The Model > Multistrata


Growth models like MGM are typically built with permanent sample plot (PSP) or other datasets from relatively small (<1000 m2) plots. However, a consequence of the alternative regeneration standards process in Alberta (ASRD 2005) means these models may be used to make projections of much larger areas, such as assigning a yield curve to a regenerating cutblock. Scaling up to the block (~20ha) scale, means consideration must be made for the heterogeneity of stand structure across the block. If the block varies considerably in density, composition or site quality, it is not really valid to average the stand structure information from samples across the entire block, and then project the average or composite tree-list. Growth models are highly non-linear, so averaging the inputs before projection will not yield the same results as entering multiple inputs (one for each sub-block stratum) then averaging the results. The latter approach is the better one, although it is more computationally demanding.

Gaps are the simplest example of strata. A block with trees on only 50% of the area may have an overall density of 5000 sph, but would be better modeled as two stand strata, one with 10,000 sph and the other with no trees. MGM requires a gap area loss estimate to deal with gaps. The gap loss forces the tree density on the non-gap area to a proportionately higher value during processing. All volume, BA and density values are converted back to a total area basis for reporting. Protocols for gap area sampling are being developed. In the TASS model used in B.C., typical defaults are a 15% gap area at stand establishment (from imperfect regeneration) plus an additional 5% gap area developing from small-scale disturbances during stand development. MacIsaac et al. (2006) sampled postharvest aspen blocks with a minimum gap size of 100m2 and found 6-7% block area in persistent gaps (existing before harvest) in regenerating aspen stands, plus 5-8% additional gaps due to patchy re-suckering following harvest, plus 13-15% area in road and landing gaps. Recent work conducted for the MGM Group (S. Gaertner pers. comm.) indicates 30%of 10m2 plots in post-fire mixedwood stands did not contain any trees. This is likely too fine a scale, since a mature tree is >30 m2 and the minimum gap size should be at least this size. A gap loss in the order of 10-30% therefore appears to be a reasonable preliminary estimate for regenerating stands. This value will largely depend on whether companies regenerate roads and landings effectively.

Stand strata
Since blocks are patchy in density, site and composition as well as including untreed gaps, MGM has been enhanced to allow simultaneous projections of multiple strata, and generate area-weight yield (and BA and density) output. In the past, users could have stratified their cutblocks, sampled them, projected each strata under a separate MGM Crop Plan, then averaged the results with appropriate area-weighting. However, MGM2009 can now do this as part of a multistrata projection. Tutorial example 3 describes the procedure to establish and interpret a multi-strata projection in MGM.

Interactions among adjacent strata
Some types of harvest patterns, such as strip harvests designed to extract hardwoods while protecting understory white spruce, create highly heterogeneous stands with fine-scale patterns. The harvest strata in the understory protection (UP) strip system are only 3-8m wide. Adjacent stands affect their neighbours to between 0.5 - 1 tree lengths away (Voicu and Comeau 200x), so these narrow UP strips consist entirely of edge-affected area. A considerable body of literature indicates that shade is a relatively simple, predictable and measurable method of assessing adjacent stand effects on tree growth. We therefore implemented a shading calculation in MGM to deal with these fine-scale adjacent stand effects.

This adjacent stand shade calculation considers the spatial position of taller surrounding stands with respect to the track of the sun, the stands' species composition, tree crown sizes and densities in computing shade. It has been tested in the Canadian Forest Service's Hotchkiss River Demonstration Area (MacIsaac et al. 1991). The shade (seasonal light transmission) prediction is used with published growth response functions from the boreal region of northeastern B.C. (Wright et al. 1998) to reduce the height and DBH growth rates of the shaded stands. The shading and projection system is dynamic: any change in surrounding stands due to growth, mortality or harvest changes the light levels in the next projection year. In the absence of long-term monitoring data on these relatively new silvicultural systems, we believe this to be a robust approach for stand growth projections (Battaglia and Sands 1998).

The adjacent stand shade calculation is also flexible to deal with many harvest patterns. The typical UP north-south strips are realtively easy to simulate, however MGM can handle any strip orientation and even highly irregular polygons such as small patch harvests or variable retention harvests with high levels of patch retention (e.g. understory avoidance types of harvests).

The multistrata projections (with or without adjacent shade calculations) do not simulate the increased windthrow risk in partial cut systems. As with most operational systems we recommend initializing MGM with silvicultural performance survey data collected at least ten years following harvest. This allows sufficient time for most windthrow to occur and for regeneration to become well-established.

Since the shade adjustments to growth is a process-oriented approach, continued monitoring of UP blocks is important. This will help us improve MGM in the future.