Improving Stand Structural, Volume, and Species Information at Local and Regional Scales for an Enhanced Forest Inventory
The increasing availability of airborne and ground-based LiDAR and high spatial resolution optical imagery from airborne (DMSI, digital camera imagery, CASI) and spaceborne (Quickbird, Ikonos, Worldview, RapidEye) programs has rapidly increased the availability of relevant remotely sensed data for forest inventories. However, there are research questions that need to be addressed before the widespread adoption of these advanced technologies.
Developing predictive forest growth models which incorporate terrain, and site conditions, physiology, climate, LiDAR, optical imagery, and other geospatial data has the potential to provide significant advances in our capacity to estimate stand structure, wood attributes as well as other stand characteristics. The integration of airborne LiDAR and other remote sensing into inventory for these purposes are still in its relative infancy; however some researchers in Canada are building these predictive systems. In the area of ecological goods and services and habitat assessment, the increased availability of LiDAR data to management agencies, and its recent adoption by forestry companies, allows the assessment of the capacity to map and model aspects of non-timber values of the forest such as water availability, viewshed assessments, and habitat for species, such as deer, moose and caribou. As a result, the additional benefit that LiDAR data can provide, in addition to more traditional forest inventory assessment needs to be investigated.
When assessing forest health status using remote sensing technology, changes in both the spectral reflectance and physical structure of tree crowns are minor during the early stages of infestation. As a result, detection of the initial onset of attack of insects by any remote sensing method is likely to be very difficult. Detection becomes easier as trees move into more advanced stages of attack, such as leaf defoliation and discolouration (Wulder et al., 2006, and references therein). There is some potential however. First, in situations where changes in structure are apparent associated with decline in a stand, for example, changes in foliage density and tree and stand structure, then LiDAR may be able to detect these changes, through the use of cover metrics as well as the vertical distribution of leaf surfaces. Perhaps more important is the concept of loss of vigour and condition of the stand prior to infestation which in turn makes the stand more susceptible to attack. Research using optical remote sensing data, linked to climate and forest growth models, has demonstrated that stands which have less than optimum characteristics for a given the age and species composition can be identified and linked to stands of subsequent attack (Coops et al., 2009).
New structural metrics and links to stand timber characteristics and wood properties:
With the increasing availability of LiDAR and high spatial resolution optical data, forest managers have seen increasing opportunities for using these data to meet a wider range of forest inventory information needs (Nelson et al., 2003). To date however, the number of attributes derived has been limited. Moreover, predictions linking tree, stand, and terrain indices with wood attributes are a relatively nascent field of inquiry. Key attributes derived from airborne LiDAR include height, which has been found to be of similar or better accuracy than correspondingfield-based estimates (Næsset and Økland, 2002), volume, (Nilsson, 1996), biomass (Popescu et al., 2004; Hyde et al., 2007), and crown closure (Holmgren et al., 2003). Advances in LiDAR technology have resulted in a greater pulse rate and a larger number of discrete-returns from each LiDAR pulse. As well, advances in small-footprint, full-waveform systems offer even finer-scale vertical discrimination (Wagner et al., 2006). As a result, a new generation of LiDAR structural metrics are being developed and must be explored, with a likely focus on the vertical distribution of foliage within the canopy, which is a strong indicator of stand age, height, density, successional status, stand developmental stage, and competition (shading and crowding), all of which are expected to influence forest structure and stand growth. Likewise, the cumulative foliage area on a per-tree basis has been found to correlate with sapwood cross-sectional area (Waring et al., 1982).
Of equal if not more importance is the linking of the LiDAR attributes to stem dimensions, including metrics such as stem class distribution, log size classes, and lumber grade metrics. To date, some analysis of this type has been taken over study areas and published in the literature (e.g., Gobakken and Næsset 2004; Thomas et al,. 2006; Packalén and Maltamo, 2008; Ozdemir and Donoghue, 2013). However, broader scale assessment is needed, as well as an assessment of if these types of relationships hold over multiple forest types and species compositions. Lastly, the role of full-waveform vs discrete-return airborne LiDAR data needs to be more comprehensively addressed. date, a number of LiDAR providers are acquiring full-waveform data, so the compatibility of metrics derived from full-waveform vs discrete systems needs to be addressed. We are fortunate that within the New Brunswick core site, full-waveform and discrete waveform data exists, allowing an inter-comparison study to be undertaken.
