bow Basin. Unfortunately, the record began in 1966 and a hind-cast to 1952 (hydrological year) was necessary. Banff maximum summer temperature and Lake Louise snow course data were used as surrogates for summer and winter glacier mass balance, respectively. Monthly wastage proportions were estimated for 1967 1974 by using modelled values of glacial melt as a template. Glacier wastage inputs to and storage held back from the Bow River hydrograph at Banff were compared with known basin yields to assess the hydrological effects of glacier volume change For 1952-1993, the average annual wastage/basin yield ratio was found to be around 1.8%. For the extremely low flow year of 1970 this ratio increased to 13%. The proportion of flow derived from glacier wastage in August of this year was estimated to be around 56%. Although the results tend to confirm the regulatory effect of glaciers on stream flow, it was found that in some years of low flow this situation has been aggravated by water being held in glacial storage.

Using airborne lidar to assess the influence of glacier downwasting on water resources in the Canadian Rocky Mountains

Knowledge of the changing dimensions of alpine glacier surfaces is critical from both a water resources and climate change indication perspective. With the development of airborne light detection and ranging (lidar) technologies with the capability to rapidly map large areas of topography at high resolutions, there is a need to assess the utility of this technology for glacier surface change detection and water resources assessment. The study presented here compares two lidar digital elevation models (DEMs) collected 23 months apart in 2000 and 2002 over Peyto Glacier, Canadian Rocky Mountains, for the purposes of intensity image feature recognition and surface downwasting assessment. The 2002 DEM was subtracted from the 2000 DEM to quantify the total and spatial variability in surface downwasting (or growth) within the glacial and periglacial environments. It was found that there was a reduction in volume totaling 33 × 106 m3 over the Peyto Glacier surface and surrounding ice-cored moraines. This downwasting was estimated to be equivalent to approximately 22 × 106 m3 of water volume and, after extrapolation, 16% of total basin runoff. The water-equivalent contribution from ice-cored moraines was estimated to be 6% of the total glacier runoff contribution, highlighting the importance of monitoring this component of glacial melt.

The forward propagation of integrated system component errors within airborne lidar data

Error estimates of lidar observations are obtained by applying the General Law of Propagation of Variances ( GLOPOV ) to the direct georeferencing equation. Within the formulation of variance propagation, the most important consideration is the values used to describe the error of the hardware component observations including the global positioning system, inertial measurement unit, laser ranger, and laser scanner (angular encoder noise and beam divergence). Data tested yielded in general, pessimistic predictions as 85 percent of residuals were within the predicted error level. Simulated errors for varying scan angles and altitudes produced horizontal errors largely influenced by IMU subsystem error as well as angular encoder noise and beam divergence. GPS subsystem errors contribute the largest proportion of vertical error only at shallow scan angles and low altitudes. The transformation of the domination of GPS related error sources to total vertical error occurs at scan angles of 23°, 13°, and 8° at flying heights of 1,200 m, 2,000 m, and 3,000 m AGL , respectively.

Quantifying errors in discontinuous permafrost plateau change from optical data, Northwest Territories, Canada: 1947-2008.

The discontinuous permafrost zone has been subject to increased air temperatures over recent decades. Permafrost thaw can cause changes to topography, hydrology, vegetation, and trace gas fluxes, and thus it is important to monitor changes in permafrost area through time. Optical imagery can be used to generate time-series databases of near-surface spectra that may be related to permafrost area. This provides a spatial perspective on area permafrost change that is not easily obtained from field data alone. This study examines the cumulative maximum and minimum errors of aerial and satellite imagery used for change detection within the Scotty Creek watershed, Fort Simpson, NWT, Canada. The results illustrate that, unless unchanging linear features are found throughout every image used (e.g., to be used as multitemporal tie points) and radiometric normalization can be applied (problematic for film images), direct image to image comparisons (e.g., subtraction) are not appropriate. Further, measureable cumulative errors are often produced by misclassification of edges, resolution limitations, and increased landscape fragmentation. At Scotty Creek, increased fragmentation of permafrost plateaus occurred from 1947 to 2008. Cumulative maximum and minimum errors result in an approximate 8%–26% error in permafrost area when compared with the total area of the site. Rates of permafrost area reduction within the study area were approximately 0.5% every year, determined from linear correlation (r2 = 0.91, n = 5). Therefore, based on the maximum cumulative error (a worst-case scenario), approximately 21–32 years (for resolutions of 0.18–1.10 m) is required between images to approximate change within this particular site. Increased (decreased) rates of change at other sites will decrease (increase) the timing required to identify change between images beyond error bounds.