Conifer Crown Fuel Modeling: Current Limits and Potential for Improvement

Research from the wildland fire community during the past decade has targeted this area of decision support for advancement (e.g., Scott and Reinhardt 2001, Cruz et al. 2003, Keane et al. 2005). The need for better stand-level estimates of CBH and CBD prompted plot-scale intensive deconstruction of tree crowns of five western species (Reinhardt et al. 2006). That work resulted in the development of correlative relationships (Keane et al. 2005) and presumably better decision support tools (Scott and Reinhardt 2005) for managers assigning values for those parameters in fire and fuels planning software. Based on this and earlier work, crown fuel attributes are now calculated by managers with few exceptions via single-tree allometries applied to standard forest inventory plot data, most typically by using the Fire and Fuels Extension of the Forest Vegetation Simulator (FVS-FFE; Reinhardt and Crookston 2003). Yet, the intensive methodology applied by Reinhardt et al. (2006) did not permit analysis of crown and canopy features for stands of varying structures or treatment histories. Major weaknesses were exposed in employing FVS-FFE’s existing CBD and CBH algorithms in the Black Hills of South Dakota (Keyser and Smith 2010). Keyser and Smith demonstrated clearly that better models of crown fuels that include more accurate representations of vertical structure and that accommodate variations in local site and stand conditions (e.g., density and structure) are needed. Similarly, using Keyser and Smith’s Black Hills data, Cruz and Alexander (2012) found that, whereas the stand-level CBH and canopy fuel load models of Cruz et al. (2003) performed reasonably well, alternative approaches were needed to estimate CBD. Overall, it is apparent that improved models of crown and canopy characteristics are needed. These models would enable managers to more efficiently plan fuels treatments and evaluate their impacts on potential fire behavior at the project level. More recently developed mechanistic models of fire spread, such as the Wildland Urban Interface Fire Dynamics Simulator (WFDS; Mell et al. 2009) and FIRETEC (Linn et al. 2002), also require detailed characterizations of crown fuels. Linn et al. (2005) and Mell et al. (2009) applied these physics-based models of fire spread to stands simulated using geometric models of crown volume (e.g., parabolic and conic forms) and simplified models of crown bulk density; both studies found that simulated fire behavior was sensitive to crown and canopy characteristics. Adopting a more complex model of crown structure capable of describing withinand amongtree heterogeneity, Parsons et al. (2011) clearly demonstrated that the characterization of crown fuels could materially alter the simulated fire behavior in these systems. Looking beyond static models of crown fuels, silvicultural treatment effects on fuel characteristics have been simulated but not observed. In a rare study of temporal changes to crown fuel characteristics, Scott and Reinhardt (2007) used FVS-FFE to simulate the effect of various treatments on crown fuels. Yet, we are aware of no long-term study of actual treatment effects on observed crown fuel characteristics that has been conducted to date. Such a study would enable a validation test of modeling simulations. A related yet often overlooked relation of crown fuel condition to crown fire potential is the moisture content of foliage. In combination with canopy base height, foliar moisture content (FMC) determines the potential for canopy ignition (van Wagner 1977). Its effect is minor relative to canopy base height (Scott 1998b) but is an operationally significant factor in crown fire resistance, and its proportional importance is positively related to surface fire intensity (Keyes and O’Hara 2002). Studies of the FMC of numerous North American species have been published (Agee et al. 2002, Keyes 2006), in some cases reporting seasonal variations and differences between new and older foliage, and are useful to managers for assigning generalized FMC values in fire model simulations. The effects of silvicultural treatments on FMC are unknown, however, as apparently no studies of treatment response have been conducted for any North American tree species (Keyes 2006). Identification of treatment effects on FMC are necessary to determine whether changes in FMC occur, and if so, whether they offset or exacerbate changes to crown fuel characteristics (CBD and CBH) associated with hazard fuels treatment. Utility of Crown Models from Other Fields Outside of the fire and fuels domain, extensive study has been made of the architecture of conifer crowns, owing to their importance as bioenergy and carbon stocks, determinants of wood quality, and drivers of tree and stand growth. Various lines of research have investigated conifer crown structural relationships at resolutions ranging from whole-tree biomass allometries to three-dimensional