Hydrological Mapping and Productivity:
The use of small-footprint, discrete-return LiDAR has also been well demonstrated for the derivation of high spatial resolution digital terrain models (DEM), which provide a detailed representation of the Earth’s bare surface. Previous research has shown that the accuracy of DEMs varies with changes in terrain and land cover type (e.g. Adams and Chandler, 2002; Hodgson and Bresnahan, 2004; Hodgson et al., 2005; Su and Bork, 2006) however these errors are typically small when compared to photogrametrically derived digital terrain models, especially under forested canopies (Adams and Chandler 2002, Hodgson et al., 2005). A number of key terrain attributes can be derived from the LiDAR digital terrain models and can provide extremely detailed information on the underlying hydrology and access to water across forested landscapes. These layers have potential to improve forest management activities, both in terms of improving site index predictions of overall productivity and in driving more sustainable decisions around resource extraction. This will allow industry and regulators to more accurately address the presence of these small water features, road placement, trafficability and erodibility issues, as well as the placement of bridges and culverts, in the policy formulation and planning process (White et al., 2012). Wet Area Mapping offers one approach, as does the use of additional airborne LiDAR derived terrain indices, to improve our understanding of forest growth and overall site productivity. This research will investigate the integration of WAM, with other terrain data, climate and vegetation structure more holistically to produce site index predictions in New Brunswick. We will also investigate the integration of existing growth and yield, and forestry information with LiDAR derived hydrological indicators and forest structural data to help build land classification schemes which can drive estimates of potential forest growth.
Within this theme, issues associated with the accurate prediction of forest growth for yield modelling is paramount – a number of options exist. For example, the use of two LiDAR spatially registered datasets (acquired at different points in time), the use of LiDAR acquired at one time step and photogrammetric, stereo matching at a second time step, or the combination of multi-date optical imagery (to characterize depletions) fused with LiDAR data (to characterize growth). Each of these options comes with different cost and accuracy considerations (Wulder et al., 2008). Recently there is an increased interest in the generation of canopy height models from a combination of high spatial resolution digital aerial photography and LiDAR. Semi-Global Matching (SGM) is an approach which can be used to generate high density point clouds from a stereo pair of digital images (Vastaranta et al., 2012). As a result the combination of an initial DEM derived from LiDAR data with a second later acquisition of digital photos may be a cost effective way to derived forest height change and is worthy of additional research. Research is needed to better understand, across Canadian forested ecosystems, the amount of time needed for sufficient growth to exceed noise and other uncertainties within the LiDAR systems as well as better understand the impact of growth increment of different species associations and site conditions on LiDAR change in height metrics. The link of these growth estimates to traditional growth and yield curves also remains a poorly researched area.
Theme 2 Questions:
How can advanced remote sensing technologies improve stand-level attribute estimation for strategic forest management?
Focus Area 1: New structural metrics for forest stand description
- Q4: What new LiDAR stand structure metrics can be developed and linked to forest inventory attributes such as lumber grade and log size class?
- Q5: How well do existing LiDAR metrics developed in the Eastern Mixedwood Boreal forests transfer to Western Mixedwood Boreal forest types?
- Q6: How can LiDAR be used to improve pre-harvest ground verification approaches and assess optimum year of commercial thinning?
- Q7: How can ALS LiDAR be used to assess product mix based on stem class distribution in mixed boreal systems?
- Q8: How can structural metrics from LiDAR of tree crown or stand canopies be used to predict wood fibre properties?
Focus Area 2: Wet Area Mapping (WAM) and productivity
- Q9: What hydrologic indicators can be derived from a LiDAR-based DEM to support land classification schemes relevant for forest management?
- Q10: How can LiDAR terrain indices be combined with structural and climate data to derive new productivity / site classification schemes?
Focus Area 3: Growth rate
- Q11: How can Semi-Global Matching (SGM)) techniques be used, when paired with existing LiDAR terrain models, to predict forest stand growth?
- Q12: How can temporal LiDAR datasets be used to predict boreal height growth and linked into existing growth and yield models?
- Q13: Can we improve forest growth projections with LiDAR-enhanced forest inventory data?
Focus Area 4: Predictive modelling
- Q14: What is the impact of changing site conditions on timber and wood quality over time?
- Q15: How can airborne LiDAR and high spatial resolution optical imagery be used to augment conventional forest health surveys?
- Q16: What is the best approach to map new “value-based” forest attributes combining conventional forest inventory and LiDAR data?
- Q17: What landscape level ecosystem goods and service indicators can be developed with airborne LiDAR data?
- Q18: How can forest attributes derived from multi-resolution remote sensing data be used to conduct habitat assessment?