distributions of individual crown components (e.g., foliage or live branches). Although distinct crown components traditionally have been studied at disparate levels of resolution in different fields, recent work has been both more attentive to developments in other disciplines and increasingly concerned with the vertical structure of the crown. There remains, however, comparatively little information concerning the magnitude of intrinsic variation in crown architecture or on the effects of stand manipulations (but see, e.g., Brix 1981; Garber and Maguire 2005a, 2005b) In the 1970s, increased use of whole-tree harvesting techniques coupled with higher fossil fuel energy prices initiated widespread efforts to quantify crown biomass relationships. Weight scaling of merchantable timber or pulpwood had been in use in many parts of North America prior to this time but there had been relatively little interest in branch wood and foliage biomass. Crown biomass regression equations were soon developed for many conifer species across the United States and Canada (e.g., Young et al. 1980, Tritton and Hornbeck 1982, Evert 1985, Standish et al. 1985). Methodology varied, but the studies producing these equations had aims similar to those of Brown (1978) in seeking regional allometries for estimating total crown mass of individual trees from standard forest inventory measurements. Recently, with growing interest in forest carbon inventory, many of the results and data sets from this class of tree biomass studies have been revisited in meta-analytic studies aiming to develop carbon yield equations for application at national or continental scales (see, e.g., Jenkins et al. 2003, Wirth et al. 2004). Many existing crown biomass equations are of limited utility for canopy fuels modeling. Most biomass studies report separate foliage and branch wood biomass equations, but in relatively few instances is branch wood disaggregated by fuel time lag class or by live/dead status. The spatial distribution of biomass within crowns is also generally ignored, although more intensive harvesting methods are now occasioning the need for information on the vertical distribution of branch wood biomass in some regions (e.g., Tahvanainen and Forss 2008). Nonetheless, this body of research on conifer biomass allometries provides considerable information about how crown components vary systematically with tree attributes. Broadly, it is apparent that differences in dbh account for an appreciable proportion of the variation in total crown biomass. At the same time, utilizing additional information on tree height and crown length (or crown ratio) can materially improve the accuracy of 166 WEST. J. APPL. FOR. 27(4) 2012 crown biomass equations (see, e.g., Brown 1978, Evert 1985). Less information is provided concerning the conditioning effects of stand attributes. In particular, few studies examine the influence of stand density on individual tree foliage or branch wood biomass after having accounted for its concomitant effects on tree-level attributes, such as crown length. Whereas most tree biomass studies have focused on characterizing total crown weights, much more detailed representations of conifer crowns have been developed for stem wood quality assessment. Lumber grade and product recovery are strongly influenced by branch size and longevity, particularly by maximum branch diameter and branch density on the lower bole. Consequently, for a number of commercially important conifer species, the vertical distribution of branch basal area has been thoroughly examined (e.g., Colin and Houllier 1992, Maguire et al. 1994, 1999). Characteristic products of this line of research are systems of equations to jointly predict branch basal diameter distributions and numbers of branches along the bole as a function of tree dimensions, such as dbh, total height, and crown length. Some of the highest resolution branching models have been developed for pine plantations in the southern United States. Recent work by Trincado and Burkhart (2009), for example, characterizes not only the vertical distribution of loblolly pine (Pinus taeda) branch diameters but also branch orientation (i.e., azimuth) as well as branch survivorship and retention. Models of conifer branching structures have considerable potential for crown fuel modeling but presently exist for relatively few species. Developing similarly detailed models for other species and for conifers growing in stands under less intensive management regimes would require further investments in data collection. Regardless, the existing body of work has advanced highly flexible and statistically efficient modeling techniques for characterizing simultaneously the size and spatial distributions of branches within the crown. These technical contributions should not be overlooked because they improve not only the ability to credibly simulate crown structure but also the ability to identify elements of that structure in empirical data. For example, this l